The Effects of Education and Direct Contact on Explicit ...
The Effects of Education and Direct Contact on Explicit and Implicit Attitudes Towards Older People
A submission presented in partial fulfilment of the
requirements of the University of Glamorgan/Prifysgol Morgannwg
for the degree of Doctor of Philosophy
April 2011
Paul Nash
University of Glamorgan
Acknowledgements
‘At times our own light goes out and is rekindled by a spark from another person. Each of us has cause to think with deep gratitude of those who have lighted the flame within us.’
(Albert Schweitzer – Theologian)
As I am sure is the case for all PhD candidates, upon reflection a doctorate is just as much about the process as it is about the completed product. The journey to completion has encompassed both good times -that I will remember forever because they were aplenty, and not-so-good times- that I will remember forever because of the unwavering support I have received. This assiduous hike would have remained untrodden had it not been for the tremendous help and support I gratefully received from many individuals as highlighted below.
Prof. Ian Stuart-Hamilton, my Director of Studies, deserves special thanks. Not only has he provided continued academic support and advice but he has always had the right words of encouragement (most of which could not be repeated), given when needed most. Dr Peter Mayer, as my Second Supervisor, has acted as mediator and voice of reason when Prof. Stuart-Hamilton and I have been swept away at tangents intangible to anyone else. Without him I fear this thesis may have reflected a somewhat chaotic mindset. As a friend, confidante and academic sounding board Dr Rachel Taylor has never failed to be there for me both at times when requested but more often than not without any request being needed. Without the support of these three fantastic individuals I would not have known where to start, where to finish or which path to take to get there. Thank you.
Data collected in this thesis has been the culmination of efforts from a number of people. Without Jill Kneath-Jones in the nursing school at the University of Glamorgan, my access to the Nursing students population would have been impossible. She has provided continued support for the duration of the longitudinal study and for this I am truly thankful. Thanks are also extended to the University of Glamorgan and the Aneurin Bevan Health Board who have facilitated data collection at different stages of the thesis. I would also like to extend my thanks to all of my participants without whose time, none of this would have been possible.
In addition to the academic side of the PhD, my friends and family have provided support and understanding from the first days of proposal writing until the final edits before submission. I have a plethora of friends who have shown understanding when things have been hard for me. I cannot mention by name all of those who have supported me, however, special mention needs to be given to Katy Walden, Joe Draper and Dr Ian Pepper. Katy is a true ray of light, always there with a smile so infectious you can’t help but have your mood lightened by her. Joe has more recently been of support; functionally looking after me when I have been writing, ensuring that I have remained fed, watered and that a life exists outside the thesis. Ian has been there to boost my confidence and ensured that I take a realistic perspective on all things academic and personal. He has been my most ardent supporter and my fiercest critic, but above all else I know I can count on him. Thanks for keeping me sane!
Foremost however, my family have been the one constant source of support from undergraduate through post graduate studies. Despite their geographical distance I couldn’t feel closer to them. My Parents have provided financial and emotional support, never questioning my chosen path, simply ensuring they have been with me at every turn. My sister has always been close to me, knowing me better than many others and as such her understanding and guidance has been invaluable. Without my family I would not have been able to get this far and it is with their continued strength that I strive to go much further.
I have mentioned but a few, however, my thanks extend to many others whom it would be impossible to name individually. Each and every one of these people has rekindled that spark within me in their own way and without that help this accomplishment would not have been half as sweet.
Table of Contents
Page
List of Tables …………………………………………………………………… 9
List of Figures …………………………………………………………………... 9
Chapter One
Attitudes ………………………………………………………………… 10
Implicit Attitudes ………………………………………….......... 15
Chapter Two
The Implicit Association Test ……………………………………........... 20
Predicting Behaviour …………………………………………… 28
Chapter Three
Prejudice ………………………………………………………………... 34
Stereotypes ……………………………………………………... 40
Chapter Four
Ageism …………………………………………………………………... 48
Chapter Five
General Methodology …………………………………………………… 70
Materials ………………………………………………………… 70
Ethical Considerations ………………………………………….. 76
Procedure ………………………………………………….......... 77
Scoring …………………………………………………….......... 78
Chapter Six
Empirical Studies ……………………………………………………….. 81
A Longitudinal Cohort Study with Nursing and Psychology Undergraduate Students
Introduction …………………………………………….. 84
Participants ……………………………………………… 86
Methodology ……………………………………………. 87
Results …………………………………………………... 88
Discussion ………………………………………………. 92
Conclusions ……………………………………………... 100
A Cross-sectional Study Using Early Years Students to Ascertain Directional Preference in the IAT
Introduction ……………………………………………... 103
Participants …………………………………………….... 105
Methodology ……………………………………………. 105
Results …………………………………………………... 105
Discussion ………………………………………………. 106
Conclusions ……………………………………………... 110
A Cross-sectional Study Assessing Attitudes Held by Hospital Nurses working in A&E and Geriatric Medicine
Introduction ……………………………………………... 111
Participants ……………………………………………… 114
Methodology ……………………………………………. 114
Results …………………………………………………... 114
Discussion ………………………………………………. 116
Conclusions ……………………………………………... 119
A Cross-sectional Study Assessing the Effects of Higher Level Education on Implicit and Explicit Attitudes Towards Older People
Introduction ……………………………………………... 121
Participants ……………………………………………… 122
Methodology ……………………………………………. 123
Results …………………………………………………... 123
Discussion ………………………………………………. 125
Conclusions ……………………………………………... 128
A Longitudinal Study Assessing the Effects of Specific Age Education on Attitudes Towards Older People
Introduction ……………………………………………... 130
Participants ……………………………………………… 132
Methodology ……………………………………………. 132
Results …………………………………………………... 133
Discussion ………………………………………………. 135
Conclusions ……………………………………………... 141
A Cross-sectional Study Assessing the Levels of Implicit and Explicit Ageism Held by Older People
Introduction ……………………………………………... 142
Participants ……………………………………………… 145
Methodology ……………………………………………. 145
Results …………………………………………………... 146
Discussion ………………………………………………. 147
Conclusions ……………………………………………... 150
A Meta-analysis of Study One-Six Data
Introduction ……………………………………………... 152
Participants ……………………………………………… 153
Methodology ……………………………………………. 153
Results …………………………………………………... 154
Test of Counterbalancing ……………………………….. 159
Implicit vs Explicit Measures …………………………... 160
Discussion ………………………………………………. 161
Conclusions ……………………………………………... 168
Chapter Seven
General Discussion ……………………………………………………… 170
General Implicit Ageism ………………………………………... 172
General Explicit Ageism ………………………………………... 175
Effects of Education on Attitudes Towards Older People …….... 177
Effects of Direct Contact on Attitudes Towards Older People …. 183
Test Stability Over Time ………………………………………... 184
Future Research Directions ……………………………………… 188
References ………………………………………………………………………. 193
Appendices ……………………………………………………………………… 229
List of Tables
Table 1 - A tabular depiction of the IAT programme …………………… 19
Table 3 - The order of presentation for the IAT ………………………… 75
Table 4 - Table of longitudinal means …………………………………... 88
Table 5 - Table of Nursing students explicit measure correlations ……... 91
Table 6 - Table of Psychology students explicit measure correlations ….. 91
Table 7 - Table of Nursing students implicit measure correlations ……... 92
Table 8 - Table of Psychology students implicit measure correlations …. 92
Table 9 - Table of comparable means for Psychology and Nursing
Students …………………………………………………………. 106
Table 10 - Table of comparable means for Qualified Nurses and
Nursing Students ………………………………………………... 115
Table 11 - Table of aggregated comparable means for Qualified
Nurses and Nursing Students ……………………………………. 116
Table 12 - Table of means comparing the effects of education …………... 124
Table 13 - Table of means comparing Psychology of Ageing course and
General Adult data sets …………………………………………. 133
Table 14 - Table of means comparing the effects of ageing on attitudes
towards older people ……………………………………………. 146
Table 15 - Table of means for each sample population …………………... 154
Table 16 - Table of significance values between samples for the implicit
D-Score Measure ………………………………………………... 155
Table 17 - Table of significance values between samples for the FSA
Explicit Measure ………………………………………………... 156
Table 18 - Table of significance values between samples for the
non-normed implicit scores ……………………………………... 158
List of Figures
Fig 1 - A screenshot of the congruent condition IAT ………………… 22
Fig 2 - Graph illustrating explicit longitudinal data …………………... 89
Fig 3 - Graph illustrating implicit longitudinal data …………………... 90
Chapter One
Attitudes
An attitude may be defined as an internal affective orientation explaining an individual’s action (Reber, 1995). An attitude is constructed of four components; cognitive, affective, evaluative, and conative. The cognitive component refers to the opinions / schema held about an object. The affective component refers to the emotion or salience towards the attitude object. The evaluative component refers to the direction of the feeling, whether the object evokes a positive or negative emotion. Finally, the conative component of the attitude is the disposition for action (Maio, Esses & Bell, 2000). It is the combination of these components that determines the attitude held by an individual.
There are also several characteristics that define an attitude within the above framework. First, an attitude is learned. Attitudes can be learned in many ways, the most prominent being personal experience, observation of salient others and societal influence. Each of the aforementioned methods exposes the individual to attitudes and information about the attitude object, which they then appraise within their own belief systems (and those held within society and salient in-groups) to form and update their own opinions and attitudes (Ruys & Stapel, 2009). Second, attitudes are predispositions. Attitudes are inclinations and tendencies for action, thus, an attitude and its direction contain motivational qualities. If the attitude is such that the salience towards the attitude object is high and the object is encountered, then be it negative or positive, as long as the conative response is satisfied, an individual will act upon their held beliefs. However, this is a predisposition and the association with behaviour is not a causal one (Ruys & Stapel, 2009). Third, attitudes are consistent. This does not mean that attitudes cannot be changed (discussed within the Ageism chapter, pp.48), just that they have a consistency in expression and measurement over time and across contexts. The way in which the attitudes are expressed may however change, dependent on the social situation and on the pervasive attitudes of any salient others in the same social setting[1] (Krosnick, 1988). Finally, attitudes are directed towards an object that is referred to as the attitude object. This does not mean to say that an attitude is formed only with respect to a tangible item, but can also be a characteristic i.e. attitude objects can be physical like cars or non-physical like sarcasm.
There are several theories of attitude formation, but two have become prominent. These are the summation (Fishbein & Ajzen, 1974) and averaged (Anderson, 1971) models. The summation approach argues that an attitude is the sum of evaluations associated with salient outcomes of observed behaviours (Betsch, Kaufmann, Lindow, Plessner, & Hoffman, 2006). Thus, the attitude (either positive or negative) is the result of the total exposure to an attitude object / target. Using this approach, an attitude can be equally strong if the salience and outcome are high but observed infrequently, or if the salience and outcome are low but the frequency of observation is high. This approach would appear to lend support to the mere exposure effect (Zajonc, 1968; Auty & Lewis, 2004). This refers to the contact an individual has with a particular attitude object. The more frequent the exposure or contact with the object, the stronger the associated attitude becomes. The effect is observable for both positive and negative attitudes and affects both explicit and implicit attitude formation and maintenance. Another facet of the mere exposure effect is that unlike other methods of acquiring information and behaviours, it is simply the exposure that builds the effect and that there need not be any explicit reward or reinforcement.
In contrast, the averaged model proposes that attitudes are subject to a process of normalisation. Thus, the attitude is formed from the average evaluations of the attributes associated with an attitude object. Rather than the summation theory that simply adds together all of the experiences, this theory posits that the direction of the attitude is a reasoned average based on evaluations from each exposure. This theory, unlike the summation approach, would incorporate outlier responses or opinions and ‘dilute’ them into the previously held knowledge about a certain attitude object. More recently it has been established that there are, however, differences in effects of different exposure types. Prestwich, Kenworthy, Wilson and Kwan-Tat (2008) demonstrated that in terms of racist attitudes, exposure to the target group did indeed alter the attitudes held. They found that the quantity of the contact improved an individual’s implicit attitude and the quality of said contact affected the explicitly expressed attitudes. This has also been shown to be the case more specifically in the field of ageing. With an intergenerational study, Tam, Hewstone, Harwood, Voci and Kenworthy (2006) illustrated the same pattern of implicit and explicit attitude change based on quality and quantity of contact. This is important when considering that implicit attitudes reflect spontaneous behaviour and explicit attitudes are predictive of planned behaviour. In order to improve the implicit attitudes towards older people and the resulting behaviour, it is important that there is a high quantity of planned quality exposure and contact time.
In a review of the literature, Betsch, Plessner and Schallies (2004) argue that both models are only applicable in certain circumstances, and an integrated model is more appropriate. They called this the value-account model, which argues that implicit attitudes are formed by summation and explicit attitudes by the averaged procedure. Betsch et al. state that the model is applicable in any situation where the stimulus has the potential to evoke an affective reaction. The model also takes into consideration the four components of attitudes, explaining the salience and motivational aspects of attitudes whilst centring on the cognitive evaluations made by the individual, and their awareness of the process.
The value-account model (Betsch et al., 2004) appears to create an inclusive model of attitude formation, stating that intuitive evaluative judgments reflect the total value of prior encounters. But what of measures of attitudes themselves? Much of the existing body of research is based on self report measures. These are typically Likert-style questionnaires with positive and negative statements towards an attitude object (e.g. smoking, racial issues, gender issues). These measures generally have high levels of consistency and reliability. However, they only measure explicit attitudes whereas attitudes can also be implicit as well as explicit. As defined by Greenwald and Banaji (1995, p.8), implicit attitudes are ‘introspectively unidentified (or inaccurately identified) traces of past experience that mediate favourable or unfavourable feeling, thought or action toward social objects’. Conversely, explicit attitudes are defined as ‘consciously expressed actions, thoughts or feelings under the performer’s control’ (Greenwald & Banaji, 1995, p.8). Perugini (2005) demonstrated that implicit attitudes are stable and enduring, allowing prediction of spontaneous behaviour after exposure to unexpected stimuli. He also demonstrated that explicit attitudes were less enduring, more malleable and predicted only planned behaviour. He believed that because the two attitude types predicted behaviour at different times, they may in fact be manifestations of a single root attitude. This research does not suggest that explicit measures are incorrect, nor that the implicit measures are innately better. The research emphasises that the ability now exists to collect data on both forms of attitude and as such, allows collection of the complete attitude as opposed to one or other subset.
