miscellaneous Flashcards
interactions
B lines are far apart= main effect B
average of lines not horizontal= main effect of A
lines parallel= no interaction
ordinal= no crisis cross
disordinal= criss cross
same averages mask the main effect, antagonistic
factorial designs
2 or more independent variables, AND each variable has 2 or more levels
within subject
all subjects go through all conditions; requires the least number of subjects
between subject
specific subjects go through only one treatment condition, requires the most number of subjects
mixed (split-plot)
any given subject receives only two or three of six treatment conditions, depending on which factor on which factor is within and which factor is between, requires a moderate number of subjects
nested
levels of factor A found under different levels of a factor B are not the same, more econo,ical, but some interactions cannot be evaluated
inductive
theory building
deductive
theory testing
experimental design: goal, variables, issues
goal: causal relationships
variables:
independent: manipulated by experimenter
dependent: measured by te experimenter
issues:
internal validity: confounded variables
external validity: generalizability, ecological validity
correlational design: goal, variables, issues
goal: correlational relationships
variables:
predictor variables: used to make a prediction
criterion variables: predicted variables
issues:
directionality problem: A–>B, B–>A, or A<–>B
third variable problem: not AxB, but AxC
mediator variable
independent variable: action
mediator variable: consequence of independent variable
dependent variable: consequence of mediator variable
moderator variables
situational/contextual factors: when _____ is _____
individual/instance-specific factors: if ______
main effects
2 way: A, B
3 way: A, B, C
4 way: A, B, C, D
two way interactions
2 way: AxB
3 way: AxB, BxC, AxC
4 way: AxB, BxC, CxD, AxC, AxD, BxD
three way interactions
3 way: AxBxC
4 way: AxBxC, BxCxD, AxBxD, AxCxD
four way interactions
AxBxCxD
prospective ex-post factor designs
cause–> effect
retrospective ex-post factor designs
effect–> cause
counterbalancing
same result for both variables when placed first means no effect of treatment
different results for both variables when placed first means an effect of treatment
time-series design: basic time series with intervention goal
compare before and after treatment, either by cell or collapsing
time series design: interrupted time series design
similar idea to basic with intervention goal, but with naturally occurring events
pretest post-test design
pretest/post-test is the within factor, treatment/ no treatment is the between factor
group 1 (experimental): pre-test, treatment, post-test
group 2 (control): pre-test, post-test
soloman four group design
group 1: pre-test, treatment, post-test
group 2: pre-test, post-test
group 3: treatment, post-test
group 4: post-test
variations
no treatment (A) - treatment (B) - no treatment (A)
or
give the treatment early, then cancel it to see the effects: treatment (B) - no treatment (A)
or
baseline. on __ (naturally occurring) or treatment (experimental induction) is introduced. observations or measures after that point are influenced: no treatment (A) - treatment (B)
