![]() Students received anywhere from a two-response scale to a nineteen-response scale. ![]() ![]() One article that simultaneously argues against Lietz’s consensus (towards more items) but also suggests that more response items don’t induce more noise is Jacoby & Mattell’s 1971 descriptively entitled piece, “Three-Point Likert Scales Are Good Enough,” in which they gave a survey of 60 statements to 360 undergraduates. My key concern with introducing more items was that we would just be introducing noise. (To imagine a 7-point scale, imagine adding a “weakly agree” and “weakly disagree” into the mix.) But the evidence she cites isn’t unanimous, and some of it finds reliability advantages only in some samples. (Earlier, ungated version available here.) She concludes, across a range of studies comparing 5-point and 7-point scale responses, that 7-point scales are usually more reliable, with more differentiation of responses than the 5-point scale. How many response choices should we include? In 2010, Lietz put out a wide-ranging review piece on questionnaire design issues (from the reliability of frequency adverbs – e.g., never, frequently, always – to the use of negative versus positive formulations). (That’s pronounced lick-ert, and has been called by Latham (2006) “ among the most mispronounced in field.”) The traditional Likert item has 5 choices: Strongly Disagree | Disagree | Neither agree nor disagree | Agree | Strongly Agree. When you tally up the responses across the items, that’s a Likert Scale. There is a whole literature on this agree-disagree style of survey item, sometimes called a Likert item. On the other hand, how much of that variation is just noise? (Do you just agree? Or do you strongly agree? Yes, for the next statement, too.) So on the second day of piloting, we use a binary scale: Do you agree or disagree? It was much less painful, but what if all the variation in opinions was not between agreeing and disagreeing but rather in shades of agreement? Maybe this intervention takes people from mild enthusiasm for the clinic’s cleanliness to radical enthusiasm. Watching the enumerator patiently try to explain, repeatedly, the difference between “strongly agree” and “agree” seemed painful for all involved. Recently Dave was in Nigeria piloting a patient exit survey for health facilities, and we had a section similar to the one above.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |