Psychometrics Flashcards
Validity
Does the test fairly and closely show the individual’s true level of function or skill?
Observed Score = _ + _
True Score + Error
Reliability
Repeatability
Are the results repeatable? Does the test yield the same or similar score consistently?
Stability
Test – Retest Reliability
Consider duration of time between pre- & post-test
Not too close together: client may learn from the 1st test
Too far apart: maturation (children will increase in skill over time; ex: 6 month gap is bad for Developmental instruments
Equivalence
Alternate/Parallel forms
Internal consistency
degree of relation/agreement between items that measure a single concept or skill
Agreement between items on a test of reading comprehension; agreement between items on a test of QOL; extremely high reliability may indicate redundancy in this case
split half is a special case of internal consistency
>is 1st half of test similar to 2nd half?
split tool into two equal, shorter forms; also called odd-even reliability
Scorer reliability
Interrater (multile Scorers)
Intrarater (1 examiner)
TYPES OF VALIDITY
Face validity (most primitive)
Content validity
Criterion related validity
Construct validity
Ecological
Factorial
Content validity
how well the tool represents/captures all aspects of the phenomenon being measured
ex: cover all aspects of Fine Motor skills?
>grasp, manipulation, shifting
degree of accurate representation of the content
Authors usually develop a test blueprint, i.e. break down gross motor into components of mobility, transfers, balance, endurance
Gold-Standard
tool with well-established psychometrics
ex: use Pea Body tool to help make new tool
Criterion related validity
extent to which one measure is related to a similar measure; scores from one test are correlated sith another
GRE scores with college GPA
ILS with KELS
Types
predictive validity (FIM)
concurrent validity: did client perform similarly across two diff tools?
Construct validity
relationship of client performance to theory
Factorial validity
factor analysis: helps validate theory, constructs
how well factors cluster to a central idea/construct
stablishes evidence of correlation with other measures of a complex behavior; may need to compare the tool with several others
Ecological validity
similar to predictive
how well client’s test performance correlates to real-life performance
Sensitivity
proportion of client’s who have condition & test positive
Positively ID for clients who have condition
