Alternative Designs

8/3/2019 Alternative Designs

1/13

1

Alternative Research Designs

When the usual crap just isnWhen the usual crap just isnt goodt good

enough.enough.

Alternative Research Designs

I. Quasi-ExperimentsI. Quasi-Experiments

II. Single Group DesignsII. Single Group Designs

III. Case StudiesIII. Case Studies

I. Quasi-Experiments

A. When true experiments cannot be conductedA. When true experiments cannot be conducted

- naturally occurring events- naturally occurring events

- limited pool of participants- limited pool of participants

B. When possible use true experimental designsB. When possible use true experimental designs

to prevent threats to validity:to prevent threats to validity:

e.g., pre-test, post-test with a control groupe.g., pre-test, post-test with a control group

Solomon-four groups designSolomon-four groups design


2/13

2

C. The trade-off between internal

and external validity

High Internal

Validity

High External

Validity

True

Exp

erim

ents

Natur

alisti

c

Obs

erva

tion

s

Quasi

-Exp

erim

ents

Correla

tional

Studie

s

D. Quasi-Experimental Designs(in increasing levels of internal validity)

1. Single Pretest-Posttest design:1. Single Pretest-Posttest design:

OOXXOO

O= ObservationO= Observation

XX= Experimental Manipulation= Experimental Manipulation

Example: Quaker OatmealExample: Quaker Oatmeal

Worst possible designWorst possible design


3/13

3


2. Interrupted Time Series:2. Interrupted Time Series:

0000000000XX0000000000

Use this when you donUse this when you dont have complete controlt have complete control

over when the manipulation occurs.over when the manipulation occurs.

Example: boxing and homicide ratesExample: boxing and homicide rates


3. Random time series3. Random time series

OOOOOOOOOOOOXXOOOOOOOOOOOO

Example: Diet and CholesterolExample: Diet and Cholesterol

Time of manipulation randomly determined

Diet Example

130

150

170

190

210

230

250

1 2 3 4 5 6 7 8 9 10 11 12

Cholesterol

Level

Week

Start Program


4/13

4


4. Nonequivalent Control Group (before-after)4. Nonequivalent Control Group (before-after)

DifferenceDifference

Experimental:Experimental: OO11 X OX O22 OO11-O-O22

Control:Control: OO11 OO22 OO11-O-O22

Example: Diet and CholesterolExample: Diet and Cholesterol

Example: Nonequivalent Control

Group (before-after)HyggeHygge, Evans, &, Evans, & BullingerBullinger(2002)(2002)

Investigating cognitive abilities in children (n= 326)Investigating cognitive abilities in children (n= 326)before and after the opening of the new Munichbefore and after the opening of the new MunichAirport.Airport.

Experimental GroupExperimental Group:: children who were (old airport),children who were (old airport),or would be (new airport) exposed to aircraft noiseor would be (new airport) exposed to aircraft noise(p. 469).(p. 469).

Control GroupControl Group:: matched with their respectivematched with their respective

experimental groups on the basis ofexperimental groups on the basis ofsociodemographicsociodemographic characteristics.characteristics.

Number of errors on the reading word list in each of

the groups before and after the airport opened


5/13

5

II. Single Group Designs

A. Reversal Designs: ABABA. Reversal Designs: ABAB

A: baseline periodA: baseline period

B: treatment periodB: treatment period

1) treatments must have1) treatments must have locallocal effectseffects

2) rules out history, maturation, testing, and2) rules out history, maturation, testing, and

instrumentation threats to validity.instrumentation threats to validity.

Example: Kirby & Shields (1972)Example: Kirby & Shields (1972)

Example: Kirby & Shields (1972)

Subject: 13 year old boySubject: 13 year old boy

Design: ABABDesign: ABAB

A: worksheet collected, graded, and returnedA: worksheet collected, graded, and returned

toe following daytoe following day

B: when student finished, he took hisB: when student finished, he took his

worksheet to the instructor who immediatelyworksheet to the instructor who immediately

graded it, and then praised his performance.graded it, and then praised his performance.

