Chapter 8Chapter 8Experimental Design: Experimental Design:
Dependent Groups and Dependent Groups and Mixed Groups DesignsMixed Groups Designs
Dependent Groups DesignsDependent Groups Designs
Matched DesignsMatched Designs Within-Participants DesignsWithin-Participants Designs
– More than one IV – WP Factorial Design More than one IV – WP Factorial Design
Repeated Measures/With-Participants Repeated Measures/With-Participants DesignDesign
Each participant is his or her own control.Each participant is his or her own control.– Increased statistical power (the likelihood of detecting Increased statistical power (the likelihood of detecting
an effect if one is present).an effect if one is present).– More economical since fewer subjects are neededMore economical since fewer subjects are needed
Often the DV is assessed at multiple time points Often the DV is assessed at multiple time points such as before and after a “treatment”such as before and after a “treatment”– Repeated MeasuresRepeated Measures
Repeating treatment conditions enables the Repeating treatment conditions enables the participants to identify what is being participants to identify what is being manipulated.manipulated.– Demand CharacteristicsDemand Characteristics
Concern over demand characteristics prevents more widespread Concern over demand characteristics prevents more widespread use of repeated measures designsuse of repeated measures designs
Repeated Measures/Within-Repeated Measures/Within-Participants DesignParticipants Design
The potential for increased economy and statistical The potential for increased economy and statistical power is weighed against the potential threats to power is weighed against the potential threats to internal validityinternal validity– Confounds - conditions that vary systematically with Confounds - conditions that vary systematically with
changes in the level of the independent variablechanges in the level of the independent variable When present in an experiment, it is impossible to When present in an experiment, it is impossible to
tell whether changes in the dependent variable tell whether changes in the dependent variable resulted from the different levels of the resulted from the different levels of the independent variable or from the different levels of independent variable or from the different levels of the confounded variable.the confounded variable.
– HistoryHistory– MaturationMaturation– TestingTesting
HistoryHistory Anything that happens between the pretest Anything that happens between the pretest
and posttest that is not part of the and posttest that is not part of the experimental situationexperimental situation
The longer the interval between pretest and The longer the interval between pretest and posttest, the greater the potential for history posttest, the greater the potential for history effecteffect
It is important to know that this is not It is important to know that this is not specifically time passage, but specifically time passage, but eventsevents that that occur during that timeoccur during that time– Equipment issuesEquipment issues– Life eventsLife events
To Minimize History EffectsTo Minimize History Effects
Shorten the interval between the Shorten the interval between the pretest and posttest, so things will pretest and posttest, so things will not happennot happen
Control the environment the pretest Control the environment the pretest and posttest as much as possibleand posttest as much as possible
MaturationMaturation
Internal processes that occur as a Internal processes that occur as a function of the passage of timefunction of the passage of time– Growth and aging processesGrowth and aging processes– Motivational effects such as practice and Motivational effects such as practice and
fatiguefatigue Especially a problem in:Especially a problem in:
– Longitudinal researchLongitudinal research– Educational researchEducational research– Therapy researchTherapy research
To Minimize Maturation To Minimize Maturation EffectsEffects
Minimize the interval of time Minimize the interval of time between pretest and posttestbetween pretest and posttest
Keep all experimental conditions Keep all experimental conditions identical during pretest and posttest identical during pretest and posttest
TestingTesting
Taking a test once affects scores on Taking a test once affects scores on the second testthe second test– Taking a pretest affects scores on a Taking a pretest affects scores on a
posttestposttest StroopStroop
To Minimize Testing EffectsTo Minimize Testing Effects
Use alternate forms, if availableUse alternate forms, if available Lengthen the interval between Lengthen the interval between
pretest and posttestpretest and posttest
Determining the Levels of Determining the Levels of the Factorthe Factor
There should be enough levels to There should be enough levels to represent the range of values of the represent the range of values of the treatment variable.treatment variable.
There should be enough levels to show the There should be enough levels to show the exact nature of the relationship being exact nature of the relationship being tested.tested.
