What is one way researchers can establish temporal precedence in their correlational studies?

Unformatted text preview: Chapter 9: Multivariate Correlational Research Reviewing the Three Causal Criteria ● ● Longitudinal designs, multiple­regression designs and the pattern and parsimony approach are multivariate designs. ○ Extremely useful when experiments are impossible to run. The three criteria for establishing causation are covariance, temporal precedence, and the internal validity. ○ Review definition for all three of these. Establishing Temporal Precedence With Longitudinal Design ● ● ● ● Longitudinal design : can provide evidence for temporal precedence by measuring the same variables/preferences in the same people at several points in time. Usually used in developmental psychology. Design is used to test causal claims. Ex: The Eron study on television violence and kids at ages 3 and the checked the same kids again at age 13. Used the same group across time. Also was a multivariate correlational study because it studied: preference for TV violence at time 1, preference for TV violence at time 2, aggression at time 1 and aggression at time 2. Interpreting Results from Longitudinal Designs ● Multivariate designs give several individual correlations. ○ Cross sectional correlations ○ Autocorrelations ○ Cross­lag correlations Cross­Sectional Correlations ● Cross­sectional correlations : they test to see whether two variables, measured at the same point in time, are correlated. ○ Ex: tv violence in the third grade, and aggression in the third grade. Autocorrelations ● Autocorrelations : they determine the correlation of one variable with itself, measured on two different occasions/points in time. ○ Ex: preference for tv violence in the 3rd grade, preference for tv violence in the 13th grade. Cross­Lag Correlations ● ● ● Cross­lag correlations : show whether the earlier measure of one variable is associated with the later measure of the other variable. MOST IMPORTANT TO RESEARCHERS Ex: Tv violence in the third grade, aggression in the 13th grade. Or aggression in the 3rd grade link to preference for tv violence in the 3rd grade. ● ● Cross lag correlations help establish temporal precedence because you can establish which variable came first in time. Many correlational patterns can be shown from cross lag correlations ○ May even see that both correlations are significant and that A causes B and B causes A. Longitudinal Studies and the Three Criteria for Causation 1. Covariance : Significant relationships in longitudinal designs helps establish covariance. a. When two variables are significantly correlated (as in the cross sectional correlations) there is covariance. 2. Temporal Precedence : two variables are measured at two points in time, they know which came first. Comparing the the strength of the two cross lag correlations, the researchers can tell which relationship is stronger 3. Internal validity : Longitudinal studies do not help out the third variable problem. Why Not Just Do An Experiment? ● ● ● ● Sometimes people cannot be randomly assigned to a variable/preferences. It could be unethical to assign some people, especially children, to a condition which they are forced in. When a design is not practical or ethical, a longitudinal design is a good option. Longitudinal designs just measure already set preferences and track their correlations with other variables over time. Ruling Out Third Variables With Multiple­Regression ● Multiple Regression : which can help rule out some third variables. Measuring More Than Two Variables Using Statistics to Control for Third Variables ● ● ● ● Using a multivariate design, researchers can evaluate whether a relationship between two key variables still holds when they controlled for another variable. When using controlled for, researchers are asking whether, after they take the relationship between two variables into account, is there still a portion of variability in the other variable. May use subgroup analysis. Ex: Income level example in book! Use it! FILL IN BETTER DEFINITION FROM LECTURE Regression Results Indicate If a Third Variable Affects the Relationship Criterion and Predictor Variables ● ● ● When researchers use multivariate designs, they are usually measuring three or more variables. They first choose a variable that they are most interested in understanding or predicting, that variable is known as the criterion variable or dependent variable . The criterion variable is almost always labeled in the top row of in the title of a regression table. ● Predictor variables (independent variable): the rest of the variables that are measured in a regression analysis. Using Beta to Test for Third Variables ● ● ● ● ● There will be one beta value for each predictor variable. ○ Beta is similar to r, but it reveals more than r does. A positive beta means there is a positive relationship between the criterion variable and predictor variable when all other predictor variables are statistically controlled for. ○ Same way for the negative betas. Betas denote the direction and the strength of the relationship. ○ The larger the beta, the stronger the relationship. ○ Remember to read over the sign b in the text. When a p­value is less than 0.05, the beta (the relationship between the criterion and predictor