PSY 3100: Ch. 5 - Identifying Good Measurement

conceptual variable

stated/described in abstract/non-specific terms

things researchers think they are measuring/manipulating

operational variable

variable stated in specific terms

things the researchers know they are measuring/manipulating

Steps to measure a conceptual variable

1.) generate list of properties on CV

2.) Determine behaviors that correspond to those properties

3.) Create measured (operationalized) variable by specifying operational def

Self-report measure

ask ppl to provide info to respond to questions

Observational measure

observe what ppl do when engaged in a task

physiological measure

record biological activity during a given task/while observing behavior

Observational measures

measure directly observable behavior along these/other dimensions (frequency, duration, intensity, speed)

physiological measures

measure of a physiological process related to psychological processes (heart rate, brain activity)

nominal scale

difference in kind only

Categorical, identity

ordinal scale

relative difference in rank/amount (size of diff)

(Identity + magnitude)

Ex.) Likert scale

interval scale

equal-sized difference in rank/amount (measured amount CAN’T be 0)

(Identity + magnitute + equal interval)

Ex.) 60 degrees isn’t twice as much as 30 degrees (30 degrees warmer)

ratio scale

equal-sized difference in rank/amount (measured amount CAN be 0)

(Identity + magnitute + true zero)

Ex.) A is 3 times greater than B

categorical variable

numbers represent differences in category only (nominal)

quantitative variable

numbers represent numerical differences (ordinal, interval, ratio)

constructability

does the operational variable represent the construct?

Subjective

Judgement

Empirical

gather data and observation

convergent validity

extent to which your measure is related to measures of similar constructs

Ex.) similar constructs to RAS: passionate love, commitment, relationship investment

Discriminant validity

degree which measure of construct is distinct from unrelated constructs

Ex.) RAS does not equal to permissiveness and self-disclosure to friends

Criterion validity

extent which your measure is related to a relevant behavioral outcome

Ex.) what types of behaviors should relationship satisfaction be related to? (eye contact, time together, hand holding, NOT breaking up)

content validity

extent which a measure captures all parts of a concept

Ex.) multiple components of relationships: feelings of partner

face validity

extent which an instrument measures what it claims to

Reliability

does the instrument provide consistent measurement?

Consistency

construct validity

Does the operational definition capture the construct of interest?

Accuracy

test-retest reliability

correlation between scores on the same measure administered on two diff occasions

Ex.) measure RSE, wait, measure again

Consistency overtime

inter-rater reliability

to what extent do 2+ observers see the same thing?

Ideally match on an individual level; must ordinally match

Consistency across raters

Internal consistency

extent which items on a scale “hang together” or measure the same construct

consistency across items

jingle fallacy

falsely assuming 2 tasks measure the same construct cuz of the same name

jangle fallacy

assuming 2 tasks measure different constructs cuz it has diff names