Personnel Psychometrics

Construct

Theoretical, cannot be directly measured

Trait

specific type of descriptive label, more stable and consistent over time

label

using numbers to keep track of individuals or things (the numbers mean nothing)

nominal

using numbers to categorize groups

ordinal

using numbers to order from most to least

interval

using numbers to represent how much of a variable is present

ratio

using 0 to represent a lack of a characteristic

Reliability

consistency; time, place, form, or rater

Validity

accuracy; how well does the test measure what it is supposed to?

T/F: Reliability is a necessary but not sufficient condition for validity

True

Validity is measured using ___

R-squared

How to interpret R-squared

if .64, 64% of variance in the outcome is explained by the predictor

How to interpret r

correlation coefficient (measures strength and direction of linear relationship)

Content validity

coverage of the construct; are all relevant aspects included?

Construct validity

theoretical soundness; does it measure the intended psychological idea?

criterion-related validity

correlation with outcomes; does it predict or relate to real-world outcomes

face validity

surface-level appearance; does it looks like it measures what it should

Reliability is measured by the

reliability coefficient; tells you the consistency of a test or tool

how to interpret reliability coefficient

ranges from 0 to 1; ideally .8-.9; <.7 is unacceptable

test-retest reliability

stability over time; ex: same test, 2 weeks apart

inter-rater reliability

agreement between raters; ex: 2 judges scoring a speech

internal consistency reliability

consistency within the test; ex: cronback’s alpha for survey items

parallel-forms reliability

equivalence of test versions; ex: two forms of a test

split-half reliability

consistency between halves; ex: odd vs even items on a quiz

convergent construct validity

correlation with similar constructs; ex: anxiety test correlates with other anxiety tests

discriminant construct validity

lack of correlation with unrelated constructs; ex: anxiety test does not correlate with creativity test

predictive criterion validity

predicts future outcomes; ex: SAT predicts college outcomes

concurrent criterion validity

matches current outcomes; ex: new test matches clinical diagnosis

postdictive criterion validity

correlates with past outcomes; ex: test explains past job performance

incremental validity

adds predictive value (new test improved prediction beyond old test)

ecological validity

adds real world applicability (lab test reflects real-life behavior)