L3: Predictors & Outcomes Flashcards
What is a predictor?
An independent variable used in regression analysis
What is a criterion variable?
Dependent variable used in non-experimental situations
What is explanation in research?
Explaining events in terms of causal relations between variables
What is description in research?
Identifies regular occurring sequences of psychological events
What is prediction in research?
Follows patterns (predictable)
What is application in research?
Informs actions & decisions for future events
Hamaker et al., (2020) found prediction to be the most undervalued goal which demonstrates researchers jump from description to causation
What is correlational research useful for?
It’s useful for making predictions
Prediction is only possible when strong correlations exist
Regression allows you to use a predictor variable (X) to predict criterion variable (Y)
If a correlation is not significant, a regression SHOULD NOT be done
Longitudinal studies accurately demonstrate description to causation
Why should studies of predictor outcomes precede intervention designs & RCT’s?
It’s important to know the outcome relation because a trial in a RCT provides information about the effect of intervention under evaluation compared to control.
What is multiple regression?
Used the analyze the relation between 1 DV and multiple IV’s
Object is to use the IV’s to predict the value of a single DV
EX. of multiple regression
Alloway & Alloway., 2010
Measured the extent to which working memory & IQ (IV’S) predict academic attainment (DV)
Conducted forced-entry then hierachal regression
What is a forced entry multiple regression?
All predictor variables (IV’s) are entered into the model
What is a step-wise multiple regression?
Predictor variables (IV’s) are entered using a semi-partial correlation
Most common use is multiple variables studied at the same time
What does multiple regression aim to do?
Compare the importance of predictors against eachother
What is hierarchal regression?
Predictors are entered into blocks
Each block represents one step of the model
A way of showing if your variables of interest explain a statistically significant amount of variance on your DV
What is dominance analysis?
Considers importance of predictors
Considers the change in semi partial correlations for all possible subset models
Comprehensive approach to distinguishing between predictors
What are the issues with longitudinal data?
Often is a within subjects design
Scores across within subjects are often correlated which goes against the point of multiple regression is to compare importance of predictors (violates assumption of independence)
What is a mixed effects model?
When to Use? – Studies that obtain multiple measurements over time (longitudinal, time-series) or multiple trials per participant (within subjects) lend themselves well to mixed model analyses
Statistical test which is used to predict a single variable using 2 or more other variables
Useful to determine the numerical relationship between 1 variable and others
Useful when our variables have 1 or more source of random variability e.g., If data is within subjects & over time
Mixed effects models have both fixed & random effects
What does a mixed effects model assume?
Assumes each ppt has unique pattern of change & stability
What are random effects?
ANOVA/Regression: Refers to how particular variables behave
EX. Weather day to day
What are fixed effects?
ANOVA/Regression: Refers to how particular variables behave
Either fixed like skin color
Example of mixed effects model
Peter et al., (2019)
Interested in predicting language growth (growth = mixed effects models)
Predictor: Vocab
Criterion: Measurement point
Measured same thing over and over to show growth
Findings: Slower processors have lower vocab in every single point in time
Start with one model & add multiple models to get mixed effects model
What did Dadvand et al., (2015) measure?
Measured whether green spaces will lead to better cognitive development growth
For example if you have lower income it may be more common to have less green spaces and also more density
Mixed effect models are able to measure the multiple variables to test their hypotheses
More green spaces = better cognitive growth
What is survival/failure analysis?
Family of techniques dealing with the time it takes to happen
What predicts the timing of an event e.g. the time a person lives on avg. until death
Survival analysis is good for preparing predictors
What did Chen et al., 1998 measure and find regarding survival analysis?
Looked at why people continue to smoke marijuana and why people stop
Across time every single cohort across time had decreased use of marijuana
Looked at loads of different predictor variables
A predictor variable of depression wasn’t a predictor in continued marijuana use
One of the most prevalent variables was becoming a mother
It doesn’t have to be about cessation it could be about the onset of something but survival analysis is good at looking at both onset and cessation