Assessment and Testing Flashcards
Measurement
process of determining dimensions of an attribute or trait
assessment
processes and procedures for collecting info about human behavior
eg: tests, inventories, interview data, observation, rating scales
appraisal
implies going beyond measurement to making judgments about human attributes and behaviors
- used interchangeably with evaluation
Interpretation
making a statement about the meaning or usefulness of measurement data based on the counselor’s knowledge or judgment
measures of central tendency
distribution of scores can be measured using:
mean: symbolized by M or X (with horizontal line on top)
median: middle score
mode: most frequent scor
these 3 fall in same place when distribution is symmetrical
skew
(see iphone pic)
refers to the degree to which a distribution of scores is not normally distributed
mode=top curve
median=middle score
mean=pulled in the direction of the extreme scores (which is represented by the tail)
a negative skew is with the tail pointing to left; positive is pointing to right (think of how values increase/decrease on horizontal axis)
Standard Deviation
- describes the variability w/in a distribution of scores
- is essentially the mean of all deviations from the mean
- an excellent measure of dispersion of scores
- Use ‘SD’ to signify standard deviation from a sample
- use sigma (think cursive ‘o’ without the first part) for population variability
variance
the SD squared (SDˆ2)
normal bell curve
distributes scores into 6 equal parts–3 above the mean, 3 below–such that:
34% & 34%=68%, comprises 1 standard deviation
13.5% & 13.5%=95%, comprises 2 standard deviations
2% & 2%=99%, comprises 3 standard deviations
standardized scores
are scores converted from the individual’s raw score that allows for comparison bw individuals and bw the same individual’s various scores (ie vocab and math)
they basically represent the person’s distance from the mean in terms of standard deviation
two most commonly used standardized scores:
z-score: the mean=0, standard deviation =1, rnge for SD is +3/-3 [the ‘z’ =zero]
T score: mean=50, standard deviation=10. Transforming this score eliminates negative numbers (unlike Z score)
[the “T”=Ten]
two most commonly used standardized scores
z-score: the mean=0, standard deviation =1, rnge for SD is +3/-3 [the ‘z’ =zero]
T score: mean=50, standard deviation=10. Transforming this score eliminates negative numbers (unlike Z score)
[the “T”=Ten]
stanine
from STAndard NINE
converts distributioin into 9 parts, with 5 in middle and SD of ~2
correlation coefficient
- measures reliability
- ranges from -1.00 to 1.00 (perfect correlation
-shows the reln’p bw two sets of #s, but nothing about cause and effect - if the reliability coefficient is high (>=.70), then it’s reliable
bivariate vs multivariate
- correlation bw 2 variables=bivariate
- ” bw 3 or more variables=multivariate
reliability
- a necessary psychometric property of tests and measures
- consistency of a test or measure
- the extent to which a measure is free from error (if the instrument has little error, it’s reliable)
stability
- test-retest reliability using same instrument
- 2 weeks is sufficient bw test administrations
Equivalence
- alternate forms of the same test administered to same group
- comparable forms of the tests, intervening events, and experiences will influence reliability
spearman-brown formula
may use this to see how reliable a split half test would be had you not split it in two
other name for the spearman-brown formula?
prophecy formula
internal consistency
- this is a split-half method where the test is divided into halves and the correlation bw these halves is calculated
- determined by measuring inter-item consistency. the more homogenous the items the more reliable the test
what are the different formulas used to determine internal consistency and when are they used?
Kuder-Richardson formula used if the test has dichotomous items (ie true/false, yes/no)
Cronbach alpha coefficient is applied for nondichotomous items (ie multiple choice, essay)
what is used to determine reliability?
correlation coefficient
- if the reliability coefficient is high (>=.70), then it’s reliable
Kuder-Richardson formula
Kuder-Richardson formula used to measure internal consistency; if the test has dichotomous items (ie true/false, yes/no)
Cronbach alpha coefficient
Used to measure internal consistency
- is applied for nondichotomous items (ie multiple choice, essay)
True vs error variance
Coefficient of determination
Coefficient of non determination
If 2 tests given and the correlation bw them is .9 (for example), then the true variance measured in common is .9^2=81%.
coefficient of determination=degree of common variance (81%)
coefficient of nondetermination=the unique variance, not common (19%=error variance)
Standard error of measurement (SEM)
Another measure of reliability helpful in interpreting test scores
- helps determine the range in which a persons score probably falls
- aka “Confidence Band” or “confidence limits”
Ex:
A person scores a 92 on a test, and Sem is 5. On a normal curve, 1SD away will be 97 and 1 below will be 87, which is where his score will be 67% of time. 95% of time his calcite will be be 82 and 102 (2SD’s away from his mean of 92).