Research designs have specific objectives they strive for have different strengths and limitations have validity considerations 3 Experimental Research Tries to Establish Cause and Effect. Internal Validity is threatened whenever there exists the possibility of un-controlled extraneous variables. External validity trade-off between internal and external validity.
Research designs have specific objectives they strive for have different strengths and limitations have validity considerations 3 Experimental Research Tries to Establish Cause and Effect. Internal Validity is threatened whenever there exists the possibility of un-controlled extraneous variables. External validity trade-off between internal and external validity.
Research designs have specific objectives they strive for have different strengths and limitations have validity considerations 3 Experimental Research Tries to Establish Cause and Effect. Internal Validity is threatened whenever there exists the possibility of un-controlled extraneous variables. External validity trade-off between internal and external validity.
Experimental Research Ex Post Facto 2 Types of research design experiments Chapter 8 in Babbie & Mouton (2001) Introduction to all research designs All research designs have specific objectives they strive for Have different strengths and limitations Have validity considerations 3 Experimental Research Tries to Establish Cause and Effect Selection of a good theoretical framework Application of appropriate experimental design Use of correct statistical model and analysis Proper selection and control of independent variables Appropriate selection and measurement of dependent variables Correct interpretation of results 3 4 Three Criteria for Cause and Effect 1. The cause must precede the effect in time 2. The cause and effect must be correlated with each other 3. The correlation between cause and effect cannot be explained by another variable 4 5 The logic of causal social research in the controlled experiment Explanatory rather than descriptive Different from correlational research - one variable is manipulated (IV) and the effect of that manipulation observed on a second variable (DV) If then . E.g. "Animals respond aggressively to crowding" (causal) "People with long engagement experience have more stable marriages" (noncausal) 6 Reviewing Important Terms Independent variable Dependent variable Categorical variable-(age, sex=not controlled) Control variable-exclude Extraneous variable
6 7 Types of Validity Internal validity External validity Trade-off between internal and external validity Series of experiments (studies) 7 INTERNAL VALIDITY Internal validity is threatened whenever there exists the possibility of un-controlled extraneous variables that might otherwise account for the results of a study 8 9 Threats to Internal Validity History-during Maturation Testing Instrumentation Statistical regression Selection bias 9 10 Threats to Internal Validity, contd Experimental mortality Selection-maturation interaction Expectancy 10 11 Threats to External Validity Reactive or interactive effects of testing Interaction of selection biases and treatment Reactive effects of experimental arrangements Multiple-treatment interference
11 12 Controlling Threats to Internal Validity Randomization Real randomization Matched pairs (but not matched groups) Randomizing treatments or counterbalancing Placebos Blind setups Double-blind setups 12 13 Uncontrolled Threats to Internal Validity Reactive effects of testing Instrumentation Experimental mortality 13 EXTERNAL VALIDITY Concerned with whether the results of a study can be generalized beyond the study itself 14 15 Controlling Threats to External Validity Selecting from larger population Participants Treatments Situations Ecological validity-emulates the real world? 15 Types of design - experiments 16 The generic experimental design: R O 1 X O 2
R O 3 O 4
The IV is an active variable; it is manipulated The participants who receive one level of the IV are equivalent in all ways to those who receive other levels of the IV 17 Types of Designs: Pre-experimental Designs (invalid) One-shot studies T O
One-group pretest-posttest O 1 T O 2 Statistical analysis? Dt
Static group comparison (one shot w/2 groups) T O 1 --------------- Statistical analysis? It O 2 18 Types of Designs: True Experimental Designs Randomized-groups design R T O 1 Statistical analysis? It R O 2
Extending the levelsrandomized-groups design R T 1 O 1 Statistical analysis? Ab R T 2 O 2
R O 3
19 Types of Designs: True Experimental Designs, contd -1 interaction -2 main effects 20 Types of Designs: True Experimental Designs, contd A factorial design with a categorical factor (B) R A 1 O 1
B 1 R A 2 O 2
R A 3 O 3 Statistical analysis? M ---------------------------- R A 1 O 4
B 2 R A 2 O 5
R A 3 O 6
Types of Designs: True Experimental Designs Pretest-posttest randomized-groups R O 1 T O 2
R O 3 O 4 Statistical analysis? MFA
Extending the design on the RM factor R O 1 T O 2 T O 3 Statistical analysis? MFA-RM R O 4 O 5 O 6
21 22 Types of Designs: True Experimental Designs, contd Extending the pretest-posttest randomized groups design on both factors R O 1 T 1 O 2 T 1 O 3 R O 4 T 2 O 5 T 2 O 6 R O 7 O 8 O 9 Statistical analysis? AN (covariate)
23 Types of Designs: True Experimental Designs, contd Solomon four-group designpurpose=pretest effects R O 1 T O 2
R O 3 O 4
R T O 5
R O 6
Statistical analysis (factorial ANOVA) No treatment Treatment Pretested O 4 O 2
Unpretested O 6 O 5
24 Quasi-Experimental Designs: Time Series (slope) D.T. Campbell and J.C. Stanley, Experimental and quasi-experimental designs for research. Copyright 1963 by Houghton Mifflin Company. Reprinted with permission. 25 Quasi-Experimental Designs: Reversal (return to baseline) 26 Quasi-Experimental Designs: Ex Post Facto (discriminate) This is one of the pre-experimental designs, but with the treatment not under the control of the experimenter. T O 1
