Hello all,
I wanted to get a total/average score from combining a 5-point Likert scale (1–5) with three rating questions (1-10).
Can I combine? if yes, which method below is acceptable to combine?
Option 1: Normalize each score to a 0–1 scale using: (raw–min)/(max–min)
this option is applied to all the questions
Option 2: Convert the 1–10 slider to a 1–5 scale using:
(B - A) * (x - a) / (b - a) + A, where [a,b] = [1,10] and [A-B] = [1,5]
this option applied to rating scale [1,10] to convert them to [1, 5]
I plan to use Cronbach’s alpha to check internal consistency.
I would appreciate any input.
Thank you!
Bahar
I recommend using min-max normalization to standardize the scales. Here's the formula:
x' = [x - min(x) ] / [max(x) - min(x)].
You can run this on both scales to make them both range from 0 to 1. Or, if you'd prefer you can convert one scale to match the other with this:
x' = (newmax - newmin) * (x - oldmin) / (oldmax - oldmin) + newmin.
Here's a good explanation: https://www.ibm.com/support/pages/transforming-different-likert-scales-common-scale
Thanks, Louis, for your response. If I want to use the following formula to convert scores:
x' = (newmax - newmin) * (x - oldmin) / (oldmax - oldmin) + newmin.
Would it be better to keep the resulting decimal values, or is that better to convert them to rounded numbers using the approach below?
1, 2 → 1
3, 4 → 2
5, 6 → 3
7, 8 → 4
9, 10 → 5
I recommend keeping the decimal values so as to not remove any information from the wider scale.
- Badges
- Science topic
- Similar topics
- Social Science
- Media
- Journalism