214 International Journal of Statistics in Medical Research, 2016, 5, 214-218 E-ISSN: 1929-6029/16 © 2016 Lifescience Global Addressing the Challenge of P-Value and Sample Size when the Significance is Borderline: The Test of Random Duplication of Participants as a New Approach Jose-Gaby Tshikuka 1,2,* , Mgaywa G.M.D. Magafu 1,3 , Mooketsi Molefi 1 , Tiny Masupe 1 , Reginald B. Matchaba-Hove 4 , Bontle Mbongwe 4 and Roy Tapera 4 1 Department of Public Health Medicine, Faculty of Medicine, University of Botswana, Private bag 00713, Gaborone, Botswana 2 Department of Health Sciences, National Pedagogic University, Kinshasa I, DRC, Republic of the Congo 3 Department of Global Health, University of Washington, Seattle, USA 4 School of Public Health, Faculty of Health Sciences, University of Botswana, Private Bag 0022, Gaborone, Botswana Abstract: The issue of borderline p-value seems to divide health scientists into two schools of thought. One school of thought argues that when the p-value is greater than or equal to the statistical significance cut-off level of 0.05, it should not be considered statistically significant and the null hypothesis should be accepted no matter how close the p-value is to the 0.05. The other school of thought believes that by doing so one might be committing a Type 2 error and possibly missing valuable information. In this paper, we discuss an approach to address this issue and suggest the test of random duplication of participants as a way to interpret study outcomes when the statistical significance is borderline. This discussion shows the irrefutability of the concept of borderline statistical significance, however, it is important that one demonstrates whether a borderline statistical significance is truly borderline or not. Since the absence of statistical significance is not necessarily evidence of absence of effect, one needs to double check if a borderline statistical significance is indeed borderline or not. The p-value should not be looked at as a rule of thumb for accepting or rejecting the null hypothesis but rather as a guide for further action or analysis that leads to correct conclusions. Keywords: P-value, Sample Size, Statistical Significance, Borderline Significance, Participant Random Duplication. INTRODUCTION The p-value that indicates that an effect under study is statistically significant is a value that was established arbitrarily and by convention it should be < 0.05 [1,2]. This means that under the null hypothesis of no association, the probability of observing an effect as large as that found in the study population by chance and by chance alone is less than 5% [1-3]. This is the same as saying that chance is an unlikely explanation of the outcome [1-3]. But if the p-value is found to be ≥ 0.05, it is said that chance cannot be excluded as the likely explanation for the outcome, in which case the null hypothesis is not rejected and often the conclusion is that there is no effect or real association [4,5]. That is to say that when interpreting results from a study, we only have one of the two alternative conclusions – either the findings show that the explanatory variable under investigation has a statistically significant effect on the outcome (p < 0.05) or the explanatory variable does not have a statistically significant effect on the outcome (p > 0.05) [1-6]. *Address correspondence to this author at the Department of Public Health Medicine, Faculty of Medicine, University of Botswana, Private bag 00713, Gaborone, Botswana; Tel: +2673554603; E-mail: josegaby.tshikuka@mopipi.ub.bw A number of authors staunchly defend this viewpoint [7,8]. They argue that when the p-value is greater than 0.05, the null hypothesis should be simply accepted and that the explanatory variable shows no effect on the outcome and no other logical assertion should be considered. However, recent developments in biostatistics have shown that more experts are increasingly using such terms as “borderline significance”, “approaching significance”, “nearing significance”, and the like for outcomes whose p-values are slightly greater than or equal to 0.05 [9-12]. The criteria used to support p-values which are slightly greater than 0.05 as of borderline statistical significance or as of no statistical significance at all remain unclear to date and therefore the need for debating and clarifying this issue. This paper discusses the challenge of p-value and sample size when the statistical significance is borderline and lays ground work for scientific reasoning to avoid misinterpretation of p-values when they are slightly greater than or equal to the statistical significance cut-off level of 0.05. Correct interpretation will prevent erroneous conclusions and misguided actions that may follow.