Page 1 of 1
4-point scale in questionnaire for analysis in SmartPLS
Posted: Fri Feb 12, 2021 5:18 am
by Wiley
I have a factor about susceptible to motion sickness in the questionnaire, with options including:
1=Not at all
2=Slightly
3=Moderately
4=Very much so
Is it ok to use it for my questionnaire, which can then be analyzed as an independent variable later in SmartPLS?
Thank you.
Re: 4-point scale in questionnaire for analysis in SmartPLS
Posted: Wed Mar 24, 2021 8:38 am
by jmbecker
PLS algorithms use normal (Pearson) correlations and regression. Thus, they assume metric or at least quasi-metric data.
Most researchers assume quasi-metric data for their questionnaire responses if it has sufficient categories (usually at least 5 or better 7 or more) and has equidistant categories (e.g., Likert scales).
Whether you can and want to assume that your variable is quasi-metric depends on your judgment (and the judgment of your reviewers after making a good argument for your decision). While 4 categories might not necessarily be a problem (even if it is not optimal), the bigger problem in your case seems that your categories are not equidistant. The perceived difference between the categories "not at all" and "slightly" might be closer than the distance between "moderately" and "very much so".