 |
|
Cruise Scientific Visual Statistics Studio Visual Statistics Studio Guide |
Item Analysis
Select (Designs, Measurement and Scaling, Item Analysis)
|
|
Select (Analysis III, Item Analysis) and click the Select All and the Accept commands
 |
to obtain the output panel of the Item Analysis application
|
|
Since the index of item discrimination for the I4 is negative, delete the Item 4
 |
and analyze the reduced data
|
|
Deletion of Item 4 increased the internal consistency reliability from .308 to .883.
Cronbach's Alpha and Kuder-Richardson's K-R-20
The step-by-step computation of the Cronbach's alpha coefficient of internal consistency reliability
 |
is as follows. Select (Operations, Add Variables) and add items 1, 2, and 3. Name the resulting variable O (Obtained Scores)
 |
Select (Transfers, Descriptors - Scalars), mark True Variance, and click on the Select All and the Accept commands.
 |
Compute the k / (k-1) term of the above equation as 3 / 2 and add the variances of the I1, I2, and I3 variables (.24 + .16 + .24 = .64). Compute the second term of the above equation (1.44 - .64) / 1.44) and multiply the first (1.5) and second (.556) terms (.833)
 |
For the binary scored items (agree-disagree, true-false) where the item variance is computed by the pq formula, the Cronbach's alpha and the Kuder-Richardson's reliability (K-R-20)
 |
are identical.