Explicit attitude testing uses self report measures and as such is prone to self presentational bias (Goffman, 1959). When reporting information about oneself, an individual is more likely to tell the other person: (i) what it is that they believe they want to hear; and/or (ii) what will place the speaker in the most socially acceptable light. Jones and Pittman (1982) observed self-monitoring and impression management in self reporting measures. They demonstrated that when reporting details about themselves, participants were careful about what information they volunteered and monitored the answers given to provide a consistently positive view.
Explicit attitudes can be measured in several ways, including Likert scale self report questionnaires, semi-structured interviews, and focus groups. The latter two methods allow for further probing into answers and as such, by using a qualitative approach, provide a deeper insight into proclaimed attitudes, but socially desirable rather than truthful responding is still possible. These findings support the theory of reasoned action (Fishbein & Ajzen, 1975) which posits that in addition to the individual’s belief, an attitude is heavily influenced by the subjective norm. The subjective norm is the belief that attitudes are often influenced by the expectations of others as well as the individual’s desire to engage in actions that are favourably viewed by significant others.
The other subset of attitude is implicit. This is different to the explicit attitudes as it is not under the conscious control of the individual and as such is not as easily measured, assessed or altered.
Implicit Attitudes
‘I cannot totally grasp all that I am… For that darkness is lamentable in which the possibilities in me are hidden from myself: so that my mind, questioning itself upon its own powers, feels that it cannot rightly trust in its own report’
St. Augustine, Confessions
Despite being written around 397AD the above quotation from the Confessions of St Augustine could have been taken from work within social psychology of the 21st century. He suggests that there are parts of himself that he is not totally aware of and that this in turn means that his self report (explicit attitudes) cannot be wholly believed. From more recent works this idea has been expanded as understanding grows and much more is understood about the implicit side to attitude formation and expression.
Implicit attitudes are “introspectively unidentified (or inaccurately identified) traces of past experience that mediate favourable or unfavourable feeling, thought or action toward social objects”
Greenwald & Banaji, 1995 p.8
Greenwald and Banaji illustrate here that an implicit attitude may indeed reveal thoughts, feelings and emotions that may normally be explicitly rejected as they conflict with outwardly held values and beliefs. Alternatively, they may be hidden as their explicit expression may have negative social consequences. More likely, however, is that implicit attitudes reveal information which is not available to the individual through introspection however motivated or able a person is (Wilson, Lindsey & Schooler, 2000). An example of this can be seen in ageism. Where an individual genuinely believes they are not ageist and outwardly expresses accepting explicit attitudes, they may harbour negative implicit attitudes based on subtle reinforces from their social interactions and environments.
Implicit measures by their very nature are not subject to the same shortfalls as explicit measures. In part this is because implicit attitude measures are often relatively opaque. By this it is meant that unlike explicit tests where it is often clear what is being measured and what socially acceptable responses are, implicit tests are less obvious. In addition, implicit measures assess automatic social cognition, and participants are unaware of the causal underpinnings of their performance even if they can guess that their performance was influenced by underlying associations. They have been shown to be less biased by deliberate attempts to conceal the attitude and in turn reflect the attitude which may indeed be unknown to the individual (Dovidio & Fazio, 1992; Greenwald & Banaji, 1995). It is exactly this quality of the implicit test that enables them to measure thoughts and processes outside of the conscious control of the individual (De Houwer & Moors, 2007).
There are several types of implicit measure that have been developed, namely: the Affective Priming Task where primes are introduced to assess the valence of decisions and influence the evaluation of target concepts (Fazio, Jackson, Daunton & Williams, 1995). The Implicit Association Test where response latencies are measured for pairings between congruent and incongruent pairings of targets and evaluations (IAT – Greenwald & Banaji, 1995). The Extrinsic Affective Simon Task (EAST) which is similar to the IAT, however, in affective Simon studies, participants are asked to choose between a positive or negative response on the basis of a non-evaluative feature of valenced words. For instance, they might be asked to respond “GOOD” whenever a person word is presented and to respond “BAD” when an animal word is presented (De Houwer & Eelen, 1998). Also there is the Go/ No-go Association Task (GNAT) which is again similar to the IAT, however, the IAT requires that an attitude toward one category (insects) be assessed relative to a second category (flowers). With the GNAT, experimenters can vary whether insects are evaluated in the context of a single category (flowers), a superordinate category (animals), a generic category (objects), or with no context at all (Nosek & Banaji, 2001).
Nosek, Greenwald & Banaji (2007) conducted a review of research using implicit measures in the seven years following the release of the IAT. They identified that each test varied in different measures of reliability and consistency mainly depending on the type of stimuli being used e.g. pictures, words and pairing categories. They concluded that when no comparable opposite category to the attitude object can be found e.g. attitudes towards television (single category assessments) the Go/No-go association task or the Extrinsic Affective Simon Task provide superior results even when compared to the IAT. When there are comparable categories (e.g. old/young) the IAT provides results superior to the others in power, reliability, validity and replicability. Some evaluative priming measures demonstrated weak internal consistency with a split-half r = .06 (Olson & Fazio, 2003). The Go/No-go Association Test (GNAT) has also demonstrated similarly weak reliability when implementing its signal detection method (split-half r = .20; Nosek & Banaji, 2001). The IAT, however, has observed split-half internal consistency scores consistently recording r = .69 (Bosson et al., 2000). On average, the internal consistency estimates for the IAT have been shown to range from .7 to .9 (Schmuckle & Egloff, 2004). When looked at in direct comparison, the IAT significantly outperforms the EAST in measures of reliability (EAST a = .19; IAT a > .75; Teige, Schnabel, Banse & Asendorpf, 2004). It is clear from this that the IAT far outperforms other latency-based implicit measures and as such demonstrates far greater reliability.
In a review, Monteith, Voils & Ashburn-Nardo (2001) found the IAT to be different from the other implicit measures in that participants were often aware of the difference in time taken in responding to congruent and incongruent pairings. Their results revealed strong implicit racial biases that were moderately related to explicit prejudice but unrelated to proneness to discrepancies. By this it is meant that IAT scores were moderately linked to the explicit measures but not to the differences between what participants explicitly stated and their IAT scores. The majority of participants detected this bias, and they felt guilty about it to the extent that they attributed the bias to race-related factors. Participants with smaller discrepancies were more prone to misattribute their IAT bias to non-racial factors and not feeling guilty. These latter findings suggest that people who typically experience success at avoiding prejudiced responses might, paradoxically, be least likely to detect subtle racial biases when they do occur. Thus, the IAT not only allows researchers to assess implicit biases but also gives participants the ability to ‘observe’ their own biases. Due to this and other qualities discussed below, this study and review will centre on the IAT.
Chapter Two
The Implicit Association Test (IAT)
The IAT as developed by Greenwald, McGhee and Schwartz (1998) is a tool for measuring the strength of associations between concepts of an attitude. By this it is meant that response latencies are measured between category pairings within an attitude object. In this case response latencies between categorisations of old and young faces. When a score has been calculated (described below) then the strength and direction of the association can be identified thus illustrating the strength and direction of the attitude held.
The test involves sorting stimuli (words and/or pictures) from four concepts (two sets of opposite concepts i.e. old/young and good/bad) using two response options, each assigned two of the concepts. A representation of the IAT programme is displayed in Table 1. The premise behind the IAT is that sorting stimuli in categories where the concepts are more strongly associated will be quicker than those where the association is weaker.
Table 1: A tabular depiction of the IAT programme
|Block |No. of Trials |Items Assigned to Left Key |Items Assigned to Right Key |
|B1 |12 |Old Faces |Young Faces |
|B2 |16 |Good Words |Bad Words |
|B3 |28 |Old Faces + Good Words |Young Faces + Bad Words |
|B4 |28 |Old Faces + Good Words |Young Faces + Bad Words |
|B5 |12 |Young Faces |Old Faces |
|B6 |28 |Young Faces + Good Words |Old Faces + Bad Words |
|B7 |28 |Young Faces + Good Words |Old Faces + Bad Words |
The IAT requires participants to categorise target concepts with descriptors representing positive and negative poles of an attribution dimension. When a pairing appears between a concept and a congruent descriptor, mapping them to the same response key is considerably easier than if the pairing is incongruent. An example of which would be a quicker response time for pairing a picture of a young person into the category ‘young/good’ rather than ‘young/bad’. It is also important to note that there are equal numbers of congruent and incongruent pairings in each of the trail blocks.
Fig 1: A screenshot of the congruent condition IAT
[pic]
Figure 1 shows a screen shot taken from the IAT which illustrates the way in which participants would see the categorisation options. The stimulus (picture of a young or old face or a positive/good or negative/bad word) is presented in the centre of the screen and the categories are static in the top left and right corners of the screen. The example above is that of an old face presented in a congruent condition where young and good are categorised together.
The response latency difference measures the extent to which positive and negative evaluations are attributed to the target concepts. This is calculated by summing the response times for the congruent and incongruent pairings and then taking the time for the congruent pairings away from the incongruent condition. The larger the time difference, the more negative the attitude is towards the target concept.
The IAT has been widely accepted in social psychology as a measure of implicit attitudes. It is not however a tool used only within this field; it is a measure that has been used in a variety of disciplines. It has been used in cognitive psychology where the underlying constructs measured in implicit cognition are identified (Fazio & Olsen, 2003). The IAT has been used in developmental psychology where implicit pro-White/anti-Black bias was evident even in the youngest group, with self reported attitudes revealing bias in the same direction. In 10-year-olds and adults, the same magnitude of implicit race bias was observed, although self reported race attitudes became substantially less biased in older children and vanished entirely in adults, who self reported equally favourable attitudes toward Whites and Blacks (Baron & Banaji, 2006). Further it has been used in clinical psychology where the IAT was identified as an accurate tool for measuring fear-related automatic associations (Teachman, Gregg & Woody, 2001). Neuroscientists have also used the IAT where it was established that even though amygdala activation to Black versus White faces is correlated with performance on indirect measures of race bias, the amygdala is not critical for normal performance on the IAT and as such expression of implicit attitudes (Phelps et al, 2000). Finally it has been used extensively in health psychology where strong implicit anti-fat attitudes and stereotypes were demonstrated using the Implicit Association Test, despite no explicit anti-fat bias (Teachman et al, 2003). It has also shown its use in practical applications outside the spheres of psychology, most prominently in the market research domain where IAT test results have been used to shape marketing campaigns based upon the positive or negative associations by consumers (Maison, Greenwald & Bruin, 2001).
These studies have illustrated that the IAT assesses internal constructs that are often distinct from the corresponding constructs measured using explicit self report measures. It is this distinction that has proven to be the IAT’s strength as it measures the implicit cognition associated with those unidentified or unreported implicit attitudes (Fazio & Olsen, 2003; Teachman et al, 2003; Baron & Banaji, 2006). It is however, worth noting that the IAT does not set out to, or propose that it does, measure the beliefs held by an individual. The IAT simply measures the associations involved in certain beliefs and it is from these associations that indirect evidence is collated for the presence of certain beliefs and attitudes (De Houwer, 2002). Equally, it has been said that as the IAT measures ‘only’ associations, these measurements can be reflections of societal exposure to stimuli and even if attitudes are reflected, it is the societal attitude reflected, not that of the individual. De Jong et al. (2000) found that the scores reported on an IAT for spiders were equally low for those people who were fearful and for those people who were not. From this they concluded that the negative association was a societal norm and not reflective of the attitude held by the individual. What de Jong et al. failed to consider was that the level of fear was self reported so could have been flawed, as well as the possible desire for some of the participants to appear unafraid of spiders as this was what they believed to be acceptable. Most results suggest that the IAT measures more than just societal views. Banse et al. (2001) examined attitudes towards homosexuality in heterosexual and homosexual men and women. Despite the report that there was a negative general attitude towards homosexuality in 2001, they found that IAT scores of the homosexual participants demonstrated a significantly more positive attitude than those heterosexual participants. This, they concluded, demonstrates that the IAT measures at least in part the attitudes of the individual rather than the societal norm.
The design of the IAT itself requires careful selection of the category labels that define the concepts for measurement as well as of the stimuli materials. To ensure that the test measures what it sets out to do, Nosek et al. (2007) outline the criteria that must be addressed. The categories for each stimulus must be clear. If the stimuli used exhibit characteristics for both of the categories then the participant might not categorise the stimuli using the correct characteristic and as such the test would not measure what it set out to.
The stimuli thus need to be distinctive. Each stimulus used must have a single distinctive characteristic for categorisation and not allow for cross-categorisation using different characteristics. As such it must be difficult to distinguish the two categories of a single nominal dimension (e.g. men or women) using any other characteristic except the nominal feature (gender). Therefore, controls are put in place for attractiveness, gender, race, expression, age etc when using faces, depending on what the measurement of the IAT is.
