PSYC 421 Week 3 SPSS Assignment Part 2 Example

SPSS Assignment Part 2: Assessing and Interpreting the Reliability and Validity of a Campus Stress Inventory

Purpose

Part 1 described the 250 students whose responses form the inventory's norm group. Part 2 asks two further questions. First, are the inventory's scores consistent, both across its 40 items and over time? Second, is there evidence that the scores reflect stress and not something else? Cohen and Swerdlik (2018) emphasize that reliability is necessary but not sufficient for validity, so both questions must be answered before the inventory's norms can be used with confidence.

Reliability

Internal Consistency

Internal consistency was estimated with Cronbach's alpha using the Scale, Reliability Analysis procedure in SPSS, with all 40 items entered for the 250 participants. Cronbach (1951) introduced alpha as an estimate of the consistency of items intended to measure a common construct. The inventory's alpha was .89, indicating strong internal consistency for a research and screening instrument. Tavakol and Dennick (2011) note that alpha increases with the number of items, so a value this high for a 40-item test does not by itself show that every item is sound.

The item-total statistics table addressed that concern. Most corrected item-total correlations fell between .35 and .62, but two items fell below .20: item 17, about worry over a car breaking down, and item 33, about the cost of textbooks. Removing either item would raise alpha slightly, to .90. These items may reflect financial circumstances more than general stress, and they are candidates for revision.

Test-Retest Reliability

A subgroup of 60 participants completed the inventory again two weeks later. The Pearson correlation between the two administrations was r = .82, p < .001, suggesting that scores are reasonably stable over a short interval. Some change is expected, since stress responds to events, so a perfect correlation would actually raise doubts about whether the inventory is sensitive to real change. Cohen and Swerdlik (2018) point out that the appropriate interval for a retest depends on how stable the construct is expected to be. Two weeks is short enough that most students' circumstances would not have changed much, but a retest timed across midterm week might show lower stability for reasons that reflect the construct, not a flaw in the measure.

Validity

Convergent Evidence

The data file also contained scores on the 10-item short form of the Perceived Stress Scale, a measure that Cohen et al. (1983) first published in a 14-item version to assess the degree to which people appraise situations in their lives as stressful. The correlation between the inventory and the Perceived Stress Scale was r = .71, p < .001. Because the two instruments were designed to measure closely related constructs, this strong positive correlation is convergent evidence that the inventory measures perceived stress. The correlation is high but not so high that the inventory merely duplicates the older scale; its campus-specific items may capture stressors the general measure does not.

Discriminant Evidence

Scores were also correlated with a short social desirability scale, which measures the tendency to present oneself favorably. The correlation was r = .12, p = .06, small and not statistically significant. Cohen and Swerdlik (2018) describe this kind of result as discriminant evidence: a test should correlate weakly with measures of constructs it is not intended to measure. The low correlation suggests that inventory scores are not driven mainly by participants' wish to appear well adjusted.

Interpreting an Individual Score

Reliability also matters for individuals. Using the sample standard deviation of 18.4 and alpha of .89, the standard error of measurement is 18.4 multiplied by the square root of .11, or about 6.1 points. For a student scoring 120, a 95% confidence band is roughly 120 plus or minus 12, from about 108 to 132. A counselor using the inventory should treat a score of 120 as an estimate within that range, not as a precise point, especially when deciding whether a student has crossed a cutoff.

Summary

The Campus Stress Inventory shows strong internal consistency, good short-term stability and a pattern of correlations consistent with measuring perceived stress rather than favorable self-presentation. Two items with weak item-total correlations should be reviewed. The evidence supports using the inventory to screen undergraduates for elevated stress and to compare groups in research. Item revision should come first, followed by a fresh reliability analysis to confirm that alpha holds. The inventory does not yet support diagnostic decisions, since no criterion such as clinical diagnoses or counseling center use was examined, and future analyses should add that evidence.

References

Cohen, R. J., & Swerdlik, M. E. (2018). Psychological testing and assessment: An introduction to tests and measurement (9th ed.). McGraw-Hill Education.

Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24(4), 385-396. https://doi.org/10.2307/2136404

Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16(3), 297-334. https://doi.org/10.1007/BF02310555

Tavakol, M., & Dennick, R. (2011). Making sense of Cronbach's alpha. International Journal of Medical Education, 2, 53-55. https://doi.org/10.5116/ijme.4dfb.8dfd