Monday, 22 October 2018

The Last Review I'll Do For the Open Journal of Statistics

The following review is for a paper that is currently published in the Scientific Research Publishing's Open Journal of Statistics (link omitted intentionally). I accepted the responsibility to review it, found it unfit for publication, and returned a review less than a month later only to find that the paper had already been published as is.

I won't be reviewing for this journal or this publishing house again. To ask for my review, get my assent, and not wait a reasonable time (a month, really?), for the review before going ahead and publishing is disrespectful of my time. It also smacks of predatory journal behaviour. 
Was there another peer reviewer and were they qualified to perform the review? Read my review, and the paper, and judge for yourself.

Review for ‘Has Tom Brady Passed His Prime?’, Paul M. Sommers, submitted to the Open Journal of Statistics 2018.
Review by Jack Davis.
The subject matter of this paper is interesting, especially in the context of Brady’s recently published book about maintaining peak performance over a long period.
The statistical content of this paper consists of the formulae for the sample mean, standard deviation, and coefficient of variation, followed by three sets of t-tests on seven variables.
In all tables, a t-test is appropriate for at most four of the seven variables.  The other three, completion percentage, touchdowns per game, and intercepts per game, deserve different statistical tests.
Completion percentage is a proportion and it needs a proportion test, a 2x2 chi-squared test, or an odds ratio test in order to see a difference between the two proportions, not a t-test.
Touchdowns and intercepts are rare count events. They follow a Poisson distribution much more closely than a normal distribution. A much more appropriate test for equality of touchdown and interception rates is the Likelihood Ratio Test.
In Table 2, how were the years chosen for the before-and-after comparison?

In Table 3, are there significant differences between the coefficients of variation?

Rather than just comparing years, why not do a regression analysis of the trends in each of these variables over time? With plots of these statistics over time, we could see if there is a decline in recent years. Finally, allowances should be made for multiple testing.
Finally, the title has a grammar error. It should read either “Has Tom Brady Passed his Prime” or “Is Tom Brady Past his Prime” but not “has . . . passed”.
With more statistical details and rigour, this could be an excellent manuscript. However, in its current state, it feels incomplete and not ready for publication.
Recommendation: Major revisions or reject.


Chica wonders if 'passed' was supposed to be wordplay.