Exam 3 study guide Flashcards
interaction effect
occurs when the effect of one IV depends on the level of another IV
2 types of interaction effects
crossover interaction
- “it depends”
- graph makes a X
spreading interaction
- “only when”
- graph is 75 degree angle facing to the left
how many cells in a 2x2 factorial design
4 cells
participant variables
any demographic characteristics, things the experimenter cannot manipulate
when the graph shows two parallel lines, this represents …
no interaction effect
how do we know there is a large effect shown in a 2x2 factorial
calculate the marginal means for each level of the 2 IV’s (should end up with 4 means) and then compare the marginal means in each IV
interaction effect calculation
- find marginal means (a, b, c, d)
- subtract marginal means in each IV (a-b =x, c-d=y)
- find the difference within IV’s (x-y)
when a study shows both a main effect and an interaction, the interaction is almost always…
more important
you can have ____________ main effects and a ________ interaction effect
non-significant; significant
3 factorial variations
independent-groups factorial design
within-groups factorial designs
mixed factorial designs
independent-groups factorial design
both IVs are studied as independent groups aka between subjects
each cell of the 2x2 has a different group of participants
within-groups factorial design
both IVs are manipulated within groups
only one group of participants in all four cells
mixed factorial designs
one IV is manipulated as independent-groups and the other is manipulated within-groups
2 cells share the same group while the other two cells share a different group
the same group cells are for the independent-groups IV
for a 2x2x2 factorial, there would be …
3 main effects
3 two-way interactions
1 three-way interaction
2 ways to identify factorial designs in empirical journal articles
the method section will describe the study
- __x__x__
the results section will examine whether the main effects and interactions were significant
2 ways to identify factorial designs in popular press articles
look for “it depends” or “only when” to highlight interaction
look for participant variables (age, gender, ethnicity, etc.)
bivariate correlations
associations that involve exactly two variables
ex. level of happiness and days spent on vacation
think quantitative or categorical
cohen’s guidelines
.10 or -.10 = small, or week
.30 or -.30 = medium, or moderate
.50 or -.50 = large, or strong
mean
arthmetic average
t test
a statistic to test the difference between two group averages
do you know how to plot points on a graph then determine if the correlation is positive, negative, or neither?
obviously
construct validity: ask about the construct validity of _____ variable
each
effect size
describes the strength of an association
statistical validity
_____ effect sizes give more accurate predictions
larger
the logic of statistical inference with p < 0.05
< 0.05 what you found in the study isn’t found in the real world, type I error
do you know how to read about significance in journal articles?
yes, table interpretation on slide 32 of bivariate correlational research part 1
outlier
small sample more affected by outliers
changes slope of the line, making correlation appear stronger than it actually is
restriction of range impacts …
strength of correlation
the pearson correlation coefficient is used to measure … and not measure …
the strength of a linear association between two variables
curvilinear associations
4 examples of non-linear relationships
exponentials
free time through life
friendliness of waiters
more money doesn’t equal more happiness at a certain point
3 causal criteria for internal variability (review)
covariance
temporal precedence (directionality problem)
internal validity (third-variable problem)
third variable problem/internal validity
is there a third variable (c) that is associated with variables A and B independently? If so, then we can’t infer causation
identify the third variable:
murder rates and ice cream sales are highly positively correlated
weather, the hotter it is = more people out and about
external validity is more important with _______ claims
frequency
moderating variables
3rd variables are confounds, moderators purposely added to help us understand correlation
ex. extraversion –> happiness, moderating variable: setting
identify the moderator:
“recent research shows that growing up in stressful economic conditions can disrupt brain development, alter behavior and challenge emotions. But for boys, the outcome is worse.”
gender
multivariate designs
such as longitudinal and multiple regression designs, involve more than two measured variables
longitudinal designs
collecting data from some group over long period of time
measuring at multiple time points
cross-sectional correlations
relationship between 2 variables at one time point
can you establish temporal precedence with cross-sectional correlations?
no!
autocorrelations
looking at correlation at each variable of itself across each time
can you establish temporal precedence with autocorrelations?
no!
cross-lag correlations
correlation between one variable and the next time period
what does n.s. stand for, especially when looking at cross-lag correlations?
not significant
what correlation is the closest to establishing temporal precedence?
cross-lag correlations
if both cross-lag correlations are significant, what does this entail?
neutrally reinforcing but don’t know which is causing the other
longitudinal designs can provide SOME evidence for causation by fulfilling three criteria:
covariance
temporal precedence
internal validity
multiple regression
increases the internal validity of study
- help rule out 3rd variable problem
controlling for confounds while predicting one variable w/ another one
3 things to look at to see if regression results indicate if a third variable affects the relationship
criterion variable
predictor variable
use beta to test for third variable
what does testing for beta do?
tell us direction and strength in variables but no cut-offs/guidelines like cohen’s d or r
only used to compare effect sizes
can only compare beta strengths within a …
single regression table
what if beta is not significant?
3rd variable is explaining the link!
can the strongest beta be negative?
yes!
when interpreting a regression table, the answer should have …
the other variables that you controlled for/mentioned in the table
4 regression terms used in popular press articles
“controlled for”
“taking into account”
“correcting for”
“adjusting for”
multiple regression is not a _____ way to rule out all kinds of third variables
foolproof
regression does not establish …
causation!
multiple studies showing same pattern in results = …
increase in internal validity/closer to causal claims
parsimony
study characteristics
the simpler the better
journalists do not always fairly represent ____ and _____
pattern; parsimony