how to calculate total score not subscore in rating scale ?

[bhr-q]

Hello,

I have the survey including the below answer, in which the answer to 4 questions is like this:

0 (No pain) :             

1

2

3

4

5

6

7

8

9

10                                    (Pain as bad as you can imagine)

And the answer for the fifth question is like:

0%  

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

My question is should I convert the fifth question to be like other question and then do like below to do comparison pre and post as I want to use Wilcoxson signed rank test.

                pre_ Q1   ………..pre_Q5         total_pre                 post_Q1 ………….. post_Q5       total_post

1 

2

3

.

.

.

10

Thanks

[mkeintz]

This is not really a SAS question, right?

It appears to me that you really have what I would call a psychometric question.  Namely, how similar is a putatively interval-based measure (0% to 100% in 10% increments) to a simple ordered rating measure (0 to 10)?  I'm sure there's a lot of psychometric literature available on this question.

I guess if your fifth question responses were already 0 to 11, you would go ahead and add those values across all pre questions and over all post questions to compare pre and post totals.    That's what your desired output table looks like.  But why would you generate a statistical test for the totals instead of a test for each individual pre-vs-post question?  Should those questions really have equal significance in the study (as implied by adding those scores).  That's another question of the psychometric literature, or perhaps for the research on whatever topic your data addresses.

But if you MUST do a comparison of a summary of all pre vs all post, I suppose you could recognize that the percent response offers 11 ordered values, just as the rating response does, and make a 1 to 1 conversion of the percentages to rates scored as 0 to 10.  

[bhr-q]

Thank you so much for your answer; the researcher asked me to consider the total score, not the sub-score, that's the reason I used the total score, and yes, my question is more about how to calculate the total score, I meant should  I add-up the answer of  ALL the question/item for the first patient like the design above and then again add up all the question/items for the section patent , etc? this is the correct way of calculating total score?

[ballardw]

Percentage as an answer implies a numerator/denominator or ratio sort of value. Not at all the same as an ordinal perception scale.

I would not provide any answer involving combining that number with 4 ordinals without seeing the questions and any instructions involved in responses.

Have you even looked at the data to see if there is any correlation between that percentage question and the others? That might tell you if people are treating them similarly enough to attempt this.