Question #5: Yes. Small samples are totally justifiable however, the statistical tool selected to analyze the data must be able to accommodate small sample size. The statistical techniques employed for small sample size is what applied statisticians call "robust statistics" or certain parametric types such as the t Test. Although the parametric tool is more robust that the non-parametric counterpart, parametric statistical tools with real small samples may produce misleading information because there is no way to determine if the data came from the Guassian population. That is to say, increasing the number of participants from 15 and 14 respectively to at least 30 and 30 might possibly produce result changes. However, there is no real way of knowing unless the study were replicated with a larger group. The advice to any research investigator is to have a sample size of each group of 25 to 30 at least. When this is not achievable always test the statistical significance of the main statistical value received; i.e., test the statistical significance to the t test or F test value received if that was the statistical test used by the research investigator.
Question #6: Yes and No. Any time there is a gender specific research situation the investigators must be sensitive to their audience. As bulimia is generally thought of to be a female disease, which it is not, it is important to study the disease of at risk people from a very neutral position. Therefore moderator gender is important and only female moderators should be used. Using male moderators with a female at risk audience may possibly bias the results or at least influence the verbal informational dissemination process. As race has no bearing on bulimia the issue is moot. However, without appearing to be stigmatic this reviewer would highly suggest that the moderators not only be female but average in weight and without observable weight issues in order to reduce participant frustration and anxiety which can influence test results for which there is no accounting.
Question #7: Post and pretest research investigative situation are usually configured to permit the research investigator to predict some phenomenon. As such sufficient time must lapse between the pre-test and post-test situation. Knowing that psychological dysfunctions do not appear nor disappear over night the length of time between testing situations is difficult to garner. Not knowing why the research investigator establish a one-eek posttest and a five-week follow-up it is not possible to determine or comment upon the validity of the time interval. However, previous studies in psychology suggest that any treatment program must allow for sufficient time for the treatment to be effective. Suggestions with respect to the time span ranges from 6 months for minor disorders to two years for serious disorders. If this gage is to be used then the researchers did not allow a sufficient amount of time to pass between the pre and post test situations and the result might possibly be that of week results. In order to counteract the time problem the research investigator would have been wise to continue the study on a longitudinal basis to determine if the differences and effect were more permanent.
Question #8: Internal validity refers to an experiment's outcome as a function of the variables being measured, controlled or manipulated. Altering the variables can impact the internal validity of a study. Possible factors influencing the internal validity of the present investigation into high-risk bulimia participants includes, but not limited to, communication between treatment and control group outside the experimental situation; pretest/posttest practice effect; possible fatigue during treatment; possible lack of instructional standards between mentors; and demand characteristics. With respect to external validity contamination include, but not limited to, the Hawthorne effect (subjects knowing that they are in an experiment); treatment carry-over effects between pre and post testing; and a situation wherein the pretest interacts with the treatment and results in an effect that will not be generalized (i.e., sensitization). With reference to the present situation internal validity seems to have been more closely accounted for than external validity.
Question #9: Unfortunately the results of the present research investigation cannot be generalized to the larger population of bulimics with respect to the treatment intervention. The reasons are as follows:
Sample size is too small for acceptable statistical analysis without testing the statistical significance of the main statistical tool.
Substantiation for the pretest - posttest interval not adequately established.
Pretest - posttest situations are not designed for population inference only for treatment vs. control group inferences.