Implementing Logistic Regression Analysis to Identify ...
Implementing Logistic Regression Analysis to Identify Incentives for Agricultural Cooperative Unions to adopt Quality Assurance Systems
Achilleas Kontogeorgos[1]*, Panagiota Sergaki1, Euthimios Migdakos[2], & Anastasios Semos1
Abstract
The purpose of this paper is to examine the factors that determine the decision made by agricultural cooperative unions and businesses to implement a Quality Assurance System (QAS) such as ISO 9001 and HACCP. A questionnaire was distributed to 122 second-degree agricultural cooperative unions and businesses throughout Greece. A total of 88 cooperatives responded to the survey, across a range of locations, representing a response rate of approximately 70 per cent. Results reveal that 43 have implemented a QAS. The findings are reported with logistic regression analysis to profile the criteria that affect the incentives for quality assurance systems adoption. However, a significant percentage of cooperatives reported that they either do not know what the QAS are or that they know of them but they have not implement one. The findings derived from the logistic regression analysis generally suggest that there is a whole range of criteria such as size, perceptions concerning QAS and the cooperatives’ activities that affect the degree of a coop’s involvement in a QAS. The research explores the incentives for implementing a QAS, which can significantly aid a coop to draw up a development strategy. However, the research was conducted only on second decree agricultural cooperative unions. Therefore, it may not adequately illustrate the incentives of investor owned firms with similar characteristics. This is the first effort that has been made to detect the situation of the Greek cooperative unions and businesses with respect to quality assurance systems and it provides a potential base to carry out further research on the incentives that can lead other small and medium size firms to adopt a quality assurance system.
JEL codes: Q130, C250
Key Words: Agricultural Cooperative Unions, Logistic Regression Analysis, Quality Assurance Systems
1. Introduction
In the existing literature there are a lot of studies examining, on the one hand, the relationship between ISO 9001 and its effect on company performance (either operational and business performance or financial performance), or its relationship with the TQM implementation (for more details about ISO papers see: Van der Wiele et al., 2005). On the other hand, the majority of papers examining the HACCP implementation basically focuses on the advantages generated by the HACCP adoption such as the minimization of hazards that can take place in foods (i.e.. microbiological, chemical or natural hazards; e.g. Henson et al., 1999, Maldonado et al., 2005, Eves & Dervisi, 2005). An important number of studies have generally examined, how effective a food safety system is (e.g. Van Der Spiegel et al., 2005, Zugarramurdi et al., 2007, Cormier et al., 2007). Consequently, there have only been a few studies with the exception of Herath et al., (2007) who investigated the incentives for the Canadian food processing sector to adopt HACCP and food safety controls, which have exclusively examined the relationship between the incentives to adopt a QAS and the firms’ characteristics.
The adoption of a quality system can play an important role in a firm’s performance, profits and costs. Nevertheless, specific impacts differ from one firm to another depending on their characteristics and the activities in which they are engaged (e.g. Holleran et al., 1999; Henson and Holt, 2000). Thus, the importance of different incentives is likely to vary among the firms with respect to their adoption of quality systems. The main objective of this paper is to illustrate whether there is a relationship between firm characteristics and the trend to adopt quality systems in the context of the Greek agricultural cooperative unions.
It is assumed that food companies, such as the Greek agricultural cooperative unions, strive to maintain or improve both food safety and quality attributes and such efforts are closely interrelated and are most likely managed as a whole. If there are such synergistic effects between food safety assurance systems and quality management systems, it is assumed that firms are more likely to adopt a broader array of quality systems to improve both food safety and quality attributes. Thus, public interventions that are exclusively promoting food safety improvements, basically through HACCP adoption, should recognize such synergies and perhaps broaden the scope of intervention to facilitate these effects (Herath et al., 2007).
Henson and Holt (2000), suggested that it is not possible to generalize the impacts of a set of incentives on the level and/or type of food safety controls that are adopted by particular firms, since they have different characteristics and objectives that vary according to the type of product manufactured and the environment in which they operate. The incentive variation of individual firms is probably a result of the relationship between the firms’ characteristics and their propensity to adopt quality systems. Indeed, there is evidence from other industries of a relationship between firm characteristics and adoption of certain business practices; for example, Shavell, (1987) suggests that a firm’s incentives to supply safe products will be affected by its size, organization, and structure of its market.
The paper is organized as follows. In the next section, a methodological consideration about logistic regression analysis is presented illustrating how a company would decide to adopt a new strategy i.e. the implementation of a quality assurance system. Then, the data collection and description of the empirical variables used in the analysis are outlined. The following part presents the results of the analysis followed by the summary of the results and its discussion.
