1/44
Looks like no tags are added yet.
categorical and numerical correlation
t-test (2 levels), anova (3+ levels)
categorical and categorical correlation
chi square
construct validity
are variables good measure of construct?
external validity
are results generalizable?
statistical validity
are the stats accurate and reasonable?
moderating variable
changes relationship between other two variables depending on its level
confidence intervals
used to test statistical significance, if it doesn’t contain null hypothesis (usually 0), it is statistically significant
autocorrelation
correlation of one variable over time
crosslog correlation
correlation between earlier measure of one variable and later measure of second variable, helps establish temporal precedence
cross sectional correlation
correlation between two different variables measured at the same time
multiple regression
determines relationship between two variables while controlling for others
criterion variable
variable being predicted, dependent variable
predictor variable
explains variance in criterion variable, independent variable
β0
y intercept of multiple regression model
β1
slope of multiple regression model, slope while keeping other variables constant
mediator variable
intermediate variable that helps explain relationship between two variables
mediator vs 3rd variable
mediator: holds interest to the researcher, first variable causes this which causes second variable
3rd variable: both other variables are independently related to 3rd variable
onegroup pretest/posttest
bad design
no true independent variable, 6 major threats as a result
history threat
event occurs during study that leads to change in participants
maturation threat
natural development of participants over time that happens between tests
regression threat
natural regression to the mean
attrition threat
participants drop out over time
testing threat
practice threat
instrumentation threat
what or who is taking measurement changes over time, natural wear of materials
observer bias
researchers’ expectations influence how they interpret results, fixed by double blind
demand characteristics
bias created when participants figure out experiment, fixed by double blind
null effect
IV did not cause difference in DV
can be caused by not enough variance between-group difference or by too much within-group variance
ceiling effect
task is made to easy
floor effect
task is made too difficult
measurement error
over or under estimate of true value
situation noise
external distraction that could cause variability
power
likelihood that experiment will show statistical significance effect when IV has effect on DV
systematic variability
levels of confounding variable coincide with levels of experiment
design confound
experiment changes two variables at the same time, making it impossible to know what is causing outcome
can fix this with control variables
selection effect
participants in one condition systematically different that participants in second condition
result of non random sample
matched pairs
participants from each condition matched with each other based on similarities
independent groups
between subjects
within groups
within subjects, repeated measures
post test only
participants tested on DV only once
between subjects design
pretest/posttest
participants tested on dependent variable twice, before and after experiment
between subjects design
extra sure of random assignment
order effects
being exposed to one conditions affects response to second condition
practice, fatigue, carryover (caffeine)
counterbalancing
presents conditions to participants in difference sequences
helps fix order effects
latin square
every condition appears in each possible position during counterbalancing
concurrent measures design
participants exposed to both conditions at the same time
within subjects design
manipulation checks
add in question of extra DV to confirm experiment creates correct condition