Professional Documents
Culture Documents
Quasi-experimental designs
Quasi-experiments
Measurement studies
Retrospective designs Evaluate existing groups or program
Next
True experiments:
Emphasize internal validity Assess cause & effect (in
relatively artificial environment)
Quasi-experiments:
Emphasize external validity Describe real / naturally occurring events Clear to exploratory hypotheses
Participants assigned to Existing or non-equivalent experimental v. control groups groups Random or matching Non-random assignment Participants & experimenter Participants not blind Blind to assignment Control group not possible?
Control study procedures Control often not possible Create / manipulate independent May not be able to manipulate variable the independent variable Control procedures & measures Partial control of procedures & measures
Psychology 242, Dr. McKirnan Week 12-13, quasi-experimental designs. Back
Home page
Next
Research Ethics
Back
Home page
Next
Measurement studies
Study question
(Outcome)
Stress -> immune system Stress & coping
Crime / trauma
Iraq service, 9 / 11 / 01 PTSD & treatment Contrails & climate change Voting patterns Health behavior
Back
Home page
Historical event
9/11 & air travel ban Economic collapse
Next
Independent variable:
Predictor variable (e.g., natural disaster) often assessed after the event (post-hoc). Researcher has little control over dose / type of predictor
Participant selection
No control over who is exposure to event Some control over selection of sample (e.g., via targeted
sampling)
No control with archival data Some control with surveys Use retrospective (measured) variables to clarify interpretation of outcomes or test hypothesis.
Week 12-13, quasi-experimental designs. Back
Home page
Next
Event
(Predictor variable) earthquake [v. control city?] crime / trauma
[v. control people?]
Outcome variable
stress & coping mental health voting patterns
Personal attitudes
historical event
Demographics
cultural event
health behavior
Back
Home page
Next
Next
Existing groups
Existing groups:
Single self-selected group; no comparison possible
users of psychotherapy (or any product) members of group or cult [contrast with demographically
matched controls?]
Next
Group
May control selection of study group, or must use Convenience sample. Other data may be available about group.
Dependent Variable(s): May or may not have control over measures (e.g., surveys v. archival measures).
Typical use: Surveys or measures after an event. Heuristic value: generating hypotheses for later study
group.
No control group; uncontrollable event, or other groups
Back
Home page
Next
10
Research questions: Does psychotherapy work from consumer view? Who gets therapy / what does it consist of? Do consumer responses vary by type of therapy? Research approach: One shot case study / survey Sampling frame: Any therapy or psychological service user No real information re: population of therapy users. Sampling procedure: 4,100 Consumer reports readers responding to in magazine mail-back survey form
Psychology 242, Dr. McKirnan Week 12-13, quasi-experimental designs. Back
Home page
Next
11
Experimental Controls
Negatives:
Selection bias no control over who got therapy (self-selection) of those who got therapy, no control over who
returned a survey (secondary self-selection) Cursory outcome measures: satisfaction rather than mental health
Positives:
Huge, national sample Wholly anonymous, 3rd party data collection; less bias Real world assessment of product quality
Psychology 242, Dr. McKirnan Week 12-13, quasi-experimental designs. Back
Home page
Next
12
Do the specific ingredients (or theory) of this treatment validly induce the key outcome?
Next
13
(Consumer reports 4)
Efficacy experiment v.
One specific diagnosis Rigorous control group no treatment condition attention control Random assignment Manualized / uniform treatments High Fidelity to treatment method Fixed number of sessions. Well operationalized outcomes, e.g., clinician-diagnosed disorder Standard / validated self-report symptom scales
Effectiveness research
Multiple diagnoses & severities No control group, 2nd controls Archival, via pt. characteristics Self-selection; shopping Multiple / mixed treatments Highly tailored to patient # sessions is patient based. Diverse, self-referenced outcomes Subjective sense of wellness Lessening of problem behaviors or moods Personal assessment of functioning Self-rated: cannot be "blind" Diverse times since treatment Retrospective rather than prospective
Back
Home page
Next
14
Back
Home page
Next
15
MDs gave medications to 83% of patients MH pros; 20% drug treatments 50% who got drugs got no counseling 20% got no information about side effects 40% of anti-anxiety drugs given > 1 year
Week 12-13, quasi-experimental designs. Back
Home page
Next
16
Group
Dependent Variable(s): Combine survey or other measures with archival or ancillary data as Control variables.
Next
17
Sampling frame: - Randomly selected residents of geographic area around TMI Outcomes: - Blood draws for immune markers - Self-reports of stress
Findings:
Next
18
Quasi-controls:
Findings:
- Long-term immune suppression - Social support buffered stress - impact of bereavement > other stressors
Week 12-13, quasi-experimental designs. Back
Home page
Next
One shot designs; Summary One Shot designs: no control over independent variable(s), only partial control over measurement:
19
Virtue:
Assess naturally occurring or uncontrollable socially or politically important events Provides real world look at processes that are typically studied in experiments: Effectiveness v. Efficacy data Archival data can help interpret the findings / control some alternate interpretations.
