What is a Likert Scale?

and How Do You Pronounce "Likert?"

    I have always restricted the use of the term "Likert scale" to those scales where respondents express their STRENGTH OF AGREEMENT with each of several statements , typically with an odd number of response options varying from "strongly disagree" to "strongly agree." I have, however, increasingly seen "Likert scale" used to describe items with five or seven ordered response options but where the response scale is not in terms of strength of agreement. For example, where the stem is "I think about the Flying Spaghetti Monster" and the response options are "never," "rarely," "sometimes," "often," and "all the time." In your opinion, are such scales appropriately referred to as being "Likert scales?"

Karl Wuensch


    Excellent question! I am not a statistician, although I pretend to be one whenever I write up manuscripts, so I would say YES, if you are then applying numbers to those responses and analyzing those numbers--note careful use of language in that last phrase.

Annette Taylor


    Concerning Likert scales: I make a point of telling students that Likert actually scaled the category labels he used. If the general format is used (1-5) but no scaling has been done to determine the anchors, I would prefer that students refer to this as a "Likert-type" scale. Likert scaled the anchors so as to approximate equal intervals.

Dr. James Thomas


   As James Thomas observes, Likert used a very specific procedure: rating of agreement/disagreement empirically-derived scales (Note: his preference being for 5 points, unweighted). Moreover, the central point indicates neutrality, and the ratings were made to unidirectional attitude statements.

    Strictly speaking, then, these are the conditions for a Likert scale. I have had recent correspondence on this issue with Marion Aftanas at the University of Manitoba, his concern being the the terms “Likert: and “Likert-type” are used rather inappropriately and somewhat indiscriminately.

     In answer to Karl’s question, I do not think that a scale should be called a Likert scale unless it meets Likert’s conditions. “Likert-type” might be better, but it does not specify the degree to which the procedure differs from Likert’s I prefer the neutral term “ordered response scale" for scales that do not meet Likert’s conditions.

 Stuart McKelvie


 The Likert scale was introduced as a scale of attitudes in Likert's "A Technique for the Measurement of Attitudes," Archives of Psychology, No.140, 1932.  It was a bipolar scale running from one extreme through a neutral point to the opposite extreme.  I think it is the bipolar nature of the scale that makes it a Likert scale, not so much whether it is measuring attitudes .  Such a scale was not new with Likert, though. Titchener used the same kind of descriptive scale to analyze feeling experiences,  running from pleasant to unpleasant through a neutral point.  Wundt did something similar even earlier on the descrption of  feelings.  Still, Likert used a numerical coding for his scale, something Wundt and Titchener did not do and used it, of course, for attitudes.
    A scale that is unipolar going from low agreement to higher agreement  would be more properly called a Guttman scale, since it was popularized by Guttman in 1944.  I am sure it has earlier incarnations, but he used it first, I believe,  in attitude scaling.  So, the linear scale with no central zero point you use in your example is a Guttman scale.

Rand Evans


    Karl, my understanding is that the Likert scale has five points from strongly agree to strongly disagree.  Those with 6 or 7 or 8, etc. are Likert-type scales.  The Guttman scale is unidimensional.    Likert, R . (1932). A technique for the measurement of attitudes.  Archives of Psychology, 140, 52.

Bill Grossnickle 


    Karl, I agree with your take on this. The Likert Scale is a measure of attitudes, preferences, and subjective reactions by eliciting a response along the lines of strength of agreement with the scale items. I don’t think the mere use of questions that have multiple response items are necessarily Likert Scales.

 Mike Brown


    I have checked several books and don’t find a succinct definition of a Likert scale but would agree with your first example.  However, I am probably guilty of using the term Likert or Likert-like for the second case.

 Michael Poteat


    As far as I know, Likert himself used strength of agreement ratings for his scales.  However, I’m not sure I see a difference between the two examples.  If the statement is reworded as “I often think about the Flying Spaghetti Monster” and I rate strength of agreement, wouldn’t the scale provide the same information as the never/rarely/sometimes/often/all-the-time scale?  I suppose it’s an empirical question to which I don’t have an answer.  If you find the answer, I’d like to hear it.

 Jon Reed

 P.S. Andrea and I are jonesing for some more peter peppers.

 P. P. S. I’ve been thinking of the Flying Spaghetti Monster obsessively as of late.


    This site list terms for "Likert" ratings of agreement (many variations) but also include terms for frequency, importance, quality, and liklihood. http://www.gifted.uconn.edu/siegle/research/Instrument%20Reliability%20and%20Validity/Likert.html .  Is there a more appropriate term, other than Likert scale, to apply when using these other options in this format?

 blaine peeden


    I can't locate the textbook that was on my shelf that explained this, but the equal intervals also result from the fact that a true Likert scale is a summative scale, made up of a number of items that, together, are internally consistent.

Michael T. Scoles, Ph.D.


I think we've become accustomed to using "Likert" to reference multipoint bipolar scales that assess agreement, frequency, and so forth. Likert-type is also used to reference them. You might check Bert Green's article on the measurement of attitudes in the Handbook of Social Psychology from the 70's (Bill Grossnickle has a copy and referenced it in his Social Psyc grad class until fairly recently).

 Linda Allred.


go check out rensis likert's phd dissertation ...a technique for the measurement of attitudes ... archives of psychology ... #140, June 1932.  likert's thesis was about trying to simplify thurstone's equal appearing intervals approach ...the data are from a larger work done by gardner murphy ... and looked at attitude scales about 1) international relations, 2) race relations, 3) economic conflict, 4) political conflict, and 5) religion.  response formats were:  yes, ?, no  mc format where options went from bipolar opposites sa to sd in 5 points  this is an excellent work to have students examine  bastardization of his work has occurred over the 70 years  if you want to see what the bugger actually looks and sounds like ... check these out .. this was in his "other" career
http://ollie.dcccd.edu/mgmt1374/book_contents/3organizing/org_process/Likert.htm
 

Dennis Roberts


Likert. Is the first syllable "like" or "lick"?
===============================================================
Date: Tue, 17 Feb 1998 10:37:40 -0500 (EST)
Sender: owner-edstat-l@eos.ncsu.edu
From: Robert McGrath <mcgrath@alpha.fdu.edu>

This question actually came up on another list a few years ago. The author collated results, and found Americans preferred like-ert, while
Europeans tended towards lick-ert. However, one of the respondents who went for lick-ert happened to be the son of R. Likert, so I've changed my pronunciation since then.

Bob McGrath
===============================================================
Gary McClelland told me (KLW): I met Rensis Likert at Michigan and he definitely introduced himself to me as "Lick-ert." My colleague, Rosina Chia, also a Michigan graduate, told me the same.

Back to the Stat Help Page


Contact Information for the Webmaster,
Dr. Karl L. Wuensch

This page most recently revised on 4. October 2005.