Midterm Flashcards
(19 cards)
Correlation strength:
0-.3 - weak
.3-.6 - medium
.7-1.0 - strong
What is the coefficient of determination?
aka shared variance - the sqrt of the correlation coefficient of two variables - a statement about the factors that underlie the variables and account for their relationship…
Content validity…
…is the content of the test valid for the kind of test it is?
Criterion-related validity
- concurrent validity - does the instrument relate to another criterion now
- predictive validity - does the instrument relate to another criterion in the future?
Standard error of estimate:
… using a known value of one variable to predict a potential range of scores on a different variable
Construct validity:
The extent to which an instrument measures a hypothetical construct
Methods of gathering evidence for construct validity:
- experimental design
- factor analysis
- convergence with other instruments (convergent validity)
- discrimination with other instruments (discriminant validity)
Three ways of determining reliability:
- test-retest
- alternate, parallel, or equivalent forms
- internal consistency (alpha coefficient, Kuder-Richardson with tests that have right/wrong answers, split-half or odd even0
Internal consistency reliability
- how do individual items relate to each other and the test as a whole? Three types: - split half/odd even - Cronbach's alpha - Kuder-Richardson
Item Response Theory
Another way of looking at reliability:
- looks at the amount of error in the entire test
- each item is assessed for its ability to measure the trait being examined
- each item is examined for its ability to discriminate based on the trait being measured
Types of derived scores:
- percentiles
- standard scores
- developmental norms
Types of standard scores:
- t-scores
- z-scores
- deviation IQs
- stanines
- sten scores
- National Curve Equivalents
- SAT, GRE, etc.
z-scores run from … to …
-4.0 to 4.0
Formula to find a z-score:
(X - M) / SD
Formula for standard error of measurement:
SEM = SD*sqrt(1-r)
Formula for standard error of estimate:
SEest = SDy * sqrt(1-r^2)
4 approaches to the construction of personality tests:
- logical/content validation
- Empirical Criterion Keying
- factor analysis
- construct validity (combines all of the above)
Pros and cons of content validation:
pro: good face validity with test takers
con: easy to fake good or bad scores
Empirical Keying procedure:
- administer items to a “criterion” and control group
- select items that distinguish between the groups, regardless of the item content
