jump to navigation

Modifying a scale + pilot studies March 18, 2012

Posted by thinkphd in : Miscellaneous , trackback

I recommend not modifying a scale used in your research unless that is your only (or best) option. There are several things you must do if you choose the modify route. All of these things take time and require specific reporting in your dissertation.

If you find that you must modify the scale, you must state how the original instrument was validated and what your modifications are (were). You must conduct a pilot study with the “new and improved” instrument – you cannot assume since the original instrument was OK, your new one is, too. It isn’t. The original scales were valid and reliable as a “whole” not the sum of its parts. Your pilot testers should be “experts” in the field. Do not pick other students or your professors (unless they represent a sample in the population).

Once you have received feedback from your expert panel, you must report what they said, and make changes to your instrument. Depending on their comments, you might need to re-pilot the instrument. Next, use Chronbach’s alpha reliability coefficient or similar test(s). You will report all of these results, and base your reasons to use the “new and improved” instrument on what you found out.

From time to time someone will ask if just changing a word or two is OK. Consult with your chair because it really depends on what words are changing and the context. For example, I used a survey with British spellings of a certain word (“s” in place of a “c”) and this was deemed to be not sufficient alteration. However, I did complete a pilot study. Here is a quote from my dissertation:

“The pilot test was conducted to provide feedback and advance warning about any corrections, deletions, or additions to be made before extending official survey invitations to participants. In addition, the pilot test invited comments about the relevance of questions as related to the research intent. Pre-testing the instrument in this way contributed to ensuring reliability and validity of the instrument relative to the contemporary sample population.”

Be Sociable, Share!


no comments yet - be the first?