Subscribe Bookmark RSS Feed

creating a new data table from p-value on all reports

Hi,

Is it possible to create a table of P-Values using tabulate? I want to be able to view a matrix or table with all of my questions listed down the left hand side and all of my factors along the top. My ultimate goal is to have a table of only P-Values for all of my survey questions by factor to easily determine which ones are significant.

Thanks!
6 REPLIES
mpb

Super User

Joined:

Jun 23, 2011

It may well be possible with Tabulate, but I'll illustrate how to do it using Transpose with the Fitness data as an example.

1. Open the Fitness data in the sample library.
2. Fit Model: Oxy, Runtime, RunPulse -->Y and Age, Weight -->Model Effects, click Run Model
3. In the Oxy report right click in the Parameter Estimates segment and click on Make Combined Data Table
4. With resulting table current, Tables>Transpose: Y-->By, Term-->Label, Prob>|t|-->Transpose Columns, click Ok
5. In resulting table delete the column labeled “label” and change Y to Question if you want
6. If you want, select factor columns and, in the column panel right click in the selection and choose Standardize Attributes and change the format to “Pvalue”

This could probably be scripted but I didn't try.
Thanks! That worked great.

However. I am using the p-value or prob> I Z I value under the Wilcoxon Non-parametric test and i have run separate reports for each of my questions to make the data more manageable but now i want all the p-value's in one report and each time i create a new combined data table and transpose it, it creates a new table for each question and i actually need them side by side on the transposed table. So there are 37 questions and 42 factors and i ran each question by factor in a separate report. So that would create 37 separate transposed tables and i just want one. Any thoughts???
mpb

Super User

Joined:

Jun 23, 2011

I would need to see an example of your data layout and what a report looks like, and what the transposed table looks like to understand how to proceed. Can you post a sample?
This is what my transposed table looks like for each report. The column labeled Visual Privacy is 1 of 37 questions. The numbers below it are the P-Values for that question. For each question i ran a wilcoxen non-parametric by factor which is the column on the far left. This generates a total of 37 separate reports. Since the factors are constant for each report i want to be able to click on each report and add the p-values from that report to the already transposed table. Even if i have to delete the label column and duplicate factors column for each report it would still save time.

Factors VISUAL PRIVACY P-Value
A1 0.37560279
A2 0.42390397
A3 0.38341393
A4 0.32826622
B1 0.85647084
B2 0.17206041
B3 0.15005356
B5 0.3370321
B7 0.89310189
B9 0.667621
C1 0.01416655
C2 0.15005356
C3 0.23030115
H1 0.05251884
H10 0.61734792
H11 0.31361689
H16 0.52032395
H17 0.52032395
H18 0.61734792
H19 0.61734792
H20 0.20776643
H9 0.52032395
L1 0.04733372
EU L1 0.2177708
L16 0.10927859
L17 0.21542451
L19 0.79681504
L2 0.35198162
L2 2 0.6390059
L24 0.52760671
L26 0.13311734
L3 0.49908901
L3 2 0.12334048
L30 0.13311734
L31 0.05802239
L32 0.08728506
L4 0.66750991
L4 2 0.84167263
L9 0.07805609
M22 0.84884207
That didnt post correctly. Is there any better way to post the example so you can view it?
mpb

Super User

Joined:

Jun 23, 2011

What should it look like? You might try placing it between lines like this:


Does your original data look like this (assuming 3 questions and 2 factors as an example)?

F1 F2 Q1 Q2 Q3
0 1 5 3 7
0 0 3 2 5
1 1 6 8 6

etc.