BUSI 820 Week 6 Discussion Forum 6 Main Thread Example

Discussion Forum 6: Main Thread and Forum 5 Reply

BUSI 820 Quantitative Research Methods, Discussion Forum 6 Main Thread (with the Forum 5 reply)

Thread by [Student Name]

The data set is illustrative, generated for this page with a fixed random seed and analyzed with the statistics SPSS reports for an independent-samples t test; the bank, the classmate and the prompt are composites standing in for the course's own; the books are real; the week, points and readings follow Liberty's public course guide for BUSI 820.

The third research question asks whether intention to leave differs between tellers and personal bankers. Each employee holds one role, so the groups are independent; the independent variable is dichotomous and the dependent variable is the summated intention scale, which was approximately normal in Forum 2. Morgan et al. (2020) point to the independent-samples t test for this design.

In the illustrative data, the 113 tellers had a mean intention score of 13.88 (SD = 4.26) and the 73 personal bankers a mean of 12.84 (SD = 3.85). Levene's test was not significant, F = 0.50, p = .48, so the equal-variances row of the output applies. The difference was not statistically significant, t(184) = 1.70, p = .090. Tellers averaged 1.04 points higher, and the 95 percent interval around that gap stretched from -0.17 to 2.26, and Cohen's d was 0.26, a small effect by the benchmarks in Cohen (1988). Because the teller group is larger and the scale has a floor and ceiling, I also ran the Mann-Whitney U test; it agreed, U = 4,649.5, p = .14, with medians of 14 for tellers and 13 for personal bankers.

The confidence interval tells the bank more than the p value does. The interval runs from a tiny advantage for tellers to a gap of more than two points, so the data neither confirm nor rule out a meaningful difference. A two-point gap on a 5-to-25 scale would be worth acting on, since tellers are the larger group and the most expensive to keep replacing. With a d of 0.26, a study would need roughly 240 employees per group to detect an effect this size with 80 percent power, more than this bank can provide in one year.

For now, the honest summary is that tellers in this sample reported slightly higher intention to leave than personal bankers, but the difference was small and could be due to chance. That is consistent with the earlier forums: the clearest signal so far is the modest association with supervisor support, not job role or work arrangement.

Reply to Lauren (Forum 5)

Lauren, your Forum 5 thread correlated sales training hours with quarterly sales for 48 representatives and found r = .41, a medium effect. You also mentioned that you have each representative's sales for the quarter before training and the quarter after. If your next step is to compare those two quarters, an independent-samples t test would treat the before and after scores as if they came from different people. Morgan et al. (2020) would point to the paired-samples t test, which analyzes each representative's change and removes the large differences between strong and weak sellers from the error term. If the change scores are skewed, the Wilcoxon signed-ranks test would be the alternative. Would you be able to add a comparison group of untrained representatives over the same quarters?

References

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Lawrence Erlbaum Associates.

Morgan, G. A., Barrett, K. C., Leech, N. L., & Gloeckner, G. W. (2020). IBM SPSS for introductory statistics: Use and interpretation (6th ed.). Routledge.