Topic 4 Quasi Experiments

Question 1

Validity of Quasi Experiments

Answer 1

1. Uncontrolled or confounding variables reduces internal validity
   a. Lack randomized groups by definition
2. Weaker statistical conclusion validity (rely on random assignment)
3. External validity may be higher
   a. May study subjects or settings that are more appropriate to a question of interest than a true experiment could
   i. Ex: group differences in clinical studies
   b. A true experiment is better but a quasi is better than none

Question 2

Quasi-Experimental designs:
Nonequivalent control-group
(NOXO and NO-O)

Answer 2

a. Have both experimental and control group
i. Subjects are not randomly assigned to groups
b. Comparing to groups not equivalent at the start
i. Vulnerable to selection bias- prior differences may affect outcome

Question 3

Quasi-experimental designs:
One-group pretest protest
(O1 X O2)

Answer 3

Baseline assessment, treatment, outcome measure

a. Lack of control group creates multiple threats
i. Ex: history, maturation, regression, testing, instrumentation, mortality

Question 4

Quasi-experimental design:
Static comparison group
(XO1 and -O2)

Answer 4

Both groups aren’t pretested

a. Lack of random assignment and pretest make it difficult groups were equivalent
i. Selection bias- If O1 and O2 differ could be by recruitment of people
ii. Mortality- people dropping out selectively alone can cause the difference

Question 5

Quasi-experimental design:
One-shot case study
(XO)

Answer 5

Group is exposed to the treatment then observation is made

a. Very difficult to interpret the results:
i. No pretest- makes it difficult to rule out threats to internal validity
ii. No control- makes it difficult to rule out threats to internal validity

Question 6

Quasi-experimental design:
Interrupted time-series
O1 O2 O3 O4 X O5 O6 O7 O8

Answer 6

Periodic measurements on some group or individual with introduction of an experimental change (X) into the time series of measurements

a. Rate prior to 1987 increasing, decreasing after
i. Idea that changing the way its media reports suicide decrease number of suicides
ii. Other variables may account for results (internal validity)
1. History- is it X or some other simultaneous event?
2. Maturation- invalid if change is not regular across measurements
3. Instrumentation- depends on the reliability of the instruments
4. Selection- ruled out if all the same people are used at all O

Question 7

Quasi experimental design:
Repeated- treatment
(01 X 02- 01 X 02)

Answer 7

Treatment is withdrawn and then presented 2nd time

a. Improve validity by presenting the treatment more than once
i. Limitation
1. Treatment can be withdrawn without causing complications

Question 8

Designs to test developmental changes: Longitudinal study

Answer 8

test a single cohort over period of time

i. Disadvantages
1. Researcher has to wait for cohort to age
2. Confounds age with time of testing
a. Secular trend- change that takes place in general population