Module 10: Multivariate Statistics Flashcards
What is linear regression used to determine?
A straight line fit to the data that minimizes deviations from the line
What other word is regression used synonymously with?
Correlation
When are multivariate statistical methods used?
When there are three or more variables of interest
What is simple linear regression?
One IV is used to predict one DV
What levels of measurement should both the IV and DV in a simple linear regression be?
Interval or ratio level (continuous)
Because correlation between 2 variables is rarely perfect what time of regression is used to improve predictions of the DV by including multiple IVs?
Multiple linear regression
What levels of measurement should both the IV and DV in a multiple linear regression be?
Interval or ratio level (continuous)
The R value in multiple linear regression represents what?
The multiple correlation coefficient or the strength of the relationship between multiple IVs
0.00-1.00
R(squared) represents:
The proportion of variance in DV accounted for by simultaneous impact of all IVs
To equally compare variables in multiple linear regression variables are transformed into standard scores so they can be equally compared, these standardized regression coefficients are called:
Beta weights
3 strategies for entering predictor variables into regression equations include:
- Simultaneous
- Hierarchal
- Stepwise
How is the simultaneous strategy of entering predictor variables into the regression equation executed?
All variables are entered into the regression equation at the same time. This is used when there is no basis for considering any particular predictor as causally prior to another
How is the hierarchical strategy of entering predictor variables into the regression equation executed?
Entering predictors into the equation in a series of steps controlled by the researcher based on theoretical considerations.
How is the stepwise strategy of entering predictor variables into the regression equation executed?
Predictor variables are entered into the regression equation in a stepwise fashion using statistical criteria or power.
What are general sample size guidelines?
At least 50 ppl + an additional 8ppl/variable
What are general sample size guidelines used with simultaneous or hierarchical strategies?
At least 20ppl/variable
What are general sample size guidelines used with stepwise strategies?
At least 40ppl/variable
What is the role of logistic regression?
Odds ratio, predicts the likelihood of something happening
What levels of measurement are the IV and DV in a logistic regression?
IV: any level
DV: nominal/categorical (usually dichotomous/bimodal; 1=smokes 0=doesn’t smoke. Can have multinomial when DV has 3 or more categories)
The odds of doing something(odds ratio) equals:
The probability of doing something divided by the probability of not doing it
If there is a strong correlation between variables in a simple linear regression what does this mean?
This indicates a strong prediction of the outcome
What does a logistic regression ultimately predict?
Categorical (nominal) DV
Pearson’s r does not indicate causality and is not used in what type of study design?
Quasi-experimental
Experimental
What test is used to compare the means of two or more groups while controlling for confounding variables?
ANCOVA (analysis of covariance)
What levels or measurement are covariates used in an ANCOVA usually?
Interval/ratio level
An ANCOVA may increase ______ validity when randomization is not possible.
Internal
ANCOVA may still be used in randomized experiments
When is an ANCOVA applicable?
When researchers are using nonequivalent control group designs to test an intervention and researchers need to consider whether pre-existing group differences affected results. An ANCOVA can be used for post hoc stat testing
What statistic does an ANCOVA result in?
F statistic
What test is used to test the significance of differences in group means for multiple interval or ratio level DVs which are considered simultaneously?
MANOVA
A MANOVA should be used instead of two or more separate ANOVA tests when there are multiple DVs to limit Type ____ error
I
What statistic does a MANOVA result in?
F statistic
Which multivariate statistical tests compare means and which predict outcomes?
Compare means: ANCOVA, MANOVA
Predict outcomes: simple linear, multiple, and logistic regression
If the population multiple correlation coefficient is equal to 0, what does this represent?
The null hypothesis (there is no relationship between the variables)