Example: Kirby & Shields (1972)


6/13

6


A. Reversal Designs (cont.)A. Reversal Designs (cont.)

Modifications:Modifications:

ABACABAC

AB ----AB ----


B. Multiple-Baseline Designs:B. Multiple-Baseline Designs:

use when the initial state cannot be recovereduse when the initial state cannot be recovered

1) Choose 2 or more:1) Choose 2 or more:

settings, behaviors, materials . . .settings, behaviors, materials . . .

2) Insert manipulation in each setting at different times2) Insert manipulation in each setting at different times

Example: structured lessons and contingent free timeExample: structured lessons and contingent free time

(Long & Williams, 1973)(Long & Williams, 1973)

(Long & Williams, 1973)

Compared the influence of structured lessons,Compared the influence of structured lessons,

and group contingent free time.and group contingent free time.

Participants: n = 32 African AmericanParticipants: n = 32 African Americans enrolleds enrolled

in 7th grade in a metropolitan area ofin 7th grade in a metropolitan area of

TennesseeTennessee

Used two classes in a multiple baseline designUsed two classes in a multiple baseline design

first math, then geography.first math, then geography.


7/13


8/13

8

Nomothetic research

drawbacksdrawbacks::

1) individual differences treated as error1) individual differences treated as error

2) grouped data may obscure psychological2) grouped data may obscure psychologicalprocesses.processes.

Example: primary images in flashbulb memoriesExample: primary images in flashbulb memories

Question: What is your most memorable image fromQuestion: What is your most memorable image fromthe 9/11 attacks?the 9/11 attacks?

0

0. 1

0. 2

0. 3

0. 4

0. 5

0. 6

P la ne s C ra shi ng B ui ld in gsCo l l a p s i n g People Jum ping O th e rs

I m a g e

PerentRepordFi

S e p t e m b e rNovember On lyN o v e m b e r \ S e p t e m b e r

These results suggests that people haveThese results suggests that people haveremarkable stable memories for the events.remarkable stable memories for the events.

But, what happens if we look at individualBut, what happens if we look at individualparticipants consistency in reporting overparticipants consistency in reporting overtime?time?


9/13

9

Consistency in primary images

across a two month delay.

Consistent 22%Consistent 22%

More Specific 15%More Specific 15%

More General 27%More General 27%

Inconsistent 33 %Inconsistent 33 %

Clearly, grouped data can be misleading!Clearly, grouped data can be misleading!

Ideographic Research

GoalGoal: to understand an individual: to understand an individual

MethodsMethods: case studies: case studies

Emphasis on the uniqueness of the individualEmphasis on the uniqueness of the individual

DrawbackDrawback: what can case studies tell us about: what can case studies tell us about

psychological processes inpsychological processes in others?others?

ShontzShontz (1965, p. 236)(1965, p. 236)


10/13

10

A chemist who wishes to study the properties of aA chemist who wishes to study the properties of a

compound or element need not concern himself withcompound or element need not concern himself with

the number of samples of the substance on which histhe number of samples of the substance on which his

tests are run, as long as he is certain that he knowstests are run, as long as he is certain that he knowsthe identity of the particular material on which he isthe identity of the particular material on which he is

working and as long as his procedures are explicitworking and as long as his procedures are explicit

and carefully followed . By the same token, aand carefully followed . By the same token, a

psychologist who wishes to study an importantpsychologist who wishes to study an important

personality process, such as anxiety, need notpersonality process, such as anxiety, need not

concern himself greatly with the size of his sample,concern himself greatly with the size of his sample,

provided that his subject is appropriately selectedprovided that his subject is appropriately selected

and that he has procedures that enable him toand that he has procedures that enable him to

recognize (that is, measure) anxiety when it occursrecognize (that is, measure) anxiety when it occurs

in the person he is investigating.in the person he is investigating.

B. Case Study as Evidence

1) Confirmatory Evidence1) Confirmatory Evidence

case study can provide confirmation of morecase study can provide confirmation of more

systematic research, providing convergingsystematic research, providing converging

evidence.evidence.

Example: H.M. and the role of the hippocampusExample: H.M. and the role of the hippocampus

in memory.in memory.

