4.1. Consumer Support for CSR
Consumer support (CS) of CSR was measured by a five-item instrument, as shown in the Appendix A
. A Cronbach alpha test was also performed to determine the reliability of questionnaire instrument. A general rule is the indicators should have a Cronbach’s alpha of 0.6 or more [60
]. For CS, the resulting reliability coefficients were 0.624 for Pakistan and 0.737 for Sudan samples, respectively, which shows that the internal consistency of constructs for the five items in the questionnaire is reliable. For Pakistan and Sudan, the mean and standard deviations were 5.64 (1.12) and 5.91 (0.86), respectively. ANCOVA was used to test the mean differences of CS among the two countries with gender, age, education and income entered as covariates. Gender, age, education and income were all insignificant; the city factor was also insignificant, implying that the two different countries did not have a significant influence on the degree of CS, when including control over the effect of respondents’ demographic characteristics. Besides, compared to the CS results of Western, developed countries (4.95, 5.19 and 4.40 for France, Germany and the US, respectively) [12
], the scores for the two Muslim countries are much higher. If compared to the results of Chinese consumers (5.574) [1
], the means of CS for two Muslim countries are still higher. However, we notice that the direct numerical comparison may have a potential problem because of the time gap between this and other studies. Therefore, this is just a preliminary exploration, and further research needs to refine it.
4.2. Consumers’ Perceptions of CSR
Respondents’ perceptions of CSR were based on their views of what they believe businesses must do. Sixteen items were employed—four each for every dimension of CSR. For example, each item states that “I believe that businesses must…” Although identical sets of factors were applied in the research of consumers’ perceptions of CSR in Western developed countries like Germany, France and the US [12
], exploratory factor analyses on our samples revealed some strong cross-loadings (>0.55). By excluding the two sub-items (LEG2 and ETH3) with cross-loadings in the Pakistan sample, an identical set of factors as per Carrol’s four dimensions emerged from our samples study. For better understanding, the varimax rotated component matrixes are reported in panel A of Table 2
of this study.
To know the strong correlations among different items, Bartlett’s test of sphericity test was run, and results are significant at the 0.001 level for the two samples, which implies the presence of non-zero correlations among the 16 items. Kaiser–Meyer–Olkin (KMO) values of the two samples are both above the cut-off point of 0.5. These results showed that the basic requirements for factor analysis were satisfied [61
]. Across the two samples, the item statements were consistently regrouped into four factors: ECO, LEG, ETH and PHI, measuring the degree to which respondents believed that businesses must perform their functions in economical, legal, ethical and philanthropic ways, respectively. Cronbach’s alpha test was performed, and reliability scores of all scales of four factors for both samples are above the usual cut-off level of 0.6 (see panel A of Table 2
). The resulting factors appear to be positively interrelated (see panel B of Table 2
for the correlation coefficients among factors). Therefore, in this study, the four-factor model of Carrol could be used for two Muslim countries. The results indicate the answer to the second question in the aims of this study.
The correlation statistics reported in Table 2
(panel B) also highlight the relationships among the four types of CSR and the overall responsibility item included in the survey: “I believe that businesses must make efforts to behave in a socially responsible manner.” For the Sudanese sample, four dimensions were found to significantly correlate with the overall responsibility item. This implies that Sudanese consumers consider all four types of responsibilities to be part of social responsibility. Meanwhile, in the Pakistani sample, only legal responsibilities are not significantly related to the overall responsibility item.
What is more, regression models were used to confirm our findings, on top of the descriptive statistics above. However, a multiple regression model that included all four dimensions of CSR as predictors of the overall responsibility item was not possible due to serious multicollinearity problems, so we chose the nested modeling strategy instead. In that method, we need to pay attention to the change in adjusted R2 from a nested model only including demographic variables, to a full model, including a selected CSR dimension. The change in adjusted R2 confirms whether the selected responsibility adds significant predictive power to the nested model. If the change in adjusted R2 is significant, the selected dimension is shown to have a significant influence on overall responsibility item; that is, this dimension is an important part of social responsibility for local consumers.
For better understanding, assume that there are two regression models. Model A regresses the overall responsibility item on gender, age, education and income. Model B regresses the overall responsibility item on gender, age, education and income, and a selected CSR dimension, such as ECO (economical). Thus, Model B is the full model while Model A is nested in Model B. The goal is to examine the change in the R2
from Model A to Model B. In this situation, the change in adjusted R2
shows how much explanatory power the selected CSR dimension adds to the nested model. Results for the nested regression models are shown in Table 3
of this study. Just like the previous results of correlation analysis, except for the legal responsibilities of the Pakistan sample, other dimensions in both countries are all highly significant, which indicates that these significant dimensions of CSR have more important influences on the degree of overall support for CSR. For the Pakistani sample, legal responsibilities are only significant at the 90% level (p
= 0.093), which is often regarded as a marginally significant relationship, that is, very close to insignificant statistically, while other three responsibilities are significant at 99% level. For the Sudan sample, all four dimensions are significant at the 99% level. But the above results are not enough to answer the third question in our aims: what is the relative importance among these four dimensions? For example, the four dimensions are all significantly important, but we did not know which one is more important and has significant differences from the others.
Therefore, we adopted ANCOVA again to test the differences among these four dimensions with demographic characteristics (gender, age, education and income) as covariates using each sample and the combined sample (both samples together). Subsequent LSD post-hoc tests were then conducted again to compare the four dimensions for each country and the combined sample. The results derived from the three types of samples are shown in Table 4
. From the results of ANCOVA, the differences among these four dimensions are strongly significant, which implies that local consumers expect companies to perform a specific type of responsibility relatively more than others.
In addition, the results of LSD post-hoc tests indicate different orders for the two countries. For the Pakistani sample, legal responsibilities seem to be relatively less important, while this dimension is the most important one for Sudanese consumers. Similar to the situation of legal responsibilities, philanthropic responsibilities are also less important for Sudanese consumers, whereas for Pakistani consumers, the dimension is a more important one than legal responsibilities and ethical responsibilities. For the combined sample, the results indicate that the economic responsibilities rank most important, while ethical responsibilities rank least important. It is noticeable that philanthropic responsibilities at the peak of Carroll’s pyramid are relatively important compared to legal responsibilities and ethical responsibilities. Thus, we got the answer to the third question in our aims, regarding the relative importance among the four dimensions between Muslim countries and Western, developed countries. The answer to the three research questions will be discussed in detail in the next part.