For example, I am using a 5 point Likert scale of
Never
Almost Never
sometimes
almost every time
every time
and for 2nd questions
5 point scale of strongly agree, Agree........Strongly disagree with the same questionnaire?
There would be two separate subscales according to their theoretical framework that can be proposed, that is to say two differential factors because otherwise it would be more advisable to have the same type of likert option.
Considering the complexity of any construct or dimension it may have been application inside questionnaire in association of comprehensive justification (very solid).
I would not recommend it, unless these two Likert scales are used for different purposes in your questionnaire, say one of the scales is adopted in a stand alone question/s about the demographics or initial information required in your research and the second Likert scale is based on the theoretical framework or model of your research. Besides, one Likert scale in your example measures frequency of an event (e.g., the number of times that you have purchased online from Amazon.com) and the other quantifies and eventually operationalise an emotion/feelings towards a topic/themes. The important point in this example is that you apply a consistent scale in measuring your constructs, which is usually "strongly disagree,...,strongly agree".
If you ignore this important difference in your questionnaire, you will face issues in analysing and interpreting the findings, or your results will be biased.
Hope it helped and all the best!
You can but It might be confusing for respondents and later for statistical test selection, better follow John-Kåre Vederhus's advice on "established measures"
Adeel Luqman, my hunch is that it would be quite permissible to do what you are suggesting, i.e., having different wording for the response options offered to participants.
The most important thing is that the response options are appropriate to the topic at hand.
Completely acceptable. Even if you are measuring a single latent construct, it does not mean that all the items must have similar Likert scale response anchors. The anchors do vary depending on how the items are phrased. Also, see in your data analysis if these items with different response anchors load on different factors; not necessarily.
Hi Adeel Luqman
you can use two Likert scales with different type of response categories as long as you find it relevant for the purpose of measuring the constructs. However, while converting the responses into the scores, it might be little troublesome for you. Though, that can be dealt with as well.
It could be acceptable, but it depends on your variables/constructs in your questionnaire/model. You should have the same Likert scale for each variable/construct to avoid any biases in your analysis.
For example, many of cyclists NEVER used bicycle sharing system. However, most of them are STRONGLY AGREE with usefulness of sharing mobility.
@Danial In your example of cyclists, probably the items of practice and attitude load on different components. It may not always be possible to have the same anchor terms across items.
Yes, it can be done, depending upon the profile and concerns of the respondents.
Definitely acceptable, please refer to the below article; Author uses different scales for each variable.
Using the same scale for all questions facilitates the completion of the questionnaire by the respondents and the interpretation of the results by the researcher. Moreover, five-point scale is just as good as any, and that an increase from five to seven or nine points on a rating scale does not improve the reliability of the ratings.
referencres:
(Sekaran & Bougie, 2010)
(Hair, Black, Babin, & Anderson, 2010).
- Badges
- Science topic
- Similar topics
- Statistics
- Survey