The IAT has shown to be a reliable test of implicit cognition with split half internal consistency being reported where r = .69 (Bosson et al, 2000) and internal consistency estimates consistently ranging from .7 to .9 (Greenwald & Nosek 2001). As already noted, these figures are more impressive when taken in the context of alternative latency-based measures where Fazio and Olson (2003) report r = .06 and Nosek and Banaji (2001) report r = .20 for the Go/No-go test. Further to this, Schmukle and Egloff (2004) conducted a thorough test of internal consistency and test-retest reliability. They concluded that the IAT demonstrated satisfactory test-retest results (r = .56) and showed evidence of both trait and occasion specific variation. In a recent review of existing literature, Perugini (2005) has illustrated that the IAT has shown consistently high internal consistency scores (α = 0.80). In addition to this he also highlighted the fact the IAT has consistently given reasonable test-retest values (R = 0.60).
The IAT has also been shown not to be susceptible to several possible confounding extraneous influences which have been criticisms of other implicit measures. Greenwald et al. (1998) highlight the fact that whether a category is assigned to either the left or right response key, the results show very little difference. This is also supported by Greenwald and Nosek (2001) who identified that it made no difference to response times and subsequent results whether the participant was either left or right handed. It is however of note that best practice adopts the procedure of counterbalancing as a failsafe to prevent this as a variable for any individual study. The IAT is often delivered in tandem with an explicit measure either for comparison or for a combined result. One possible influence is the delivery order of measures. In a meta-analysis of IAT studies, Hofmann et al. (2005) found that there were no order effects observed.
There are three variables which have been proved to influence the first version of the IAT. These variables are: cognitive fluency, participant age and IAT experience. Greenwald et al. (2003) developed the improved algorithm (as detailed in the thesis methodology) called the D-measure which overcame each of these variables or at least reduced their effects. Cognitive fluency is a phenomenon where those participants who perform the task more slowly overall tend to show a higher IAT effect (either positive or negative). Those slower responders have higher raw scores on all of the measures (congruent and incongruent), resulting in larger overall scores which can exaggerate effects. By using block means and standard deviation calculations Greenwald et al. (2003) demonstrated that this effect can be minimised. Similarly, older participants tend to show larger IAT response times which can again be reduced and the results made more comparable when using the D Score calculation (Greenwald et al, 2003). Greenwald and Nosek (2001) identified the decline in effect magnitude with repeat administration of the IAT. When using the D Score this effect also decreases, however this is one extraneous variable that should be considered if using multiple IATs in a single session or repeated measures IAT over a short period of time.
Nonetheless, the IAT is a robust measure of attitude. For example, it is difficult to fake, in part because automatic responses are a key part of the participant response and these are hard to control. For example, Kim (2003) tested the controllability of the IAT in two studies using racism as the measure and flowers and insects as a control. He found that in both cases the results from the IAT could not be controlled / faked even if under instruction to do so. Kim demonstrated that the only way to control answers was to be told how to do so (responding slowly to a subset of the stimuli). He also identified that even after multiple trials, participants did not spontaneously discover the strategy for controlling their responses. From this he concluded that the IAT was a robust measure, and is clearly more so than explicit measures which have been shown to be susceptible to self presentational bias as well as capturing more subtle biases that may be undetectable using self report measures (Dasgupta et al., 2000). These findings have been supported by Banse et al. (2001) who identified that the IAT could only be reasonably ‘controlled’ by participants upon instruction on how to do so and with multiple exposures to the IAT format. It can further be argued that as the IAT often reveals associations not explicitly endorsed by participants that it is resistant to deliberate alteration and manipulation. This point is demonstrated by Nosek (2002) who demonstrated that many white participants showed a consistent implicit preference for black relative to white despite the explicit desire not to do so. To cement this point, the same study also showed that many of the black participants do not show implicit preference for black relative to white despite the explicit desire to do so. Nosek et al (2007) further posit that as IAT measures are often so weakly correlated to explicit attitude measures it is likely that deliberate faking is not a substantial issue under typical study conditions.
Predicting Behaviour
The IAT has been shown to be a valid and reliable measure of associations and from that an indicator of relative implicit attitudes held. This in itself is a useful tool for measuring and assessing the attitudes held by individuals, groups and the wider society. However, how predictive of behaviour can IAT results be? Steffens, Schulze and Konig (2006) used IATs to measure the ‘Big Five’ personality traits and tested the IAT’s ability to predict spontaneous behaviour. They tested participants using both implicit and explicit measures and found that IATs were predictive of spontaneous behaviour where explicit measures were not. They also identified explicit measures as being related to self ratings or behaviour where IATs were not.
These results can be explained when the processes behind the behaviour are examined further. Automatic processes are based on an associative network that operates in a fast, effortless and unintentional manner where controlled processes are in contrast based on higher order mental processes of reasoning that influence behaviour in a slower more intentional way (Evans, 2008). The opportunity and motivation to control behaviour can mediate the amount of automatic and controlled process for any behaviour. As controlled processes are time and resource intensive they will only guide behaviour if the opportunity is given and if the individual is sufficiently motivated to do so. If there is the absence of either opportunity or motivation, automated processes will have a greater impact. As implicit measures assess automatic processes, they should be successful in predicting behaviour to the extent that automatic processes drive behaviour. As controlled processes are able to override automatic processing, the predictive validity of implicit measures will decrease as motivation and opportunity increase. Linked to the cognitively demanding nature of controlled processes, those individuals with a higher working memory capacity (WMC) should be better at overriding their automatic processes even in high demand situations. Thus, the predictive validity of the IAT would be lower in these circumstances. Barrett et al. (2004) similarly found that those with higher WMC were more successful in enacting controlled goal-directed processing than individuals low in WMC. It can therefore be suggested that those with higher WMC are able to moderate the predictive ability of the IAT and other implicit measures but not the attitude/associations that they measure. However, this said, the moderation can still only occur when there is the motivation, opportunity and sufficient WMC, otherwise the automatic processes return and the associated behaviour can be observed.
Another personality trait that can moderate the predictive ability of the IAT is control over non-verbal behaviour. Controlling non-verbal behaviour is more difficult than controlling verbal behaviour as it is an automatic response not requiring conscious processes. However, control over these processes is not impossible (DePaulo, 1992). Indeed, whereas most people are unaware of their body language, there are those who are more aware and have become practised at controlling these cues (Dasgupta & Rivera, 2006). From this it can also be implied that in addition to an individual’s WMC, the more aware and higher the person’s ability to control their non-verbal behaviours, the less predictive of non-verbal behaviour the IAT becomes.
Further to these individual traits, the following situational moderators also affect the predictive ability of the IAT - cognitive capacity, processing time and alcohol consumption. Cognitive capacity refers to the finite amount of processing capacity. If there is increased cognitive load from multiple drains on the capacity, the influence of controlled processes on behaviour will decrease and consequently the automatic processes which are less susceptible to capacity constraints will increase. This has been demonstrated succinctly by Friese, Hofmann and Wanke (2008) who conducted a study using self report measures and IATs for chocolate and fruit. Half of the participants were instructed to remember a one digit number (retaining their cognitive capacity) and the other half were instructed to remember an eight digit number (reduced cognitive capacity). Following the tests, participants chose either 5 pieces of fruit or chocolate as their reward for participation. The IAT as expected predicted the choice behaviour well for those people whose cognitive capacity had been diminished and the explicit measure was predictive for those who retained their cognitive capacity.
Processing time refers to the time available in which to process the information, make a decision and act upon it. Similar to cognitive capacity, the shorter the time available the increased cognitive demand and the more reliance on time and memory saving strategies. Schemas, stereotypes and more easily available cues are used to make decisions on behaviour than if under unconstrained conditions (Dijker & Koomen, 1996). When the available time is decreased the opportunity to engage in controlled processing also decreases and as such it would be expected that the IAT would be a better predictor of behaviour when the participant is under time constrained conditions and reliant on automated processes rather than when no time pressure is applied. This has been demonstrated when Friese, Wanke and Plessner (2006) found that when placed under time constrained conditions, more than 60% of participants followed their implicit preference as shown using an IAT. Conversely when no time pressure was applied they found nearly all of the participants behaved as expected from their explicit measure.
Finally, alcohol consumption also affects the predictive ability of the IAT. Alcohol impairs functioning across many domains including working memory, attention and self regulation (Giancola, 2000). That said, it has been shown that despite affecting controlled processing, it leaves automatic processing largely unchanged (Fillmore et al, 1999). From this it can be hypothesised that the IAT would be a better predictor of behaviour for those individuals who have consumed alcohol when compared to those who have not. Hofmann and Friese (2008) measured eating restraint using both implicit and explicit measures using both participants who had consumed alcohol and those who had not. They found that those who had consumed alcohol followed the pattern of eating behaviour predicted by the IAT whereas those sober participants were more likely to follow the predictions from the restraint questionnaire.
Thus, empirical evidence suggests that opportunity, motivation, situational and individual factors all play a role in the predictive validity of the IAT. Each of these factors can be linked to the central executive and how its impairment of function leads to greater predictability using the IAT. The central executive is responsible for the distribution of cognitive resources and information processing (Baddeley, 1996) and to fulfil these functions must harness controlled processing. Time constraints, alcohol consumption, low capacity etc are drains on the central executive and as such impair controlled processes where automatic processes remain unaffected (Baddeley, 1996; Fillmore et al, 1999; Hull & Sloane, 2004). Implicit measures will primarily predict behaviour under conditions of low opportunity or motivation to control cognitive processes (Friese, Hofmann & Schmitt, 2009). It is the lack of conscious control that underpins spontaneous, unplanned behaviour that gives the IAT high predictive validity in this area. It is in the area of planned behaviour when the individual is able to and motivated enough to control their processes that explicit measures provide a better predictor of behaviour.
The reason why the IAT is more successful than other implicit measures at predicting this spontaneous behaviour is the dissonance caused by switching between congruent and incongruent conditions and the increased time lag associated with the latter due to increased cognitive load. When compared to its closest ‘rival’ (Extrinsic Affective Simon Task – EAST), De Houwer and De Bruycker, (2007) concluded that the IAT is a better measure of inter-individual differences in attitude than the EAST. This is an especially pertinent point when considering that De Houwer was the psychologist responsible for the construction of the EAST. This they concluded when analysing consistency, reliability and predictability of future behaviours.
However, it is also important to note the main limitations of the IAT. Further to the fact the IAT predicts only spontaneous rather than planned behaviour, the IAT is also malleable in that it can be influenced by indirect means. Foroni and Mayr (2005) demonstrated that by using a priming story, the IAT effect could be shifted and the results changed. By this, what is meant is that participants’ attitudes could be changed on the IAT if they were first presented with positive or negative priming stories for the attitude object. This they found to be more effective than explicitly telling the participant deliberately to alter their results. This should not cause problems in controlled conditions but where additional stimulus material is presented even in the form of an explicit tool with positive and negative statements, measures should be in place to prevent this becoming an extraneous variable. To this end, the studies in this thesis will present the explicit measure after the completion of the IAT so that the implicit measure cannot be coloured.
In summary, the usefulness of the IAT is due to a number of factors: resistance to self presentational bias (Egloff & Schmuckle, 2002); its adaptability to numerous forms of concept measurement (Greenwald & Nosek, 2001); and its lack of dependence on introspective access to the association strengths being measured (Greenwald et al., 2002). Because of these, the IAT can be used as a flexible yet powerful measure of implicit attitude. This flexibility gives researchers the opportunity to access implicit attitudes and stereotypes over numerous categories, overcoming some of the limitations presented by using self report methods.
Chapter Three
Prejudice
So far our focus has been on attitudes. However, although attitudes are central to the formation of prejudicial thoughts and ultimately treatment, it is arguably the stereotypes derived from attitudes that are the key factor in shaping behaviour. Eagly and Diekman (2005) highlight role incongruity as the basis for prejudicial behaviour. They propose that prejudice itself derives from the dissonance between beliefs about the stereotyped attributes associated with a group and the beliefs about the attributes that allow success in valued social roles. Prejudice is as the name suggests the process of pre-judging a particular object (person, place or thing) by one or more of its characteristics. It implies the arrival at a judgement before sufficient evidence has been gathered. Nowicki (2008) presents prejudice in a similar manner to that of an attitude in that it can be broken into component parts; cognitive, affective and conative. Cognitive prejudice refers to the belief that an individual holds, that a certain opinion or attitude is correct and true. Affective prejudice describes what the individual likes or dislikes. Conative prejudice is similar to the conative component of attitude, in that it refers to the propensity of the individual to act on their prejudice. It is the inclination or predisposition and direction of action that it is aimed to measure when assessing the attitudes held by an individual. Psychologically, these components can be drawn together succinctly to define prejudice as:
“…the holding of derogatory social attitudes or cognitive beliefs, the expression of negative affect, or the display of hostile or discriminatory behaviour towards members of a group on account of their membership to that group”
(Brown, 2001, p.8)
This highlights the fact that prejudice is the social orientation towards either a whole group of people or towards individuals due to their belonging to a certain group. Indeed, when judgements and assertions are made about certain social groups, the person making the assertion is using one or more social categorisations as a starting point to infer attributes towards the people in question. It is this process of social categorisation that is so central to the operation of prejudice. Indeed Allport (1954) argued it is the condition without which prejudice could not exist. Although central to prejudice, categorisation is a process that occurs not only in unusual circumstances but as Bruner (1957) posited, is an integral part of human existence.
Social categorisation is a useful phenomenon as it allows for the grouping of similar characteristics enabling the world to be processed in a less arduous and more efficient way. We do not have the processing capabilities to assess each characteristic belonging to each person and object that we come across in our daily lives. As such we rely on these grouping and categories to ease cognitive load. Put succinctly, “…categories are nouns that cut slices through our environment” (Allport, 1954, p.174).
