Hi everyone,
I have one construct, speaking proficiency, for example, and five aspects of vocabulary, grammar, pronunciation, fluency and communicative ability to measure my construct (speaking proficiency). As far as I know convergent validity which is a sub-type of construct validity is used by taking two measures to measure the construct (speaking proficiency) of respondents. I kindly want to know if I can be using those five measures instead of two measures only to measure speaking proficiency by using convergent validity in SPSS. I want to know if those five are interrelated so that I would be able to claim convergent validity for my instrument.
Thank you very much in advance.
Karwan,
Dear Karwan,
Your question is very interesting. The most efficient way to analyze your construct is to test a structural model with the different factors you point out. You can test how all these measures (vocabulary, grammar, pronunciation, fluency and communicative ability) as related to the construct "proficiency". You can do this by means of using structural equation modeling with different software (LISREL, AMOS, EQS, MPlus, etc.). This approach is more efficient than a only correlational ones by SPSS.
I hope it could help you.
Best.
Prof. Albert Sesé
I concur with my colleagues, but want to add a caution that in order to compare different measures, you need to make sure that the respective scales of measurement are compatible with one another. Likewise, consider whether the scores yielded by the various measures have similar psychometric properties. It is not advisable to compare standard scores with t scores , for example. If the scores used for the various measures feature different psychometric properties, consider standardizing them.
- Badges
- Science topic