2. Methodological considerations
According to Karshenas and Stoneman (1993), the expected profit gain by adopting a new technology in a firm in a given industry will depend on the characteristics of the firm (rank effect), the number of other adopters (stock effect), and the firm’s position in the order of adoption among the competitors (order effect). While the stock effects and order effects are important in determining the dynamic diffusion path of technology adoption, rank effect would determine the cross-sectional difference in new technology adoption behaviour among firms (Madlener and Wickart, 2004). The categorization of firms into “adopters” and “non-adopters” is based on the dichotomous outcome of the adoption decision, which characterizes the dependent variable (Y). Thus, a firm is defined as an “adopter” where Yi = 1 or as a “non-adopter” where Yi = 0
In this paper, the binary logistic regression analysis will be used to classify the agricultural cooperatives in adopters and “non-adopters”. Binary logistic regression is most useful in cases where we want to model the event probability for a categorical response variable with two outcomes. Since the probability of an event (QAS adoption or not) must lie between 0 and 1, it is impractical to model probabilities with linear regression techniques, because the linear regression model allows the dependent variable to take values greater than 1 or less than 0. The logistic regression model is a type of generalized linear model that extends the linear regression model by linking the range of real numbers to the range 0-1. In this study, the adoption decision is based on a set of cooperative level incentives, which are related with the cooperative’s specific characteristics. These characteristics would finally determine the cooperatives’ decision of whether or not to adopt a QAS. However, these characteristics could have a multiple or multidimensional effect on a cooperative’s decision. Thus, a given characteristic may be associated with many incentives related to the decision to adopt a QAS.
Using the logistic regression model, the probability of adopting a QAS can be described as:
πi= [pic] where: pi is the probability that the ith case will adopt a QAS and zi is the value of the unobserved continuous variable for this ith case.
The model also assumes that Z is linearly related to the predictors (the cooperative’s characteristics). Thus, zi=b0+b1xi1+b2xi2+...+bpxip where xij is the jth predictor for the ith case, bj is the jth coefficient and p is the number of predictors.
Finally, the regression coefficients are estimated through an iterative maximum likelihood method. Table 1 presents the dependent and independent variables that are developed using the information collected in the research stage and Table 3 depicts the analysis used.
Table 1: Variable description
|Variables |Description |Range |Mean |Std. |
| | | | |Deviation |
| | | | | |
|Dependent Variables | | | | |
|QAS Implementation |At least one HACCP or ISO 9001, |1 |0,49 |0,503 |
| |implemented = 1*3 | | | |
|HACCP Implementation |HACCP implementation with certificate or |1 |0,43 |0,498 |
| |in progress to | | | |
| |certify *1 = 1*3 | | | |
|ISO 9001 Implementation |ISO 9001 implementation with certificate |1 |0,35 |0,480 |
| |or in progress to certify*1 = 1*3 | | | |
|Independent Variables (cooperatives’ characteristics) | | | | |
|Turnover (million €) |Turnover in million € (2006) |76,09 |10,82 |13,281 |
|Exports (million €) |Export value in million € (2006) |61,49 |3,51 |10,282 |
|Trademarks |Registered*2 trade marks = 1*3 |1 |0,41 |0,494 |
|Highly-educated personnel |Personnel with a university or a |74 |11,97 |10,976 |
| |technical school degree | | | |
|Commercial strategy to |Managers’ statement for their commercial |1 |0,73 |0,448 |
|increase market share |strategy = 1*3 | | | |
|“QAS are just more |Managers’ statement for the use |1 |0,72 |0,503 |
|bureaucracy” |of a QAS= 1*3 | | | |
|“QAS are improvement tools” |Managers’ statement for the use |1 |0,53 |0,454 |
| |of a QAS = 1*3 | | | |
|Number of activities |Different activities and products |10 |5,44 |1,911 |
|Dairy – cheese products |Dairy subsector = 1*3 |1 |0,20 |0,406 |
|Wine |Wine subsector = 1*3 |1 |0,14 |0,345 |
|Olive oil |Olive oil subsector = 1*3 |1 |0,33 |0,473 |
|Fruit & vegetables |Fresh Fruits and vegetable subsector =1*3|1 |0,27 |0,448 |
Sample size: 88 agricultural cooperative unions (2nd degree cooperatives)
*1 within a 6 months period since the survey according to cooperative managers’ statement
*2 Source: ICAP Business directories, 2006 (icap.gr)
*3 Otherwise = 0
3. Data analysis
The examined sample consists of the members of PASEGES – the Greek Federation of Agricultural cooperatives. One hundred and twelve agricultural cooperative unions (second degree cooperatives) and 14 cooperative businesses are the members of PASEGES (Pahellenic federation of agricultural cooperatives unions). A questionnaire was mailed to these agricultural cooperative businesses in order to determine which QAS they implemented and more specifically, to identify their perceptions about the implemented QAS. Finally, a valid response rate of 72,13% was achieved. This rate is quite satisfactory and representative of the entire population of agricultural cooperative unions and businesses.
Finally, the participating cooperatives were classified into 3 groups depending on the level of knowledge and application of the QAS:
1st group: 43 cooperatives (48,9%) that apply at least one QAS.
2nd group: 29 cooperatives (33,0%) that are familiar with the meaning of a QAS, but do not implement one.
