To Art, and others: On Sun, 07 Jul 2002 18:50:04 GMT, "Arthur J. Kendall" <[EMAIL PROTECTED]> wrote:
> Some of the confusion is because the original use of the term refers to a > summative scale made up items with that response scale. Some people use the term > Likert scale to refer to a single item with that kind of reponse scale. > [ snip, an example using 'Disagree' - which is what Likert used, and what most textbooks' examples consist of.] Generally speaking, an item can have a scale, but a 'Likert scale,' in my little experience, has always been a summation. The items - in my preference - should be scored symmetrically around a neutral center. I have also seen 'Likert' used here as a synonym for 'additive scale' - even for a scale made up of 0/1 items. In reviewing an article, I would object if that (a total of 0/1s) was called Likert; or, calling a single item a Likert scale. A single Likert-scaled item is a pretty darn feeble construct, so I think that someone has to be explicit about the limit, if that is what they test. That is, I do think they ought to call it an 'item' and not a scale. However: R.P. McDonald (in Test Theory, 1999) says, "It seems inevitable that we follow the accepted usage and refer to an integer score for an order-category item as a *Likert score*." - I grant that 'score' is better than 'scale' but I still prefer 'item.' How far 'gone' is the terminology? I think McDonald concedes too much, unless the 'accepted usage' is really that wide-spread.... -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
