construct bias Flashcards
what is test bias?
arises when there is a factor inherent in a test that systematically prevents accurate, impartial measurment
what is construct bias?
occurs when a test has different meaning for two groups in terms of the construct it measures
what does construct bias concern?
the relationship of observed scores to true scores on a test
when would we conclude there is bias on a test?
if the relationship is systematically different for different groups
what type of situations does construct bias cause?
two groups having the same average true score but different averages on observed scores
when does predictive bias occur?
when a tests use has different implications for two or more groups
how do you internally evaluate a test?
it is bias if people belonging to different groups respond in different ways to the item and if it could be shown that these responses were not related to group differences associated with the psychological construct
what is a discrimination index?
reflects the degree to which the item is related to the total test score
what does a high discrimination index show?
that people who answer an item right tend to do better on a test as a whole than people who answer it wrong
what does a strong discrimination index show?
an item is highly similar conceptually to most other test items
what is factor analysis?
used to evaluate the internal structure of test separately for different groups
what would it mean if a factor analysis revealed that the structure was unidimensional?
the test doesnt suffer from construct bias
what would it mean if a factor analysis revealed that the structure was multidimensional?
the test has different internal structure for the two groups and does have construct bias
what is differential item functioning analysis?
an approach to the assessment of construct developed within the context of item response theory
what does DIF assume?
that we can estimate peoples trait levels directly from test data
what does DIF mean for traits?
that they are in essence, participants true scores for the construct being measured
if the trait levels and item responses for the groups don’t match up doing DIF what does this mean?
theres no evidence for construct bias
what is rank order consistency?
calculate the means associated with each item separately for each group then caculate the spearman rank correlation between the item means
if the correlation is low for a rank order what does this mean? (less than .80)
there is evidence to suggest construct bias
how can predictive bias be explained?
do the scores from a particular test predict a criterion with equal accuracy for two or more groups?
what is step one of predictive bias anayses?
determine whether the test scores predict the DV to start with
what is step two of predictive bias analyses?
determine whether the test scores predict the DV equally enough across groups
whats bivariate regression?
using one continously measured variable (X) to predict another (Y)
what is intercept?
the expected value of Y when X is 0
what is slope?
the expected increase or decrease in the DV as a function of an increase or decrease in the IV
what will the results of a regression be if there is no bias?
it will be equally applicable to different groups
what is the common regression equation?
the regression equation derived from the total sample
what does it mean if the group level estimates dont match the common regression equation?
test is biased
what is outcome score bias?
the supervisor who provides the ratings is biased in favour of one group
what if there is differences in reliability?
this is a form of construct bias that can have an impact on predictive bias