In order for categorisation to work, however, there must be clear distinctions in the characteristics held by each of the groups. This is also the basic premise behind the formation of prejudice. As the categories help us to order and simplify groups of stimuli/people, they also aid in the discrimination between those who do and do not belong. Tajfel (1959) defined the formation of prejudice through categorisation, creating two hypotheses regarding the consequences of this categorisation. First, he stated that when a category is created which incorporates a set of stimuli (be they objects or people etc.) which cause some of the stimuli to fall into either one group or another, this will enhance any pre-existing differences between the two categories. Second, as an extension to the first hypothesis, he stated that members of different groups will be seen as more different from each other than they may truly be, with members of the same group appearing more similar and therefore exaggerating differences. Both facets of this were illustrated by Eiser (1971) who asked participants to judge the permissiveness of a series of attitude statements concerning recreational drug use. In the experimental condition, half of the statements were attributed to one newspaper and half to another, whereas in the control group only one source was stated. The perceived difference between the permissive and restrictive statements was significantly greater in the experimental condition. This finding clearly illustrates the importance of the relatively arbitrary assignment of categories. McGarty and Penny (1988) used a similar design and repeated the findings for both category differentiation and assimilation within groups. More recently, in a meta-analysis of categorisation literature, Bigler and Patterson (2007) highlight the importance of in-group bias as well as implicit attributions in the development of stereotypes and prejudices towards and for salient social groupings.
Social categorisation and intergroup discrimination were demonstrated in a striking way by Tajfel, Flament, Billig and Bundy (1971). The aim was to establish whether simply belonging to a group might be enough to initiate behavioural prejudice (establishment of in-group and out-group behavioural differences). It was during these experiments that the researchers established the Minimal Group Paradigm where the only difference known to the participant was that of the arbitrary grouping with no categorical contexts. They found that over 70% of participants made choices that favoured their own group as well as identifying the adoption of the ‘maximising difference’ strategy (Turner, 1983) where intergroup differences are maximised often to the detriment of the out-group. Similar results were reported by Sturmer, Snyder, Kropp and Siem (2006) when looking at in-group empathic concern. They found support for ‘in-group helping’ whereby helping behaviour was initiated for the members of a perceived in-group and not the out-group members. Moreover, the level of in-group empathy was directly linked to the level and salience of perceived similarities of in-group members. These findings are important as they clearly demonstrate the cognitive processes behind prejudice formation and the apparent normalisation of this process. It is important to note, however, that personality perspectives of prejudice formation (conative and affective) are more useful in explaining the extremes forms of prejudice, a point covered later.
Addressed above are the results of categorisation along one domain (e.g. Old/Young, Male/Female, British/Non-British). From a theoretical perspective what happens when the activation of two of these domains is present and people span categories making them partly in-group and partly out-group dependant on the dimension observed? Doise (1976) argues that when two categories cut across each other (e.g. age and gender) any difference in initial categories will be reduced because of simultaneous between- and within-group effects on both dimensions. Essentially, the processes of assimilation and differentiation should cancel one another out, reducing or removing bias in terms of either age or gender. This theoretical finding has been explored and supported on numerous occasions (Brown & Turner, 1979; Diehl, 1990; Vanbeselaere, 1991; Pickett, Silver & Brewer, 2002). These results, however interesting in the laboratory, could potentially provide significant inroads to reducing prejudice in the real world. They seem to suggest that if situations can be arranged so that two or more dimensions cross one another, the likelihood of prejudice between groups being persistent is reduced. But is this the case when tested outside the laboratory?
When tested in a real world environment, this model does not seem to work as well as with basic laboratory-based concepts. The salience of one of the crossed categories tends to take precedence over the other so that an equilibrium can be reached over the incongruence caused by no clear cut in-group/out-group category. In essence this means that one of the crossed categories has higher salience and to resolve any disequilibrium, it is this category through which discrimination and group categorisation occurs. Hewstone, Islam and Judd (1993) demonstrated this where the crossing groups were religion (Muslim/Hindu) and nationality (Bangladeshi/Indian). In this as with other similar studies (Brewer, Ho, Lee & Miller, 1987), if people were seen to be sharing the same religion, they were always evaluated in a more positive light. If the person did not share the same religion, regardless of nationality or language spoken, they were always rated significantly lower. Similar findings were also reported by Kang Fu (2007) who identified cultural salience as the overarching mediating factor in inter-race/cultural/religion/class marriages in the USA. These studies demonstrate that unlike in the laboratory, category salience is dependant on the locale and the cross category encountered which in turn influence the evaluation of in-group and out-group members.
Two fundamental aspects of the categorisation process are the exaggeration of the intergroup differences and the enhancement of intragroup similarities. When looking at everyday situations such as sporting clashes it is surprisingly the out-group homogeneity that is emphasised rather than that of the in-group. For example, Hamilton and Bishop (1976) observed the categorisation of ethnicity in these terms. When speaking to residents of communities in America into which new families moved, they found that six months after the new family moved in, only 11% of the community knew the surname of the new family if they were black compared to 60% if they were white. The remaining 89% referred to the new family as “the blacks” with no distinguishing characteristics, whereas the white family were perceived more individually. Similarly, Hutchinson, Jetten, Christian and Haycraft (2006) found that higher in-group identifiers presented with a homogeneous in-group perceived more in-group homogeneity when their group was under threat than did low identifiers. This has also been found in the field of gerontology where, when questioned, both residents of an older persons’ residential home and members of a younger student group each rated the other’s age group as more homogenous than their own (Linville, Fischer & Salovey, 1989). The process of out-group homogeneity is, however, not universal. Brown (2001) emphasised that out-group homogeneity is observed when the in-group is not just larger but is not a minority group. When minority groups were observed, in-group homogeneity was significantly more salient and reported as such by members. This is not to suggest that group size is the factor at play (Bartsch & Judd, 1993) but instead that those in minority groups identify more strongly with their in-groups than do members of less salient larger groups. An example of this would be the BME communities within larger cities where their cultural background as a minority is more salient than the sense of ‘Britishness’ is for the larger general out-group population. The reason for this is that when in the minority, people strive to make themselves more similar to their conception of when stereotypical in-group member should be (Turner, Hogg, Oaks, Reicher & Wethrell, 1987).
In sum, it is fundamental for people to categorise due to the size and complexity of the daily information processed. One outcome of this categorisation process is the accentuation of intergroup differences and the reduction of intragroup differences, both of which affect both evaluation of the out-group and intergroup perceptions, attitudes and behaviour. When two or more of these categories cross, in a real world environment, the more salient category is dominant. Once categorisation has occurred, intergroup differences are accentuated with homogeneity dependent on the minority status of the in-group. It is the salience attributed to the in-group membership and often stereotyped differences between categories that determine the basis for prejudice.
Stereotypes
Stereotypes are the belief that members of the same group indeed also share a certain attribute. This assumption arises directly from the categorisation process through the assimilation of in-group differences and as such the promotion of out-group homogeneity. A cultural example of this can be seen in football supporters where followers of a particular team will assume a group identity and norms differentiating themselves from supporters of rival teams. Further to this they will also attribute a different set of characteristics to the out-group supporters to further differentiate themselves. Allport (1954) identified some key ways in which stereotypes were formed simply through the way in which we are raised within our culture and environment. He identified family socialisation, exposure to images in books, television and newspapers as key contributing sources for potential prejudicial stereotypes. Obviously in today’s world, the Internet and seemingly barrier-less technology all feed into the process of stereotype formation (Kvieskaite, 2007). Devine and Sherman (1992) propose an alternative to this view in that attitudes are formed to serve an ideological function. It is in this function that they justify the way in which certain groups of people are treated and endorse the dominant group’s right to its privileged position. Both of these viewpoints suggest that stereotypes are not only rooted in the cognitive processes of the individual but moreover are integral to social constructs. Further to this, it has also been identified that stereotypes can be formed from a cognitive bias, resulting in a perception-based correlation between minority groups and infrequently occurring attributes (Stroessner, Hamilton & Mackie, 1992; Rupp et al., 2005). The reason for this being adopted as a stereotypical attribute of the minority group is due to both a categorical distinctiveness process (increasing the in-group/out-group difference) and the distinctiveness of the attribute occurring. An example of this would the perception of increased crime rates attributed to members of BME communities in otherwise Caucasian majority areas.
So when do stereotypes become activated and when do they influence our behaviour? Darley and Gross (1983) concluded that we use stereotypes not indiscriminately, but rather we use them to create a platform from which we then seek out further information. It is in the absence of additional information that we apply these stereotypes, however hesitantly we may do this. It has been shown that the use of stereotypes (especially gender) do affect people’s judgements even when additional information is presented about the individual character of the person being judged (Glick, Zion & Nelson, 1988). It has further been shown that both of these hypotheses can be correct in that people tend to use the stereotype as a platform on which to base their assumptions of a person. However, rather than seeking information generally about the person, information is sought to confirm the stereotype rather than to cast doubt on it (Stangor & Ford, 1992). There is still the ability, however, for individuals to search for information to contradict the stereotype (Macrae, Millne & Bodenhausen, 1994). As discussed previously the use of stereotypes, as with other heuristic techniques, allow for increased processing capacity of other information presented simultaneously. Linked to this, it has also to be considered that stereotype use will also increase if people are cognitively or emotionally preoccupied with other concerns. The reason for this is the inverse to that presented by Macrae et al. (1994) in that these distractions create a higher cognitive load thus making it more efficient to employ labour-saving stereotypes (Brown 2001). Further to this, Huntsinger, Sinclair, Dunn and Clore (2010) identify not only emotional preoccupation but positive mood state as a greater predictor of stereotype activation.
When observing stereotypes rather than influencing them, they can be used to make sense of both in-group and out-group behaviour. Ross (1977) proposed the notion of ‘fundamental attribution error’ which states that people assume internal causes for others behaviour but external causes for their own. Pettigrew (1979) expanded this theory including group phenomenon to the ‘ultimate attribution error’. This posits that negative behaviour by out-group members will be attributed to internal causes whereas those behaviours by in-group members will be rationalised by external influences. Similarly, when the out-group expresses positive behavioural traits it will be seen as the exception that proves the rule instead of in-group members where it is seen as an in-group characteristic. Ultimate attribution error has been shown in a research setting by Beal, Ruscher and Schnake (2001) whereby they demonstrated that subsequently presented mitigating explanations for negative acts did not temper impressions of out-group members, and subsequently presented crediting explanations for positive acts did not enhance impressions of out-group members. Brown (2001) highlights from this that the more abstract a general stereotype construct is, the more resistant to change in the light of new information it may be, whereas concrete representations are more easily disconfirmed by one or two contrary instances. He further explains that positive in-group and negative out-group stereotypes lean towards abstraction whilst negative in-group and positive out-group images are usually more concrete. From this it has to be of concern how negative stereotypes can be addressed and changed.
Stereotypes can be changed through the presentation of contradictory information, but how that information is presented (concentrated examples or sporadic) and the affective nature (positive or negative) of the stereotype undergoing change are integral factors to the extent and level of success of the modification. A growing body of research (Allport, 1954; Cook, 1962, 1978; Pettigrew, 1979; Stephan & Stephan, 1984; Paolini, Hewstone, Cairns & Voci, 2004; King, Winter & Webster, 2009) has shown that contact between groups can alter stereotypes and reduce prejudice provided that it takes place under certain conditions.
The first of these conditions states that there should be a framework of social and institutional support for the interventions designed to facilitate contact (Allport, 1954). By this it is meant that those in authority (actual or perceived) should demonstrate unwavering support for the goals of the intervention. In doing this, people are encouraged to act in a non-discriminatory way that should eventually lead to them internalising these behaviours into their own attitudes. Festinger (1957) posits this is because most people have a need to bring into line both their behaviour and their beliefs so as to alleviate any possible dissonance.
The second of these conditions refers to the acquiescence potential. For contact to be successful, it needs to be of sufficient frequency, duration and closeness to facilitate meaningful relationships to develop between members of the groups concerned. As Cook (1978) highlighted, this can be in stark contrast to the short burst of casual contact in many intergroup contact situations.
Third, it is necessary for the two groups to meet where both are considered to have equal status. Many of the stereotypes of out-groups are held as they are considered lesser or inferior in a multitude of different ways. If, when meeting, this unequal status is maintained then the likelihood is that existing negative stereotypes for both groups will be strengthened rather than weakened (Blanchard, Weigel & Cook, 1975).
The final condition identified by Allport (1954) was that of co-operation. This follows on from realistic group conflict theory (Sherif, 1954). Members of different groups are dependent on one another for the achievement of a jointly desired goal and as such must co-operate towards this from a unified platform. Ultimately, any solution to reducing prejudice must be through minimising the destructive potential of division maintenance whilst still allowing each of the groups to maintain their distinctive identities.
Recent research (Tam, Hewstone, Harwood, Voci & Kenworthy, 2006; Prestwhich, Kenworthy, Wilson & Kwan-Tat, 2008) has shown that this is not such a straightforward process. Both studies have found that direct contact does indeed affect the attitudes held. However, they have established that the perceived quality of the contact affects the explicit attitude held where the quantity of contact is that factor affecting the implicit attitudes expressed.
