Please enable JavaScript.

Coggle requires JavaScript to display documents.

Lecture 7 - Convergent and discriminant validity (and how to investigate…

- - - - How to perform the method: (1) meta-analysis, where you thereby find the overall/average validity coefficient (correlation) based on many studies.
        
        Demonstrate the variability of validity coefficients
        
        All this enables the investigation of the causes of differences in validity coefficients
        
        Conclusion: You then use the results and the variability to adjust and improve your test
        
        We want as little as possible variability (as small of a SD as possible) in our tests. Little variability = the test is robust.
  - - - Example where we use nomological network: We look at the Big 5 personality traits. We want to know if the correlations are consistent with expectations/theory
  - - - Breakdown of what the model's title means: Traits = latent variables, methods = raters(self, interview, teacher, parent) or a type of questionnaire (EPQ and NEO-PI-R), multi = many latent variables (extraversion and conscientiousness), and many methods/ways of measuring
- - - - Example of conclusion in relation to validity:
        
        "There is weak support for convergent validity: .50 and .52 (we would ideally like these values to be higher in order to see a strong support for convergent validity)
        
        "There is no support for discriminant validity: .43 and .48 (for there to be support for discriminant validity we would need the values to be way lower)