Chapter 10: Non & Quasi experimental strategies

Nonexperimental Strategies

•Compare scores without limiting confounding variables

•Lack manipulation, random assignment, or control

Quasi-experimental strategies

Compare scores while limiting confounding variables

Nonexperimental Designs

Groups produce a set of scores, groups are not manipulated

• Manipulation can be unethical or impossible

Test differences between preexisting groups

Researcher cannot use random assignment to create groups

Interpretation confounded by individual differences

Groups differ but not why they differ

Cohort effect

Differences in behavior, attitudes, or outcomes between groups of people that result from being born and raised during different time periods (rather than from age itself).

Differential research design

Compares scores from preexisting groups (age, gender, personality, etc.)

Posttest-only nonequivalent control group design

• Administer treatment to one group

• Measure and compare both groups (posttest)

Pretest-posttest nonequivalent control group design

•Measure and compare both groups (pretest)

•Administer treatment to one group

•Measure and compare both groups (posttest)

One group of participants measured before and after treatment

No control group for comparison

Pretest-posttest design

Each participant measured once before and once after treatment

Time-series design

Series of observations before and after treatment

Cross-sectional research design

Different groups of participants for each age group

Cross-sectional research design: Strengths

•Observe how behavior changes with age without waiting for aging

•No attrition (i.e., participant dropout)

Cross-sectional research design: Weaknesses

Factors other than age may differentiate groups (i.e., cohort effects)

Longitudinal research design

Observe a group of participants over time

Longitudinal research design: Strengths

Absence of cohort effects

Longitudinal research design: Weaknesses

• Time-consuming and expensive

• High dropout rates (i.e., participant attrition)

• Early measures could influence later ones