SPSS for DATA ANALYSIS - Part 4

SPSS for DATA ANALYSIS – PART 4

In the third blog, we were able to graphically observe some relationships between the customers’ “Income” and “Education” by “Offer Acceptance.” In this section of SPSS for Data Analysis, we shall be investigating the correlation between the offers made in the campaigns. In other words, the aim of performing this analysis is to see if we can infer some pieces of information as to the trend of the offer acceptance by the customers.

Refresh your memory by glancing through the earlier articles via the link Part 1, Part 2, and Part 3.

Correlation analysis

In the SPSS, click on the “Analyze” toolbar and select the “Correlate” option in the drop-down. Then click on the “Bivariate…” which implies that we are carrying out pair-wise correlation, the correlation between offer acceptance in the first campaign with other campaigns. NOTE: It is assumed that you have a basic understanding of some statistics terms used in this article! Correlation does not mean causality!

After you click on “Bivariate…” a page similar to Figure 31 should appear where the variables for correlation analysis are selected; in this case, we are selecting the offer acceptance for all the (5) campaigns to observe the linear relationship between them.

After selecting the variables for correlation analysis, they are put in the “Variables…” box by clicking the arrow (5). Automatically, the arrow will reverse when you have placed the variables in the box, which tells you that you can modify the content in the box; to remove or substitute.

Options can be selected by clicking (6) in Figure 32, displaying Figure 33. You may choose to select other statistics such as “Means and standard deviations” for analysis. We are not selecting any statistics other than “Pearson Correlation Coefficients” and “Two-tailed” test of significance. Additionally, we are asking SPSS to flag significant correlations, which will be indicated by “*.”

Click “Continues” in the “Bivariate Correlations: Options” window then click “OK” in the “Bivariate Correlations” window.

Get this cleared! Pearson correlation coefficients are not sufficient to explain relationships between variables; thus, the significance test on the coefficients. The essence of this is to investigate the reliability of the results.

In Figure 34, the correlation coefficients and the significance level of one variable with another can be seen clearly. Although the correlation coefficients are pretty small, all the coefficients are statistically significant. In other words, there is a high chance of having similar customers' responses to the subsequent offers. Using the classification of Pearson correlation coefficients reported by Parvez Ahammad, it can be concluded that the offers acceptance has negligible correlation coefficients with one another except “AcceptedCmp1” with “AcceptedCmp5” and “AcceptedCmp4” with “AcceptedCmp5” that fall within the “Low” correlation coefficient. Notably, here appears a stronger relationship between “AcceptedCmp1” with “AcceptedCmp5” and “AcceptedCmp4” with “AcceptedCmp5” when compared with others. This could imply that customers that accepted the offers in the first campaign also accepted in the fifth campaign, and probably there were some similarities in the campaigns that attracted the customers.

Conclusion

In this section, correlation analysis has been performed on the offer acceptance columns of the marketing campaign, and it has been observed that from the first campaign to the last (fifth) campaign, there were negligible Pearson correlation coefficients but statistically significant.

NEXT: Compare means analysis (T Test)