Psych 325 1st Edition Lecture 24 Chapter 9: Choosing the Right Research DesignPart III: Within-subjects Design-Within-Subjects DesignoEach condition has the same group of peopleoBig advantages-High statistical poserFewer participants neededNo group differences = less noise-Groups are perfectly equatedUltimate form of matchingNo person confoundsoBig problem-Sequence or carryover effects-Can use counter-balancing to address this problemoEach participant is her/his own control group (perfect matching!)-Still need a lot of participants to increase external validity-Similar to a between-groups design in that a one-way or factorial can be usedoUse if:-Studying change in time for the same participants (i.e. depression therapy)-Expect large individual differences (i.e. past sexual experience, extroversion, sex education)-None of the tasks has a permanent effect (i.e. adult scene not too stimulating)-Problems with Within-Subjects DesignsoSequence effects: fatigue or boredom affects later conditions -Reduces experimental realism-Participants are human: they can only do so much at one timeoCarryover effects: one condition affects response to others-Order effect: new condition takes on different meaning after prior conditionSee text ex: Stapel & Schwartz (1998) study of attitudes: Republican party (Powell-Dole) (p. 260)-Practice effect: enhances performance; like testing effectSee Strack (1988) study of dating frequency and life satisfaction (p. 260) -Dating items 1st = strong correlation/Dating items 2nd = correlation not significant-Interference effect: prior condition interferes with laterOne word answer task (p. 261)oParticipants may figure out experimenters' hypothesesoWear our participants by giving too many tasks (did the frustration or the IV cause the outcomes?)These notes represent a detailed interpretation of the professor’s lecture. GradeBuddy is best used as a supplement to your own notes, not as a substitute.-Order-the positioning of specific tasks overly influences the role of other tasks on outcomes-Practice: participants not used to certain tasks: takes time to figure out-One task has such a big effect that it screws up responses to other tasksoOverriding theme?-Giving too many tasks or specific orders of those tasks may limit the internal validity of the IV-DV relationship-Addressing Problems in Within-Subjects DesignoPre-training on an experimental task reduces practice effectsoRest periods reduce sequence effectsoCounter-balancing distributes carry-over effects evenly, but doesn't eliminate them-Systematically vary order of IV's-Complete counterbalancingEvery possible order usedImpossible if using lots of conditions-Incomplete CounterbalancingoReverse counter-balancing-Create a single order and its reverse-Problem: middle treatmentsoPartial counter-balancing-Randomly select a subset of possible orders-Problem: no guarantee of balanceoLatin-square design-Each treatment occurs with equal frequency in each position-# or orders = # of treatments-# or rows = # of columnsoReverse CB is the simplest kind of incomplete CB-Guarantees that average serial positions are equaloRandom selection approach: good but not perfectoWarning: Latin square does not work with an odd # of unique