Some research on prejudice indicates a change from blatant to more subtle forms of prejudice from the fear of reprisal as social norms change towards what might be loosely termed ‘political correctness’ (Gaertner & Dovidio, 1986; McConahay 1986). Specifically, with little conscious awareness or intent on action, individuals’ negative associations that are consciously renounced can be rekindled and used as a benchmark for responding to members of a stereotyped group (Bargh, Chen & Burrows, 1996; Devine, 1989; Fazio, Jackson, Dunton & Williams, 1995; Greenwald, McGhee & Schwartz, 1998; Legault, Green-Demers & Eadie, 2009; Mendoza, Gollwitzer & Amodio, 2010). Now that these negative attitudes have been internalised and become implicit, the attitude holder may indeed no longer be immediately aware that they hold these attitudes. It is with the lack of conscious awareness that the negative attitude now becomes the basis for unplanned responses to the attitude target. This phenomenon can be observed with ageism in that the explicit bias is not expressed, however, behaviour and language used precipitates the negative stereotypes assimilated. This has been highlighted as a more dangerous form of prejudice as even those outwardly expressing the best of intentions have difficulty trying to avoid negative responses that are generated by implicit processes (Bargh, 1997). Since the advent of the IAT, it has been possible to accurately measure the implicit attitudes held by individuals. It is this technique that has led to the belief that racial prejudice is in fact no less prevalent than it was in the 1940’s even if it is not explicitly expressed (McConahay, 1986, Devine & Monteith, 1999).
Monteith, Voils & Ashburn-Nardo (2001) conducted a study with 79 participants using both implicit (IAT) and explicit (Modern Racism Scale and a 32 point discrepancy questionnaire) measures. The majority of their sample exhibited negative implicit racial attitudes that were only moderately linked to their explicit measures. They have suggested that this may not only show that the participants have negative implicit attitudes but also that they either control their explicit attitudes to appear socially acceptable or (as the discrepancy measure indicates) that the majority are unaware that they harbour negative attitudes. The gulf between implicit and explicit measures has also been highlighted by Fazio & Olsen (2003), who concluded that implicit and explicit measures each report aspects of attitude that are unique and interact as a predictor of behaviour in any given setting. By this they meant that both measures of implicit and explicit attitudes are valid in so much as they measure distinctly different concepts. It is only when combining data on both subsets of attitude (implicit and explicit) that a holistic picture of attitudes can be obtained. This is a troubling finding when it is recalled that implicit attitudes have been shown to be accurate predictors of spontaneous behaviour and the explicit measure only as a predictor of conscious deliberate behaviour (Perugini, 2005). Both points keenly demonstrate the need to address not only one aspect of attitudes held but for researchers to be mindful of both before conducting research and disseminating findings.
Chapter Four
Ageism
In order to examine how realistic stereotypes of ageing are, we need to investigate and identify what ageing constitutes. Sonnenschein and Brody (2005) predicted that by 2050 almost 50% of the population will live past 85, whereas in 1900 only 25% could expect to reach 65. The picture illustrated suggests that the average life expectancy has and will continue to increase albeit at a diminishing rate. Despite prejudice often being a largely irrational, with regard to older people some of the prejudice can be based in biological and observable declines. There are both physical and psychological losses associated with ageing, a fact that is universally accepted.
From a psychological point of view the key losses are associated with memory and brain function. The subsets of dementia affect many older people but are not necessarily associated specifically with ageing or indeed with affecting all older people and as such will not be covered. One of the main areas of decline both studied and associated with older people is that of intelligence. Intelligence is generally seen as declining in later life. What truth is there in this? It should be stated at the outset, that on most measures there is evidence for a decline, but the key question is the size of this decline.
‘We are ... in the unpleasant and illogical condition of extolling maturity and depreciating age’
Dewey’s paradox of ageing (Dewey, 1939, pp.4)
The problem with ageing is that societally it is seen as being simultaneously a time of wisdom and a time of stupidity. Looked at more closely, what we are saying is that people believe older adults are wiser and more knowledgeable, but that they are slower and less efficient at dealing with the new and/or when they have to think on their feet. Horn and Cattell (1967) identified an age related decline in fluid intelligence (problem solving) whereas crystallised intelligence (facts) remained stable. It is worth noting, however, despite the fact younger people do significantly better than older people at fluid intelligence tests, fluid tests are generally against the clock whereas crystallised tests have no time limit so people can take as long as they like to produce their answers. In addition to this, it is the lack of practice of mental skills causes the neural systems to deteriorate through lack of use rather than specifically the ageing process (Stuart-Hamilton 2006).
When referring back to Dewey’s Paradox it seems that there is some truth in the commonplace observation that old age brings wisdom at the expense of a loss of wit, but the size of this change is open to debate. But the above points are still academic in nature because should it or does it matter if in older age you are unable to analyse problems in the same way as younger people? Should it or does it matter if you are no longer able to remember what a Lammergeier is? To most people these things are not important and to purpose that they are because they are important to ourselves takes a narrow view of society. Klaczynski and Robinson (2000) demonstrated that older people shift more to heuristics rather than addressing each problem as novel as a way of coping with everyday tasks. It is in solving problems in everyday life that predict an older person’s everyday functioning better than any traditionally used psychometric test (Allaire and Marsiske, (2002).
In addition to cognitive declines, there are also agreed physical declines associated with ageing. Vision is something often taken for granted as we go through life but about 33% of people aged 65+ have one or more diseases relating to their eyes and vision (Quillan, 1999). The most common decline is the inability to focus at different distances due to the lens in the eye losing some of its elasticity. However, despite observing the marked acuity decline associated with age, Corso (1981) noted that the problem could be alleviated to that of a younger adult if visual displays had high contrast in luminance (black writing on white display).
Similar to that of visual deterioration, auditory decline has been linked to ageing throughout the adult lifespan too with people as young as 50 being impaired in at least some circumstances (Bromley, 1988). Research has identified that of those people aged 70-80, 32% would have serious hearing impairment but further, of those aged over 80, 50% would have severe auditory impairment (Herbst, 1982).
It is clear that both physical and psychological changes occur as one enters later life. What is generally believed and what forms the basis of stereotypes and societal norms might at times be at variance with this. Ageism appears to have a base in physiological and psychological fact, however, little or no account is taken of the compensation method adopted by older people to minimise the effects of age related loss. Similarly little importance is placed on the positive aspects of ageing which are similarly integral to an older person but counter the existing accepted heuristics and as such are often overlooked or cast aside. From these disparate viewpoints, a legitimate question arises in whether the objective views of ageing (which are generally negative) are therefore likely to cue negative attitudes in students who take ageing-related courses. This is a distinctly under-researched area but one of great importance. If attitudes become worse the more educated a person becomes in the field of ageing, what does this say for the state of care for older people? What has to also be reflected on is whether these negative views of ageing are necessarily ‘wrong’ since the wealth of objective evidence is generally negative. With old age inevitably leading to some degree of cognitive and physical decline, is it not warranted to view old age in a negative light especially when compared to youth? This thesis examines in detail both the effects of specific and general higher level education on attitudes towards older people. Amassing a research base for training and policy decisions in this area is crucial in the delivery of key objectives surrounding dignity in care in later life.
The term ageism is one originally coined by Butler (1969) to describe the observed discrimination and prejudice directed towards older people specifically because of their age. Palmore (1999) furthered this concept by explaining that ageism involved both cognitive (stereotypes and attitudes) and affective (prejudice and discrimination) processes. Like other forms of prejudice (e.g. racism and sexism), ageism is a way of pigeon-holing people into heuristic categories that ignore individuality and apply accepted groups’ norms to all group members. Cummings et al. (2000) have identified that common forms of modern ageism include devaluing the contributions made by older people and viewing the pathologic processes sometimes associated with later life as normal components of the ageing process. Angus and Reeve (2006) have further identified that this socially ingrained ageism actively promotes stereotypes of social isolation, physical and cognitive decline, lack of physical activity and economic burden. Gerontophobia is a narrower band of ageism that specifically refers to a phobia (irrational / unreasonable fear) of older people. Lynch (2000) identified ageing anxiety as a major component of gerontophobia and more widely ageism. Ageing anxiety was explained as the “combination of people’s concerns or fears about getting older” (Lynch, 2000. p. 533). These fears were based on concerns over loss of social contact, reduction in cognitive abilities, changes in physical appearance, declines in overall health and financial hardships that are themselves stereotyped characteristics of the ageing process. In addition to these somewhat irrational fears, it is the knowledge that simply by living life we will become a member of this out-group, a process and transition whose path cannot be altered or avoided. The inevitability of the transition itself is stressful and causes anxiety amongst those who hold these negative stereotypes. Cummings et al. (2000) found that people with a greater knowledge of the ageing process actually held fewer of the stereotyped opinions of the ageing process and as such exhibited significantly lower levels of ageing anxiety and worries about later life.
The prevalence of all forms of ageism has increased past those of sexism and racism (Banaji, 1999; Kite & Wagner, 2002) although it is typically harder to measure due to the implicit ways it is conducted (Levy & Banaji, 2002). This has increased to such a degree due to the change in societal ethos regarding prejudice. It is now much less socially acceptable to be sexist, racist or homophobic, however, to be ageist is not seen in the same way. Ageism is prominent in advertising, media, comedy and in the way in which older people are generally viewed. It is due to this prominence that acceptance surrounding ageism has occurred. It is often seen as humorous and based in some degree of fact, this negating any negative effects or outcomes. Due to the ‘humorous’ nature of ageism, the aforementioned negative effects on the older person (self-esteem, disablement, self-isolation) are generally ignored by the wider public. A problem arises, however, when trying to measure the ‘colloquially’ clear prevalence. As previously mentioned, on explicit measures, people will present themselves in what they see as the most socially acceptable light. Because of this self presentational bias, the explicit measure of ageism (unless very subtle) is not truly capturing the full picture of ageism. Spontaneous behaviour and that not consciously thought about is influenced by implicit attitudes. Due to the difficulty of accessing the implicit constructs of attitudes, these are often overlooked. However, as discussed, they form an integral part of the overall attitude. Implicit tests often require computer programmes (IAT, EAST, GNAT) to measure accurate response latencies and as such it is prohibitive to conduct them in public spaces.
So, why should ageism be singled out as an important area for research? Ageism is different from all other prejudices in one key way. A racist will never change skin colour, a misogynist will never change sex, but an ageist person will become that which they hate should they live long enough. As such, every person should be conscious of the fact that if discrimination against older people is tolerated, one day it could be directed towards them. This causes issues around self-esteem and wellbeing in older adults, as well as resulting in costs to the wider society as well as the individual. There are negative consequences both personal and societal associated with prominent ageist attitudes. In the workplace discrimination due to age is increasing (McCann & Giles, 2002) which affects the company and the victim alike. The practice of discrimination has been shown to cause lowered self-efficacy, decreased productivity and cardiovascular stress (Levy, Ashman & Dror, 2000). This is costly to the individual and to the company and due to the symptoms described (decreased productivity), can lead to the perpetuation of ageist attitudes. For the companies involved there can be costly lawsuits for age discrimination, ranging from a few thousand dollars up to $58.8 million (McCann & Giles, 2002) as well as having the associated costs of decreased morale, segregated workforces and decreased per capita productivity.
Most of the early research conducted into ageism was conducted using scales which measure commonly held opinions or a person’s knowledge about ageing. These range from the infrequently used Old People Questionnaire (Tuckman & Lorge, 1953) to the still often used Facts on Ageing Quiz, or FAQ (Palmore, 1977). The FAQ consists of 25 true or false items which measure a person’s knowledge of the ageing process. As it measures knowledge rather than individual opinion it cannot be classed as a pure attitude measure, however, it is useful in gauging the actual level of knowledge and the level of misinformation upon which attitudes can be based. More frequently used as an explicit attitude measure is the Fraboni Scale of Ageism – FSA (Fraboni et al., 1990).
The FSA was developed to fill the hole Fraboni et al. (1990) believed existed in that the previous scales only measured the cognitive components of ageism. The FSA aimed to be a more complete measure of ageism including items to measure antagonistic, discriminatory attitudes and the tendency toward avoidance. The scale includes three factors for a complete measure: Antilocution (antagonism and apathy fuelled my misconceptions and misinformation), Avoidance (withdrawal from social contact with older people) and Discrimination (discriminatory opinions regarding the political rights, segregation and activities of older people) (Rupp et al., 2005). There had been sparse empirical evidence to support the validity and reliability of the FSA further than the establishing theory. However, Rupp et al. (2005) conducted an in-depth analysis of the scale and concluded that the scale was a valid and reliable multidimensional measure of ageism. They conducted a confirmatory factor analysis that supported the multidimensional nature of FSA scores that emphasised both cognitive and affective facets of attitude measurement. Finally they also identified that younger individuals and men had significantly higher ageism scores on the FSA than older individuals and women. These were similar to the findings by Kalavar (2001) who had previously identified that male college students displayed more ageist attitudes than female college students. In a practical context, Mueller-Johnson et al. (2007) provided 94 undergraduates with the same testimony given by a 79-year-old male or female witness, and they were then asked to complete the FSA and the Ageing Semantic Differential. Participants who evidenced stronger ageing stereotypes on these measures rated the witness less favourably than did participants who were less prejudiced. This experiment demonstrated a clear link between perceived credibility of older adults and ageist attitudes.
Discussed above are two established ageism measures. This study, as with the majority of other current studies, has used the Fraboni Scale of Ageism. The reasons for this are succinctly demonstrated by Stuart-Hamilton and Mahoney (2003) when they collected data from two hundred British participants in an age awareness workshop. The participants were given the Palmore Ageing Quiz (PAQ) and the Fraboni Scale of Ageism (FSA) immediately prior to the workshop and one month after the event. Post-event testing demonstrated a significant improvement in PAQ scores, reflecting retained knowledge acquired at the workshop. Scores on the FSA were unaltered, though there was a significant lowering in scores on the Antilocution section of the test. In combination, these results indicate an increased awareness of ageing issues may alter factual knowledge and increase awareness of politically correct language, but attitudes toward older people and ageing remain essentially unaltered. This and similar studies (Gething et al., 2004; Rupp et al., 2005) question whether the PAQ is a valid choice of tool for measuring changes in attitudes to ageing, since PAQ performance can improve independent of measures of attitudinal change.
