Complex Research Designs PDF

Chapter 8: Complex Research Designs Multiple Dependent Variables In many cases, psychologists are interested in observing the effects of an IV on several DV e.g. Multiple Dependent Variables Related issues: 1) Carryover effects of testing participants through multiple DVs: 2) Manipulation ch...

Chapter 8: Complex Research Designs Multiple Dependent Variables In many cases, psychologists are interested in observing the effects of an IV on several DV e.g. Multiple Dependent Variables Related issues: 1) Carryover effects of testing participants through multiple DVs: 2) Manipulation check: necessary when constructs are manipulated indirectly (emotions and other internal states) Need to determine if manipulation was effective Multiple Dependent Variables 3) Multiple measures of the same construct: e.g. Effect of exercise on stress: -self-report & physiological measures as DVs Can consider each separate item on a rating scale as a dependent variable. But better to treat responses collectively as a multi-response measure of a single construct of stress Need to ensure that test items (dependent variable) correlate with each other for internal consistency (i.e. calculate Cronbach’s α) Multiple Independent Variables Allows researchers to investigate whether the effect of one independent variable depends on the level of another. Factorial design Independent variable = Factor Examine how each level of one factor depends on each level of the other factor; look at the dependent variable in all possible combinations of each factor level Factorial Design Table: 2 x 2 Design Each combination becomes a condition of an experiment Number of conditions in the product of the number of Factorial Design Table: 2 x 3 Design  2 x 2 x 3 Factorial Design Factor 1. Treatment (2 levels): Behaviour Modification, Psychotherapy Factor 2. Dosage (2 levels): 100 mg, 300 mg Factor 3. Treatment setting (3 levels): Inpatient, Day treatment, Outpatient Multifactorial design It is unusual in psychology to design experiments with more than 3 factors with each factor having more than 2- 3 levels each e.g. Unfeasible to do a 2 x 2 x 2 x 3 factorial design 24 conditions to examine! Assigning Participants to Conditions Decision must be made to treat each factor as a between-subjects variable or a within-subjects variable Between-subjects: participants assigned to only one condition of an IV Within-subjects: participants assigned to each condition (level) of an IV Possible to treat one IV as a between-subjects factor and another IV as a within-subjects variable = Mixed factorial design 2 x 2 x 2 Factorial Design Psychotherapy Type Therapist Gender Cognitive Behavioural Two MonthsTwo Weeks Two MonthsTwo Weeks Length Female or Cognitive Behavioural Length Male  Psychotherapy Type Therapist Gender Example of ways to Cognitive Behavioural Two MonthsTwo Weeks Two MonthsTwo Weeks assign participants in a mixed-factorial design? Length Female See to the left! Cognitive Behavioural Length Male Non-manipulated Independent Variables (Can’t have random assignment) A factor that is not manipulated, but simply measured, can be considered as a between-subjects factor But no causal conclusion can be made about the influence of this non-manipulated IV Example: Test memory for health-related words and non-health related words in participants reported to have high or low hypochondriasis hypochondriasis measured for each participants (e.g. self-report scale) and then divided on severity (high or low), which is a non- manipulated independent variable Graphing Results of Factorial Experiments Two results of interest in a factorial design: 1) Main effect: the influence of one factor (IV) averaged across all levels of the other factor (IV) Is there a main effect of the influence of one factor all by itself? 2) Interaction: The effect of one factor IV depends on the level of another IV Do they influence each other?  2 x 2 x 4 design Main effect: Time of day by itself as an effect (1) Cell phone use by itself as an affect(2) Interaction: How time of day (1) and cell phone use (2) interact to affect driving performance Driving performance is overall Lower at nighttime, with both of The effects taken into mind, and cell phone use and time of day interact to affect driving performance fferent types of possible interactions fferent types of possible interactions Description Correlational studies with factorial designs Factorial design with non-manipulated independent variables equivalent to a __________________. Why? Example: Non-manipulated IV Mood and Self-esteem Measure these variables and categorize participants according to level Mood (positive, negative) Self-esteem (high, low) DV = Willingness to have unprotected sex Assessing relationships among multiple variables Examine relationships between several variables (categorical and quantitative) Observations: Assessing relationships among multiple variables Factor analysis: a statistical technique that groups variables into “clusters”; reduces variables into factors Clusters: Examples of psychological factors derived from a large set of measurements Mathematical Intelligence (arithmetic, quantitative estimation, spatial reasoning, etc.) Verbal Intelligence (grammar, reading comprehension, vocabulary, etc.) Big Five Personality Factors Extraversion (warmth, gregariousness, activity level) Exploring Causal Relationships Can examine possible causal relationships in complex correlational studies by controlling possible third variables Multiple regression: a statistical model used to explain the relationship between one continuous dependent variable and two or more independent variables. Examples of questions that can be answered by Multiple Regression Do age and IQ scores effectively predict GPA? Do weight, height, and age explain the variance in cholesterol levels? Sample output of multiple regression analysis (important to understand how to read these tables!)