Validity

Question 1

What are the 4 types of validity?

Answer 1

1. Statistical conclusion validity
2. Internal validity
3. Construct validity
4. External validity

Question 2

What are the 8 possible threats to statistical validity?

Answer 2

1. Low statistical power.
2. Violated assumptions of the test statistics (for example the lack of independence)
3. Fishing through the data
4. Unreliability of measures. Higher noise in the measures attenuates bivariate relationship, higher samples size to detect the effect
5. Restriction of range; floor and ceiling effects attenuate bivariate relationships
6. Unreliability of treatment implementation
7. Extraneous variance of settings
8. Hetereogeneity of units: it increases variance so to obscure covariation

Question 3

What are the 8 possible threats to internal validity?

Answer 3

1. Ambiguous temporal precedence (reverse causation and bi-directionality)
2. Selection of units (omitted-variable bias)
3. History (external events affect the outcome independently of the treatment)
4. Maturation (outcome change because of time)
5. Regression artefact (regression to the mean). If individuals are selected
   because their extreme scores (high or low) there is tendency for them to
   score extremely less on other measures
6. Attrition (experimental mortality), it is a special subset of selection bias
   after the treatment, not control by random assignment
7. Testing (taking a test once may influence scores when a test is taken again)
   Instrumentation (a stock exchange index; how answers are reported in face
   to face interviews)