I want to use pls-sem model in smart pls in order to find the relationships between variables. I’m going to collect data through questionnaire. So I want to know whether I should use the same measurment scale for all of the questions in my questionnaire to get appropriate results?
Our recent research suggests that item format probably doesn't matter.
http://paulspector.com/research-methods/item-format-method-variance/
Hi Sahar
Let your research question decide which methodology is suitable. Not the otherway.
Cheers!
Yes, you can except demographic information. 5-point and 7-point both are fine.
Yes you can. Its advisable to use continuous scale or interval scale to meet the requirement for parametric statistical anslysis.
5 point likert scale is good, expect where the research question need to give the respondent more choice we go with 7.
Five point Likert scale data is widely used in Smartpls. However, when it comes to using it for all your scales, there is a caveat. It is usually recommended to vary the scale properties (such as the scale anchors), especially for the predictor and the criterion measures, in order to avoid the common method variance owing to item formats. If the format for the responses remain the same across the various scales in the questionnaire, it is likely to lead to stylized response patterns that, in turn, may show spurious relationships among variables.
In case all the scales are standardized and changes are not possible, one may go for the other procedural and statistical remedies listed by Podsakoff and others (2012).
Article Sources of Method Bias in Social Science Research and Recomm...
However, as Prof. Paul E. Spector has mentioned, the item format method variance might not really be as serious a trouble as researchers would expect it to be. I, for one, am surely going to cite him next time on this!
you can use 5-p likert scale .however, for demographics you don't have to use that.
Dear Sahar Mortazaee
You can use 5 or 7 point Likert scale or 7 point semantic differential scale to measure the responses of your questions. However, its better to use different scale for different variables of the model particularly for the independent and dependent variables just to avoid the possibility of Common Method Variance (CMV) issue. That means for independent variable you can use 5-point Likert scale whereas for dependent variable you can go for 7-point Likert scale or semantic differential scale. Using different scale point is not a problem because the software considers the standardized values. Hopefully you will find it useful.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey