2: 4 Vailidites

Question 1

Construct validity

Answer 1

-adequacy of operational definition
-does OD tap into conceptual variable being studied
-evaluate quality of operational definition
-abstract variables must infer from objective measurements.
Important for all

Question 2

Statistical validity

Answer 2

Do the numbers support claim
-are appropriate stats used
-margin of error
-what is the strength of an association
-is conclusion stat. Significant. 
Type 1 error: false positive; low probability that finding is due only to chance
Type 2 error: false negative; null result was a miss 
Important for all

Question 3

External validity

Answer 3

Genealizability to other people/contexts
-threatens by anything that may make data unrepresentative
-threats to generalizing beyond small: biased sampling, self-selection/volunteer bias
-threats to generalizing beyond the context: subject reactivity, highly controlled lab setting, unique measures or manipulating
More important when generalizability is emphasized(frequency or association)

Question 4

Internal validity

Answer 4

Causal relationship can be inferred
-no possible alternative causal explanations
ESSENTIAL FOR CAUSAL CLAIMS
-degree to which ensures only IV manipulation is responsible for changes in DV
-control EV: eliminates as alt factors(confounds), CV threaten internal validity

Question 5

Interrogating frequency claims

Answer 5

Construct: how well was variable measured

Stat: how large sample size and margin of error?, confidence true value of a proportion or avg lies within range

External ESSENTIAL: generalize from sample to intended population? Sample representative of population? How was it acquired?

Question 6

Interrogating association claims

Answer 6

Construct: how well was each variable measured?

Stat: how strong is association? Is stat. Significant? Avoid type 1 and 2 errors

External: to what can we generalize? Does apply to other? Consider representativeness of sample and setting.

Question 7

Interrogating causal claims

Answer 7

Construct: how well was IV manipulated? How well was DV measured?

Stat: how big is diff? Is stat. Significant? Avoid Type 1 and 2 errors

External: (many not be prioritized)
To whom or what generalize effect? Does apply to other people or occur in other contexts?
Consider representativeness of sample and setting

Internal ESSENTIAL: required to establish causation
How well have EV been controlled.

Question 8

Trade off between internal and external

Answer 8

-steps to increase internal may decrease external

Increase internal but external will suffer

Question 9

3 criteria for causation

Answer 9

Covariance: variables associated
Temporal precedence: manipulate A before measuring B
Internal validity: eliminated possible alt. Causal explanations by controlling EV