150 likes | 309 Views
Quasi-Experiments. Research Objective. Study relationships among variables for existing groups. Show direct cause & effect. Explain outcomes after the fact. Type of Design. Cross-Sectional Longitudinal. True Experiment Quasi-Experiment. Explanatory Case Study Exploratory Case Study.
E N D
Research Objective Study relationships among variables for existing groups Show direct cause & effect Explain outcomes after the fact
Type of Design Cross-Sectional Longitudinal True Experiment Quasi-Experiment Explanatory Case Study Exploratory Case Study
What Are Quasi-Experiments? • Quasi-experiments are just like experiments except for one thing. • They violate the condition of random assignment of test subjects to treatment and control groups.
Why Do This? • There are two main reasons why researchers fail to achieve the random assignment assumption: • It is difficult or impossible to achieve random assignment because the participants are embedded in groups • Example: students in classes • There are ethical reasons for assigning some participants to the treatment group • Example: potentially life-saving new drug
The Classic Experiment • Select representative sample • Randomly assign to control and treatment groups • Measure the outcome (dependent) variable for both groups • Do something to the treatment group • Measure the outcome variable again for both groups
Quasi-Experiment Did you select a representative sample of test cases from a single homogeneous population? YES No Not a quasi-experiment Not a quasi-experiment Not a quasi-experiment No No Did you ASSIGN test subjects to treatment & control groups NON-randomly? YES Did you then do something to treatment groups that you did not do to the control group? YES Quasi-Experiment
Non-Equivalent Group Designs • One of the most common reasons for going to a quasi-experiment is that your potential test subjects are naturally embedded in some larger groups. • Students in classes • Hospital patients in wards • Participants in an intervention program • Monkeys in troops
Subjects Embedded in Groups • There is a new set of experiential learning materials about environmental education for high school students. • You believe that this way of learning about the environment is more apt to produce change in behavior than traditional methods. • Things like recycling, turning off the lights, not letting the water run while you brush your teeth.
Your Subjects • You get three of seven science teachers in a school district to agree to try the new approach. • The other 4 will use the old approach. • Your test subjects are students in their classes. • You will measure behavior (how often they recycle, etc.) before and after the education, new or traditional.
The Problem • This looks just like a true pre-post measurement experiment. • The problem is that your test subjects, the students, come in groups. • You did not assign Student A to control and Student B to treatment.
The Threat • The big problem with this design is that you cannot be really sure that the control and treatment groups are equivalent. • For example, what if the 3 teachers who agree to try the new approach are just better teachers and their students are therefore more motivated?
Key Concepts What’s the same and what’s different between a true and a quasi-experiment.
The DIFFERENCE • Quasi-experiments differ from true experiments in only one way: there is no random assignment of individual participants to treatment and control groups. • This makes it much harder to control non-treatment variance. • It greatly increases threats to internal validity. • It can also increase threats to external validity. • It reduces explanatory power.