Table 3, coefficient (Y is consistently very high, ranging from .82 to .93,
with a median of .90 (Cronbach, 1951). These results compare favorably
with most attitude measures (cf., Smith et al., 1969).
Item analyses (correlations between each item of the commitment scale
and the total score less the item) are reported in Table 4. The last column
in Table 4 reports the average item-total score correlations across six
samples. In three of the six samples for which these data are available,
a nine-item short-form of the instrument utilizing only positively worded
items was used. The average item-total score correlations reported for the
negatively worded items are therefore based on three samples while the
average correlations for the remaining positively worded items are based
on six samples. A review of the correlations reported in Table 4 indicates
that each item had a positive correlation with the total score for the OCQ,
with the range of average correlations being from .36 to .72, and a median
correlation of .64. In general, the negatively worded items correlate less
highly with the total score than the positively worded items, although this
difference is not great. These results suggest the 15 items of the OCQ are
relatively homogeneous with respect to the underlying attitude construct
they measure.