# How to calculate validity of questionnaire in SPSS?

I have 3 types of questionnaires,

1. A 15 item scale with each item on a 7-point Likert type scale.
2. Six subscales each containing different number of items
3. 21 subscales each containing 4 items.

I have collected a sample of size 100.

• How can I use this sample to determine the validity of these questionnaires?
• How can this be done in SPSS?
-
(Psychometric) questionnaire validation is a complex area, discussed in a large literature and not really a well-defined statistical question that could be addressed by some SPSS procedure. What do you want to do with this “validity”? Where do the questionnaires come from? Standard scales? Previous research? Ad hoc scales you created yourself? –  Gala Jun 6 '13 at 10:09
Also a brief point of terminology, to help you understand any material you might read on the topic: A question or statement in a questionnaire (but not a group of questions) is what is called an “item”. Strictly speaking, a scale or subscale is a group of items selected to measure a particular construct but there is a lot of confusion in the usage of the term “Likert scale” (what Rensis Likert did was indeed combine several items, sometimes with different formats but what many people call a “Likert scale” is in fact an item, not a scale). –  Gala Jun 6 '13 at 10:14
@Gael Laurans. thanks for your comments. actually those questionnaires are standard scales. they are called as instrument. after conducting the pilot survey, now i am interested to check whether they are valid or not. how can i check their validity? –  Ina Jun 6 '13 at 11:30
If those are standard scales, given in particular the sample size, I would say that it does not make sense to try to run sophisticated analyses to assess their validity. There are certainly no simple procedure that could give you a number characterizing their “validity” or a binary OK/not OK decision. Just go on with your research, using them as intended. –  Gala Jun 6 '13 at 11:57

As Gaël mentions, assessing scale validity is a large and complex topic.

Here are just some basic suggestions to get you started:

• Check the factor structure of the test to evaluate whether items load most on the theorised scales. You could start with exploratory factor analysis and then later on build up to confirmatory factor analysis.
• Assess the reliability of the test given that reliability is necessary but not sufficient for validity. I.e., measure internal consistency reliability and test-retest reliabilility.
• Correlate your scales with a wide range of other variables.
• Do your measures correlate with other existing measures of the same construct?
• Do your measures not correlate or correlate to a lesser extent with things that theoretically should not correlate with the variable?
• Do your measures predict theoretically relevant and important variables?
• Do your measures correlate with alternative ways of measuring the variable (e.g., other report, behavioural measures, etc.)?
• Assess whether experts in the domain consider the items to accurately and adequately reflect the domain.

Acquiring this information typically takes multiple studies.

Furthermore, you may find it useful to think of validity as a property of the application of a test to the formation of a conclusion, rather than being a property of the test per se. Thus, you may want to reframe your question as one of saying whether the conclusions you want to make from the test are valid.

-
....thanks.....i wl try to go through this.... –  Ina Jun 6 '13 at 17:30