“Noferror’s rebuttal of my use of medians was valid, but some of the comments made by other statisticians were just plain mean,” said Stribution.
Photo by Maria Dhilla
In what was called “a significant breakthrough” by statisticians worldwide, Dr. Sam Pling and Dr. Didi Stribution have recently published a research paper in the Journal of Applied Statistics titled “Phenomenally Phe-nominal: A Diatribe on How to Make Data Points Feel Good About Themselves Even if They Are Not Significant.” It highlights their new methodology for classifying extremely special, worthy, and top-notch kinds of data: the phe-nominal scale.
Dr. Stribution elaborated further on the reasoning behind their research: “When a researcher finds out that their data is not statistically significant, it can be soul crushing,” wrote Stribution. “It really makes you feel 78% sure that 63.3% of the time 45% of your existence on 3.6% of this Earth is 99.9% meaningless. Dr. Indie Pendent, my statistically significant other, told me that I’m going through a midlife crisis, however, I know that the field of statistics is truly responsible for my inevitable demise. That’s why Sam and I decided it was time for a change.”
As explained in an interview with Dr. Pling, the nominal scale is one of four existing scales, and is used to assign events, objects, or other qualitative data into discrete, unordered categories. “For instance,” said Pling, “I once went to the doctor because I punched through the drywall after my research got rejected again, and there I filled out a form denoting my gender and blood type. Those are examples of nominal categories! Does that ring a bell curve? The phe-nominal scale, on the other hand, would create categories for ‘whatever data that the researcher deems important,’ which is usually all of it.”
However, other statisticians (sample size n = 128) like the renowned Dr. Margi Noferror, have brought up an average of 3.4 issues per sentence (σ= 24.1) within the paper. Dr. Noferror wrote a letter to the Journal of Applied Statistics, claiming that Dr. Pling and Dr. Stribution’s methodology for testing the phe-nominal scale was faulty. “In the paper, my disreputable colleagues listed their confidence level for this new scale as 100%, which may sound great, but it’s actually 100% bad news,” wrote Noferror. “The field of statistics is confounding like that.”
Other researchers pointed out that, instead of rejecting the null hypothesis, the authors rejected the idea of having a null hypothesis altogether, writing “Null hypothesis? More like dull hypothesis.” Cora Lation, a statistics graduate student, commented, “I know Dr. Pling is an advocate for alternative hypothesis testing, but this is just three standard deviations too far from the norm.” Another graduate student mentioned that “the paper says Cohen’s d was very large, but I don’t think they realized that d must be between 0 and 1; they gave some other measure of d instead.”
When asked to explain whether or not their research is correct, Pling and Stribution instead stated: “There may be a 1 in 1,072,469 chance that we encountered a Type 1 error, but it’s 2905.005 times more likely that critics are extrapolating beyond the scope of our research.” Dr. Pling and Dr. Stribution further responded to criticism, saying: “We don’t appreciate that the statistical community is basically telling us ‘your data is full of anomalies,’ and ‘you must be getting out, liars!’”
The Journal of Applied Statistics has released a statement about the burgeoning controversy, saying that “on average, the mean comments about this published work are not our preferred mode of handling the wide range of critical reviews. We hope that this sample of the wider population of statisticians will take ANOVA look at this paper to determine its significance.”
M. Q. Dhilla is the embodiment of the newspaper come to life, but that's a secret, so don’t effin’ snitch.
