How to Validate a Survey or Questionnaire

A survey that hasn't been validated is a serious weakness in a proposal — committees want evidence that your instrument actually measures what you claim it measures, not just that the questions look reasonable. Here's what validity actually requires.

Use an existing validated instrument if one exists

Before building your own, search for an existing instrument that's already been validated for a population and construct similar to yours. Using an established tool saves you the validation process entirely and gives your study more credibility — cite the original validation study and confirm the instrument is appropriate for your specific population (validity doesn't always transfer perfectly across cultures, age groups, or clinical contexts).

If you need to develop your own instrument

Content validity

Does the instrument cover the full construct you're measuring, without irrelevant items? Established through a literature review informing item generation, followed by expert panel review — typically reported as a Content Validity Index (CVI) based on expert ratings of each item's relevance.

Construct validity

Does the instrument actually measure the underlying concept it claims to? Usually established through factor analysis (confirming items group together as theoretically expected) or by testing hypothesized relationships with other measures.

Criterion validity

Does the instrument correlate with an established "gold standard" measure of the same construct, either at the same time (concurrent validity) or predicting a future outcome (predictive validity)?

Reliability is a separate question

Validity (measuring the right thing) and reliability (measuring it consistently) are different properties — an instrument can be reliable without being valid. See our guide to Cronbach's alpha and reliability testing for the reliability side of this.

Pilot testing

Before full data collection, pilot the instrument with a small sample from your target population to check for confusing wording, completion time, and any items that don't perform as expected — this is expected practice, not an optional extra step.