home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!paladin.american.edu!auvm!WATSON.BITNET!FOLGER
- Return-Path: <Folger@OLIVAW.watson.ibm.com>
- X-External-Networks: yes
- Message-ID: <9209081754.AA0313@OLIVAW.watson.ibm.com>
- Newsgroups: bit.listserv.stat-l
- Date: Tue, 8 Sep 1992 13:20:24 EST
- Reply-To: "Davis A. Foulger (914) 945-2077 (t. 862-2077)" <Foulger@WATSON>
- Sender: "STATISTICAL CONSULTING" <STAT-L@MCGILL1.BITNET>
- From: FOLGER@WATSON.BITNET
- Subject: Re: continuous likert scales
- In-Reply-To: <9209081636.AA0104@OLIVAW.watson.ibm.com>
- Lines: 34
-
- I must agree with Frank Dane that hardly anybody ever uses "Likert Scaling"
- correctly anymore. What was depicted in the graphical scale query by
- MGRANAAS@CHARLIE.USD.EDU was actually more of an agree/disagree semantic
- differential. The graphical scale certainly was arguable continuous, but
- was certainly not ratio (no zero point). The only question of advantage,
- then, would appear to be one of reducing measurement error by providing a
- more open scale.
-
- On the basis of the research I've seen to date, I would question whether a
- graphical scale will actually make much of a difference to anybodys results.
- The seven plus or minus two rule of scaling didn't just happen. It
- reflects (and continues to do well in the face of continuing) research that
- has shown that there are limits to the extent to which people can reasonably
- differentiate constructs. The latest run of such research has surrounded
- so-called ratio scaling methods in which people are given a baseline and
- asked to make a comparison (e.g. if black and white are 100 units
- different, how different are Clinton and Bush), and the results are
- fascinating. It turns out, for instance, that people estimate true ratio
- measurements with about logarithmic accuracy (e.g. the bigger the measure,
- the bigger the error; such that plotting the two against each other on
- log-log paper results in a linear relationship). My point ... people make
- measurement errors and the scale of those errors appears to expand as a
- function of the openness of the scale. Making the scale wide open using a
- graphical scale simply expands the scale of the errors that will be made.
-
- Given this expansion of error with scale, the question to be asked, I think,
- is one of cost ... cost to the participant in time spent answering
- questions ... cost to the researcher in time spent preparing,
- administering, and scoring the questionaires. It is unclear to me that the
- additional costs ... especially in terms of inconvenience to the subject
- ... come close to covering the fairly modest gains on error that expanded
- scales (graphical, 100 point, or whatever) MIGHT provide.
-
- Davis
-