Despite there being apparently equal amounts of positive and negative stereotypes pertaining to older people (Hummert et al., 1994), the pervasive attitudes present in research suggest higher negative attitudes than positive ones (Kite & Johnson 1988). This finding is also repeated in measures of implicit attitudes (Perdue & Gurtman, 1990). Isaacs and Bearison (1986) have shown that children as young as six exhibit ageist prejudices present in their cultures. Unlike other prejudices mentioned, ageist attitudes are still openly prevalent in society. In television comedies, elderly people are depicted, defined by stereotyped negativities regarding physical decline and both physical and mental incompetencies (Zebrowitz & Montepare, 2000). When the age stereotypes have been acquired, they will be easily activated by the presence of an elderly person (Banaji & Hardin, 1996; Perdue & Gurtman, 1990). This will result in the generalisation of elderly people to the stereotyped schema held. Once acquired, these attitudes are maintained and strengthened when encountering elderly people even if they do not exhibit characteristics associated to the stereotypes held (Levy et al., 2000; Murphy, Monahan & Zajonc, 1995). Challenging these negative attitudes also proves harder for ageism than other forms of prejudice. Hill et al. (1990) demonstrated that even when encountering contradictory evidence, attitudes towards older people were resistant to change and in most cases did not alter. As with most negative associations, if younger people can avoid spending time with older people and encountering evidence that may either support or challenge their beliefs about them, they will do. Purdue and Gurtman (1990) observed that young people are actively seeking ways to refrain from engaging in social meetings with elderly people. This avoidance only reinforces the implicitly held beliefs as it prevents the individual from having ‘meaningful’ encounters with elderly adults that may in fact cause inconsistencies in schema to be noticed and re-evaluations of attitudes to take place.
Much of the ageing literature on attitudes relates to the explicit attitudes held (Cummings, Kropf & DeWeaver, 2000; Catterall & Maclaran, 2001; Depaola, Griffin, Young & Neimeyer, 2003). As already discussed, this is acceptable when measuring attitudes to predict behaviour in planned situations but in novel situations or with prolonged exposure, what attitudes or behaviours would be expressed? Levy and Banaji (2002) conducted a review on implicit ageism illustrating a pervasive and wide reaching proliferation of negative ageist attitudes. They found that as well as older examples of ageism such as within fairytales like Hansel and Gretel where older people are portrayed as sinister and evil, more modern equivalents exist. They highlighted the fact that in modern media, older adults appear in comical roles or where stereotypes of ageing such as physical decline or incompetence are used (Zebrowitz & Montepare, 2000). This highlights the fact that ageism has deep roots in the social unconscious and from an early age is instilled into the thoughts of younger children. The continual reinforcement of these implicit stereotypes is not questioned by children as they see it having no relevance to their emerging selves as they are not referring to their in-group (Levy & Banaji, 2002). Due to the lengthy process of ageing, children’s concept of older people is more abstract and when little contradictory evidence is presented, the negative characteristics are assumed. It is then continued exposure to negative stereotypes throughout the lifespan that causes this preponderance of ageism. Repeated exposure to the primes/negative associations increase the strength of the negative implicit attitude whilst often leaving the explicit attitude unaffected (Levy et al., 2000).
Further to the review by Levy and Banaji (2002), research has continually demonstrated that implicit attitudes towards older people demonstrate a wide reaching ageist bias. In the same year, Jelenec and Steffens (2002) found that in using both community and university samples, implicit ageism was recorded at a significantly high level. They further analysed the data produced during the IAT and demonstrated that not only were younger target stimuli judged more positively but also that those data pertaining to older people were judged more negatively. This trend has continued to become more evident over the years, most recently demonstrated by Turner and Crisp (2010). They found that the pervasive implicit attitude towards older people was negative. In addition to this, they found that both with actual and perceived contact with older people, the implicit attitude improved. This is in support of Tam, Hewstone, Harwood, Voci and Kenworthy (2006) who also found contact to improve implicit attitudes towards older people.
These findings were not only in the young. Negative implicit ageist attitudes are held by older people themselves (Levy & Banaji, 2002). The reasoning given for this is that elderly people have acquired the same implicit prejudices throughout their lives and have not had sufficient time or opportunity to develop the mechanisms to defend against this.
There are negative consequences for elderly people as a result of the ageist attitudes held as they are not only subjected to ageist prejudices from others but also internalise these implicit biases (Nemmers, 2005). As people progress through the life span their age schema become more elaborate as more information both about others and themselves becomes incorporated. As they age, the number of traits, categories and subcategories they have within the schema grows, however, core elements are still retained. Research does support this developmental approach, finding that despite having a more complex picture of ageing, older people do not necessarily hold more positive views. Hummert et al. (1994) found that older people did indeed hold more stereotypes about older people but they had more negative ones as well as having more positive ones. Some studies further report that older people do judge their age category more favourably than younger people do, however, that is only more positively as a comparison but both groups have generally negative attitudes towards older age (Kite et al., 1991). Levy (1996) found that elderly people who exhibited higher negative implicit attitudes also performed significantly worse on memory tasks. She found that the perceptions of older adults could also be affected by implicit self stereotyping. She concluded that implicit age stereotypes can influence the views of older adults both towards others as well as upon themselves.
Levy’s (1996) research supports previous research findings that when elderly people adopt these societal stereotypes, they see decline as inevitable and that becoming a less active member of society as the only option (Rodin & Langer, 1980). Similarly, Butler (1987) found that when adopted, these stereotypes became a self-fulfilling prophecy, reinforcing stereotypes through the inaction and deficits resulting from their initial belief and internalisation. Although some physical decline through arthritis, heart disease, etc, is related to ageing, much of the widely held belief about age changes in health are over-pessimistic (Rowe & Kahn, 1987). Ageing should not be viewed in simple terms of losses and gains but more within the Selection, Optimisation, Compensation (SOC) model proposed by Baltes and Baltes (1990). In this model it is accepted that losses are associated with ageing but that successful ageing is less to do with these and more reflected by the way in which the older person adapts. The model suggests that first, salient goals are selected, strategies and techniques for achieving these goals are optimised and those losses associated with ageing that may hinder goal attainment are compensated for. In a review of literature, Rowe and Kahn (1987) also found that psychological wellbeing and social interactions affected the physical wellbeing of the elderly. They showed that with decreased social support often associated with old age, mortality and morbidity rates increased as did the adherence to previously well maintained health-promoting regimes. One of the most common prejudices against the elderly comes from the perceived cognitive decline associated with ageing. The most important concept concerning cognitive ability is that of perceived control. Older people see physical and mental deterioration as something they cannot control; it is this perception that results in the reduction in active coping mechanisms (Rodin, 1986). He also demonstrated that this perceived helplessness is directly associated with the decrease in motivation and self esteem and eventually with the increase in illness, mortality rates and memory problems. Similarly, Rodin (1986) showed that the opposite is true; with the provision of activity to increase perceived self control a marked improvement in memory, alertness, activity and physical health was observed. This point has been echoed by Coleman and O’Hanlon (2008) who highlight that successful ageing / optimal ageing is achieved not only through acknowledgement of the associated losses (as with other stages of the lifespan) but also the successful adaptation and coping with the stresses and changes in life. It is taking control of those challenges and adapting through acquired life skills that is fundamental to mental, psychological and emotional health at all stages in the lifespan.
With an increasingly ageing population this pervasive negative attitude clearly has negative effects on a growing proportion of UK residents. Over 65’s account for a substantial proportion of the hospital admissions in England and Wales. In the year April 2008 – March 2009 22.8% of the total admittance to NHS A&E departments in England was for those people aged 60 and over (Hospital Episode Statistics, 2010). With there being a pervasive negative attitude towards older people, it is possible that the care of these individuals may indeed be less than that given to a younger person. Gatz and Pearson (1988) believed that despite negative attitudes towards ageing not being global within the healthcare profession, there were specific biases that affected the way in which older patients received care. Duerson, Thomas, Chang and Stevens (1992) indicated that the attitudes held by staff can affect the treatment received and the way in which elderly people are treated. This was succinctly demonstrated where older people were not receiving the same diagnosis based on the same symptoms as younger people where the only differentiating feature was that of the patient’s age. Not only that, but they also refer to findings that health care workers in general have been shown to hold negative attitudes towards older people. They cite Coccaro and Miles, (1984) who conducted a series of explicit measures demonstrating that ageism was a widespread issue even in the early 1980’s. James and Haley (1995) reviewed ageism in the German healthcare system and found several cases where clinicians considered psychotherapy with depressed elderly patients to be ineffective with no assessment other than initial categorisation by age. Similarly, Filipp and Schmitt (1995) found that medical professionals were refraining from treating patients with mental impairments because due to their age the conditions were considered irreversible. It is apparent that the unfavourable and stereotypic view of older people held by some clinicians is the crucial factor behind the inadequate treatment often being received. This is an enduring phenomenon and continues a long tradition which is by no means new in origin. Freud was reluctant and often refused to treat older patients because he didn’t believe that they had the remaining life span in which to experience the benefits (Woodward, 1991).
As with the preponderance of research in this area, Duerson et al. (1992) conducted a series of attitude tests to medical students using explicit measures (Palmore’s Facts on Ageing Questionnaire (FAQ i&ii)). Primarily this questionnaire is employed to assess the knowledge of ageing but it is also accepted as an explicit measure of ageist attitudes. Duerson et al. (1992) found that the scores were no higher than those reported in the general population and no real increase was measured from pre to post educational training. They concluded that this was a reflection on the lack of specific geriatric training. Duerson suggested that in order to improve this knowledge and sensitise them to the growing needs of the elderly, medical students required more specific geriatric training not simply training which focussed on the losses associated with ageing. She also highlighted that direct contact with older people and patients would help student clinicians improve their perceptions. This is a point also highlighted by Tam et al. (2006) who identified direct contact with older people as a factor improving implicit attitudes towards older people. Further they established that quality of contact can be the precursor to changing the explicit attitudes held. They identified that simply encountering and being exposed to older people could improve the implicit attitudes towards older people. This was under the condition that during this exposure, no overtly stereotypical experiences occurred such as a frail older person falling over. It was only when quality time was spent with an older person such as in conversation or a prearranged activity that an individual’s explicit attitudes were improved. This is a revision from Gatz and Pearson (1988) who identified that simple exposure to elderly people and ageing issues had been shown to reduce ageism.
This point was reiterated by Grant (1996) who suggested that ageing texts often referred only to the problems rather than the successes with ageing and describing elderly people as suffering from multiple handicaps. She draws on the fact that this specific education should be given as a matter of course in institutions offering health care courses to increase the level of understanding and care given to elderly patients.
The question remains, if the attitudes presented are clear within society and specifically within the health care system, can they be altered? It is difficult but not impossible to change attitudes and there are several ways in which attitudes can be changed or modified which fit into different categories.
Weakly held and less salient attitudes are easier to change than strongly held attitudes and as such, stronger attitudes are developed in areas which an individual (or in-group to which they belong) considers to be of higher salience. These strongly held attitudes can be either positive or negative but are usually clearly polar. In areas of limited or questionable importance, attitudes tend to be weakly held, ambivalent or neutral which means that they are more susceptible to change.
There are six basic categories of attitude change strategies:
• Changing the basic motivational function. These strategies are based on the theory that attitudes serve four basic functions: utilitarian, ego-defensive, value-expressive and knowledge. By changing the basic motivational function, the attitude towards the product can be changed (Sarnoff & Katz, 1954).
• Associating the attitude object with a special group, event or cause. Attitudes can be altered by indicating the attitude object’s relationship to particular groups, events or causes. Concern for the environment has been one cause used recently (Petty, Wegener & Fabrigar, 1997).
• Relating to conflicting attitudes. Generally people do not like dissonance and strive to reconcile this. If they can be shown that their attitude towards an attitude object is in conflict with another attitude, they may be induced to change one of the attitudes (Petty, Wegener & Fabrigar, 1997).
• Altering components in the multi-attribute model. These strategies attempt to change the evaluation of attributes by upgrading or downgrading significant attributes; change beliefs by introducing new information; and by adding an attribute, or by changing the salience ratings (Mitchell & Olson, 1981).
• The elaboration likelihood model. This model suggests that attitudes can be changed by either central or peripheral routes to persuasion. In the central route, attitude change occurs because an individual seeks and evaluates additional information about the attitude object. In this case, motivation levels are high and the individual is willing to invest the time and effort to gather and evaluate the information, indicating a high level of involvement. Multi-attribute models are based on the central route to persuasion as attitudes are believed to be formed on the basis of important attributes/features and beliefs. In the peripheral route, individuals are either unwilling or unable to seek additional information. Involvement is low, so they must be offered secondary inducements in an attempt to influence attitude change. (Jones et al., 2003)
It is of note that all of these strategies take the traditional view that attitude precedes behaviour, and use the relationship between attitude and behaviour to effect attitude change. Also each of these strategies can be used in an educational setting to impart knowledge and cause disequilibrium in the current schema held to force a re-assessment of existing attributes and evaluations to modify the existing ageist attitude. This has been shown by Kite et al. (2005) who demonstrated that upon providing information about a person that countered existing negative ageist stereotypes, the negative attitudes can be diminished. An example of this would be providing participants with pictures or video clips of older people participating in competitive sport. These images would be contrary to those stereotypes held and cause a re-evaluation to occur. More recently in an educational context, Westmoreland et al. (2009) demonstrated that through a well structured educational based on the psychological principles above, attitudes towards older people can be changed. They demonstrated, using a longitudinal study, that attitude change was possible and that reflexive accounts also illustrated a conscious awareness of this on the part of the students involved.
It is in the utilisation of these strategies in the training of nurses and people involved in the care of older people that the pervasive negative attitudes can be challenged. In challenging these stereotypes and commonly held misconceptions the inequalities in care can also be addressed. The first step in this process is to assess the current standing and efficacy of existing gerontological awareness/training. From here, future studies can identify the gaps in training and trial interventions can be developed, tested and implemented.
