Non-true-experimental Designs, cont. PSY 231 Research Methods in Psychology
Announcements Journal Article Summary #2 (or research participation alternative) due in labs this week bring results from group projects to lab GAs will go over plans of analyses with you
today Chapter 12—The Correlational Research Strategy Chapter 13—The Descriptive Research Strategy Chapter 10—The Nonexperimental and Quasi- Experimental Strategies: Nonequivalent Group, Pre-Post, and Developmental Designs Chapter 14—Single-Subject Research Designs Chapter 15—Statistical Evaluation of Data
balance of the validities internal—extent to which a research study produces a single, unambiguous explanation for the relationships between two variables external—extent to which we can generalize the results of a research study to people, settings, times, measures, and characteristics other than those in that study balance depends on level of experimental control
threats to internal validity assignment bias history maturation instrumentation testing effects
nonexperimental research vs. quasi-experimental research nonexperimental—intended to demonstrate a relationship between variables but does not attempt to explain quasi-experimental—uses some rigor and control of true experiments but contains a flaw that prevents cause-effect remember—true experiments include a control group manipulation of independent variable(s) random assignment to groups
nonexperimental research vs. quasi-experimental research both use comparisons between groups or conditions both use nonmanipulated independent variables attempts to control internal validity threats nonexperimental—no quasi-experimental—yes
nonexperimental research vs. quasi-experimental research two types between-subjects designs nonequivalent groups within-subjects designs pre-post designs
nonequivalent groups designs posttest-only nonequivalent groups assignment bias, no control of time effects nonexperimental participants program measure participantsno program measure
example smoker participants training program measure smokersno training program measure
nonequivalent groups designs pretest-posttest nonequivalent control design assignment bias, some control of time effects quasi-experimental participantsmeasureprogrammeasure participantsmeasureno programmeasure
example no t.v. aggression aggression participantsmeasure t.v.measure t.v. aggression aggression participantsmeasure (no change)measure
pre-post designs one-group pretest-posttest design little control of time effects nonexperimental participantsmeasureprogrammeasure
example smoker participantsmeasureprogram measure
pre-post designs time-series design some control of time effects quasi-experimental measuremeasuremeasuremeasure program measuremeasuremeasuremeasure
example
developmental designs nonmanipulated independent variable—age or time three types cross-sectional longitudinal sequential
cross-sectional design different participants, at different ages, studied at same time year of birth age at study year of study2008
cross-sectional design advantages time efficient no long-term cooperation disadvantages no individual changes cohort effects year of birth age at study year of study2008
longitudinal design same participants, at different ages, studied at different times year of birth1988 age at study year of study
longitudinal design advantages no cohort effects assesses individual changes disadvantages time consuming long-term cooperation year of birth1988 age at study year of study
sequential design different participants, at different ages, studied at different times age at study year of study
single-subject designs originally used in behavior modification often seen in clinical and applied settings not used very often in traditional research
single-subject designs different from traditional designs usually conducted with one person or a small group much more flexible require continuous assessment
single-subject designs baseline observations—no treatment administered baseline phase—series of baseline observations treatment observations—treatment administered treatment phase—series of treatment observations
reversal design also called ABAB design behavior is observed during baseline phase (A) treatment phase (B) baseline phase (A) treatment phase (B)
reversal design A B A B
multiple baseline designs reversal designs may not work need to show treatment is effective across circumstances multiple baseline across subjects behaviors situations
single-subject designs advantages establish cause-effect between treatment and behavior integrate experimental research with clinical practice allows flexibility
single-subject designs disadvantages cause-effect established for only one person needs multiple observations no statistical control
feedback thinking about my 2 lectures… please write down 1 thing you liked please write down 1 thing that you didn’t like/need to improve turn it in on the table along with Rating Sheet #1