Kyriazos et al (Construct validity-MTMM) Flashcards
1
Q
Construct validity
A
How well a test measures the intended construct. Ongoing process.
2
Q
Multitrait-Multimethod (MTMM) Matrix
A
Method to assessconstruct validityby evaluating convergentand discriminant validityusing multiple traits and methods.
3
Q
MTMM evaluates the validity of ____ using different ___.
A
traits, methods
4
Q
Traits def
A
Hypothetical constructs (e.g. cognitive abilities).
Can be assessed with methods like self-reports or interviews.
5
Q
Four sets of correlations in MTMM
A
- Monotrait-monomethod
- Monotrait-heteromethod
- Heterotrait-monomethod
- Heterotrait-heteromethod
6
Q
Campbell and Fiske’s 4 rulesfor evaluating MTMM matrices
A
- MONOtrait-HETEROmethod correlationsshould be statistically significant and strong.
- HETEROtrait-HETEROmethod correlations should be weak or similar to HETEROrotrait-MONOmethod correlations.
- HETEROtrait-MONO method correlations should be smaller than MONOotrait-HETEROmethod correlations.
- The pattern of trait intercorrelations should be CONSISTENT across ALL blocks to ensurediscriminant validity.
=> I.e. The way DIFFERENT traits (like anxiety, self-esteem, depression) are related to each other should stayCONSISTENT whether you’re using the same method or different methods.
-> This helps confirmdiscriminant validity, meaning the test isn’t mixing up or confusing different traits with each other.
7
Q
Trochim’s 5 guidelines
A
- HIGHEST correlations should be with same trait (mono-mono).
- Validity diagonal should be positive and strong. Shows convergent validity
- Validity coefficients should be the biggest in their block -> In each group of numbers (method block), the diagonal number (comparing the same trait across methods) should be the biggest. Shows that the trait is being measured accurately.
- Off-diagonal elements should be smaller. Proves that the test isn’t confusing different traits— key for discriminant validity.
- Consistent relationships between traits
8
Q
Validity diagonal
A
Same trait measured with different methods