In summary, previous research has found that there are pervasive and ingrained negative implicit attitudes towards older people (Levy & Banaji, 2002). It has further demonstrated that those negative attitudes are also recorded on explicit measures of ageism (Depaola, Griffin, Young & Meimeyer, 2003). Exposure to older people can improve attitudes, however it occurs in differing ways. Contact in and of itself can improve implicit attitudes whereas only quality contact time can improve a person’s explicit attitude towards older people (Tam, Hewstone, Harwood, Voci & Kenworthy, 2006). These attitudes are not solely confined to the general populace but are also present in the medical profession where caring for older people is integral to their daily routines (Duerson et al., 1992). Filipp and Schmitt (1995) also highlighted that medical professionals were refraining from treating older patients with mental impairments because due to their age the conditions were considered irreversible. It has been shown that these attitudes (implicit and explicit) can be altered in the favour of older people. Westmoreland et al. (2009) demonstrated that through a well structured educational based on the psychological principles above, attitudes towards older people can be changed.
Despite there being research on attitudes towards older people, the quantity is representatively small and restricted in nature. Most of the studies use either implicit or explicit measures and as such only collect data on one facet of attitudes towards older people. The studies that have been done have generally used students and have used this group as a representative sample of the wider population without making due consideration for the effects of education on individuals’ implicit and explicit attitudes. Considering that there are clear behavioural implications in the manifestations of attitudes, there is surprisingly little research on care professionals providing care for younger and older people on a daily basis. Studies one and three will assess the attitudes of nurses in training and qualified nurses working in hospital environments to try and build a knowledge base in this gap. The studies will assess the effects of nurse training on attitudes towards older people as well as looking at the progression from this stage, post-training, when nurses are in situ caring for the general populous and those specifically caring for older people. This should aid the understanding of the training required to ensure that attitudes towards older people are more positive and that the care older people receive is of the same standard expected for other patient groups. Following this, two educational classroom based courses will be assessed, looking at the effects of education about older people and ageing (Study Five) and of younger people (Study Two). This will assess the current psychological course content on two specific student groups enabling conclusions to be drawn about the current levels of educational interventions and the direction that future interventions may need to take. This is an important step because although there have been successful intervention measures in other countries, none have been conducted in the UK and none have to date influenced the content of training received by healthcare professionals in the UK. Further, Study Four will provide information on the effects of general higher level education on the implicit and explicit attitudes towards older people. As previously mentioned studies are often conducted with student samples and then generalise findings to the general populous. This is a potentially problematic procedure because as yet there is little known about the impact of general higher level education on attitudes towards older people. Finally as each of the studies has used the same data collection instruments and procedures, a meta-analysis will be conducted to assess the impacts of each of these variables on both implicit and explicit measures of ageism. This is a unique angle as to date no study has looked at both implicit and explicit measures of ageism with homogeneous and heterogeneous populations to provide a comprehensive picture of ageism.
This thesis will present the following studies in order to address the gaps in existing research presented above:
1. A longitudinal study assessing the impact of a three year undergraduate nurse training degree on implicit and explicit attitudes towards older people.
2. A cross-sectional study investigating the impact of an educational intervention promoting younger people on implicit and explicit measures of ageism
3. A cross-sectional study investigating the impact of contact with older people in a nursing context on implicit and explicit measures of ageism.
4. A cross-sectional study assessing the impact of general higher level education on implicit and explicit attitudes towards older people
5. A longitudinal study assessing the effects of a Psychology of Ageing module on implicit and explicit attitudes towards older people
6. A cross-sectional study to assess the levels of implicit and explicit ageism of older people
7. A meta-analysis of all sample subsets
For each of the above studies the same implicit and explicit measures were used so that there was parity between the scores returned in each of the samples. As the same tests were used, a meta-analysis could be conducted to give a holistic picture of attitudes towards older people. Further to this, it also allowed for a test-retest analysis to be conducted on the longitudinal studies, thus enabling reliability and consistency measures to be taken. It was important for there to be comparable scores so that the differences that occurred between groups could be more easily compared and the effects of any interventions more accurately reflected. Details of the procedure are provided in the next section.
Chapter Five
General Methodology
All studies followed the same procedure, testing materials and controls. To this end, a single comprehensive methodology section will be included to save repetition in the description of each investigation. However, for each study, separate participant information will be provided plus additional information regarding populations or variations in the test manipulations.
Materials and Procedure
Two measures were used: an explicit measure, the Fraboni Scale of Ageism (Fraboni, Saltstone & Hughes, 1990) and an implicit measure, the Implicit Association Test (Greenwald, McGhee & Schwartz, 1998).
Explicit Measure:
Aspects of the FSA are described here for the sake of clarification, but a full consideration of the relative merits is presented in Chapter Four.
The Fraboni Scale of Ageism (Fraboni, Saltstone & Hughes, 1990) was chosen as a well established and well accepted explicit measure of ageism (Rupp, Vodanovich & Credé 2005). The Fraboni Scale of Ageism (FSA) was developed in 1989 as a way of building on past ageism scales which focused on age stereotypes and myths. The FSA is a 29-item Likert scale which uses a specific operational definition of ageism based on three of Allport's (1958) five levels of prejudice: "...antilocution (mere antipathetic talk); avoidance (avoiding members of the disliked group); and discrimination (excluding members from the disliked group). The FSA requires participants to respond to either negative or positive age-related statements using a 4-point Likert scale (1=Strongly Disagree to 4=Strongly Agree). The total scores that can be recorded on this scale range from 29 – 116 with higher scores illustrating higher levels of ageism. Any score over 58 is illustrative of explicit ageism. To avoid habitual response errors, some items are reverse coded. For example, item 26 reads: "Most older people should not be allowed to renew their drivers licence." This was reverse coded so that a response of 1 (strong agreement) became 4, to reflect a more ageist response.
The Fraboni Scale of Ageism (FSA) has been shown to have good levels of internal consistency with a Cronbach’s alpha coefficient of .86. (Fraboni et al., 1990). An exploratory factor analysis supported the antilocution, avoidance, and discrimination factors, these accounting for 23.3%, 7.2%, and 7.0% of the variance, respectively (Fraboni et al., 1990). Reliability has been shown on each of the factors mentioned with the following Cronbach’s coefficient alpha reliabilities: Antilocution (.76), Avoidance (.77), and Discrimination (.65).
A copy of the scale presented to participants is in Appendix 1. The scale is an untimed paper and pencil measure and the items are always presented in the same order. This is in accordance with standardised instructions.
Implicit Measure:
Aspects of the IAT are described here for the sake of clarification, but a full consideration of the relative merits of different implicit measures is presented in Chapter Two.
Several implicit measures have been developed, the principal of which are: the Affective Priming Task (Fazio, Jackson, Daunton & Williams, 1995), the Implicit Association Test (IAT – Greenwald & Banaji, 1995) and the Extrinsic Affective Simon Task (De Houwer & Eelen, 1998).
The IAT has scored well against other measures of reliability. In a recent review, De Houwer & De Bruycker, (2007) demonstrated that the IAT has shown consistently high internal consistency scores of .83. In addition to this Schultz et al. (2004) also highlighted the fact the IAT produced stable test/re-test correlations of 0.46 over immediate, 1 week and 4 week periods. As already noted in Chapter Two, the r values for the IAT are significantly better than those returned on other comparable tests.
The IAT has also shown to be more of a robust measure of attitude (Dasgupta, 2000; Kim, 2003, Nosek et al., 2007). Its supporters argue that it is difficult to fake, in part because automatic responses are a key part of the participant response and these are hard to control. As has already been noted, Kim (2003) tested the controllability of the IAT in two studies using racism as the measure, and flowers and insects as a control. He found that in both cases the results from the IAT could not be controlled / faked even if under instruction to do so. Kim demonstrated that the only way to control answers was to be told how to do so (responding slowly to a subset of the stimuli). He also identified that even after multiple trials, participants did not spontaneously discover the strategy for controlling their responses. From this he concluded that the IAT was a robust measure, and is clearly more so than explicit measures that have been shown to be susceptible to self presentational bias as well as capturing more subtle biases that may be undetectable using self report measures (Dasgupta et al., 2000).
The usefulness of the IAT is due to a number of factors: resistance to self presentational bias (Egloff & Schmuckle, 2002); its adaptability to numerous forms of concept measurement (Greenwald & Nosek, 2001); and its lack of dependence on introspective access to the association strengths being measured (Greenwald et al., 2002). Because of these factors allowing the IAT to be used as a flexible yet powerful measure of implicit attitude, it was chosen as the implicit measure for this investigation.
The IAT requires participants to categorise target concepts with descriptors representing positive and negative poles of an attribution dimension. It is argued that when a pairing appears between a concept and a congruent descriptor, mapping them to the same response key is considerably easier than if the pairing is incongruent. In other words, if the participant believes that the image and the description match, they will respond faster than when there is a mismatch. For example, for most people there would be a quicker response time for pairing a picture of Adolf Hitler into the category man/bad rather than man/good. The response latency difference measures the extent to which positive and negative evaluations are attributed to the target concepts. This is calculated by summing the response times for the congruent and incongruent pairings and then taking the time for the congruent pairings away from the incongruent condition. The greater the time difference, it is argued, the more negative the attitude is towards the target concept (Greenwald et al., 2003). In essence, when translated into real term scores produced from the IAT, positive scores equate to negative implicit attitudes towards older people and the higher the number, the more negative the attitude expressed.
The IAT used in these studies was presented using proprietary software (E-prime created by Psychology Software Inc). The images and words used for the IAT were supplied and previously tested for suitability and effectiveness by Banaji and her research team. Their contribution and assistance is gratefully acknowledged.
The IAT consisted of both practice and test datasets. Initially the participants were presented with simple pleasant/unpleasant categories and similarly old/young categories: this was to familiarise both with the concepts being introduced and also with the responses required from them. After the familiarisation, pairings were made of either good/ young or good/old. For the purpose of internal counterbalancing, the categories were then reversed and presented to participants with both practice and test blocks. Each test had identical words and images; these were presented in a random order of the programmes generation. Again as previously, a test set was presented first to familiarise the participants with the concept of response pairings before the measurement sets were presented. Table 2 is a representation of the order of presentations used in the IAT programme (the order is reversed in the counterbalanced trials).
Table 2: The order of presentation for the IAT
|Block |No. of Trials |Items Assigned to Left Key |Items Assigned to Right Key |
|B1 |12 | Old Faces | Young Faces |
|B2 |16 | Good Words | Bad Words |
|B3 |28 | Old Faces + Good Words | Young Faces + Bad Words |
|B4 |28 | Old Faces + Good Words | Young Faces + Bad Words |
|B5 |12 | Young Faces | Old Faces |
|B6 |28 | Young Faces + Good Words | Old Faces + Bad Words |
|B7 |28 | Young Faces + Good Words | Old Faces + Bad Words |
A copy of the IAT is available upon request.
To produce scores for the implicit test, the D Score algorithm has been used. This divides the differences between congruent and incongruent pairings by the individual standard deviations to produce a normed implicit score. Full details of the algorithm can be found in Appendix 4 on page 236.
Participants completed the implicit and explicit tasks either singly or as part of a group with each participant working in a separate and isolated cubicle or booth. The order of presentation of items was counterbalanced. The participants provided their responses anonymously, but were given anonymised identification numbers so that: (1) potential order of presentation effects could be measured; and
(2) it was possible to match participant responses in the longitudinal study component of the research.
Ethical Considerations
Prior to the commencement of any research, the programme of study was submitted to and approved by the University of Glamorgan Ethics Committee. It was considered that there would be no psychological harm or distress caused to participants so no special measures were observed. However, the following considerations were implemented.
Confidentiality: Participant records were anonymised. All information collected will be held in a secure location for a five-year period after publication of results. This will then be destroyed in accordance with accepted University procedures.
Right to withdraw: Participants were fully informed of their right to withdraw at any stage, without penalty. They were also informed of their right to refuse to answer any questions.
1) Informed consent: Participants were informed prior to commencing the study about what they would be asked to do. They were not fully informed about the hypotheses of the study in advance, because of concerns that this might bias explicit responses. All groups were capable of providing their own informed consent. No deception occurred at any point and all participants were debriefed at the cessation of testing.
2) Full debrief: Participants were debriefed after completing both the implicit and explicit measures to share with them the objectives of the research and to allow discussion of their experiences. These are discussed within Chapter Seven.
3) Potential distress: There was little chance of the study creating distress. Participants were asked to complete accepted IAT measures - no overtly unpleasant or threatening imagery was involved. However, participants were permitted to withdraw at any stage, as already stated.
4) Co-operation with participating institutions: The experimenter agreed to comply with any further checks or other requirements imposed by participating institutions (e.g. hospital, medical surgery and residential care homes).
Procedure
Upon entering the computer laboratory, participants were spaced one computer apart from each other so that they each had their own work space and maintained the individual test area integrity. Each participant sat in front of a terminal with the testing software pre-loaded and the IAT instruction screen visible. The participants were initially given verbal instructions advising them how they should complete the test, stressing that the computerised implicit section be completed first. This instruction was included as it was considered that completing the explicit measure first may indeed affect the participants due to the content of some statements. Further on screen instructions were provided detailing the manner in which completion of the IAT should occur. Participants were advised that if they were unclear of the procedure after reading this additional information, clarification should be sought from the experimenter in the room.
