CHP. 9 - Multivariate Correlation Flashcards
Multivariate Correlation
Correlation between more than 2 variables
Longitudinal Designs
Used to establish temporal precedence
- often used to study changes over time
Typically measure 4 variables
- Variable 1 @ Time 1 & Time 2
- Variable 2 @ Time 1 & Time 2
3 Types of Correlations with Longitudinal Designs
- Cross-sectional
- Auto Correlations
- Cross-lag
Cross-Sectional Correlation
Measures both variables at the same time
- Is the correlation between variables consistent over time?
Auto Correlations
Examines correlations of a single variable between each of the time points
- Is the variable consistent over time?
Cross-Lag Correlations
Can be used to establish temporal precedence
3 possible outcomes:
1. A causes B
2. B causes A
3. A & B are mutually reinforcing
Multiple Regression Analysis
Statistical technique used to rule out some 3rd variables
- establishes internal validity
Steps for Multiple Regression
- Scientist chooses the variable they are most interested in understanding (criterion variable)
- Rest of the variables measured in the regression analysis are called predictor variables
- Calculate beta for each of the independent variables
Beta
Relationship between predictor & criterion variable
- positive b = positive relationship
- negative b = negative relationship
- beta = 0 = no relationship
Parsimony
Degree to which good theory provides the simplest explanation of a pattern of data
Pattern & Parsimony Approach to Establish a Causal Claim
- Come up with a theory that specifies a mechanism for the causal path
- Theory would have a set of predictions
- By showing evidence for each part & each sub-theory you defend the overall theory
Mediator
Some factor that is internal to the causal variable
- is a type of causal claim
Requirements for Mediation
*Directionality is important!
1. “C” must be significant
2. “A” must be significant
3. “B” must be signficant
4. Regression analysis
Mediators vs. Moderators
“Why are these 2 variables linked?” vs. “Are these 2 variables linked in every situation?”