I am conducting a survey using self developed questionnaire. I have established its face validity and content validity. Some expert reviewer has suggested I must establish the validity by comparing it with a "GOLD STANDARD" measure. I am unable to find any single gold standard measure to compare with as my questionnaire is a mix of various aspects and they are usually measured using different questionnaires. Which is the minimum level of validity and reliability which I should check for before using it to collect data? What can I do to validate my survey questionnaire?
Vivek Ramanandi First, the questionnaire is presented to the arbitrators in the field of specialization. Second, the validity and reliability tests are conducted using the SPSS.
I believe that you should present your specialist instrument. Then do a pilot study to verify that a sample of individuals understands the items on your instrument well.
If most experts claim that your instrument has good items for measuring variables, and most individuals in the pilot study understand the content of the items well, you should apply your questionnaire to a significantly large sample.
With the data collected and tabulated, you should start using traditional measures of validity / reliability, like the alpha of Conbrach and Omega of McDonald.
Subsequently, you must verify the dimensionality of the instrument using methods of polychoric correlation, parallel analysis and exploratory factor analysis. All of these methods have indexes that, when reached, indicate the instrument's validity.
Q. "What can I do to validate my survey questionnaire?"
Jonhatan's advice is great!
I would add the following:
a. Assure that the collection of the pilot study surveys is truly anonymous to protect the identify of each and all responders. Do NOT have anyone connected to you or the survey be the drop-off point.
b. Structure the pilot audiences to include not only your prime recipients but their internal and external suppliers and clients/customers as well.
c. Once you have done the""Official Version" be certain to prepare and release a summary of the significant feedback - without analysis - to all responders within not more than 21 days.
Finally, recall that people commit strongest to those reports they help shape, not edit.
Stay Healthy!
Cheers,
Bill
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Probability
- Probability Theory
- Reliability