History: Since the study is a long term one, two years in process, other events can occur during this time which can have an effect on the experiment also need to make sure that all other factors are the same for both groups and the only difference is the special reading program.
Maturation: During the two years that the reading program is implemented, the children are going to undergo psychological and physical changes it could be that their improvement in reading is due to these other factors and not the treatment.
Testing: It could be that the children have become “test-wise” and that’s why they are reading better in the posttest, and not because of any actual improvement in reading.
Instrumentation: is not a threat to internal validity since you can measure the child’s reading ability quantitatively, and it is not an observational measurement.
Statistical Regression: (the tendency of the students who score extremely well or poorly to score better , if they did poorly first time and to score worse, if they did very well first time).
Differential Selection: is probably not a threat to internal validity because the fifty children in control group are selected randomly, and random selection is best protection against differential selection.
Experimental Mortality: Twenty three children from the special reading program dropped out and fourteen children from the control. Since the number of children giving the post test is lower in each case the final figure is not going to be representative of the true reading abilities. It might be that the really poor readers dropped out of the program leaving only those who made progress.
Selection Maturation Interaction: is probably not going to affect the internal validity because children in the test group and control are from same school and grade level, so their ages are going to be approximately the same.
Experimental Treatment Diffusion: Since the children in reading program and control are in very close proximity to each other, it could be that over time some of the material used for the special reading group finds its way in the hands of students continuing with the normal reading. Hence, we have the diffusion of materials.
Compensatory Rivalry by the Control Group: It is possible that the control reading group increased reading abilities by 1.6 years because they thought they were in competition with the test group and so they made exceptional efforts to improve.
Compensatory Equalization by Treatments: This will not come into play because the special reading group does not receive anything which will be deemed as desirable, and so the control will not have to be given any favors as compensation.
Resentful Demoralization by Control Group: If the control group felt that the special reading group is getting favors which are being withheld from them, they may not have done as well as they could have and as a result the special reading group will appear to have made more progress in comparison.
I believe that the posttest only control group design will be best suited to determine whether students learn more when they take unannounced pop quizzes or regularly scheduled, announced quizzes. The first step in designing this experiment is the selection of the test subjects. I would choose two sections of the same grade level in a randomly selected school. The test would be carried out during the same lesson of each section (for example English), and I would have the same teacher teaching both sections. This is done to make sure that only one variable (announced quizzes vs. unannounced) is changed in each section. There is the danger here that the children in the announced quiz class will tell the children in the unannounced class. That factor can be controlled by having the pop quiz early in the week and then announce the quiz in the other class a day after the pop quiz has been given. The next step would be to administer the treatment to the experimental group while the other gets no treatment. The last step would be to administer the posttest to both groups.
a) She will have to use the quasi-experiments since the administrators of the school want to maintain the school intact group and the participants cannot be assigned randomly.
b) The fact that teachers cannot be randomly assigned is the biggest threat to the internal validity. To take care of this problem we can increase the number of schools of the district that take part in this study and then allow each school to participate in one program, A or B. We can still have good internal validity provided schools are randomly assigned to group A or B, the appropriate unit of statistical analysis is maintained and other obstacles to random assignment are avoided. For example some schools may feel that the person assigning groups is favorably or negatively biased towards some schools so the assigning is not really random. To avoid this when schools are being assigned have someone all the school administrators trust present at the assigning.
Does the school allow differential services for its students? If not then we can’t have a control group and hence we will be forced to choose the single case design.
Is this program aimed at children with learning disabilities or normal children? If the program is aimed at children with learning problems then examiners can make respectable estimates of the progress expected with these children during regular instruction based on their history or maturity level and measure the gains made with the treatment and then compare the measured results with the estimated results to see if the results with the treatment are any better. In this case the single case design may be used. If it is aimed at regular students, estimating gains with regular instruction is more difficult, less accurate, so a control group becomes necessary, and the group experimental design will have to be used.
The reading material provided was the only source used to answer these questions.