I have compiled a questionnaire consisting of items which aim to measure different aspects of elevator and staircase use. The main idea when composing the items was that it would be the most informative if I could measure the direction and the extent of influence the concept has on the individual. That is why items look like this:

"When I am in a hurry, I use the

staircase -3 - -2 - -1 - 0 - 1 - 2 - 3 elevator"

I have 70 items in total, which were constructed based on 15 categories, e.g. Laziness, Speed, etc. My plan was to just sum up the standardized values of items which I think belonged together to create scales to work with. I do not think an EFA approach would be applicable here, because I do not think there are any latent variables to define.

My question is: what would be the correct way to select the best items to include in the summed scales? Do you think my idea of summing the standardized item scores is correct, and if not, what would you do differently?

Edit: My main goal is to reduce the number of variables, and use them in a cluster analysis to explore if my sample could be divided into clusters in a meaningful way and if there are any nontrivial clusters. The research question is essentially: can people be categorized based on their luggage, how tired they are, etc. (these are the scales I am trying to create), and based on in which direction these "factors" influence their decisions.

  • $\begingroup$ But you haven't told us what it is you are actually trying to do? What is your research question that you are trying to answer? $\endgroup$ – StatsStudent Jun 16 '19 at 16:07
  • $\begingroup$ Thanks for reminding me! I have edited the post. $\endgroup$ – hnandi Jun 16 '19 at 16:31
  • $\begingroup$ If you are summing item scores (raw or standardized, both are reasonable) you are actually assuming they belong to one construct. This construct could be thought as a latent factor if there are considerable correlations between the items, or it could be thought as a battery, if correlations are small or don't matter. stats.stackexchange.com/q/133492/3277 $\endgroup$ – ttnphns Jul 11 '19 at 12:41

First, 70 items is a LOT. Are people really going to keep seriously thinking about and answering question if they are answering 70 (possibly quite similar) questions?

There is a latent variable here - possibly more than one. Why do you say that there isn't? People have a propensity to use stairs or elevator (under different circumstances). That latent, by most definitions.

I'd use EFA.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.