SOAN 2120 Lecture Notes - Lecture 13: Random Number Generation, Random Assignment, Standardized Test

• School: University of Guelph
• Department: Sociology and Anthropology
• Course: SOAN 2120
• Professor: D Walters

Document Summary

Want to make sure that the treatment and control group are equivalent in their scored before the experiment begins, equal in every possible way. Then we know that the differences at the end can be attributed to the experiment. How do we establish the equivalence of two or more groups: random number generator. It is objective, equal probability of being assigned to each group. It is esse(cid:374)tial that the g(cid:396)oups a(cid:396)e (cid:862)e(cid:395)ui(cid:448)ale(cid:374)t(cid:863) Even with the random assignment (ra), the groups cannot be exactly equal (ex. If the assignment is random, we can use tests of statistical significance (manipulation vs. random variation) As the sample size increases, individual differences decline. Without ra: findings may be attributed to pre existing differences selection bias. Notation: x= independent variable, o= dependent, r= random assignment, before x is the pretest. R o3 o4: o1 is the pretest for the experimental group, o3 is the pretest for the control group.