Exam 3 Material Flashcards
What are bivariate correlations?
composed of 2 variables
used to determine the relationship between two different variables
What type of variables are found in bivariate correlations?
measured variables
1 = continuous
the other may be continuous or categorical
the variables must have greater than or equal to 2 levels
What are 3 limitations to correlational research?
- can’t randomly assign
- prediction as primary goal
- can’t manipulate
What is a construct validity concern with correlations?
How well was each variable measured?
What is a statistical validity concern with correlations?
How well do the data support the state conclusions?
What is the relationship between statistical significance and effect size?
Larger effects + larger sample size –> significant p-value
large samples are needed to get sig effects with the effect size is small
What are the implications of larger effect sizes for correlations?
more accurate predictions based on correlation
What are the two types of effect sizes?
R - 2 continuous variables
d - 2 variables, 1 is categorical
What does measurement error influence?
Reliability
unreliability reduces the observed correlation
How does restriction of range influence correlations?
It distort observed correlations
gives a biased estimate
What are the two types of restrictions of range?
the ceiling effect
the floor effect
What is desirable variability vs. undesirable variability?
Desired: we want variability in X and Y
What are 5 statistical validity concerns?
Effect size
significance testing
measurement error
restriction of range
outliers
What are the 3 criteria for causation?
Covariance
Temporal Precedence
Internal Validity
are multivariate associations better than bivariate?
not necessarily, but the do “get closer” to causal inferences
What are the 3 types of longitudinal correlations?
Cross-sectional
Autocorrelations
Cross-lag
What are cross-sectional correlations?
aka synchronous
time 1 is compared to variable A
time 2 is compared to variable B
What are autocorrelations?
aka stability
time 1 is compared to time 2
variable a is compared with variable b
What are cross-lag correlations?
time 1 is compared to variable b
time 2 is compare to variable a
tells us which came first
1 sig correlations shows directionality
What criteria for causality does multiple regression address?
internal validity
rules out 3rd variables
What types of variables are in a multiple regression?
predictor variables (>2)
outcome variables (1)
What is the effect size used in regression?
Hierarchical multiple regression: each steps adds a simultaneous regression
What is R^2?
coefficient of multiple determination
how close the data are to the fitted regression
% of outcome variable variance that is explained by the model
What is pattern and parsimony?
the results of a variety of correlational studies all support a single, causal theory
or
the most parsimonious explanation for a pattern of results across studies is a causal one
What is mediation?
WHY 2 variables are associated
implies causation
What is moderation?
WHEN or FOR WHOM are two variables associated
a moderator…
…influence the relationship between 2 variables
x <———>y
^
moderator
EXTERNAL VALIDITY
a mediator…
… explains the relationship between 2 variables
x —> mediator –> y
INTERNAL VALIDITY
What is full or partial mediation?
full: after considering the mediator variable, there is no x or y relationship
partial: some relationship exists between x and y without mediator variable
What is a spurious correlation?
when two variables appear to be correlated but are not
icecream and sunburns
What types of variables are in an experiment?
_> 1 manipulated
_> 1 measured
What are 3 potential threats to repeated-measures design? and how to control
Order effects - counterbalancing
demand characteristics
design confound - keep everything else constant
What is Internal Validity as it related to experiments?
Are there any possible alternative reasons for the change in DV other than the change in IV?
What is an example of systematic variability?
a 3rd variable/confound that regularly/consistently covaries with IV
What validity is systematic variability concerned with?
Internal validity
What are 3 examples of unsystematic variability?
- measurement error
- individual differences
- situational noise
What validity is unsystematic variability concerned with?
Statistical validity
When you control for unsystematic variability you…
… increase power of the study
Why is a manipulation check important?
provides empirical evidence of construct validity of the IV
What does a pilot study do?
It confirms effectiveness of manipulation in a simple first study
what test is associated with effect size?
Cohen’s d
What are null effects?
no group/condition difference
the IV did not covary with the dv
What is the file drawer problems?
when people won’t publish studies that find a null effect
What are 3 possible reasons for a null effect?
- IV does not affect DV
- too much within groups variability
- too little between groups variability
weak manipulations have…?
… less power