variable) is considered significant. When a p­value is greater than 0.05, the beta is considered not statistically significant. ○ We cannot conclude that beta is different from zero. Adding More Predictors to a Regression ● Adding several predictors to a regression analysis can help answer two kinds of questions. ○ It helps control for several third variables at once. ○ By looking at the beta for each predictor variable, we can get a sense of which factors most strongly affect classroom behavior problems. Regression In Popular Press Articles ● When you encounter association claims in popular press, the journalist will rarely mention p­values, betas, and predictor variables. ○ If you read carefully you can tell that multiple regression was used. “Controlled For” ● ● ● This is the most common phrase for regression analysis. A clue the study used multiple regression. In the journal, the things controlled for would be the predictor variables. “Taking Into Account” ● Means the researchers conducted multiple regression analysis. “Correcting for or Adjusting For” Regression Does Not Establish Causation ● ● ● Even though multivariate designs analyze with regression statistics a third variable, they may not be able to establish temporal precedence. The second problem is that researchers cannot control for variables they do not measure. There could be many more important third variables that the researchers did not consider. ○ Multiple regression allows you to control for variables the researcher wants to measure. ● Randomized experimental design is the best way to establish causation. Getting At Causality With Pattern and Parsimony ● ● ● ● ● ● Longitudinal designs help establish temporal precedence. Multivariate designs help establish internal validity by controlling for some third variable. A variety of correlational studies can lead to a single causal direction. Called pattern and parsimony because there is a pattern of rules that is best explained by a parsimonious causal explanation. Parsimony : the degree to which a good scientific theory provide the simplest explanation of some phenomenon. ADD IN LECTURE NOTES INFO FOR THIS SECTION. Pattern, Parsimony, and Popular Press ● ● When journalist write about science, they sometimes do not write about pattern and parsimony well. They may only report on the results of the latest study and leave out old studies that may have revealed other important information or the theory behind the study. ○ Gives off the idea that a single study can erase the findings of previous studies. ○ Does not capture the scientific process of adding onto old findings. Mediation ● ● ● ● ● ● ● Once a relationship between two variables is established, we want to figure out why. Sometimes a relationship between two variables involves a mediator in between the two variables. Experimental and correlational studies can have mediator steps between two variables. Mediation analysis relies on multivariate tools such as multiple regression. The overall relationship only exists between a and b but c is in between. ○ Test for all 3 relationships amongst each other. FILL THIS IN WITH NOTES FROM LECTURE Mediation is established only when the proposed caused variable is measured (or manipulated) first in a study, followed by the mediating variable followed by the proposed outcome. Mediators Vs. Third Variables ● ● ● Mediators are similar to third variable explanations in that both of them involve multivariate research designs and researchers can use the same statistical tool (multiple regression) to test detect them both. A third variable explanation, the proposed third variable, is external to the two variables in the original bivariate correlation. ○ May be seen as the problematic, lurking variable. Mediator variable is internal to the causal variable and often of direct interest to the researchers. Mediators Vs. Moderators ● Testing for mediation versus moderation involves asking different questions. ● ● ● ● ● When researchers test of mediating variables they ask: why are these two variables linked? When they are testing for moderating variables they ask: Are these two variables linked the same way for everyone or in every situation? Mediators ask: Why? Moderators ask: For whom? Or When? Moderating variables make the relationship between two variables less intense. LOOK AT THE DIAGRAMS Multivariate Designs and the Four Validities ● ● ● ● ● ● ● ● ● For any multivariate design it is important to ask about the construct validity of the variables in the study. Ask how well was each variable measured in comparison to the operational definition. WORK ON UNDERSTANDING THIS. We can also interrogate the external validity of a multivariate design. How well does it generalize to other populations or other situations. Ask if the group was randomly sampled. For statistical significance you can ask about the effect size, betas, and the statistical significance. They need to also look for subgroups, outliers, curvilinear relationships. LOOK AT DIAGRAM FOR THIS SECTION! AND DRAW IT! ...
View Full Document

How do researchers establish temporal precedence?

Do correlations establish temporal precedence?

What is temporal precedence quizlet?

For which type of correlation does the pattern of results allow researchers to establish temporal precedence?