3rd group: 16 cooperatives (18,1 %) that do not know the meaning of a QAS and consequently do not implement a QAS.
Table 2 presents the QAS implemented by the cooperatives. It is clear that the most widely spread quality system is ISO 9001 followed by HACCP. Thus, 38 cooperatives implement ISO 9001 and 31 cooperatives implement HACCP principles.
Table 2: The quality assurance systems implemented by the cooperatives
|Quality Assurance System |Frequency |
|ISO 9001 |38 |
|ISO 14001 |3 |
|HACCP principles* |31 |
|BRC |6 |
|IFS |4 |
|AGRO 2.1 & 2.2 & EUREPGAP |9 |
* Including the following standards: ELOT 1416, Codex Alimentarious and ISO 22000
The relationship between the implementation of ISO and HACCP must be examined in order to determine whether the implementation of these systems is independent. Table 3 illustrates the relation between the existence of HACCP and ISO in the cooperatives. Firstly, it must be mentioned that 45 cooperatives (51,1%) do not apply any QAS at all, while, from the remaining 43 cooperatives, 26 (29,5%) of them apply both ISO and HACCP. Twelve cooperatives implement only ISO, and 5 cooperatives implement only HACCP. The use of a X2 test can examine if the implementation of ISO 9001 is independent of the HACCP implementation. The Pearson Chi-Square value is 32,295 (with df=1) meaning that these 2 variables (HACCP implementation and ISO implementation) are not independent and the implementation (or not) of a QAS affects the implementation of the other. Upon calculating the index phi (2X2 table, phi value = 0,606, and significance 0,00) it can be concluded that there is a positive and relatively important relationship between HACCP and ISO implementation. This is probably due to the fact that in most cases an external consultancy agency takes over the implementation procedures for both HACCP and ISO 9001 systems. This is a common practice among the Greek food companies especially for HACCP (Semos and Kontogeorgos, 2007).
Table 3: The relationship between HACCP and ISO 9001
| |Without HACCP |With HACCP |Total |
|Without ISO 9001 |45 (51,1%) |5 (5,7 %) |50 (56,8%) |
|With ISO 9001 |12 (13,6%) |26 (29,5%) |38 (43,2%) |
|Total |57 (64,8%) |31 (35,2%) |88 |
4. Results Analysis
The main analysis was conducted in two stages. In the first stage, all the variables, describing the different cooperative characteristics selected for the analysis were used to identify their intervention with the implementation of a QAS (without examining if it is the ISO 9001 or the HACCP system). In the second stage of the analysis, a separate model was estimated for each QAS. Consequently, in order to illustrate each variable’s effect on cooperatives’ decision to implement a QAS, the entry method of variable selection of the logistic regression analysis was used. In this model the dependent variable takes on the value one (Y=1) if the cooperative implements HACCP or ISO 9001. The Statistical Package for Social Sciences (SPSS 14.00 for Windows) was used for the analysis of the results. The analysis results are presented in Table 4. The different types of R2 and the classification table (Table 5), where 4 out of 5 cases are correctly predicted, suggest that the model adequately fits the data.
Table 4: Logistic regression analysis for the cooperatives characteristics that affect a QAS implementation
|Variables (coops |B |S.E. |
|characteristics) | | |
| |Without a QAS |With a QAS | |
|Without a QAS |34 |7 |82,9% |
|With a QAS |8 |33 |80,5% |
|Total | | |81,7% |
α: The cut value is 0,5
In order to identify both cases, where the estimated model has small adaptation and cases that have an enormous effect in the model, the examination of residuals is required in the logistic regression analysis. Field (2005) provides analytical directions for a residual analysis of such models. However, the residual analysis of this model indicated that there is no need for special treatment over data in order to face extreme values or effects of specific cases in the total adaptation of the model.
Field (2005) also has proposed the examination of multicollinearity by investigating the paired cross-correlations using the process of partial correlation provided by SPSS. Despite, the existence of multicollinearity, this does not affect the values of the factors participating in the model but only their significance (Garson, 2008). The examination of partial cross-correlations and verification of the Pearson statistic showed that there is no statistically important cross-correlation between the variables participating in the model except for the case of exports and turnover where there is a partial correlation (Pearson correlation = 0,8, sig. at 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 download
- handy reference sheet hrp 259
- a new view of multivariate logistic regression
- multinomial logit sarkisian
- the kohala center
- homework assignments for msci biostatistics ii
- impact of microfinance loan on poverty reduction amongst
- differences between statistical software sas spss and
- sampling and sample size columbia university
- implementing logistic regression analysis to identify
- generalized linear models using spss
Related searches
- logistic regression for longitudinal data
- multivariable logistic regression analysis
- univariable logistic regression model
- multivariable logistic regression model
- binary logistic regression analysis
- binary logistic regression equation
- binary logistic regression formula
- binary logistic regression 101
- binary logistic regression pdf
- multinomial logistic regression assumptions
- multinomial logistic regression stata
- multinomial logistic regression in sas