Chapter 9



Chapter 9

Lesson 16: The Use of Correlational Studies in Psychology

In Lesson 15, you learned that the preferred type of scientific comparison study is the experiment. Experiments are preferred because of two procedures that they include: (1) an independent variable is manipulated, and (2) participants are randomly assigned to the experimental and control groups. Let's review why these procedures are so important for making cause-and-effect interpretations by looking again at the pellagra studies of Joseph Goldberger.

In Lesson 2, we discussed how associations between unsanitary conditions and pellagra (a disease characterized by vomiting, diarrhea, sores, fatigue, and dizziness) were used to argue that the disease is caused by an infectious agent (a "germ"). Goldberger demonstrated that this was not likely to be true: neither he nor his coworkers developed pellagra after consuming the bodily fluids of pellagra victims. Instead, Goldberger hypothesized that pellagra is caused by eating a diet low in meats and high in starchy foods (Stanovich, 2001). He tested this idea by putting one group of men (the experimental group) on a high-carbohydrate, low-protein (HCLP) diet, and a second group of men (the control group) on a diet in which they received more protein and less carbohydrates. After some time had passed, a number of subjects in the experimental group developed pellagra, whereas no one in the control group developed the disorder. He concluded that an HCLP diet causes pellagra.

Why couldn't Goldberger simply have compared people who already were eating an HCLP diet with people who already were eating a more balanced diet? This procedure would have left uncontrolled a number of extraneous variables. People who already were eating an HCLP diet were more likely to be poor, to be living in unsanitary conditions, to be living in crowded conditions, to be suffering from particular illnesses, to be experiencing a great deal of stress, and so on. In order to conclude that a poor diet is the cause of pellagra, Goldberger needed to control for the effects of these extraneous variables. Thus, he had to manipulate the independent variable by determining which diet a participant received.

In deciding who was to get which diet, why was it important that participants be randomly assigned to the experimental or control groups? For example, why couldn't Goldberger let his subjects choose which group they wanted to be in? Again, this procedure would have left uncontrolled a number of extraneous variables. If participants could have chosen which diet they were to get, a whole host of psychological and biological factors would been confounded with the effects of the diet. For instance, heavier people and people with various health problems might have chosen the high-carbohydrate diet more frequently. Random assignment, on the other hand, equally distributes important extraneous variables across the experimental and control groups. Only in this way can researchers be assured that the groups will be the same at the start of the study. Because of this similarity at the beginning of the study, any differences found between the experimental and control groups at the end of the study are likely to be due to the effects of the independent variable.

Correlational Studies

But it is not always possible or desirable to perform experiments. Sometimes it is unethical to perform experimental manipulations and, other times, the nature of the phenomenon makes experimental manipulations virtually impossible to do. For example, researchers cannot test the hypothesis that smoking causes cancer by randomly assigning participants to a smoking or a nonsmoking group. This would subject smoking-group participants to a potentially lethal manipulation, which would be unethical. Chapter 9 provides other examples of the difficulty of performing experimental manipulations. For instance, it is very difficult or impossible to manipulate variables that we think are important for the success or failure of romantic relationships. Not only would it often be unethical, but the nature of long-term relationships makes it nearly impossible to do so. You could not randomly assign single people to a married versus an unmarried group in order to measure the effects of marriage on various behaviors. People would not get married (or put off getting married) simply for the sake of an experiment. You also could not randomly assign married couples to a divorced versus an undivorced group in order to measure the effects of divorce on children. People would not do it and the experiment would be unethical.

In such cases, the best we usually can do is to perform correlational studies. A correlation is an estimate of the degree to which two variables are associated--the degree to which they change together, either in the same direction, a positive correlation, or in opposite directions, a negative correlation. An example of a positive correlation is the association between height and weight: as height increases, weight tends to increase (and vice versa). An example of a negative correlation is the association between how sleepy you feel and the amount of time it takes you to fall asleep: as you feel more sleepy, it takes less time for you to fall asleep (and vice versa). In married couples, positive correlations have been found between spouses in age, education, physical attractiveness, attitudes, values, and vulnerability to psychological disorders. In married couples, negative correlations have been found between the length of courtship and the likelihood of divorce as well as between a couple's financial situation and the likelihood of divorce. Couples who date for a longer time before marrying are less likely to get divorced; and couples in higher socioeconomic classes are less likely to get divorced.

Interpreting Correlations

Finding a correlation between two variables tells us that they change together, but we can't be certain what is causing them to do so. This is because correlational studies neither manipulate an independent variable nor randomly assign participants to different groups that receive different levels of the independent variable. Instead, when performing a correlational study, researchers measure the variables as they already exist in nature. Thus, by the time they measure the variables, any factor or factors that have caused the variables to become associated typically have already had their effects (that is, the causes typically occurred at some point in the past before the observations began). Because these causal factors were neither observed nor manipulated by the researchers, they cannot determine the effects of the factors. For example, eye color and hair color are positively correlated: people with darker hair tend to have darker eyes. Whatever factors caused the eyes and hair to have these colors had their effects sometime in the past. All that we are able to determine with a single correlational study is that, at the present time, the two variables are associated in a large group of individuals.

