Pre and Quasi-Experimental 1 What We Will Cover in This Section Pre-Experimental Quasi- Experimental. Summary 2 Internal Validity Revisited 3
Common Threats to Internal Validity 1. History. 2. Maturation. 3. Testing. 4. Instrument Decay. 5. Statistical Regression. 4 1. History ny event that occurs between the first and second dependent measures that is not manipulated by the experimenter. Delay 5 2. Testing Participation in the pre-test may cause changes in the person. - Reactivity - Memory Delay 6
3. Maturation Changes in the individual over time that are not associated with the independent variable. Placebo Delay Delay 7 4. Instrument Decay Changes in the measuring instrument over time. Observer gets bored. Test becomes obsolete. Machine wears out. 8 5. Statistical Regression Occurs when participants are placed into groups based on extreme scores. Extreme scores tend to move(regress) toward the mean. Delay 9
Pre-Experimental 10 One-Shot Case Study Group Independent (s) controlled by the experimenter Dependent ment(s) made after the treatments are applied. What is missing from this design? 11 Benefits and Issues Benefits. 1. OK for preliminary research. Issues. 1. History. 2. Maturation. 3. Regression. 12
One-group Independent Group 1 What is missing from this design? 13 Benefits and Issues Benefits. 1. OK for preliminary research. Issues. 1. History. 2. Maturation. 3. Regression. 4. Testing. 5. Instrument decay. 14 Quasi-Experimental 15
Definition that have all of the controls of a true experiment but do not duplicate the true experimental design. Weakness in one or more areas. Weakness leads to confounding. Gives more information than nothing at all. 16 Ex Post Facto pproach Independent Dependent Group 1 Group 2 Groups divided based on some pre-existing condition. ment(s) made after the assignment to groups 17 Benefits and Issues Benefits. 1. May be the only way to study some influences. 2. May be OK for preliminary research. Issues. 1. Ss not randomly assigned to treatment conditions. 2. If a person is unusual on one characteristic he may be unusual on others. 18
Non-Equivalent Control Group, Pretest: Independent SS 1 SS 2 SS 3 1 SS 4 SS 5 SS 6 2 19 When Used When participants cannot be assigned at random. When you have to use pre-existing groups. 20 Issue and nswer Issue. The groups may differ on the dependent variable at the beginning of the experiment. nswer. Compare groups on the pre-test to see if there are differences. 21
Other Issues and Fixes 1. Replicate the study. 2. ssign multiple groups to treatment conditions. 22 Time Series Similar to single subject design but with one group. Usually have Baseline measurement on dependent variable. Some manipulation or event. Second measurement on the dependent variable. 23 Interrupted Time Series Design Baseline () Event Post Test 24
Number of Fatalities 290 280 270 260 250 240 230 220 210 200 Connecticut Traffic Fatalities Speeding Crackdown 1951 1952 1953 1954 1955 1956 1957 1958 1959 Year 25 When Used ssess the impact of some naturally occurring event. ssess the impact of some broad treatment within an existing group. 26 Control Series Design Uses existing groups but attempts to match the treatment group with an equivalent control group. 27
Control Series Design Baseline Baseline Event Event Post Test Post Test 28 Issues Potential confounding from History Maturation Instrumentation changes. 29 Connecticut Traffic Fatalities 15 Fatality Rate 12 9 6 1951 1952 1953 1954 1955 1956 1957 1958 1959 Year Connecticut Control 30
Thought Problem Petal D. Stamen was interested in the influence that flowers would have on women s affection toward men. Petal sent a dozen roses to a random sample of women then asked them to fill out a well researched affection survey. 1. What kind of design is this? 2. Is this a good or bad design? Why? 3. How could this study be improved? 31 32