Please enable JavaScript.
Coggle requires JavaScript to display documents.
Chapter 7: Scale Reliability & Validity - Coggle Diagram
Chapter 7: Scale Reliability & Validity
Realiability
Consistency or dependability of a measure
Same construct should give similar results
Reliability does not guarantee accuracy
A Measure can be reliable but invalid
Sources of Unreliability
Observer/researcher subjectivity
Imprecise or ambiguous questions
questions respondents cannot answer well
clear, simple wording improves consistency
Reliability test
Inter-rater
agreement across observers
test-retest
stability across time
split half
consisency between halves
internal consistency
Cronbach alpha
Validity
Does the measure represent the constuct?
Often called construct validity
Needs theoretical + empirical assesment
Reliable and valid measures are both needed
Translational Validity
Face validity
makes sense on its face
Content validity
covers the full domain
expert judges can assess items fit
q-sort supports theoretical assessment
Criterion-related Validity
Convergent
agree with the same construct
Discriminant
differs from other constructs
Concurrent
relates to current criterion
Predictive
forecasts future criterion
Measure Error
Observe score = true score + error
Random error adds variability/noise
Systematic error creates directional bias
Random hurts reliability, systematic hurts validity
Integrated Validation
Conceptualize constructs and dimensions
Create/refine items with judges
Pilot test + correlations/factor analysis
check reliability and predictive relationships