Chapter 7: Single Factor Designs

Exam I Results

Problems with Biasing Subject bias Hawthorne effect Effect of knowing one is in a study “Good” subjects Participants tend to be cooperative, to please the researcher Evaluation apprehension Participants tend to behave in ideal ways so as not to be evaluated negatively Demand characteristics Cues giving away true purpose and study’s hypothesis Controlling for participant bias Effective deception Use of manipulation checks Field research

Ethical Responsibilities of Participants Be responsible Show up for scheduled appointments, or inform research of cancellation Be cooperative Behave professionally when participating in research Listen carefully Ask questions if unsure of your rights or of what you are asked to do Respect the researcher Do not discuss study with others Be actively involved in debriefing Help the researcher understand your experience

Notes about counterbalancing Reverse counterbalancing, the experimenter simply presents the conditions in one order, and then presents them again in the reverse order. A B C D D C B A Block Counterbalance: every condition occurs once before any condition is repeated a second time. C D A B A C B D

Chapter 7. Experimental Design I: Single-Factor Designs Chapter Objectives Identify and understand the defining features of the four varieties of single-factor designs Know when to use an independent samples t test and when to use a dependent samples t test Describe two different reasons for using more than two levels of an independent variable Decide when to use a bar graph versus a line graph to present data

Chapter Objectives Understand why a 1-way ANOVA is the appropriate analysis when examining data from single-factor, multilevel studies Know when to use a one-way ANOVA for independent groups versus a one-way ANOVA for repeated measures Construct an ANOVA source table for a 1-factor ANOVA for an independent groups design

Chapter Objectives Understand why post hoc statistical analyses typically accompany 1-factor ANOVAs for single- factor, multilevel studies Understand the logic behind the use of three special types of control groups: placebo, wait list, yoked Understand the ethical issues involved when using certain types of control groups

Single-Factor—Two Levels Between-subjects, single factor designs Independent groups designs Manipulated independent variable (separate groups) Random assignment to create equivalent groups Matched Groups Designs Manipulated independent variable (separate groups) Matching to produce equivalent groups Nonequivalent groups design (ex post facto designs) Subject variable as an independent variable Deliberate attempts to select Ss to reduce nonequivalence

Single-Factor—Two Levels Within-subjects, single factor designs Also called repeated measures designs Manipulated independent variable (all Ss participate in all levels of the independent variable) Famous historical example  Stroop

Single-Factor—Two Levels

Single-Factor—Two Levels Analyzing single-factor, two level designs t test assumptions Interval or ratio scale data Data normally distributed Homogeneity of variance t test for independent samples, for Independent groups designs Nonequivalent groups designs t test for dependent samples (paired, repeated measures) for Matched groups designs Repeated measures designs

Single-Factor—More Than Two Levels Between-subjects, multilevel designs Advantage #1  ability to discover nonlinear effects RT study with 2 levels (1 and 3 mg of caffeine) Adding levels (2 and 4 mg)  possible nonlinear effect

Another nonlinear example…

Single-Factor—More Than Two Levels Between-subjects, multilevel designs Advantage #2  ability to rule out alternative explanations

Multilevel Designs If the balloons popped, the sound wouldn’t be able to carry, since everything would be too far away from the correct floor. A closed window would also prevent the sound from carrying, since most buildings tend to be well insulated. Since the whole operation depends on a steady flow of electricity, a break in the middle of the wire would also cause problems. Of course, the fellow could shout, but the human voice is not loud enough to carry that far. An additional problem is that a string could break on the instrument. Then there could be no accompaniment to the message. It is clear that the best situation would involve less distance. Then there would be fewer potential problems. With face to face contact, the least number of things could go wrong. (Bransford & Johnson, 1972, p. 392)

Single-Factor—More Than Two Levels Left: context sketch Right: partial context sketch

Single-Factor—More Than Two Levels Between-subjects, multilevel designs Effects of practice ruled out (1 rep = 2 reps) Context has to accurately reflect content (“partial context” condition poor) Context must be there when studying content (“context after” condition poor)

Single-Factor—More Than Two Levels Within-subjects, multilevel designs Research Example: Debunking the Mozart effect Multilevel repeated measures IV  listening experience Listening to Mozart Listening to gentle rainstorm Control – no listening DV  recall of digits Results  No “Mozart” effect Significant practice effect instead

Single-Factor—More Than Two Levels Presenting the data Sentence and paragraph form Table form (e.g., for the balloon study)

Single-Factor—More Than Two Levels Presenting the data Graph form Continuous variable – unlimited intermediate values exist e.g., drug dosage level Line graph preferred, but bar graph OK Discrete variable – no intermediate values e.g., the five levels of the context experiment Use a bar graph, line graph inappropriate, for example:

Single-Factor—More Than Two Levels Analyzing single-factor, multilevel designs Multiple t tests inappropriate Increases chances of Type I error 1-factor Analysis of Variance (ANOVA) “1-factor” = 1 IV “2-factor” = 2 IVs (factorial design – Chapter 8) Once overall significant effect found, then post hoc testing Comparing each level of IV against each other level

Special-Purpose Control Group Designs Placebo control groups Placebo – inactive substance Ss think they are being treated but they are not Placebo effect When performance of placebo group = experimental group, Ss expectations explain the effect of treatment Wait list control groups To insure equivalent groups in a study of program effectiveness Wait list group later administered treatment only if shown to be effective (unethical to deny treatment)

Special-Purpose Control Group Designs Research example 15: Placebo + Wait List IV  exposure to subliminal recordings Experimental  weight loss recording Placebo control  dental pain recording (but told was weight loss tape) Waiting list control  no tape until wait was over Double blind procedure used DV  weight loss Results  equal amounts of weight loss for all three groups

Special-Purpose Control Group Designs Yoked control groups Each control group subject “yoked” to an experimental group subject. In experimental designs in which members of an experimental group and a control group are paired, the yoked control group members receive the same stimuli, reinforcements, or punishments as the experimental group members but without the possibility of influencing these effects through their own behavior. Example: Reward and test performance.

Summary Depending on your empirical question, you may choose which type of design to use: Between-subjects vs. within-subjects Single factor, two level vs. Single factor, multi-level Special-purpose control group designs Depending on the type of design, you will choose the appropriate statistical test to test your hypothesis e.g., independent samples t-test, dependent samples t-test, 1-way ANOVA + post-hoc tests Once you have your results, you share them with others both in writing and in visual form (tables, graphs)