Experimental Research Neuman and Robson Ch 9 Introduction

Experimental Research Neuman and Robson, Ch. 9

Introduction l l Experiments are part of the traditional science model Involve taking “action” and observing consequences of this action Can collect data using rigorous control A good example is Albert Bandura’s Bobo doll study: “Transmission of Aggression Through Imitation of Aggressive Models” (1961) l l Click on the following link to read this classic paper: http: //psychclassics. yorku. ca/Bandura/bobo. htm

Topics Appropriate for Experiments l l Excellent for hypothesis testing For explanatory research Small group interaction Two types: l Laboratory settings (traditional) l l Used in psychology Natural settings (field experiments) l More typical in sociology

The classical experiment l Three pairs of components l Independent and dependent variables l l l Pre-testing and post-testing l l IV is manipulated DV is observed for change To compare variation in DV before and after treatment Experimental and control groups l Experimental group receives “treatment” and is compared to control group (no treatment

The Double-Blind Experiment l Neither researchers or subjects know who is experimental group l “blinding the study” l To reduce experimental bias

Selecting Subjects l l Generalizability important Probability sampling where possible Random assignment essential to ensure no difference between experimental and control groups Sometimes “matching” used l l Individuals are matched on important characteristics Problem of “volunteer” subjects

The Classical Design and Variations l “Pre-experimental” Designs (no random assignment l l A. One shot case study B. One group pre-test-post-test design C. Static group comparison “True Experimental” Designs (random assignment) l l l D. Pretest-Post-test control group design E. Post-test only control group design F. Solomon Four-Group design

“Pre-experimental” Designs l l No random assignment used A. One shot case study: Tx (IV) l B. One group pre-test-post-test design: Obs 1 l Obs (DV) Tx Obs 2 C. Static group comparison: Exp. Grp. Ctrl. Grp. Tx Obs

“True Experimental” Designs l l Use random assignment D. Pretest-Post-test control group design (classical experimental design): Exp. Ctrl. l Obs 1 Tx Obs 2 E. Post-test only control group design Exp. Ctrl. Tx Obs

“True Experimental” Designs (cont. ) l F. Solomon Four-Group design (Bandura’s Design): Exp. Ctrl. Obs 1 Tx Tx Obs 2

Problems of Internal Validity l Is change in DV actually caused by IV? l A. Problems concerning Subjects l l 1. Selection bias 2. Experimental Mortality 3. Rivalry 4. Demoralization

Problems (cont. ) l B. Problems with experimental procedure l l l 1. Testing 2. Instrumentation 3. Causal time order 4. Diffusion 5. Compensation to control group C. Problems with time l l History Maturation

Other Problems in Experimental Research l l l Interaction of selection and other factors Problem of statistical regression External validity or generalizability Hawthorne Effect Modelling Effect Sample problems l Volunteers vs non-volunteers

The Hawthorne Effect: l l l The Hawthorne effect – refers to change in behaviour produced by the “stimulus” of being singled out and made to feel important during the experimental process. First observed 1925 -1928 at Hawthorne plant in Illinois during series of experiments on effects of lighting on worker productivity Explanation for increase provided by Professor Elton Mayo (“founder” of human relations theory)

The Hawthorne Experiments l Four experiments: l l A. Used Three experimental groups and no control group. Found productivity increased when lighting improved. B. Experimental and control group used. Experimental group had increased lighting. Both groups showed increased productivity. C. Experimental and control group used. Experimental group had decreased lighting. Both groups showed increased productivity. D. Two women studied under varying lighting conditions. Subjects increased productivity according to which lighting conditions (more or less) researchers said were “good”

Strengths and Weaknesses of Experimental Design l Strengths l l l Establishing causality Control Longitudinal research Replication Weaknesses l l l Artificiality Experimenter effect Sample size

The Field Experiment l l Used in sociology Natural setting Uses statistical control For an excellent example of a field experiment, see The Impact of television: a natural experiment in three communities by Tannis Macbeth Williams available in library. Call # PN 1992. 6. I 5

Strengths and Weakness of Field Experiments l Strengths l l l Takes place in “real world” Very good generalizability Weaknesses l l Lack of internal validity Difficult to establish causality Time consuming Subject to researcher bias