Liability:
lack of control group creates multiple threats to internal validity No pre-measure makes interpretation (e.g., of change) difficult.
Week 12-13, quasi-experimental designs. Back
Home page
Next
20
Non-equivalent groups
Time series designs
Psychology 242, Dr. McKirnan Research Ethics Back
Home page
Next
21
Event or intervention May or may not be controllable by researcher, e.g., policy change.
Next
22
Confound
People leave the experiment nondrop-out randomly (i.e., for reasons that may affect the results). Week 12-13, quasi-experimental designs. Back Next
Home page
23
Sampling frame: - Participants in testing centers Study structure: - Baseline retrospective interview at testing session - Follow-up interview 3 months later Quasi-controls: Outcomes:
Psychology 242, Dr. McKirnan
Next
24
Findings:
- Significant shifts toward safety - Few demographic predictors of risk or risk change
Threats - Self-selection into testing group to internal - Mortality: non-random drop-out(?) validity - History: general shift in norms & behavior during study time may account for observed change - Instrument change; people may answer more conservatively during a follow-up interview
Psychology 242, Dr. McKirnan Week 12-13, quasi-experimental designs. Back
Home page
Next
25
Sampling frame & - Longitudinal data across multiple Study structure: years in target school grades. - No control group possible.
Quasi-controls:
Outcomes:
Psychology 242, Dr. McKirnan
Next
26
Findings:
Modest, statistically significant increase in scores Usual demographic predictors of change; more affluent, better schools..
Internal validity?:
Reactive measures; teachers & students do better when measured; (they also cheat; see Houston Miracle article)
Instrumentation: kids get better at taking standardized tests, teacher better at teaching them History: General cultural shift
Education more prominent in city More affluent families sending kids to public schools
Week 12-13, quasi-experimental designs. Back
Home page
Next
27
Virtues:
provide data on naturally occurring socially or politically important events Pre-measure allows researcher to interpret change & examine status of groups at baseline. History maturation statistical regression reactive measures mortality / drop-out
Back
Home page
Next
28
(No baseline)
Contrast group
Groups are not equivalent at baseline, due to.. Self-selection Non-random assignment Use of existing groups Participants not blind
Assessments may or may not be controlled Survey or interviews Archival / existing data, e.g., clinic records, grades Intervention or event may or may not be controlled by researcher; Existing program Experimental intervention Naturally occurring event (..9/11..)
Home page
Back
Next
29
Non-equivalent groups Self-selection Non-random assignment Use of existing groups Participants not blind
Observation1 used to
Assess equivalence of groups at baseline Test for threats to internal validity:
Reactive measures History, mortality effects Regression effects
Psychology 242, Dr. McKirnan Week 12-13, quasi-experimental designs.
Back
Home page
Next
30
Example Non-equivalent control group design: Effects of condom distribution on sexual safety
Intervention: - Condom education & distribution in High School health classes
Sampling frame: - Schools in New York & Chicago - Schools matched for SES, race, size Study structure: - NY = intervention schools, Chicago are contrast schools. - Baseline, sexual health programming, end of year Follow-up Outcomes:
Psychology 242, Dr. McKirnan
Next
31
Findings:
NY (intervention) students; lower STD rate, safer sex NY and Chicago students; similar levels of sexual activity Thus; sexual health classes appeared to increase safety without increasing sexual activity. Reactive measures; Study is not blind; NY students know they are the intervention group
Internal validity?:
Next
Non-equivalent designs
32
Group 1
Observe1 Observe1
Group 2
Group 3 Group 4
Intervention
Contrast group
Observe2
Observe2
Groups 1 & 2:
Observation1 used to Assess equivalence of groups at baseline Test threats to internal validity
Groups 3 & 4:
Post-test only tests for reactive effects of assessment Compare 1+2 versus 3+4 Test interaction of treatment group x pre- post- versus post- only
Psychology 242, Dr. McKirnan Week 12-13, quasi-experimental designs.
Back
Home page
Next
33
Virtue: Study natural / real world interventions Contrast group lessens major threats to internal validity Liability: non-equivalent groups = possible confound.
Next
34
Non-equivalent groups
Next
35
Group
Test effect of intervention or event on ongoing series of measurements. Intervention may be experimental or observed
Policy shift, e.g., educational policy
Uncontrolled event; e.g., 9/11/01, Media event
Next
36
Group
Measure1
M2
M3
M4
M5
M6
Multiple baseline
Intervention or event
Demonstrate highly stable effect long-term crime rates disease prevalence economic performance
Show steady rate of change
Shift in stable rate after intervention Increase / decrease in rate of change after intervention
Next
37
Core question: Do baboon troops develop and transmit a learned culture? Baseline: Long-term observational data on aggressiveness in a specific baboon troop. Intervention: Tuberculosis outbreak due to infected food. Dominant / aggressive males fed first
are selectively infected are naturally culled from troop Naturally occurring event in >20yr. ongoing
Psychology 242, Dr. McKirnan
field study.
