Let me ask you a question: If someone reading this is a disinterested third party, why should s/he believe that you, a medical doctor, know more about statistics than a Professor of Statistics at a major university?
I don't think anyone should, or rather, instead of attempting to decide whether or not something is correct based on the personal characteristics of the person presenting the information, one should focus on the information presented instead.
I have no reason to doubt that Jessica Utts knows more about statistics than me, although I will admit that I am depending upon an element of authority in order to make that call. If I read that article and the author had an undergraduate, rather than a graduate degree, or was working at a community college instead of an established university, I'd wonder just how much they knew about statistics. But because of her background, I am willing to make the assumption that she has a deep understanding of statistics. So rather than allowing me to have an inflated sense of my own understanding of statistics (which I presume is quite limited in comparison), what my experience with statisticians has taught me over the years is that my understanding of what it is that statisticians are expected to know about was wrong. It is my understanding of their scope of practice which I have had to modify. So I no longer expect statisticians to know how to weigh evidence, develop valid outcome measures, understand research design methods, test inferences, choose statistical techniques, formulate prior probabilities or display any knowledge about the subject which is undergoing testing. I do expect them to be able to know what kinds of statistical tests can be applied to specific kinds of data, how to calculate p-values and confidence intervals, how to model different kinds of distributions, or even to be aware of the assumptions which underlie the use of specific tests (although, apparently this last one is pushing the envelope). So I expect that they can provide the numerical information that is used for activities such as inference testing, weighing evidence, or assessing validity, but I don't expect them to be able to carry out those activities as it has been amply demonstrated that an understanding of the numerical information does not lead to an understanding of these other issues.
What I see when I read the article by Utts is that some of the topics covered are outside of those which fall under her reported area of expertise. And I see that she has made what appear to be errors in those topics, which makes sense if you take into consideration what I said earlier. So, I point out some of those errors, and in the past, I have explained some of them in greater detail to you. It obviously gets me nowhere with you, as I am simply someone whose information can be dismissed on the basis of personal characteristics.
What is a neutral third-party supposed to think of that? No one can be expected to judge whether or not it is appropriate to use Rosenthal's fail-safe N under these circumstances. What I say here really cannot be used to decide this issue, even if I manage to come across as credible to some. It gets decided within the scientific arena, when Utts and other parapsychologists are or are not able to interest an ever-widening circle of scientists in related fields. That they have not been able to do so is really all that is needed.
Linda
