Should samples to be used on psychometric evaluation have homogeneity? Like for example same occupation? age group? etc. And should it have an rrl? thank you!
It depends what population you want your evaluation to apply to. For example, I might have an anxiety measure that I want to argue applies for US children from 6-10 years old. I should restrict the sampling to this group, but would want it heterogeneous within this group. Sometimes you will want purposeful sampling within this, so for example I would probably want my sample to have an adequate number of people with high anxiety. There are also other issues if you psychometric measure is meant to establish cut-offs/thresholds.
I agree with Daniel and Khan both. Samples have to be chosen in an interplay between pragmatic evaluation of which groups and circumstances your instrument/evaluation should apply to, and of the mathematical assumptions/demands of the statistical procedures you intend to use.
Especially if there is a component of distress or dysfunction in your research question, which is not expected to be highly prevalent in a broader population.
It can be nice to have breadth in order to assess the generaliSability of your instrument, but in it is sometimes safer to operate within strict parameters.
I was thinking of health care providers (nurses, caregivers) however I do think that I have to control the age group. I will be evaluating Ryff's PWB scale