Biostatistics and the genius (Mozard) from Bauru (fob/Usp)

This comment is my opinion about scientific Works that do not present statistic analysis of the results.

_____________________________________________________

In spite of the idea that “if your experiment needs statistic, you ought to done a better experiment,” several other designs only based in the routine microscopic exam without utilization of histomorphometry and statistical analyze of the results are inaccurate, generators of dogma, mite, lucubration, discussions and equivocate conclusions. Of course, reading some scientific works that did not present statistical analysis of the results thought: “As he/she may make this affirmation without statistic analysis?” For example, I cite the report of the discovery of a new oral lesion (Collagenous fibroma) that presents similar histopathological aspect to yet well known other oral lesions (Fibrous reactional hyperplasia) such as describe by professor at the FO/USP, Cavalcante de Araújo et al. The inexpert reader would ask me: why? Because the histochemistry examination of these lesions is similar; and the simple subjective qualitative analysis of the pattern of distribution of collagens fibers is variable from observer to observer. In addition, this pattern of distribution of collagens fibers is also variable from region to region exanimated at the microscopic.  In other words, is possible to identify in some regions of inflammatory reactional hyperplasia (mature or "old" lesions) areas similar to collagenous fibroma as well as at a collagenous fibroma secondary infected areas of inflammatory hyperplasia fibrous such as I showed using different microscopic fields of the reactional hyperplasia presented by Cavalcante de Araújo et al.. Therefore, a statistical difference at the volumetric of point of collagen fibers among fibrous lesions of the oral cavity might indicate a different biological behavior, and so, a new lesion (collagenous fibroma) that according to Cavalcante de Araújo et al. would have a more aggressive behavior. For me, who I am a normal person is difficult to establish the difference at the histopathological aspect without histomorphometric study and statistical analysis of the results. For the genius is not. He/she sees the lesion and observes the difference that normal person without statistic analysis does not identify. This is a characteristic of the geniality. During my post-doctoral study at the Bauru School of Dentistry at the Sao Paulo University (fob/USP), Bauru, Brazil, at a class ministered by the genius (Mozard) of Bauru, hearing several times his affirmation: “I do not like to use statistical analyzes in my works because changing the test obtain different result”, incited me another think : “would be truth or truly?”. However, the opinion of a genius (he/she) must always be considered because they visualize what that normal persons do not. Of manner that for years, systematically, what erudite think of my ex-professor at fob/usp, Professor Consolara, come back to brain torturing me. Was necessary for me confirms it or not.  Afterward my demission at the Federal University of Santa Maria (UFSM), Brazil - by a fraudulent administrative process for not justified lack during my post-doctoral study at the fob/usp - disposing of time, decided to verify the veracity of the affirmation of my ex-professor at the fob/usp who is considered a genius for the Brazilian Oral Pathology. For that, used the data of my doctoral thesis comparing the results of the statistical analysis that there was been made by same professor of statistic that also there yet was made the statistic analysis of my master thesis at the Federal University of Rio de Janeiro (UFRJ) and who works at the biostatistics division at the Cancer National Institute (Inca), RJ, Brazil, with different statistical parametric test using Excel program for me programed.  As at my post-doctoral thesis there was been used non-parametric variance analysis of Friedman more the test of multiple compare, I compared the results obtained with these tests with the considered the most potent parametric test (Test F) and student's test for comparing difference between only two groups. In addition, as our work there also was been realized in mice, I used these results individually and in conjunct with the data obtained with the study at rats selecting one group (Methotrexate); identifying similar results what obtained with non-parametric variance analysis of Friedman. In conclusion, the results to the group methotrexate when compared with the control were similar using non-parametric variance analysis of Friedman more multiple compare or different parametric tests (variance analysis + Student's test). So, the different tests pertinent to statistical analysis of determinate data - when well selected - produce similar statistical analysis. Therefore, the results of my post-doctoral thesis at the fob/usp, when submitted to different statistical tests (parametric and not parametric tests)  indicated to data presented similar results, contrasting with the erudite affirmation of the professor at fob/usp, Alberto Consolaro (the genius of Bauru). Indeed, I also do not agree with the initial think: “if your experiment needs statistic, you ought to done a better experiment.” In my opinion the best experiment is what that may also be statically proofed. Therefore, you (who is not a genius) ought to make a better experiment always that different statistical tests - indicated to the study - present different levels of significance. For the genius, these differences are irrelevant front to increased visual acuity. For us that are normal biostatistics helps to have critical sense in the interpretation of the results.