Questions From Lectures II Flashcards
Atminimum, howmany variables are there in an association claim?
.An association that involves exactly two
variables.
What characteristic of a study’s variablesmakes a study correlational?
They are measured, not manipulated.
Sketch three scatterplots: one that would show a positive correlation, one that would show a negative
correlation, and one that would show a zero correlation.
.
Sketch three bar graphs: one that would show a positive correlation, one that would show a negative
correlation, and one that would show a zero correlation.
.
When do researcherstypically use a bar graph, as opposed to a scatterplot,to display correlational
data?
. can guess, maybe ask
. most likely if one of the variables is categorical
. most likely not to use it if both variables quantitative
In one ortwo briefsentences, explain how you would interrogate the construct validity of a bivariate
correlation.
.Does the measure have good reliability?
-Test/Retest, Internal Reliability, Interrater Reliability
.Measuring what it intends? What is the evidence for its face validity, its concurrent validity, its discriminant and convergent validity?
-Face/Content Validity
-Predictive/Concurrent Validity
(e.g. Do mothers’ answers to this question correlate with their actual employment history? for maternal employment)
-Convergent Validity
-Discriminant Validity
What are five questions you can ask aboutthe statistical validity of a bivariate correlation?Do all ofthe
statistical validity questions apply the same way when bivariate correlations are represented as bar
graphs?
.What is the effect size?
.Is it statistically significant?
.Subgroups within the sample? Is the relationship spurious? Is there a third variable?
.Are there outliers?
.Is the relationship curvilinear? If slope of pattern is not just a straight line, r does not describe pattern well.
Which ofthe three rules of causation is almost alwaysmet by a bivariate correlation? Which two rules
might not bemet by a correlationalstudy?
.Covariance
.Temporal precedence or internal validity
Give examples ofsome questions you can ask to evaluate the external validity of a correlationalstudy.
.Can the association generalize to other people,
places, and times?
Must consider who the participants were and how
they were selected.
The size of the sample does not matter as much as
the way the sample was selected from its
population
Why can’t a simple bivariate correlationalstudymeet allthree rulesfor establishing causation?
.No time difference between measures!
Explain how longitudinal designs are conducted. Why is a longitudinal design called amultivariate
design?
.Because each measure of one variable at different times is a different variable right.
Like, TvViolence2001, TvViolence2011
AND, TIME is a THIRD variable
So no matter what you do, it will always be multivariate
Identify the three types of correlations in a longitudinal correlational design: cross‐sectional
correlations, autocorrelations, and cross‐lag correlations.
.Cross-Sectional: TvViolence2001 & Aggression2001
.Autocorrelations: TvViolence2001 & TvViolence2011
.Cross-lag: TvViolence2001 & Aggression2011
Interpret different possible outcomesin cross‐lag correlations, andmake a causal inference fromeach
pattern.
.ASK
Explain howmultiple‐regression designs are conducted.Describe in your own words whatitmeansto
say thatsome variable “was controlled for” in amultivariate study.
.LOOK IN TEXT
.Control for: Holding p aotential third variable
steady while investigating the association
between two other variables.
Researchers are asking whether, after they
take the relationship between the third variable
and the outcome (effect) into account, there is
still a portion of variability in the outcome
(effect) that is attributable to the predictor
(cause)
Define dependent variables and predictor variablesin the context ofmultiple‐regression data.How
many dependent variables are there in amultiple‐regression analysis?Howmany predictor variables?
.Criterion: Researchers most interested in understanding or predicting (also called DV in this case)
.Predictor: Used to explain variance in the dependent/criterion variable (also called IV in this case.
.only ONE criterion/dependent variable
.unlimited predictor/independent variables i assume
Identify and interpret data fromamultiple‐regression table and explain, in a sentence, what each
coefficientmeans. What does a significant betamean? What does a nonsignificant betamean?
When you have only one predictor variable in your model, then beta is equivalent to
the correlation coefficient between the predictor and the criterion variable. This SPSS for Psychologists – Chapter Seven 209
equivalence makes sense, as this situation is a correlation between two variables.
When you have more than one predictor variable, you cannot compare the
contribution of each predictor variable by simply comparing the correlation
coefficients. The beta regression coefficient is computed to allow you to make such
comparisons and to assess the strength of the relationship between each predictor
variable to the criterion variable.
Give atleastthree phrasesthatindicate that a study used amultiple regression analysis.
.rl between x & y is negative, even when z IS CONTROLLED FOR
.rl between x & y is negative, INDEPENDENT OF the proportion of z
.rl between x & y is negative, even when z IS HELD CONSTANT
.rl between x & y is negative, and is NOT ATTRIBUTABLE TO THE THIRD VARIABLE OF z, because it holds even when the proportion of z is held constant
What are two reasonsthatmultiple regression designs cannot completely establish causation? Explain
why experiments are superiortomultiple‐regression designsfor controlling forthird variables.
- Even though multivariate designs analyzed
with regression statistics can control for third
variables they cannot establish temporal variables, they cannot establish temporal
precedence. - Researchers cannot control for variables that
they do not measure.
A well-run exp yy erimental study is ultimately more
convincing than a correlational study.
The power of random assignment would make the
groups likely to be equal on any possible third
variable.
A rand i d i t i till th ld domized experiment is still the gold
standard for determining causation.
Multiple regression allows researchers to control for
potential third variables, but only those that they
choose to measure
Explain the value of pattern and parsimony in research.
.An approach which allows researchers to
investigate causality by using a variety of
correlational studies that all point in a single,
causal direction.
-pattern of results best explained by parsimonious causal explanation
-parsimony: simplest explanation of a pattern of data
-several diverse predictions are tied back to one central principle = parsimony
-does not work for a single study
Consider why journalistsmight preferto reportsingle studies,ratherthan parsimonious patterns of
data. What problemsresultsfromthistendency?
.Trying to find news, and flashy headlines
.They usually only report the latest finding.
They selectively present only a part of the
scientific process.
Identify amediation hypothesis and sketch a diagramofthe hypothesized relationship.Describe the
stepsfortesting amediation hypothesis.
TESTING FOR A MEDIATING VARIABLE
Kenny (2008):
1.Test for relationship c.
2.Test forrelationship a.
3.Test for relationship b.
4.Finally, run a regression test, using both the predictor and mediator variables to predict the criterion, to see whether relationship c goes away
OR
test for relationship c, then a, then b
-run regression test
-relationship btw IV and DV should drop significantly or become zero when mediator is controlled for
MULTIVARIATE CORRELATIONAL RESEARCH (look up and understand more if time)
Articulate the difference betweenmediators,third variables, andmoderating variables.
.med: “why are these two variables linked?”
mod: “are these two variables linked the same way for everyone, or in every situation?”
THIRD VARIABLE
internal validity rule, when you can come up with an alternative explanation for the association between two variables, that alternative explanation is the third variable
Give an example of a question you would ask to interrogate each ofthe four validitiesfor amultivariate
study.
.Longitudinal designs help establish temporal
precedence, and multivariate provide evidence
for internal validity
Should interrogate the construct validity (i.e.,
how well each variable was measured) external measured), external
validity (i.e., how well the results generalize),
and the statistical conclusion validity (i e the and the statistical conclusion validity (i.e., the
effect size and statistical significance).
What are theminimumrequirementsfor a study to be an experiment?
.A study in which one variable is manipulated and the other is measured.