chapter 5 Flashcards
there must be evidence that it has validity
for the Particular situation_ to which it is being applied.
__valid measure
It is necessary to show some proof that a test is valid for the purpose for which it is being used.
Validity
can be defined as the _agreement__ between a test score or measure and the quality it is believed to measure.
_Face__ Validity
The test appears to measure what it says it’s measuring.
Three primary types of evidence
Construct-related validity
Criterion-related validity
Content-related validity
Construct __underrepresentation
not all important aspects of the construct are sampled;
Construct-irrelevant variance
scores are influenced by factors irrelevant to the construct ..such as the effect of anxiety or reading problems on test-taking
particular criterion
the correlation between a test and a well-defined criterion measure
predictive_validity test_
Using a test to forecast something is actually a type of criterion validity
Strong Interest Inventory
compares people’s interests with those of individuals who are in a variety of occupations and are happy in them.
Validity coefficient.
The relationship between a test and a criterion is usually expressed as a correlation
Hamilton
is a rating of a patient by a professional
Beck_
self-report scale of depression
What does the criterion mean?
The criterion itself must be Reliable and valid.
The best predictor of future
behavior is past behavior!
Consider _differential__ prediction:
predictive relationships may not be the same for all groups.
criterion
itself must be the most accurate measure of the phenomenon being observed
construct_
s something totally abstract, an idea built by thinking about something and trying to come up with a description of a quality we think is “there.
Construct validity evidence
a researcher simultaneously defines some construct and then develops intrumentation__ to measure i
love factors
Attachment
Caring
Intimacy
_convergent evidence
When a measure correlates well with other things thought to be measures of a certain construct, `
divergent validation
At the same time that your test should be correlated with other measures of the construct you are proposing it measures, it also should not correlate with measures of unrelated (or especially opposite) constructs.
criterion-referenced test
is one that wants to predict to a particular criterion
norm reference/ opulation-referenced tests
, where you look not at whether the test predicts a certain criterion but where this score falls in the population being measured
The maximum validity coefficient between two variables is equal to
the __square root_ of the product of the reliability of the test and the reliability of the criterion.
All validity is ultimately construct validity!
it is logically impossible to demonstrate that an unreliable test is valid!