A true control condition is one in which the A true control condition is one in which the treatment variable is absent.treatment variable is absent.
Certain problems arise when manipulating Certain problems arise when manipulating the IV – Carry Over Effectsthe IV – Carry Over Effects
Carry Over EffectsCarry Over Effects
Because each participant Because each participant experiences multiple treatment experiences multiple treatment combinations, experiencing the early combinations, experiencing the early treatments can affect responses to treatments can affect responses to subsequent treatments subsequent treatments – Also called “transfer effects”Also called “transfer effects”
Order EffectsOrder Effects Differential Order EffectsDifferential Order Effects
Dealing with Order EffectsDealing with Order Effects
Randomly determine the order of Randomly determine the order of treatmentstreatments– Difficult to do unless you have a large number Difficult to do unless you have a large number
of participants.of participants. Completely counterbalanced approachCompletely counterbalanced approach
– Requires that each condition occurs equally Requires that each condition occurs equally often, and precedes and follows all other often, and precedes and follows all other conditions the same number of times.conditions the same number of times.
Incomplete counterbalancingIncomplete counterbalancing– Requires that each condition occurs equally Requires that each condition occurs equally
often.often.
CounterbalancingCounterbalancing
ParticipantParticipant Order of ConditionsOrder of Conditions
11 LALA MAMA HAHA
22 MAMA HAHA LALA
33 HAHA LALA MAMA
44 LALA HAHA MAMA
55 MAMA LALA HAHA
66 HAHA MAMA LALA
Differential Order EffectsDifferential Order Effects
When some levels of the IV may When some levels of the IV may irreversibly influence the DVirreversibly influence the DV– Certain orders may permanently change Certain orders may permanently change
the individuals the individuals Example – A teaching technique Example – A teaching technique
Cannot be controlled by Cannot be controlled by counterbalancingcounterbalancing– Need to do a Between-Subjects DesignNeed to do a Between-Subjects Design
Repeated Measures DesignsRepeated Measures Designs
ADVANTAGESADVANTAGES Require fewer subjectsRequire fewer subjects
Take less time to completeTake less time to complete
More powerful than More powerful than between subjects designs between subjects designs because each participant is because each participant is compared with him/herselfcompared with him/herself– Error variance is reducedError variance is reduced
DISADVANTAGESDISADVANTAGES Problems posed by history, Problems posed by history,
testing, and maturationtesting, and maturation Problems posed by Problems posed by
multiple treatment multiple treatment interference effectsinterference effects– Differential Order EffectsDifferential Order Effects
Example – THC, Alcohol & Example – THC, Alcohol & DrivingDriving
Many studies have shown that both Many studies have shown that both THC and alcohol impair driving ability.THC and alcohol impair driving ability.
No studies have compared the two No studies have compared the two drugs, nor have any studies examined drugs, nor have any studies examined subjective experiences of the drugs subjective experiences of the drugs while drivingwhile driving
How do we do this study with a within How do we do this study with a within subjects design?subjects design?
Starting with the TitleStarting with the Title
Data Analysis - Partitioning the Data Analysis - Partitioning the VarianceVariance
The treatment effect is estimated The treatment effect is estimated from differences within subjects from differences within subjects rather than between subjects.rather than between subjects.
Between subjects variance reflects Between subjects variance reflects differences between subjects, NOT differences between subjects, NOT due to the treatment.due to the treatment.
Analysis of Variance Summary TableAnalysis of Variance Summary Table
Between subjects variance is listed first Between subjects variance is listed first and then removed from further and then removed from further considerationconsideration
Within subjects variance is partitioned Within subjects variance is partitioned into the variance due to the treatment into the variance due to the treatment and error variance (variance due to and error variance (variance due to chance factors).chance factors).
The The FF ratio compares the variance due to ratio compares the variance due to the treatment (numerator) to the error the treatment (numerator) to the error variance (denominator).variance (denominator).
