I've created a five point Likert scale for a survey. Using the results I can easily convert to a weighted average (which I've done). What I want to know is, if it's okay to covert the result to a percentage by scaling it up.
I can make the percentage by just dividing by five and multiply the result by 100, but It just feels wrong as I think it misrepresents the respondents answer. I'll explain.
Essentially I'm scaling a 5 point Likert scale up to a 100 point scale. Where the values a respondent gives become equal to 1 = 20, 2 = 40, 3 = 60, 4 = 80 and 5 = 100. However this only gives the respondents 5 choices out of what should be 100. Where it should really be 1 = 1 to 20, 2 = 21 to 40, 3 = 41 to 60, 4 = 61 to 80 and 5 = 81 to 100.
So say someone answers a question with 4 and I convert this to fit a 100 point scale. I'd be changing it to 80, but if I had given this person a 100 point scale he may have answered 61, 80, or anywhere in between. And in a weighted average the difference between 61 and 80 could have quite an impact.
This of course ignores that the person may never have even selected an answer between 61 and 80 when given a greater choice.
I feel like doing this misrepresents and misuses the person's response to the survey. Is this correct, or have I just gone down some rabbit hole? Thank for any help in advance!
Best Answer
This answer is predicated on your survey items having some possibility of being an interval-like scale. That is to say, the difference between two responses is comparable in size.
First, you probably don't want to divide by 5 for a scale from 1 to 5. Instead, you probably want to code from 0 to 4, and then divide by the maximum. Then, 0% is the lowest possible option, 100% is the highest possible option. The advantage to a strategy like this is that it allows you to rescale responses on scales with different granularity (e.g., a 5-points vs a 7-point scale).