In general, when we find a correlation between two variables, A and B, there are three possible causal relationships among the relevant variables:

• Changes in Variable A may be causing changes in Variable B.

• Changes in Variable B may be causing changes in Variable A.

• Changes in Variable C (an unmeasured and unknown third variable) may be causing changes in both Variables A and B.

In other words, correlational studies have two major problems that make it difficult for us to infer anything about the cause of the correlation. (1) The directionality problem refers to the possibility that the first variable is causing changes in the second variable, or that the second variable is causing changes in the first variable. (2) The third-variable problem refers to the possibility that there is an unmeasured third variable that is causing changes in both the first and second variables. A third variable is an extraneous variable that causes changes in both of the correlated variables.

Let's apply this discussion to an example. Let's suppose that there is a correlation between the kind of car one drives and whether or not one has cancer: people who drive sports cars get less cancer, on average, than people who drive other kinds of car. What is causing the correlation? We can't know for certain based on this correlation alone, but there are several possibilities. With regard to the directionality problem, it could be that driving a sports car causes people to get less cancer (perhaps they feel better about themselves and this positive attitude results in less cancer). On the other hand, it could be that getting cancer causes people to choose automobiles other than sports cars (perhaps their treatments cost so much that these people can't afford sports cars). With regard to the third-variable problem, it could be that there is an extraneous variable, such as age, that has causal effects on both the kind of car one buys and one's chances of getting cancer. For instance, people who are older are less likely to buy a sports car and they also are more likely to get cancer.

Correlations Refer To Averages

When two variables are correlated, this tells us that the two variables tend to change together, on average, in a large group of individuals. But this does not mean that, for each person in the group, the two variables are related in the same way. For example, although height and weight are correlated (taller people tend to be heavier than shorter people, and vice versa), some short people are heavier than some tall people. Based on the correlation, we might predict that a particular tall person is heavier than a particular short person. But we will not know for certain until we actually weigh the two people.

Thus, it is extremely important to keep in mind, as you read about the correlational studies in Chapter 9 and the rest of the textbook, that finding a correlation between two variables tells us nothing for certain about a particular individual. For example, there is a negative correlation between age at the time of a divorce and the likelihood that one will remarry: the older one is at the time of divorce, the less likely it is that one will remarry. But there are many older people who eventually get remarried and many younger people who never remarry.

Thus, when learning about the existence of a correlation, one must be careful not only about making cause-and-effect interpretations, but also about inferring anything about a particular individual.

Critical Thinking Questions

Question 16-1

In the 1970s, a team of researchers in Taiwan were interested in discovering the variables that best predicted who would make use of birth-control methods. The researchers looked at many variables and found that people who had more electrical appliances (such as toasters and fans) in the home were more likely to make use of birth control (Stanovich, 2001). Given what you have learned in this lesson, how would you interpret this finding?

Suggested Answer

Question 16-2

Determine whether the following correlations are positive or negative.

• couples with higher levels of marital satisfaction tend to have better adjusted children

• the more committed a spouse is to a career, the lower is the level of marital satisfaction

• the more that wives work outside of the home, the more time that fathers devote to child care

• the greater the amount of stress, the lower is the level of marital satisfaction

• the more that mothers work outside of the home, the greater is the self-reliance and sense of responsibility of children

• the more comfortable spouses are sharing information with each other, the more satisfied they are with their marriage

Suggested Answer

Question 16-3

At least one study has found a positive correlation between a wife's income and the likelihood of getting divorced. Couples in which the wife earned between 50% and 75% of the family income are more likely to get divorced than are couples in which the wife earned less than 50% of the family income. Assuming that this finding is accurate, what is the best way to interpret it? Explain your answer.

Suggested Answer

Question 16-4

A positive correlation exists between marital satisfaction and the number of positive messages communicated between spouses; and a negative correlation exists between marital satisfaction and the number of negative messages communicated between spouses. When Taresha and William, a married couple, talk with one another, they often make a larger number of negative than positive comments. Based on the reported correlations, do Taresha and William have an unsatisfactory marriage?

Suggested Answer

Question 16-5

In studies of predictors of marital success, researchers have found that people are more likely to divorce if their parents have divorced. How would you interpret this finding? That is, what is the cause of divorce in people whose parents have divorced?

Suggested Answer

Bibliography and References

Ricker, J. P. (2002). An introduction to the science of psychology. Boston: Pearson Custom Publishing.

Stanovich, K. E. (2001). How to think straight about psychology (6th ed.). New York: Longman.

Weiten, W., & Lloyd, M. A. (2000). Psychology applied to modern life: Adjustment at the turn of the century (6th ed.). Belmont, CA: Wadsworth.

RETURN TO TOP

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download