A computer simulation was designed to investigate the degree to which coeff
icient alpha is affected by the inclusion of items with different distribut
ion shapes within a unidimensional scale. Theoretical scales were generated
that combined different numbers of normal, moderately skewed, extremely sk
ewed, and leptokurtic items. This process was repeated using different numb
ers of response categories (3, 5, 7, and 9), and different levels of interi
tem correlation (.25, .50, and .75). Results indicated that reliability did
not decrease dramatically as a result of utilizing differentially shaped i
tems within a scale; the largest decreases came about when utilizing highly
skewed items. However, these results were moderated by the correlational s
tructure and by the number of response categories. Utilizing differentially
shaped items brought about the largest decreases in reliability when the i
nteritem correlations were low and when the number of response categories w
as small. Conversely, reliability was least affected when the interitem cor
relations were high and when the number of response categories was large. I
t was suggested that the practice of deleting items from a scale on the bas
is of distributional nonnormality may adversely affect scale validity with
no appreciable increase in reliability.