Back
Home page
Next
38
Next
39
Data show stable baseline over multiple observations Timing of intervention precise relative to data collection
Uncontrollable, naturally occurring event Tests hypothesis re: modeling effects in health behavior
Initial increase expected Hypothesis tested by longer-term shift in base rate, available due to archival time-series data Effect found for both genders.
Week 12-13, quasi-experimental designs. Back
Home page
Next
40
Magics Announcement
Initial spike New, higher base rate Low & variable baserate of testing
Tesoriero, J.M., Sorin, M.D., Burrows, K.A., LaChance-McCullough, M.L. (1995). Harnessing the heightened public awareness of celebrity HIV disclosures: Magic and Cookie Johnson and HIV testing. AIDS Education and Prevention, 232-250.
Psychology 242, Dr. McKirnan Week 12-13, quasi-experimental designs. Back
Home page
Next
41
Intervention or event
Hypothesis; tested by
interaction of blocking variable by repeated measure:
Is shift in stable rate ( rate of change) greater in one group than another?
Back
Home page
Next
Blocking variables
42
Testing blocking variables in the HIV testing time-series data. Core questions:
Both heterosexuals and Ethnic minorities had low HIV testing rates
May feel HIV is not relevant to them it is a white gay problem. They may lack resources or venues for testing.
Will having a prominent African-American Heterosexual disclose HIV+ status may change those perceptions? Heterosexuals will respond more strongly to the Magic Johnson media event than will gay/bisexual men.
African-American and Latino men and women will respond most strongly.
Psychology 242, Dr. McKirnan Week 12-13, quasi-experimental designs. Back
Home page
Hypotheses:
Next
43
Back
Home page
Next
44
The hypothesis that heterosexuals would be more affected by the Magic announcement was supported by the interaction of Time x the blocking variable of sexual orientation.
Psychology 242, Dr. McKirnan Week 12-13, quasi-experimental designs. Back
Home page
Next
45
Back
Home page
Next
46
HIV testing among Whites was similar to African-Americans & Hispanics at baseline, They showed stable, much higher testing rate after Magics HIV announcement.
Back
Home page
Next
47
A series of measures before & after an event allows us to clearly identify patterns of behavior, and to test group differences (via blocking variables). The hypothesis that ethnic groups would differ was supported by interaction of Time x the blocking variable of ethnicity (but in a direction that was not predicted: Whites showed more change).
Psychology 242, Dr. McKirnan Week 12-13, quasi-experimental designs.
Home page
Back
Next
48
Time series is most common with archival data: existing, standard records collected for other purposes. Used where: The hypothesis concerns changes in long-term trends Typically an experiment cannot be run
Simple practicality or cost, e.g., health care issues Ethics; crime rates, rates of domestic violence, etc. The target events are not controllable.
Virtue:
Study natural / real world processes or interventions Blocking variables comparing time trends across groups -- lessens major threats to internal validity
Week 12-13, quasi-experimental designs. Back
Home page
Quick quiz
49
Back
Home page
Next
Quick quiz, 2
50
Match:
A = Mortality / drop-out
B = Maturation
C = History D = Statistical regression E = Reactive measures
Psychology 242, Dr. McKirnan
People leave the experiment nonrandomly (i.e., for reasons that may affect the results). Week 12-13, quasi-experimental designs. Back Next
Home page
Quick quiz, 2
51
Match:
A = Mortality / drop-out B = Maturation C = History
People respond to being measured or being a measured a second time.
Next
Quick quiz, 3
52
Match:
A = Mortality / drop-out
B = Maturation
C = History D = Statistical regression E = Reactive measures
Psychology 242, Dr. McKirnan
Back
Home page
Next
Quick quiz, 4
53
Match:
A = Mortality / drop-out
B = Maturation
C = History D = Statistical regression E = Reactive measures
Psychology 242, Dr. McKirnan Week 12-13, quasi-experimental designs. Back
Home page
People leave the experiment nonrandomly (i.e., for reasons that may affect the results).
Next
Quick quiz 5
54
Group
Measure1
M2
M3
M4
M5
M6
This is called a:
Intervention or event
D = ..lot of work.
Back
Home page
Next
Quasi-experiments; Summary
55
1. Study naturally occurring events that could not be brought into a lab or a true experiment.
Measurement studies
Retrospective designs
Next
Exam issues
56
Slide 2: true v. quasi experiments Threats to internal validity Basic forms of quasi-experiments
Single shot Single group pre- post- test Non-equivalent two group designs:
Self-selection (in or out [mortality]) Existing groups Non-blind Non- random assignment
Next