Most of the studies are inducating five point Likert Scale but can I use four point likert scale?
Please answer this question with relevant reference.
The original Likert scale was a 5-point scale. But there is no reason to limit yourself to precisely this number of points. You'll find everything from about 4 to 10 points on such scales (occasionally even more, though it's unlikely we gain additional information by expanding beyond 7 or so).
The most important thing to bear in mind is that odd-numbered forms always have a "neutral" response option (the middle) while even-numbered forms force the respondent to make a choice one way or the other. Each approach has its advantages.
"According to Bishop (1987), respondents in public opinion polls
tend to gravitate towards neutral responses because they try to avoid
negative feelings regarding a conflicting issue. In Nowlis, Kahn, and
Dhar (2002), this idea of neutrality of answers occurs due to the
avoidance of a cognitive effort related to the choice of positive and
negative feelings regarding an issue. Normally it is offered the possibility
to give neutral responses when the subjects are not so used or
indifferent about the research topic. A ‘neutral’ or a ‘no opinion’ response is adopted to avoid taking responses which do not reflect their
true beliefs (Johns, 2005; Krosnick et al., 2002).
Some researchers suggest removing neutral answers as an answer
option in a Likert scale (Garland, 1991; Kalton, Roberts, & Holt, 1980;
Krosnick et al., 2002). By removing neutral options respondents are
forced to do a cognitive effort (Krosnick et al., 2002). Ambivalent
persons are forced to take an ‘opinion position’ about a specific issue
(Weijters, Cabooter, & Schillewaert, 2010; Nowlis et al., 2002). Subjects
with higher ambivalence will lean towards more extreme option when
it is used a 5-point Likert scale (Nowlis et al., 2002). When neutral
options are avoided, the respondents are forced to express their true
feeling (Garland, 1991)."
You can find the references at:
Perceptual studies on air quality and sound through urban walks | Request PDF (researchgate.net)
One thing to consider is the type of statistical analyses that you intend to carry out. For items with ordered categories response format, it is common to treat them as continuous interval level measures, so long at the number of categories is about 5 or more. So it is not unusual to use procedures such as regression analysis, ANOVA, or factor analysis with responses to 5 - point Likert type scales. But you are more likely to get criticism from reviewers, when using such methods with only 4 response options. Unless you are going to use more advanced procedures such as ordinal regression, my experience is that reviewers or journal editors are likely to ask that you treat responses as categorical, which would limit you options available, particularly if your sample is small, or if there are a small number of responses to some of the categories.
بعد التحية
يمكن استخدام مقياس ليكرت الرباعي
NONE AT ALL, except to remember -at the time of statistical treatment- that the scores of the responses to the items, despite the fact that some "believe or agree that they do", ARE NOT OR ARE AT AN INTERVAL LEVEL, YES NO ORDINAL
Likert scale is also subjected to a lot of debates and controversies in regards with the analysis and inclusion of points on the scale. Auseful article pl readArticle Likert Scale: Explored and Explained
I am obviously very late in responding to this question (my bad). You have some very good responses to the question you raised. When using a four point scale you will be loosing variance in your responses. A five point scale gives you more variance which can be important in terms of the statistical analyses done. John Crawford makes this point in his response above. Several people have also raised the issue of a neutral response choice within the scale. A likert scale is a forced choice assessment , you are forcing people to choose from the options that you give them. What if the respondent does not find any of the choices that represent their thinking? This may encourage some respondents to leave the item blank. In this situation you may provide a situation where the scale is no longer useful, or you have a scale that is missing input on one item. You lose some of the variance in your responses. Some scholars would suggest that you make those respondents think, do the cognitive work and make a choice. When you force respondents to make a choice from the list you have provided are you "creating" a response rather than measuring the construct. Forcing respondents to choose among the choices you have provided you are forcing the respondent to make a choice that is not reflective of the respondent's thinking. In doing this are you not "creating" rather than measuring the respondent's thinking? In providing this neutral response I think it important that you assign a score of 0 to the neutral response. If you give the neutral item a number rather than 0, your total scale score is higher than it should be. Hope this is helpful. Ed
- Badges
- Science topic