------------------ Statistical analysis? A O 2
27 Quasi-Experimental Designs: Switched Replication Trials Levels 1 2 3 4 5 1 O 1 T O 2 O 3 O 4 O 5
2 O 6 O 7 T O 8 O 9 O 10
3 O 11 O 12 O 13 T O 14 O 15
4 O 16 O 17 O 18 O 19 T O 20
28 Quasi-Experimental Designs: Single Participant Identify participant and follow over time. Does the treatment produce the same effect each time? Are treatment effects cumulative, or does participant return to baseline? Does participants response become less variable over treatment times? 29 Quasi-Experimental Designs: Single Participant, contd Is participants magnitude of response sensitive to multiple treatment applications? Do varying intensities, frequencies, and lengths of treatment produce varying responses? Ex Post Facto How it differ from experimental research? 30 Purpose of causal comparative research Attempts to determine cause for Existing conditions Preexisting differences in groups Alleged cause and effect have already occurred Orientations Retrospective (basic): starts with an effect and seeks possible causes Prospective (variation): starts with a cause and investigates its effects on some variable Causal-comparative (ex post facto) research The independent variable (IV) is not manipulated; it has already occurred Independent variables sometimes called attribute variables Less costly and time-consuming to conduct Establishing cause-effect relationships is more difficult than in experiments Procedures in causal- comparative research Identify an existing condition or event (e.g., differences in socialization among 1st grade students) Look backwards to see what may have caused this difference/condition to occur (i.e., some attended preschool, some did not) Rule out other causal factors Sometimes confused with correlational research: Both lack manipulation of variables Both require caution in interpreting results Both can support subsequent experimental research Causal comparative vs. correlational research Causal comparative Attempts to identify cause-effect relationships At least one independent variable Two or more groups Involves a comparison Correlational No attempt to understand cause and effect Two or more variables Only one group Sometimes confused with experimental research: Both try to establish cause-effect relationships Both can test hypotheses concerning the relationship between an independent (X) and a dependent variable (Y) Both involve group comparisons
Comparison to experiments Causal comparative Individuals already in groups before study begins Independent variable has already occurred Independent variable is not manipulated Cannot be Should not be Could be, but is not Experiment Individuals randomly assigned to groups (e.g., treatment or control) Independent variable manipulated by the researcher Examples of non-manipulated independent variables Age Sex Ethnicity Learning style Socioeconomic status (SES) Parent educational level Family environment Type of school attended Design of causal-comparative research Select 2 groups that differ on some IV One group possesses a characteristic that the other does not Each group possesses the characteristic, but in differing amounts Randomly sample Ss from each group Collect info on Ss to determine equality of the groups Compare groups on the DV Difficulty in interpreting findings Establishing cause and effect requires caution! Alternative explanations: Different causal variable Order of causation Reverse causality Order of occurrence Evidence necessary to demonstrate that X causes Y: Establish statistical relationship between X and Y (i.e., correlational research); determine that X precedes Y in time (collect data over time, i.e., longitudinal research); demonstrate that other, unknown factors did not determine the dependent variable (i.e., experimental research).
Becker & Gersten (1982): Effects of Project Follow-Through
Quasi-experimental study
Ex post facto study
Problem: Are the two groups in this study comparable to one another? In order to make sure that the two groups are comparable, and to ensure that the only post-test differences between the groups are due to the independent variable (the Follow-Through intervention), data were obtained on students: family income gender language spoken in home mothers education ethnicity number of siblings.
Research Design FOLLOW-THROUGH Year 1 (1975) Gr 5 Site 1 Site 2 Site 3 Site 4 Site 5 Year 2 (1976) Gr 6 Site 1 Site 2 Site 3 Site 4 Site 5 NO FOLLOW-THROUGH Year 1 (1975) Gr 5 Site 1 Site 2 Site 3 Site 4 Site 5 Year 2 (1976) Gr 6 Site 1 Site 2 Site 3 Site 4 Site 5
Dependent variables Wide-Range Achievement Test (WRAT) reading mathematics Metropolitan Achievement Test (MAT) reading mathematics RESULTS A total of 180 comparisons of FT to No-FT students. Of these, only 56 (31%) favored FT students! Largest differences between FT and No- FT students were in basic skills areas. FT students achievement declined by grades 5 and 6 (2-3 years after end of FT). Critique of this research What are the strengths of the study? Groups are comparable to one another. Contrasted statistical with practical significance. Large sample size. Multiple replications of treatment effect. What are the weaknesses of the study? Lack of random assignment. Focus on standardized test performance. You have to creatively design your experiment!!! 48 49 What about Sampling??? 1. Selecting subjects to participate in the research Careful sampling to ensure that results can be generalized from sample to population The relationship found might only exist in the sample; need to ensure that it exists in the population Probability sampling techniques 50 Sampling 2. How the sample is divided into two or more groups is important to make the groups similar when they start off randomization - equal chance matching - similar to quota sampling procedures match the groups in terms of the most relevant variables; e.g. age, sex, and race Make sure that your measurement is VALID & RELIABLE 51 NEXT WEEK: VALIDITY & RELIABILITY 52