Hello,
I am trying to standardise survey responses with varying scales. I am looking at z scores. But some scales have 'don't know options. what is the best way to prepare this data before testing?
Example
1- 5 , strongly disagree to Strongly agree and the 6th option is 'don't know'
And
1 - 7 with 4 = 'undecided'
How should I prepare the data set in jamovi before testing relationships?
Standardising responses_Don't Know responses
-
- Posts: 1
- Joined: Wed May 31, 2023 10:02 am
Re: Standardising responses_Don't Know responses
If I could re-collect all of the data and use consistent scales ("strongly disagree", "disagree", "neutral-or-don't know", "agree", "strongly agree"), I would do that.
Otherwise, there are various ways to try to handle the situation, though some people will disagree with each method. I would recode the first so that each "don't know" response is recoded to be "3".
I would then center each scale by subtracting 3 from each value measured on the first scale (thus yielding a -2 to 2 scale) and by subtracting 4 from each value measured on the second scale (thus yielding a -3 to 3 scale).
Finally, I would multiply each value on the -2-to-2 scale by 1.5, thus yielding a -3-to-3 scale. Thus, both scales would range from -3 to 3.
Otherwise, there are various ways to try to handle the situation, though some people will disagree with each method. I would recode the first so that each "don't know" response is recoded to be "3".
I would then center each scale by subtracting 3 from each value measured on the first scale (thus yielding a -2 to 2 scale) and by subtracting 4 from each value measured on the second scale (thus yielding a -3 to 3 scale).
Finally, I would multiply each value on the -2-to-2 scale by 1.5, thus yielding a -3-to-3 scale. Thus, both scales would range from -3 to 3.
Re: Standardising responses_Don't Know responses
it may be appropriate to treat "don't know" responses as missing values; that's another option.