Assessment Midterm

from study guide

Mean

weighted balance score, arithmetic average

standard deviation

how far each score is “on average” from the mean

z score

finding the number of standard deviations a score is from the mean. a way of standardizing the deviation.

standard scores

a transformed score with a known mean and standard deviation (like a z score, to all whole numbers with positive integers)

correlation coefficient

measures the strength and direction of the linear relationship between variables.

psychological construct

a concept that’s not able to be looked at directly. Things that are inferred, a theoretical thing we can define (introversion/behavior)

operational definition

the thing that you do to measure the construct

operationalism

the philosophical position that a construct is defined by the procedures used to establish it. “you can’t let theoretical definition get too far ahead of measurement tool".”

reliability

the consistency of measurement

retest reliability

correlation of the same test given twice. the more time between tests the less test/retest reliability

alternate form reliability

correlation between 2 versions of the same test

internal consistency coefficient

estimating reliability of the sum of many part scores from the correlations among the part scores. Cronbach’s Alpha is an example of

true score

a hypothetical average of many measurements using a particular test with no carryover effects (reliability)

construct score

platonic true score, a person’s true level on a latent construct, independent of measurement

observed score

true score plus the measurement error

measurement error

deviations from the True Score. the difference between an observed score and the true score of a quantity being measured. 

split-half reliability

correlation between two halves of tests

carryover effects

practice effects and fatigue effects

construct validity

the totality of evidence supporting the use of a test

how many forms of validity? name them!

9! construct, content, test, face, concurrent, convergent, predictive, incremental, discriminate

face validity

just by inspecting the items of a measure, do they seem like measures of the intended construct?

convergent validity

does the test correlate with the variables it’s predicted to correlate with?

discriminate validity

is the test uncorrelated with variables it should not correlate with?

concurrent validity

two things trying to measure the same thing and it’s correlation. Matches results of previous tests.

predictive validity

one thing meant to predict the other. Guess the future.

incremental validity

multiple predictors predicting multiple outcomes. Does each predictor give us new information that the other one doesn’t?

statistical bias

systematic error applied to test validity. consistently off in a specific direction/wrong in a particular direction. never cancels out, compounds with each measurement. test bias does not equal test unfairness. 

central tendency bias

never giving extreme answers to anything. often seen in sophisticated adults. soften extreme response options (almost never vs never)

jingle fallacy

when things sound the same but measure different things

jangle fallacy

when things are measuring the same thing but use different terms

bloated specifics

highly similar items correlate well with each other but are mostly redundant (increases reliability but is redundant)

examples of bad test items that seem obvious?

double barreled questions. leading questions. negative words, extreme words. opting in/out. presumptuous questions. ranked responses.

response bias

systematically give inaccurate answers

leniency bias/severity bias

excessively forgiving/harsh

social desirability bias

making oneself look good. includes impression management and self-deception.

standardized test

test with known mean and standard deviation