TRPOFILE= transposing the data matrix

To transpose the rows and columns of the data matrix, select Transpose on the Output Files menu or include TRPOFILE= in your control file. Details are given at Transpose file specifications.

 

TRPOFILE= file name. This is the name of the transposed file.

This operates in association with TRPOTYPE= control characters. This controls what is written to the transposed file.

 

TRPOFILE=? opens a Browse window

 

 

 

 

 

Example 1: You want to "rack" the data in exam12lo.txt and exam12hi.txt, so that each person record contains the ratings from both files:

 

1. Analyze exam12.txt with data=exam12lo.txt

Transpose this data file only to exam12lod.txt

The "lo" items are now rows

 

2. Analyze exam12.txt with data=exam12hi.txt

Transpose this data only to exam12hid.txt

The "hi" items are now rows

Transpose the control file only to exam12tran.txt

The person labels are now shown as item labels

 

3. Analyze exam12tran.txt with data=exam12lod.txt+exam12hid.txt

The columns are the original persons. The rows are all the original items, "lo" and "hi".

Transpose this control and data file to exam12rack.txt

The persons return to the rows.

The columns are now all the original items, "lo" and "hi".

 

4. Analyze exam12rack.txt with Winsteps. This is the desired control and data file.

 

Example 2: We want to see if one group of persons have a different response "style" than another group of persons on a 6-category rating scale. To do this,
1. Transpose the Winsteps control-and-data file so that the rows become columns: Winsteps output file menu: Transpose.
2. In the transposed Winsteps control-and-data file,
GROUPS = 12112111.... to match the person groups.
3. Save the transposed file and analyze it with Winsteps.
4. Look at Table 3.2. There will be one rating scale for Group 1 and another rating scale for Group 2.

 

Example 3: We want Cronbach Alpha for the items.

Cronbach Alpha is designed for persons, but there is no reason why the same computation cannot be applied to items. In Winsteps, transpose the data (use the Output Files menu), analyze the transposed dataset, and report Table 3. Its Cronbach Alpha applies to the original items. If Alpha for the items is too low, then the likely reason is that the person sample size is too small. More persons -> higher item Alpha. More items -> higher person Alpha.

 

---

 

Empirical Person Response Function

 

Using example file Exam1.txt.

Step 1. Standard analysis

Step 2. Output files menu: Transposed

Step 3. Select the options you want

Step 4. Click on permanent or temporary file

Step 5. Click on Launch Winsteps

Step 6. Winsteps analysis of transposed file

Step 7. Graphs Window: "Empirical ICC."

Step 8. Click on "Absolute x-axis"

 

Here's the person response function for the first person in Exam1.txt: