Tweet
Dear Biomech-L readers,
Recently I posted a question regarding concerns with statistical analyses in studies with small sample sizes. The question generated a lot of interest - I received over 50 replies, not including those who were just requesting that I post a summary of replies. The sheer number of responses reflects how powerful the listserve is as a resource for information.
I tried to post the summary of responses in a single email, but I failed because the message was greater than the maximal size allowed. Therefore, I have broken the replies into two separate emails. Part 1 has a summary of all replies, and Part 2 has the actual replies by all persons.
The original post was:
Dear Readers,
I am hoping that some of you who have expertise in the area of statistics and scientific journal review can help me with the following concern.
Recently I have submitted papers to peer-review journals that describe the results of investigations performed on 'small sample sizes'. Obviously, small is a relative term. For the sake of this discussion, my samples sizes have been greater than 4 and less than 10 persons.
Multiple times I have received reviewer comments that the sample size was too small, which limited my results. What concerns me is that most of my investigations involve a repeated measures design, during which subjects are tested in two or three environments, with the objective being to determine the affect the environment has on my measure. I typically use a paired t-test or repeated measures ANOVA with a post hoc evaluation if a significant main effect is determined.
In these papers, if I were to fail to reject the null hypothesis, then I can understand the concern about power and sample size. However, in many of my papers, I have determined a significant effect relative to an a priori p
Recently I posted a question regarding concerns with statistical analyses in studies with small sample sizes. The question generated a lot of interest - I received over 50 replies, not including those who were just requesting that I post a summary of replies. The sheer number of responses reflects how powerful the listserve is as a resource for information.
I tried to post the summary of responses in a single email, but I failed because the message was greater than the maximal size allowed. Therefore, I have broken the replies into two separate emails. Part 1 has a summary of all replies, and Part 2 has the actual replies by all persons.
The original post was:
Dear Readers,
I am hoping that some of you who have expertise in the area of statistics and scientific journal review can help me with the following concern.
Recently I have submitted papers to peer-review journals that describe the results of investigations performed on 'small sample sizes'. Obviously, small is a relative term. For the sake of this discussion, my samples sizes have been greater than 4 and less than 10 persons.
Multiple times I have received reviewer comments that the sample size was too small, which limited my results. What concerns me is that most of my investigations involve a repeated measures design, during which subjects are tested in two or three environments, with the objective being to determine the affect the environment has on my measure. I typically use a paired t-test or repeated measures ANOVA with a post hoc evaluation if a significant main effect is determined.
In these papers, if I were to fail to reject the null hypothesis, then I can understand the concern about power and sample size. However, in many of my papers, I have determined a significant effect relative to an a priori p