Assessing Concurrent Validity

[DarkoG]

Hi all,

I'm fairly new to R and have been working in psychometrics for the past year or so.

I just had a quick question about using the Pearson product moment correlation.

I've developed a questionnaire that attempts to answer the question: can we evaluate soft skills based on online activity? One of those skills, for example, is communication. That variable is assigned a value of 0 (poor communication) to 5 (excellent communication), so I'm really dealing with ordinal data. What I'm looking to do is to establish concurrent validity by comparing my results with a pre-existing validated tool that measures the same domains.

I'm wondering, would it be appropriate to bring in Communication scores from that pre-existing validated instrument, along with the Communication scores from this new instrument, and correlate those using the Pearson product moment correlation? Is there anything else I'm missing?

Any and all help is greatly appreciated!

Darko

 

[hlsmith]

@spunky

 

[spunky]

Hi all,

I'm fairly new to R and have been working in psychometrics for the past year or so.

I just had a quick question about using the Pearson product moment correlation.

I've developed a questionnaire that attempts to answer the question: can we evaluate soft skills based on online activity? One of those skills, for example, is communication. That variable is assigned a value of 0 (poor communication) to 5 (excellent communication), so I'm really dealing with ordinal data. What I'm looking to do is to establish concurrent validity by comparing my results with a pre-existing validated tool that measures the same domains.

I'm wondering, would it be appropriate to bring in Communication scores from that pre-existing validated instrument, along with the Communication scores from this new instrument, and correlate those using the Pearson product moment correlation? Is there anything else I'm missing?

Any and all help is greatly appreciated!

Darko

To address this kind of questions you'd need to run a full Structural Equation Model (SEM) and see the correlation of the latent Communication factors between both scales.

 

[DarkoG]

Thank you for the guidance @spunky!

What would be the next step after that? (Apologies is that would require a lengthier answer). Would you be able to point me to resources that might be able to guide me through the process?

Thanks again!