I am designing a questionnaire by using five-point Likert scale. What is the advantage of five-point Likert scale over seven-point Likert scale?
Thank you.
$\begingroup$
$\endgroup$
Add a comment
|
2 Answers
$\begingroup$
$\endgroup$
4
In general what and how you intend to use the questionnaire makes a difference in practice. I have often only used 5 points because my subjects were often quite busy in daily life and return rates were higher with 5 points. With 7 points you get more data and I would think that might help in situations with low sample size.
I cannot find the actual paper, but this practical source might be helpful.
I might need to add, that I have not been doing many surveys in the last couple of years. The state of the art might have advanced and I am unaware of that...
-
$\begingroup$ I'm going to take issue with the claim that "with 7 points you get more data and I would think that might help in situations with low sample size." The granularity of measurement is different from the number of measurements taken. The latter is "sample size" and the former has nothing to do with it. $\endgroup$ Commented Aug 23, 2014 at 19:12
-
$\begingroup$ (continued) If anything, in small samples I'd prefer fewer scale items, from a purely statistical perspective. With coarser measurements, there is less variation in the data to be explained. This might sound undesirable, but if you have very little data it contains very little explanatory power, so coarser measurements are less of a burden on the data. $\endgroup$ Commented Aug 23, 2014 at 19:15
-
$\begingroup$ Do you not end up with an easier separation problem if people can choose from 7 instead of 5 values. That is more data to separate answers and increases the information available, albite not the sample size. $\endgroup$ Commented Aug 23, 2014 at 22:37
-
$\begingroup$ A finer scale has the potential to contain more information, but it just as well leaves more room for noise and errors to enter the data. In a small sample, unnecessary noise can be disastrous. Coarser measurements will cause the data to clump more. So you'd be correct if the data is poorly separated. But if there is good separation then it will be emphasized. And this is before you consider how survey subjects respond differently to different scale types (which is something I don't know much about, and I think will end up being more important in practice). $\endgroup$ Commented Aug 24, 2014 at 1:28
$\begingroup$
$\endgroup$
I agree that from the responder viewpoint, seven point scales look more time consuming to answer. A bigger question would be, why an odd number? An even number forces a side of opinion. Imagine questions with social context. If everyone responded 3/5, would you rather interpret it to mean that no one cares about the question or that people are ambivalent about the issue? Perhaps then the sample needs better screening? Then imagine the same sample on a four point scale where 75% of response was 2/4.
