Unai Coto Suárez

To ensure you are using valid methods while working with weighted survey data, you can consider the following:

• Besides the Survey package, consider using the Weights package for more advanced weighting techniques.
• You can perform sensitivity analysis to check how your results change with different weighting schemes.
• You ca validate your results using different statistical packages and methods to ensure consistency.

Lastly, always refer to the documentation of the packages you are using to understand how weights are handled. If possible, consult with a statistician or survey methodologist to review your approach and ensure its statistical soundness.

Do you need to weight to get internal consistency reliability?

It looks like you're weighting to get results representative of a population. Its not necessary, I think, to assess internal consistency reliability. I think the normal approach here would be to calculate omega on the unweighted scores and use the scale or subscale scores with weights in subsequent analyses.

Its possible weighting is part of the scale scores (but that's unsusual) and then I think you'd need to seek expert advice on the best way to handle that.

I agree with Thom, you don't need a "representative" sample for measures of internal consistency and/or reliability. The goal usually is to reflect the internal structure of the scale, and unweighted analysis focuses on how the items relate to each other within the collected sample.