Interpretation of the Results of ANOVAInterpretation of the Results of ANOVA
Calculated Calculated FF is compared to critical is compared to critical FF::If calculated f is equal to or greater If calculated f is equal to or greater than critical than critical F:F:
FF is significant is significant
There is a significant difference There is a significant difference among the means of the different among the means of the different treatment levelstreatment levels
Get it?Get it?
Data AnalysisData Analysis
ResultsResults
DiscussionDiscussion Describes the outcome of the research in wordsDescribes the outcome of the research in words
– Briefly summarize what you foundBriefly summarize what you found Integrates the outcome of the study with Integrates the outcome of the study with
previous research findingsprevious research findings– Here is how you data fit in to the larger Here is how you data fit in to the larger
literature baseliterature base Draws conclusionsDraws conclusions
– Based on theory or practical applicationBased on theory or practical application May present suggestions for future researchMay present suggestions for future research
– LimitationsLimitations
Discussion – Explaining the Discussion – Explaining the effecteffect
Discussion – Implications & Discussion – Implications & Meaning Meaning
Discussion – Addressing a Discussion – Addressing a limitationlimitation
Discussion – A summary of Discussion – A summary of the results that happens to the results that happens to
be a conclusionbe a conclusion
Discussion – Addressing a Discussion – Addressing a limitationlimitation
Example – 2X2 WS FactorialExample – 2X2 WS Factorial
Question/ProblemQuestion/Problem
Increase in the coadministration of Increase in the coadministration of alcohol and caffeine (energy drinks) alcohol and caffeine (energy drinks) in college studentsin college students
Little research on the interactive Little research on the interactive effects.effects.
How does caffeine influence the How does caffeine influence the effects of alcohol on information effects of alcohol on information processing tasks?processing tasks?
MethodsMethods
Procedures & ResultsProcedures & Results
Mixed DesignsMixed Designs
Have at least two independent Have at least two independent variablesvariables
At least one variable is a between At least one variable is a between subjects factor.subjects factor.
At least one variable is a within At least one variable is a within subjects factor.subjects factor.
Reasons for Using Mixed Reasons for Using Mixed DesignsDesigns
Repeated measures factors are desirable Repeated measures factors are desirable because they require fewer participants because they require fewer participants and can take less time.and can take less time.
Some variables are manipulated between Some variables are manipulated between participants to reduce or prevent fatigue, participants to reduce or prevent fatigue, interference effects, or demand interference effects, or demand characteristics.characteristics.
Sometimes participant variables are Sometimes participant variables are studied studied (e.g., gender, handedness, etc.)(e.g., gender, handedness, etc.)
Example - Our THC and STM Example - Our THC and STM StudyStudy
Problem – The effects of THC Intoxication on Problem – The effects of THC Intoxication on the ability to do occupational tasks requiring the ability to do occupational tasks requiring STMSTM
Research Hypothesis – THC intoxication will Research Hypothesis – THC intoxication will impair STMimpair STM
IV – Three smoked THC dosesIV – Three smoked THC doses– 0%, 5%, 10%0%, 5%, 10%
DV – Span test for words at different time DV – Span test for words at different time intervalsintervals– 15min, 1hr and 3hrs15min, 1hr and 3hrs
DescriptivesDescriptives
Testing the WP FactorTesting the WP Factor
Testing Levels of WP FactorTesting Levels of WP Factor
Testing the BS FactorTesting the BS Factor
Post Hoc Test on the BS Post Hoc Test on the BS FactorFactor
Matched groups designsMatched groups designs Provides the added power of Provides the added power of
dependent groups designs and dependent groups designs and eliminates the problem of carryover eliminates the problem of carryover effects.effects.
Matching variablesMatching variables – participants are – participants are matched, or made equivalent, on matched, or made equivalent, on variables that correlate with the DV.variables that correlate with the DV.
By using appropriate matching By using appropriate matching variables the error variance due to variables the error variance due to individual differences is reduced and individual differences is reduced and power is increased.power is increased.