factor causes effects

treatment changes come before outcome changes

changes not due to extraneous variables 

through use of experiments

construct validity

degree to which study measures or manipulates what it intends to

1. What does the treatment really manipulate?
2. What does the measure really measure?
3. Is their behavior genuine of merely an act?

Threats: poor measures, poor manipulation, experimenter or participant bias

degree to which results can be generalized to different participants, settings, times

1. Generalized to other settings?
2. Generalized to other participants?

Improve: replicate results with different population

1. Potential harm minimized?
1. informed consent
2. debriefing
2. Potential benefits maximized?
1. benefits to humanity
2. validity
3. Permission to conduct research?
1. Consult
2. IRB

Institutional Review Board: committee of at least 5 members (1 nonscientist) who review and monitor proposed research

violates federal law to proceed without IRB approval

8 threats to internal validity

(Campbell & Stanley)

1. Selection- preexisting group differences
2. Selection by maturation interaction- groups changed at different rates
3. Regression to the mean
4. Mortality
5. Maturation
6. History
7. Testing
8. Instrumentation

1. Selection by maturation
2. regression to the mean

1. Self selection- participants choose group, fundamentally different
2. Arbitrary selection- groups still fundamentally different on some variable

1. Idential participants dont exist
2. Cannot match participants on every variable
3. Cannot match partpnts on all relevant variables because we dont know them
4. Pretest score screening can be misleading

1. Maturation
2. History
3. Testing effects

1. Instrumentation
2. Measurement Error
3. Mortality