Graham recruited student volunteers to participate in his dissertation study. He set up several times for students to come to a
specified classroom and read various types of instructional materials and to be tested. He ran all of his control conditions first and then for each session he placed all students at the session in the same treatment condition. This is problematic because those who volunteered early are likely different than those who volunteered later. This problem is primarily due to 1. response bias. 2. instrumentation bias. 3. researcher bias. 4. sampling bias.
Bias refers a prominent problem in statistical analysis whereby one or more analytical factor are favored than the other during an analysis which should be made random. The problem. With Graham's dissertation study is the fact that he failed to randomlyvplace his subjects or observation in the study groups, favoring a particular group with non random subset. When randomization is ejected or missing from an analysis or study, it becomes less and less representative. Here, allotting early Arrivals Into the treatment group has introduced a sampling bias as those who came later, this will also leads to less reproducibility of experiment.
