|Year : 2015 | Volume
| Issue : 3 | Page : 106
Significance of statistical significance
Amit Vasant Mahuli, Simpy Amit Mahuli
Department of Public Health Dentistry, NIMS Dental College, Jaipur, Rajasthan, India
|Date of Web Publication||19-Nov-2015|
Dr. Amit Vasant Mahuli
Department of Public Health Dentistry, NIMS Dental College, Jaipur, Rajasthan
Source of Support: None, Conflict of Interest: None
|How to cite this article:|
Mahuli AV, Mahuli SA. Significance of statistical significance. J Dent Res Rev 2015;2:106
The probability that an outcome or relationship is brought about by something other than just mere random chance or extrapolation of the study results to population at large in terms of P value (probability value). Statistical significance is over-hyped phenomenon in bio-medical research, misconception that non-statistical significant results or research which accepts null hypothesis is branded as inferior research has become one of the main barriers in generating quality evidence for any research question. The understating of clinical significance and statistical significance is important in bio-medical research; clinical significance of a nonstatistically significant result can provide hope in the line of treatment of many rare diseases. The researcher's acumen to relate, P value, confidence interval (CI), α (alpha) error, β (beta) error, and power of study holds the key to successful research outcome.
Sample size and its relation to the statistical significance is immense. Sample size is influenced by factors such as effect size or time points or difference expected between variables, homogeneity of participants, acceptable risk of error that investigators consider (alpha and beta error), and rate of attrition expected during the study [Flow Chart 1]. Bio-medical research has started looking beyond P value in terms of understanding CI and clinical significance as the outcome of research. The calculation of CIs assumes the data variable being distributed normally. As in a normal curve, 68% of all observations lie within ± 1 standard deviation (SD) from the mean, 95% of all observations lie within ± 1.96 stds, and 99% of observations lie with ± 2.58 stds. When calculating the 95% CI, 1.96 is used as the multiplier. CI = 1.96 × standard error (SE), SE is std divided by the square root of the number of observations in the study, or SE= SD/√n. Research hosts wide possibilities of analysis and outcome measurements, even though P value is a robust measure, the understanding and interpretation of research can go beyond P value and present vital evidence in bio-medical research.,
| References|| |
Fethney J. Statistical and clinical significance, and how to use confidence intervals to help interpret both. Aust Crit Care 2010;23:93-7.
Burmeister E, Aitken LM. Sample size: How many is enough? Aust Crit Care 2012;25:271-4.
Pereira SM, Leslie G. Hypothesis testing. Aust Crit Care 2009;22:187-91.