Class 6 Hepp et al -15 Cha 13 Quant Descriptive Designs.
Research Designs: Kazdin Gelso
Research Designs: Kazdin Experimental Quasi-Experimental Correlational
Internal Vs. External Validity Internal Validity Experimental Control Random Assignment to Groups Manipulation of IV External Validity Generalizability
Internal & External Validity E I Experimental Field E i Correlational Descriptive e I Laboratory e i Clinical Trials
Kerlinger MAXMINCON Principle Maximize Minimize Control
Kerlinger MAXMINCON Principle Maximize Variability in constructs of interest Experimental: Treatment groups Correlational: Measured variables Minimize Error/random variance Experimental: Within treatment groups Assessment/observation procedures Participants’ characteristics not examined Control External Confounds
Quant. descriptive Designs Purpose: Describe Phenomena Types of Studies Occurrence Characteristics Relationship among variables Survey epidemiological Variable Centered Person Centered
Can Survey Research Be Used To: Test Cause-Effect Relations ? YES ------ NO Test a Hypothesis ?
Survey Research Describe, document frequency, characteristics, patterns of behaviors, attitudes, events- Epidemiological Identify variables that may explain behaviors, attitudes, events Test hypothesis about the association of behaviors, attitudes, events Generate a causal hypotheses cholera example Chap. 13 p. 289 Cook, Alegria, Lin and Guo’s Study
Quant. descriptive Designs Purpose: Describe Phenomena Types of Studies Occurrence Characteristics Relationship among variables Survey epidemiological Variable Centered Person Centered
Latent vs. Observed Constructs Cole et al. (1987) assessed social skills and depression from 4 perspectives: Self-report Researchers’ behavioral ratings Interviews Significant others’ report
Variable Centered Research Bi-variate Correlation Extent to which 2 or more continuous psychological constructs vary together Pearson correlation coefficient r- bivariate correlation: strength and direction r = .35 r = -.35 r =. 50 r2 coefficient of determination Regression multiple IV’s one DV Examine combined contribution of several IVs to one DV Examine unique contribution of each IV to the DV while controlling for the other IVs
Pina-Watson et al Hypothesis
Pina-Watson, et al. 2014
Bivariate relation (Pearson’s r) of x, y, and z to Life Satisfaction Independ. SC r = .46 Barriers r =-.32 Life Satisf. Career D Self- Efficacy r = .44
Coefficient of Determination Correlation of Life Satisfaction with ISC r = .46 r2 = .21 21% Barriers r = -.32 r2 = .10 10% CDSE. r = .44 r2 = .19 19% Variance explained (shared)
Regression Hypothesis
Examine the unique () and collective (R2) relation of predictors x, y, and z to Life Satisfaction controlling for S & G Independ. SC Barriers SES Life Satisf. Career D Self- Efficacy GEN
1 SES + 1 GEN +1 ISC + 2 Bar + 3 CSE+ C = Life Satisfaction R2 = Independ. SC r =.46 .26 SES -.18 Barriers r =-.32 Life Satisf. Career D Self- Efficacy r = .44 .29 GEN
Pina-Watson, et al. 2013
Categorical Continuous ????
Multiple Linear Regression Objective: examine the unique and collective association of several independent variables (or predictor variables) to one dependent variable (also called criterion variable). Both IVs -predictor(s) and criterion-DVs are continuous. Categorical variables with only two levels can be used as predictors-IVs (e.g. gender, depressed/not depressed; treatment/control group)
Fill in the Blanks The best research design is: What type of design is used in the randomized controlled clinical trials used to examine therapy outcome?
Multiple Linear Regression It is best when: Predictors (IVs) are not highly correlated among themselves and Predictors (IVs) are correlated to the Criterion (DV)