deck_4832830 Flashcards
Measurement Scales
Nominal- name (gender, race, political affiliation)Ordinal- rank order, intervals not equal (level of education, likert scale)Interval- equal intervals, no zero (temperature, IQ)Ratio- meaningful zero (height, weight)
Why do counselors need to know about assessment?
-Basic understanding of clients’ reports (from formal assessment contexts mentioned earlier)-Constructs measured by the tests used-How to understand standard scores-General reasonableness of interpretations and recommendations-Ethical issues -What types of testing could I refer my client for? -Why you might want to use psychological tests as part of overall assessment-Clinicians (like everyone) have biases! (example of opioid abuse)-Quick way to gather information-A different way to reach the client (written/tactile vs. all verbal)-Assessment can be therapeutic (more later)-The current trend … document! Assess!DSM-5
What is the difference between criterion-referenced and norm-referenced tests
Norm-referencedScore is based on comparison to othersIQ TestCriterion-referencedScore is compared to an established standard (criterion), not others’ scores Non-Curved math test, SAT
mean, median, mode – what they are; when each might be more or less appropriate as a measure of central tendency
mean is a true average (quick & easy), median may be better when looking for a more typical answer; not as affected by extremes in either direction (i.e. income, home price). Mode helpful in determining most common outcome
standard deviation (what does it represent?)
average deviation from the mean. In normal curve, 68% will be w/in 1 SD, 95% w/in 2 SD’s
areas under the normal curve
1 SD- 68%2 SD- 95%3 SD- 99%
what is a percentile score; be able to roughly translate these into z or T scores
% of people in the norming group who had a score at or below your scoreImagine Normal curve, can be split into SD’s (-3 - +3even), Z Scores (-4 - + 4, even), or T Scores (10-90 even). Plot percentile score on curve (50 is median), then figure out what T or Z score would be
shapes of curves … what types of situations/data would lead to a positively skewed distribution, negatively skewed distribution, bimodal, normal
Bimodal- 2 populations, a missing variable (SES on life expectancy curve)Positive Skew- bottom heavy; a few scores trail off to rt (i.e. income)Negative Skew- top heavy; a few scores trail off to lt (easy test)
know how positively and negatively skewed curves affect mean, median, and mode
mode will always be high point on curve, but median and mean will be dragged towards the tail (mean gets pulled further than median because of extreme values)
Reliability: what it is; ways to assess; when each type of reliability is/is not appropriate
-Consistency/dependability of a measurement procedure or test-necessary but not sufficient for validity-Types: test-retest, internal consistency, parallel forms (compare to alternate version of test), interrater reliability
Test scores include “truth” + error – what are common sources of error in test scores?
test taker (transient states)Context (Modivation, situation, recording errors)Test- (ambiguous questions, long = fatigue effects)
What is the relation of true variance, error variance, and reliability?
True variance is the amount of variance that occurs naturally, error variance is due to error (context, test-taker error, test error). A reliability coefficient (closer to 1 is better) estimates the variance between the true score and the observed scoreTrue Variance/(True + Error Variance) = Reliability Coefficient
What the standard error of measurement actually represents.
a type of SD that measures variation in an individual, not a group
Be able to explain what a confidence interval is – how you interpret it
level of confidence that test score = true abilityTrue score = observed score + error, so we can be 90% confident based on test reliability score, mean and SD that TS is within X # of points from the observed score.
What is meant by validity?
are we measuring the construct that we are intending to measure? Reliability is Neccessary but not Sufficient for Validity