When the participants had completed the IAT, another information screen was presented advising them to continue to the explicit paper measure. As part of the initial procedural statement, participants were informed that the paper questionnaire was to be completed by circling the most relevant answer category for each of the statement items. This was again completed in silence. When each of the participants had completed the two tests, they remained at their stations until all participants had completed the test. All participants received a verbal debrief giving contact details of experiment staff, supervisors and support services. In addition to this the relevant certification (see appendix two) and debrief (see appendix three) were also given to participants for their reference. In those longitudinal studies, participants were also all reminded that they would be required to undertake a similar experiment at the end of their academic programme and that the contact details they had provided to enable contact at this stage would remain secure.
Scoring
Explicit Measure:
The Fraboni Scale of Ageism (FSA) requires participants to respond to either negative or positive age related statements using a 4-point Likert-scale (1 representing strong agreement with the statement and 4 representing strong disagreement). In order for higher scores to reflect more ageism some items were reverse coded. The scores are then summed for each participant to enable mean group population scores to be calculated. Any score over 58 was a score denoting explicit ageism.
Implicit Measure:
Scoring for the IAT has been taken from the improved scoring algorithm devised by Greenwald, Nosek and Banaji (2003).
The IAT programme collects the response latencies for all of the congruent and incongruent pairings made by each participant. Each participant’s scores are then collated and calculated separately prior to the group aggregation. Test scores/latencies were used from Blocks B3, B4, B6 and B7 where previously only data collected in B4 and B7 were used. This change was made to double the overall amount of data collected and after testing it was established that including these blocks which were previously used only as practice blocks did not adversely affect the integrity of the data.
Initially each of the congruent and incongruent pairing response latencies were grouped so that mean scores could be calculated. In each of these groups, any categorisation errors were removed with then an additional 600ms added to the latency recorded when the correct category pairing was given. This error adjustment yielded more significant results and removed the often quicker responses recorded when a participant simply hit the keys to record a quick response without observing the categories.
Any participant who returned a latency of over 10,000ms had that trial disregarded as Greenwald et al. (2003) identified this group to have simply responded too slowly either through distraction or lack of concentration and their response was no longer reflective of their attitude.
When these data adjustments had been made, the mean congruent and incongruent response latencies were calculated for each participant. The congruent latencies were then taken from the incongruent scores and the higher the resulting number, the more preference was shown towards younger people/stronger the ageist attitudes. Normalisation of each of the participant’s scores was then undertaken whereby the score was divided by the standard deviation of each individual’s responses. This was done because magnitudes of differences between experimental treatment means are often correlated with variability of the data from which the means are computed (Greenwald, Nosek & Banaji, 2003). Using the standard deviation as a divisor adjusts differences between means for this effect of underlying variability. This produces the D score which has been shown to protect against cognitive skill confounds in participants (Cai, Sriram & Greenwald, 2004) which was previously highlighted as an issue with the IAT (McFarland & Crouch, 2002). Any positive score indicates a preference for younger over older people and a negative score reflects a preference for older over younger people. The computational scoring algorithm can be found in Appendix Four on pp. 236
Chapter Six
Empirical Studies
Prior to the advent of the IAT (Greenwald, McGhee & Schwartz, 1998) the overwhelming majority of ageism (and other attitude) research centred on self report explicit measures (Cummings, Kropf & DeWeaver, 2000; Catterall & Maclaran, 2001; Depaola, Griffin, Young & Neimeyer, 2003). These provided a picture of a person’s or group’s expressed attitude but did not access their implicitly held opinions. Since their advent, implicit measures, with the IAT being the forerunner, have enabled researchers to access this previously inaccessible attitudinal construct. This new ability was quickly adopted into many fields of research including: racism (Baron & Banaji, 2006); clinical psychology (Teachman, Gregg & Woody, 2001); neuroscience (Phelps et al., 2000); health psychology (Teachman et al., 2003) and even in a more practical setting with customer preferences in market research (Maison, Greenwald & Bruin, 2001). Despite ageing and ageism receiving a lot of interest in research, policy, practice and in the population as a whole, the majority of the research evidence base as shown in this review is rooted in explicit measure research (Kalavar, 2001; Mueller-Johnson et al., 2007). Even when implicit measures have been used, Levy & Banaji (2002) have shown these tend to be with single groups or comparing homogeneous groups. By homogenous groups, it is meant those groups with similar characteristics (i.e. age, gender, education etc). Studies that compare only these groups are generally controlling for certain factors but in doing so are not looking at factors such as age differences. By not looking at heterogeneous groups (where key characteristics between groups are different) there are central variables not being assessed thus leaving a wide research gap.
This research is not only addressing the lack of implicit evidence base in the field of ageism but is also widening the scope of the research done. From a review of literature, six main studies have been identified and conducted using eight different participant populations. These populations are used in comparison for specific investigation into the effects of direct contact and specific education on attitudes held. In addition to this, a meta-analysis has been conducted to compare each of the groups to ascertain where any group differences may lie. Further to this, as the testing included the administration of the Fraboni Scale of Ageism, these comparisons between groups can be repeated for explicit measures while clearly demonstrating the dichotomy between the implicit and explicit scores. More unique to this study, however, is the collection of data over a period of time with the same cohort. The vast majority of studies, even when they have used implicit measures have relied solely on the cross-sectional data available through single testing. This study includes two longitudinal samples, one investigating the impact of university based nurse training compared to other higher level education and the other investigating a specific gero-psychology module. The longitudinal sample is important to accurately reflect any changes in both individual and cohort attitudes (implicit and explicit) due to interventions and situational variables.
Each set of results will be presented with their literature rationale and a description of the sample, stating any deviations from the standard testing detailed in the methodology section. In brief, the studies will look firstly at the effects of three year undergraduate nursing and psychology degree programmes on attitudes towards older people. Second, the effects of a youth-centric education course will be assessed, followed by the effects of direct contact with older people from working in different nursing environments. The fourth study will look more generally at the effects of higher level education on attitudes towards older people, whereas the fifth will specifically look at the effects on attitudes of a Psychology of Ageing course. The penultimate study will look at the attitudes of older people before a meta-analysis is presented. The order of this has been chosen to present a coherent argument about how attitudes are affected primarily by education. First, a longitudinal study enables a picture to emerge on the effects of a non-specific higher education and a nursing course on attitudes towards older people. From this it is of interest to understand the directional nature of the IAT and the effects of undertaking education specifically on the stimuli diametrically opposed to that which is being measured (youth-centric course for ageism measure). Building on the literature review and the findings from the nursing students, the direct contact hypothesis was tested in a nursing environment which also allows for implications to be drawn about care of older people. With knowledge gained specifically about two university courses, it is of interest to ascertain more widely the impact of higher level education with participants who have not undertaken higher or further education. Building on this, the effects of specific age related education are assessed relative to the general adult population prior to looking at the effects of ageing on attitudes towards older people when effects of education have been controlled for. Finally a meta-analysis is presented to retrospectively compare each of the groups, something that has not previously been done and something that allows different factors to be assessed.
Study One – A longitudinal cohort study with Nursing and Psychology undergraduate students
Introduction
The effects of education on self presentational bias in explicit measures have long been established (Goffman, 1959); the more educated a person is, the more aware they become of the subjective and societal norms against which they will measure themselves and in turn be measured. How this affects the implicit attitudes held by a person is not as well established, with no longitudinal data existing that follow cohorts looking at this particular factor. Cross-sectional studies have been conducted that suggest particular student cohorts have different attitudes towards older people (Jelenec and Steffens, 2002). However, a key criticism of these studies could indeed be that they are cross-sectional. As data have only been collected at one point, there is no way of knowing for example, whether the results are due to individual differences between the types of people attracted to these courses or whether the course itself has influenced the attitudes. Similarly cross-sectional data have been collected illustrating the differences between implicit and explicit attitudes (Betsch, Plessner & Schallies, 2004) but again few give any picture of changes over time.
Ageism in nursing care has been extensively researched, specifically with regard to Nursing students (Duerson et. al., 1992; Gatz & Pearson, 1988; Grant, 1996). Research findings indicate that generally negative attitudes towards older people are held and expressed. Higgins, Van der Riet, Slater and Peek (2007) found this pervasive negative attitude to be clearly apparent in Nursing students and expressed through marginalisation, segregation and “Chinese whispers” style stereotyping. Despite the less than rosy outlook, Cheong, Wong and Koh (2009) have identified that through successful education, medical students in Singapore are reported to hold positive attitudes towards older people. This is, however, a test with explicit measures and a finding that could be culturally specific. More recently, Gonzales, Morrow-Howell and Gilbert (2010) established that this is a tide that can be turned. By implementing a bespoke training package, they have illustrated how the attitudes of medical students can indeed be made more positive through a mixture of education and contact with older people. Each of these studies has been conducted using explicit measures as the key measure and with the exception of Gonzales, Morrow-Howell & Gilbert, has employed a cross-sectional methodology. The measurement of Nursing and Medical students’ attitudes towards older people is of high importance as Duerson, Thomas, Chang and Stevens (1992) indicated that the attitudes held by staff can affect the treatment received and the way in which elderly people are treated. With increasing numbers of older people requiring hospitalisation or medical treatment due to an ageing population, the attitudes and associated behaviours of medical professionals will be of increasing importance. Previous studies have indicated that the attitudes held by medical students are predominantly negative, however, these have been single groups. It is also of importance to be able to compare these groups to other students not undergoing medical training to ascertain whether it is the medical training per se that is resulting in the negative attitudes flourishing.
The current study seeks to build upon the research base looking at the attitudes held by Nursing students. This study sought to follow two cohorts of students, one undertaking a higher education undergraduate degree level programme in psychology and the other undertaking a higher education undergraduate degree level programme in nursing. These two cohorts were chosen primarily to assess the efficacy of nurse training in reducing ageism. Psychology students were selected as a base measure as there is no specific gerontological education received during the programme, only minimal content included in a developmental context. In addition to this, the gender balance on both courses is very similar. Both implicit and explicit measures of attitudes towards older people will be taken to track any changes caused by the education programme being undertaken. The hypotheses to be tested are:
(H1) There will be a significant difference between the implicit and explicit scores returned by each of the groups at each time point.
(H2) Explicit attitudes will not be significantly changed by the delivery of either Nursing or Psychology undergraduate programmes.
(H3) There will be a significant change in the implicit attitudes of the Nursing students between commencement and completion of their course.
Participants
The initial cohort comprised 74 students. Of this sample, 40 were Psychology students (Mean age: 24.0, S.D.: 8.0) and 34 were Nursing students (Mean age: 22.6, S.D.: 7.6). At mid-point testing, the cohort comprised of 65 students. Of these, 34 were Psychology students (Mean age: 25.3, S.D.: 7.7) and 31 were Nursing students (Mean age: 23.4, S.D.: 7.4). However, by the end of the course of study, due to course attrition, withdrawal and non-attendance the sample for the full longitudinal analysis was reduced to 52 students. Of these, 26 were Psychology students (Mean age: 28.0, S.D.: 8.1) and were 26 Nursing students (Mean age: 25.2, S.D.: 7.9). Of the Psychology students, 4 were male and 22 were female with an age range of 20-47 years. Of the Nursing students, 3 were male and 23 were female with an age range of 20-46 years. Results will be reported using all participants where appropriate and when correlation analysis is being conducted only those students completing all three testing phases will be included[2]. The initial sample was an opportunity sample of two targeted undergraduate course programmes at the University of Glamorgan. In addition to standard course content, none of the participants had received any additional gerontological interventions or training prior to testing.
Methodology
The materials used and overall methodology were similar to those presented in the General Methodology section (see pp. 70). The sole difference was that this study was a longitudinal cohort study taking measures of implicit and explicit ageism at three time points rather than on a single occasion. Measures were taken at the start of the respective degree programmes, again in the second year to obtain midpoint data, and finally again at the end of the courses. This method was chosen to gain a base measure of attitudes and then to ascertain any effects caused by either the degree or specific nurse training. Contact details (name, email, telephone number) for each of the participants were taken at the start of the project and stored in a secure filing cabinet in a locked storage room for the duration of the project so that participants could be contacted again and linked to their participant numbers to allow for cohort comparisons. The contact details and participant numbers were stored separately so a theft or accidental release of a document would not lead to the identification of participants.
Results
The descriptive results for both groups at each time point are presented prior to ANOVA tests being conducted. Further to this both correlations between implicit and explicit measures and correlations over time are presented.
Table 3: Table of longitudinal means
|Group |Year |D-Score (mean) |D-Score S.D. |FSA Score (mean) |FSA S.D. |
|Psychology Students |1 |.4526 |.428 |54.4 |10.13 |
| |2 |.7588 |.759 |53.8 |7.56 |
| |3 |.7091 |.710 |54.2 |9.36 |
| | | | | | |
|Nursing Students |1 |.2631 |.474 |49.5 |9.28 |
| |2 |.7028 |.610 |47.3 |9.07 |
| |3 |.6206 |.671 |47.8 |9.57 |
The descriptive results above illustrate the mean scores received by each of the groups (Nursing and Psychology students) at each measurement time in both implicit (D-Score) and explicit (FSA) measures.
The explicit results are expressed in graphical form in Fig 2
Fig. 2: Graph illustrating explicit longitudinal data
[pic]
At time point one there was no significant correlation between the implicit and explicit measures (r= -0.094, p=0.495). Similarly at time two there was no significant correlation (r= 0.149, p=0.247). Finally this finding was repeated at time three (r= 0.069, p=0.628).
Data were analysed using a mixed ANOVA on the PASW statistical package. This showed a significant group difference in explicit scores (F(1,183)=19.157; p ................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related searches
- effects of education on health
- effects of education on society
- the effects of video games on teens
- the effects of communication technology
- the effects of online shopping
- the effects of technological changes
- the effects of advertising
- the effects of world war 2
- the effects of colonialism
- effects of education on people
- effects of education on poverty
- analyze the effects of the neolithic revolution