Chapter 9 Multivariate Correlational Research Flashcards
multivariate designs involve what?
3(+) variables
multiple regression
statistical analysis that measures the relationship b/w predicTor and criterion variables while CONTROLLING FOR OTHER PREDICTORS
how is a third variable controlled for?
holding it constant while investigating the relationship b/w 2 other variables
multiple regression can help address which validity?
internal
does multiple regression establish temporal precedence?
no
the overlap b/w 2 circles in a Venn diagram indicates what?
the degree of association b/w 2 variables
1 circle in a Venn diagram represents what?
the variability of 1 variable
what does it mean when we say we want a unique overlap b/w criterion and the predictor of interest variable?
we want a section where only these 2 variables are overlapping (w/o the 3rd variable)
3 possible outcomes of multiple regression
- predictor has no independent effect on the criterion variable
- both the predictor and 3rd variable have an independent effect on the criterion variable while controlling for the other (each have a unique overlap w/ the DV)
- the third variable has no independent effect on the criterion variable when controlling for the predictor
does a p value or a beta give significance?
p value
finding the unique overlap b/w IV and DV is a method of doing what?
controlling for the 3rd variable
beta in regression analysis
“standardized regression coefficient”, the unique overlap b/w IV and DV w/o overlap from 3rd variables
example of how beta can indicate strength and direction of relationship
negative beta = negative correlation
out of a group of predictor variables, the one with the largest beta value will be the what?
strongest predictor
multiple regression can help answer which 2 questions?
- helps control for several 3rd variables at once
- often examines the betas for all other predictors to understand which factors most strongly predict the DV
b
unstandardized regression coefficient
main differences b/w beta and b
beta: standardized, can compare within the same regression analysis (result table) but not across different ones
b: unstandardized, cannot compare within the same regression analysis
b template
for every 1 unit change in X, there is a b units change in Y, holding third variables C, D and E constant
beta template
this beta means that X is/is not associated with Y such that high/low scores on X go w/ high/low scores on Y, controlling for other predictors C, D and E.
mediator
a variable that helps explain the relationship b/w IV and DV; an outcome of the IV responsible for the DV
triangular diagram is an indicator of what?
mediation
total effect (of mediation) =
direct effect + indirect effect
what types of studies can we find mediation?
correlational and experimental
can mediation imply causation?
yes
similarities b/w mediators, moderators and 3rd variables
-involve multivariate research designs
-can be detected using multiple regression techniques
longitudinal designs
study in which the same variables are measured in the same people at dif. points in time
can longitudinal designs establish temporal precedence?
yes!
can longitudinal designs establish causation?
not on their own!
3 types of correlations in longitudinal studies
cross-sectional
autocorrelations
cross-lag
cross-sectional correlation
whether 2 variables measured at the same point in time are correlated
autocorrelations
the correlation of each variable w/ itself across different time points
cross-lag correlations
correlation b/w an earlier measure of 1 variable q/ a later measure of the other variable (X time 1 w/ Y time 2)
which correlations of longitudinal studies can establish temporal precedence if significant?
cross-lag correlations
4 possible outcomes of longitudinal studies
- significant correlation b/w earlier X and later Y
- sig corr b/w earlier Y and later X
- sig. corr. including both 1 and 2 (mutually reinforcing)
- zero correlation
which possible outcome of longitudinal studies is rare and why?
zero correlation, because longitudinal studies usually look at multiple variables so 99.9% of the time there’s some significant findings
why do longitudinal studies lack internal validity?
they cannot automatically rule out 3rd variables
advantages of longitudinal studies
may establish temporal precedence
may help study causation when its impossible/unethical to manipulate the IV (ex: parenting style/neglect)
drawbacks of longitudinal studies
- difficult to recruit people who agree to participate across a long time period
- attrition (participants dropping out)
- money and time consuming
parsimony
extent to which a theory provides the simplest explanation
pattern
same results are seen in similar research/theories
methodological pluralism
using many methods and designs to test a theory
pattern and parsimony are referred to as…
the producers of science
