Measurement/methodology/other Flashcards
(34 cards)
william wundt
first psychology lab
lewis terman
revised stanford binet test for US
between subjects
each subject is exposed to only one level of each IV (high protein breakfast vs low protein breakfast)
matched subjects
match subjects on variable they want to control (like matching IQ
within subjects
repeated measures design
same subjects get all the levels of IV (day 1 high protein breakfast, day 2 low protein breakfast)
nonequivalent group design
control group is not necessarily similar to experimental group due to not using random assignment
-educational research
experimenter bias
due to bias, the experimenter might inadvertantly treat groups of subjects differently
- used double blinding to fix this
demand characteristics
cues that suggest to a subject what the researcher expects of them
- use deception to fix this
hawthorne effect
effect that being observed has on behavior
- use control groups
measures of variability
standard deviation, range, variance
normal distribution
- 2: 2%
- 1: 14 %
0: 50%
1: 84%
2: 98%
reject the null
when we find statistically significant results
Type 1 error
reject the null, but null is true
we say there are results but there really aren’t any
Type II error
accept the null, but null is false
we say there are no results but there really are
probability of making this error is called beta
ANOVA
measures how much group means differ from each other by comparing between group variance/within group variance using the F ratio
norm referenced
assessing persons performance in terms of how that person did compared to others - “got 90 percentile on GRE”
domain referenced
criterion referenced
test takers knowledge of specific content
drivers license written test
reliability
consistency
use standard error of measurement (SEM), smaller the SEM the more reliable
content validity
test coverage of a particular skill or knowledge are that it is supposed to measure
face validity
whether or not test appears t measure what it is suppose to measure
criterion validity
how well a test can predict someones performance on a test of the same skill or knowledge area
predictive validity - future performance
concurrent - same time
construct validity
how well performance on the test fits into theoretical framework related to what it is you want the test to measure
convergent and discriminant validity
convergent validity
if two variables are highly related , people who score high on you test should also score high on the other test
discriminant validity
test performance not correlated with performance on a test measuring a theoretically unrelated variable
