Chapter 6: Experiments in the Real World Ways to control for bias of people in experiments. Single Blind: An experiment is single blind if the units are unaware of the exact treatment being imposed on them. Controls for subject bias. Double Blind: An experiment is double blind if the units and the experimenter are unaware of the exact treatment being imposed. Controls for subject and experimenter bias. (93)
Example of a Double Blind Experiment In cold studies, the doctors in charge label the treatments with a code number (such as #90210). Patients receive a treatment labeled with a code number. The nurses who give the treatments and record the responses know the treatment by its code number. Neither the nurse nor the patients know if the treatment being imposed is an experimental drug or a placebo.
Problems with Subjects (p. 94) Nonadherers: Subjects who participate but do not follow the experimental treatment. Refusals: Some subjects that we want in our study may refuse to participate. Dropouts: Subjects may start in the study and later dropout. Especially true for experiments that last over and extended period of time.
Can we generalize? “Statistically significant” is not enough Sample should represent the population Lab rats to people: Not representative Experiment should be realistic Example 5: Studying frustration (p. 95) Results should be “useful.” Expert in the field can best answer this question
More Experimental Designs More than one explanatory variable Treatment combinations Matched pairs design Block design
More than One Explanatory Variable Example 8: Effects of TV Advertising (p. 98)
Matched Pairs Design Units are matched to form pairs or each unit receives both treatments. The responses for the pairs are compared. Example: Testing effectiveness of a sunscreen lotion. Difficult to compare how the lotion works on different people due to different skin types and body chemistries. Better method: Test one lotion on one arm and the other lotion on the other arm of each unit (person). Compare the effects on each arm for all units.
Block Design A block is a group of experimental subjects that are similar in ways that are expected to affect the response to the treatments. Note: The blocks are NOT randomly assigned. Used when we have groups with similar traits (as in stratified random sampling) that may affect the outcome of the experiment. In a block design, the random assignment of subjects to treatments is carried out separately within each block. A separate randomized comparative experiment is performed for each block.
Example 10: Men, Women, and Advertising (p. 100)
Advantage of Using a Block Design Can control for the effects of some lurking variables that may be confounded with the explanatory variable. We include a potential lurking variable in the design and it’s effects can now be accounted for. We call this a blocking variable.