We have a Likert type test with scaling 0-5, developed and validate in our counrty, and wonder whether it is possible to make use of it as 1-5. We particularly want to make the scale to similar to the other scale in the questionnaire package for better comprehension especially in adolescent and pre-adolescent samples. Is this permissible? Answers and references for relevant research, if possible, is very much appreciated.
Two comments: First, and I presume you know this, if you change the number of points on the scale it will mean you can no longer say the scale is validated. So you will need to consider whether to just accept that or try to re-validate.
Second, there are good arguments that a unipolar scale should be 1- 5 points. See
Revilla, M, Saris, W, E. & Krosnick, JA 2014, 'Choosing the Number of Categories in Agree-disagree Scales', Sociological Methods & Research, vol. 43, no. 1, pp. 73-97.
Krosnick is a good source for research into other questions as well. There are lots of other arguments as well such as the value of having numbers as labels, only labeling the extremes, category specific questions and so on.
I understand from your reply you say that "it is not admissible until I re-validate the instrument". And in the validation of the "new form" 1-5 would be a reasonable option.
The 0-5 scale, I gather, is somehow item-specific (as Saris et al. use the term). Thus I think you would be well-advised to change to the 1-5 standard form, as Dr. Roberts suggests. See Saris, Willem E., et al. "Comparing questions with agree/disagree response options to questions with item-specific response options." Survey Research Methods. Vol. 4. No. 1. 2010. Revalidation would, however, be required.
Additionally, interesting response-order effects have been observed for Likert-type scales. So, may be this article will be useful for you, too.
http://www3.nccu.edu.tw/~jyjan/papers/responseorder.pdf
In practice, you can change the number of points and have the same results regardless of scale length. However, some researchers are stubborn in that they assume that if the number of response options has changed, that somehow changed the whole instrument. In some rare cases that may be so, however I have yet to find one such case.
My suggestion would be to do a pilot study, where you would present both measures (the original and the one with the adjusted scale) to two similar groups of people and compare the results. When the results show that the two measures function in the same way (which should be the case), proceed with your intended study.
Presenting all items in a battery in the same way is a good way of balancing response effects. And it would be even better if all the items in the questionnaire would be presented in random order.
See these:
- Dawes, J. (2008). Do data characteristics change according to the number of scale points used? An experiment using 5-point, 7-point and 10-point scales. International Journal of Market Research, 50(1), 61–77.
- Leung, S.-O. (2011). A Comparison of Psychometric Properties and Normality in 4-, 5-, 6-, and 11-Point Likert Scales. Journal of Social Service Research, 37(4), 412–421. http://doi.org/10.1080/01488376.2011.580697
It seems to me you have a choice between potentially confusing your respondents with two similar but different response formats, versus maintaining a strict replication of one of the two scales. In that case I would vote for changing the 0-5 scale to match the other one.
I have attached a paper by two well-known experts on survey design which includes a discussion of differences in the number of responses for Likert type items.