Education for Health

LETTER TO THE EDITOR
Year
: 2019  |  Volume : 32  |  Issue : 3  |  Page : 135-

Time to reconsider our obsession with statistical significance


Sumeet Bhatt 
 Department of Public Health Dentistry, Himachal Institute of Dental Sciences, Paonta Sahib, Himachal Pradesh, India

Correspondence Address:
Sumeet Bhatt
Department of Public Health Dentistry, Himachal Institute of Dental Sciences, Paonta Sahib, Himachal Pradesh
India




How to cite this article:
Bhatt S. Time to reconsider our obsession with statistical significance.Educ Health 2019;32:135-135


How to cite this URL:
Bhatt S. Time to reconsider our obsession with statistical significance. Educ Health [serial online] 2019 [cited 2022 Nov 27 ];32:135-135
Available from: https://educationforhealth.net//text.asp?2019/32/3/135/282874


Full Text



Dear Editor,

In 2016, The American Statistical Association recommended several changes to the ongoing statistical practices in the research field worldwide.[1] It mainly involves the misuse of value of P < 0.05 as the threshold for statistical significance. This practice is employed by researchers the world over, but it can result in researchers being more concerned about a study's “statistical significance” rather than its “practical significance.” The original purpose of statistical significance was as an indication when a result warrants further scrutiny. However that idea has lost its meaning altogether, and statistical inference is now being incorrectly equated with scientific inference,[2] thus becoming the sole criterion for judging a result's practical significance.

Furthermore, there has been observed a “publication bias,” favoring the studies with “novel and significant” results.[3] Our “publish or perish” culture is doing more harm to science than benefit. Since more journals accept the “P < 0.05 studies,” there is a typical direction toward claiming greater significance, with authors shaping the presentation of results to make it more palatable.

“It is important to identify the various misconceptions associated with P value,[4] such as the belief that P value shows us the probability that the observed effect is due to chance.” ASA recommends that P value should be reported on a continuous scale rather than categorical as a threshold value of less than- or >0.05. The results, instead of statistical significance, should include a measure of practical significance of the study, for example, effect size. The journal editorial boards should be encouraged to disallow the use of the phrase “statistically significant” in manuscripts they will accept for review. It is essential to focus on “significant sameness” (replication studies) rather than our obsession with significant differences.

There is a pressing need to seriously rethink the role of statistics in scientific advancement. This change should be incorporated in statistics taught in medical institutes as well.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

References

1Wasserstein R, Lazar N. The ASA's statement on p-values: Context, process, and purpose. Am Stat 2016;70:129-33.
2Hubbard R, Haig BD, Parsa RA. The limited role of formal statistical inference in scientific inference. Am Stat 2019;73 Suppl 1:91-8.
3Dickersin K. The existence of publication bias and risk factors for its occurrence. JAMA 1990;263:1385-9.
4Goodman S. A dirty dozen: Twelvep-value misconceptions. Semin Hematol 2008;45:135-40.