Would the value of Cronbach’s alpha increase if any items were deleted—and, if so, which items and by how much?.
Before performing a factor analysis, do a reliability analysis for the entire 20-item scale. Click Analyze ➜ Scale ➜ Reliability Analysis. Move the 16 negatively worded CESD items and the four reverse-coded items into field for items. Click the Statistics pushbutton and in the next dialog box, click Descriptives for all three options (Item, Scale, Scale if Item Deleted); Inter-Item Correlations; and Summaries for Means, Variances, and Correlations. Click Continue, then OK, then answer these questions: (a) How many cases were in this analysis? Why do you think the number is so low? (b) For the 20-item CES-D scale, what is the value of Cronbach’s alpha? Does this indicate adequate internal consistency? (c) What was the range of correlation coefficients between pairs of items on the CES-D? Does something about this range seem puzzling? Between which pairs of items are the correlations highest and lowest in value? (d) What was the mean inter-item correlation? (e) In the panel for ItemTotal Statistics, which item had the strongest corrected itemtotal correlation? Which had the weakest? (f) Which item had the strongest (and weakest) squared multiple correlation? (g) Would the value of Cronbach’s alpha increase if any items were deleted—and, if so, which items and by how much?
