Chapter 9: Multivariate Correlational Research Flashcards
Three Causal Criteria
- Covariance
- Temporal precedence
- Internal validity
Covariance
As A changes, B changes; e.g., high levels of A go with high levels of B, and low levels of A go with low levels of B (positive correlation).
Temporal precedence
A comes first in time, before B.
Internal validity
There are no possible alternative explanations for the change in B; A is the only thing that changed.
Multivariate designs
Research study designs that involve more than 2 measured variables. Includes longitudinal design, multiple regression, and “pattern and parsimony”.
Longitudinal design
A study in which the same variables are measured in the same people at different points in time. Establish temporal precedence.
Cross-sectional correlations
A correlation between two variables that are measured at the same point in time. (think “s” for same)
Labeled with double-headed arrows because we are not assuming what causes what - we are just saying that they covary so the cause could go either way.
Choosing multivariate design over experimental design
It may not be feasible to randomly assign people to particular experiences.
It may not be ethical to randomly assign people to particular experiences.
Auto-correlation
The correlation of one variable with itself, measured at two different times. (think “autonomous”)
Single-headed arrows due to the passing of time - We KNOW that A comes before B in time.
Cross-lag correlation
A correlation between earlier measure of one variable and a later measure of another variable. Addresses directionality problem and establishes temporal precedence.
Single-headed arrows that cross, make sure to label!
Multiple regression
A statistical technique that computes the relationship between a predictor variable and a criterion variable, controlling for other predictor variables. AKA multivariate regression.
Beta
Represents the relationship between IV and DV after controlling for other variables.
Interpreting beta
Positive value indicates that as IV increases so does DV.
Negative value indicates inverse relationship.
Zero indicates there is no linear relationship.
Larger beta means stronger relationship.
Smaller beta means weaker relationship.
But there are no guidelines for interpreting effect sizes (weak, moderate, strong) in the same way we do with r.
Regression in popular press articles
“Controlled for”
“Taking into account”
“Correcting for”
Control for
Holding a potential third variable at a constant level (statistically or experimentally) while investigating the association between two other variables.
Parsimony
The degree to which a theory provides the simplest explanation of some phenomenon. In the context of investigating a claim, the simplest explanation of a pattern of data; the best explanation that requires making the fewest exceptions or qualifications.
Criterion variable
Dependent variable.
German “B” and b
You can compare the strength of German “B” to other “B”s in the same regression table, but not from one table to another. This is a standardized coefficient.
You cannot compare two “b” values within the same table to each other. This is an unstandardized coefficient.
Predictor variable
Independent variable.
Mediator
A variable that helps explain the relationship between two other variables. Think “causal mechanism”
Intro to mediation
Start with an association between IV and DV.
Mediation hypotheses propose a mechanism for a bivariate relationship. Why are these two variables correlated?
Mediation hypotheses are causal statements because they specify a time sequence for the three variables (temporal precedence).
Mediators versus third variables
Both involve multivariate research designs.
Both can be detected using multiple regression.
Third variables are external to the bivariate correlation (problematic).
Mediators are internal to the causal variable (not problematic).
Mediators versus moderators
Mediators ask “why,” and moderators ask “for whom” or “when.”
Detect moderation by looking at the correlations within groups!
Multivariate designs and internal validity
Gets us closer to making causal claims because longitudinal studies help with temporal precedence, and multiple regression helps rule out third variables.
Multivariate designs and construct validity
It is important to interrogate the construct validities in all of your variables in multivariate designs. Are your measures reliable and valid? How did you measure your variables?
Multivariate designs and external validity
Consider whether the results are generalizable to other people, contexts, situations, and so forth. You would want to look at how the sample was obtained.
Multivariate designs and statistical validity
Consider the effect size of the beta and the statistical significance (p value). You might have several betas that are statistically significant, but you can compare their effect sizes within the same beta table.