Milner et al. (1968):

subject H. M.subject H. M.

temporal lobe andtemporal lobe and hippocampalhippocampal damaged indamaged in

treatment of epileptic seizures.treatment of epileptic seizures.

normal I.Qnormal I.Q

normal digit spannormal digit span

profoundprofound antrogradeantrograde amnesia (failure to learnamnesia (failure to learn

new information)new information)

Interpretation: No transfer from STS to LTSInterpretation: No transfer from STS to LTS


11/13

11

B. Case Study as Evidence

2. Disconfirmation of a General Law2. Disconfirmation of a General Law

Necessary and/or Sufficient CausesNecessary and/or Sufficient Causes

Nature of the theoryNature of the theory DisconfirmationDisconfirmation

A sufficient cause of BA sufficient cause of B A& not BA& not B

A is necessary cause of BA is necessary cause of B B & not AB & not A

Contributory causesContributory causes cannot be disconfirmedcannot be disconfirmed

with case studieswith case studies

Case studies and disconfirmation

Example 1: Motor Theory of speech perceptionExample 1: Motor Theory of speech perception

Speech Motor

programsyntax meaning

Motor Program (A) is necessary for deriving meaning (B)


LennebergLenneberg (1962)(1962)

case study of an 8 year old boycase study of an 8 year old boy

1) lacked motor shills for speech1) lacked motor shills for speech

(no motor program)(no motor program)

2) could understand and follow simple2) could understand and follow simple

commands.commands.


12/13

12


Example 2: Flashbulb memory hypothesisExample 2: Flashbulb memory hypothesis

Brown &Brown & KulickKulick(1977) unexpected events(1977) unexpected events

that trigger a strong emotional response evokethat trigger a strong emotional response evoke

aa special mechanismspecial mechanism that leads to thethat leads to the

storage of vivid and detailed recollections ofstorage of vivid and detailed recollections of

the circumstances surrounding the experience.the circumstances surrounding the experience.

Strong emotion (A) is a sufficient cause of vividStrong emotion (A) is a sufficient cause of vivid

and detailed recollections (B)and detailed recollections (B)

Example 2: Flashbulb memory hypothesis

(cont)

McCloskey,McCloskey, WibleWible, & Cohen (1988), & Cohen (1988)

Flashbulb memory and theFlashbulb memory and the ChallengerChallengerexplosionexplosion

Susan . . . Stated that she was eating lunch in theSusan . . . Stated that she was eating lunch in the

cafeteria with John, Beth, and Jennifer . . . However,cafeteria with John, Beth, and Jennifer . . . However,

Tim . . . Stated that he was eating with John when heTim . . . Stated that he was eating with John when he

learned of the explosion, but did not rememberlearned of the explosion, but did not remember

Susan, Beth, or Jennifer being there. BethSusan, Beth, or Jennifer being there. Beth

recollected that she was elsewhere at lunch the dayrecollected that she was elsewhere at lunch the day

(p. 175).(p. 175).

C. Descriptive Uses of Case Studies

1. Provide a prototypical example:1. Provide a prototypical example:

e.g.,e.g., The Three Faces of EveThe Three Faces of Eve

2. Provide a detailed account of a rare or2. Provide a detailed account of a rare or

unusual phenomenonunusual phenomenon

e.g.,e.g., LauriaLauria The Mind of theThe Mind of the MnemonistMnemonist


13/13

13

C. Descriptive Uses of Case Studies

3. Demonstrate important methods or3. Demonstrate important methods or

procedures.procedures.

e.g., description of a behavior modificatione.g., description of a behavior modification

treatment of an autistic childtreatment of an autistic child

4. As a source of hypotheses4. As a source of hypotheses

e.g., Jean Piagete.g., Jean Piagets works work

Conclusions

1 Alternative research designs can provide1 Alternative research designs can provide

important informationimportant information

quasi-experimental, smallquasi-experimental, small nn, case studies, case studies

2. But, they have clear limitations2. But, they have clear limitations

3. Researchers should use the best designs3. Researchers should use the best designs

possible, minimizing alternative explanationspossible, minimizing alternative explanations

Alternative Designs

Documents

Transcript of Alternative Designs