BUSI 614 Week 6 SPSS Quantitative Assignment 5 Example

Does Store Format Change Satisfaction? A One-Way ANOVA in SPSS With Welch and Games-Howell Tests

Research Question and Hypotheses

A composite regional grocery retailer operates three formats: large supercenters, smaller neighborhood markets and an online service with store pickup. Before deciding where to invest in service improvements, management wants to know whether overall shopper satisfaction differs across formats. The dependent variable is overall satisfaction, rated from 1 to 10 and treated as interval data, and the independent variable is format, a nominal variable with three levels. Because the question compares the means of more than two independent groups, a one-way analysis of variance is the appropriate test.

The null hypothesis is that mean satisfaction is equal across the three formats; the alternative is that at least one format's mean differs. A significance level of .05 is used.

Data and Descriptive Statistics

The illustrative file contains 180 recent shoppers, 60 per format, each rating overall satisfaction after a shopping trip. In SPSS's Variable View, format was coded 1 for supercenter, 2 for neighborhood market and 3 for online pickup, with value labels attached and the measure set to nominal, while satisfaction was set to scale. Before the main analysis, a frequency table confirmed that no ratings fell outside the 1-to-10 range and that no cases were missing.

Neighborhood market shoppers reported the highest mean satisfaction, and online pickup shoppers showed the most variation in their ratings.

SPSS Procedure

The analysis was run through Analyze, Compare Means, One-Way ANOVA, with satisfaction entered in the Dependent List and format as the Factor. Under Options, Descriptive, Homogeneity of variance test and Welch were selected. Under Post Hoc, Tukey and Games-Howell were both selected so that the appropriate comparison could be reported once the variance test was known. Normality within each group was checked through Analyze, Descriptive Statistics, Explore, with format as the factor list and Normality plots with tests selected. Clicking Paste instead of OK saved the commands to a syntax file, so the same analysis can be rerun on updated data or checked by another analyst without repeating the menu choices.

Assumption Checks

Observations are independent, since each shopper appears once. Shapiro-Wilk tests did not indicate departures from normality in any group: supercenter, W = .98, p = .366; neighborhood market, W = .99, p = .676; online pickup, W = .97, p = .105. With 60 cases per group, ANOVA is also robust to moderate departures from normality.

Levene's test, however, was significant, F(2, 177) = 3.90, p = .022, indicating that the variances are not equal; the online pickup group's standard deviation is about 40% larger than the neighborhood market group's. Although ANOVA with equal group sizes is fairly robust to unequal variances, the Welch test and the Games-Howell post hoc procedure, which do not assume equal variances, are reported as the primary results.

Results

The standard ANOVA showed a significant difference among formats, F(2, 177) = 5.95, p = .003. The Welch test confirmed the result, Welch's F(2, 115.8) = 7.22, p = .001. The effect size was η² = .063, meaning that format accounted for about 6% of the variation in satisfaction; the less biased ω² was .052. Expressed as Cohen's f, the effect was .26, close to the value Cohen (1992) described as a medium effect.

Games-Howell comparisons showed that neighborhood market shoppers were significantly more satisfied than supercenter shoppers and online pickup shoppers, while supercenter and online pickup shoppers did not differ significantly. Tukey's test gave the same pattern. Because the difference between supercenter and online pickup shoppers was small, at 0.15 points, and its confidence interval spans zero, the data give no reason to treat those two formats differently when planning improvements.

Interpretation for Management

The neighborhood markets are not merely different; they are the benchmark the other two formats should be measured against. A gap of nearly one point on a ten-point scale is meaningful for a retailer, because research on the American Customer Satisfaction Index has linked customer satisfaction to customer loyalty (Fornell et al., 1996), and Anderson and Sullivan (1993) found that satisfaction raises repurchase intentions and that falling short of expectations lowers satisfaction more than exceeding them raises it. Management should study what neighborhood markets do differently, such as shorter checkout times or more personal service, and test whether those practices can be transferred to supercenters. The wide spread of online pickup ratings suggests inconsistent execution, such as variable wait times at pickup, that is worth investigating even though its average does not differ from the supercenter's.

Limitations

The design is cross-sectional and observational. Shoppers choose their format, so differences may reflect the kinds of people who shop in each format, such as their age or household size, rather than the format itself. A follow-up analysis could add shopper characteristics as covariates in an analysis of covariance, and a field experiment could test specific service changes in selected supercenters. Satisfaction was also measured with a single item, which is efficient for a concrete overall judgment but provides no detail on which aspects of the experience drive the differences. With 60 shoppers per format, the study had good power to detect a medium effect but limited power to detect small differences, so the absence of a supercenter-versus-online difference should not be read as proof that the two are equal.

References

Anderson, E. W., & Sullivan, M. W. (1993). The antecedents and consequences of customer satisfaction for firms. Marketing Science, 12(2), 125-143. https://doi.org/10.1287/mksc.12.2.125

Cohen, J. (1992). A power primer. Psychological Bulletin, 112(1), 155-159. https://doi.org/10.1037/0033-2909.112.1.155

Fornell, C., Johnson, M. D., Anderson, E. W., Cha, J., & Bryant, B. E. (1996). The American Customer Satisfaction Index: Nature, purpose, and findings. Journal of Marketing, 60(4), 7-18. https://doi.org/10.1177/002224299606000403