On the use of one‐sided statistical tests in biomedical research

Abstract

SummaryThere is a tendency to automatically use two-sided tests to assess the statistical significance of experimental results. Yet if a theory predicts the direction of an experimental outcome, or if for some practical (eg clinical) reason an outcome in that direction is the only one of interest, then it makes sense to use a one-sided test. The use of a two-sided test in these situations will lead to too many false negatives. Consequently treatment effects that corroborate a theory or that are of practical importance may be missed. This problem becomes particularly acute in the case of borderline results.Following a nonsignificant one-sided test, the possibility of an effect in the direction opposite to that predicted or required can be assessed in an exploratory fashion by computing the odds in favour of such an effect. Anyone is then at liberty to pursue this possibility as they see fit. The question of whether to use a one-sided or two-sided statistical test should always be decided on logical grounds not statistical ones, and suspicions regarding the motives of the investigator(s) should be disregarded. On the other hand, this choice can be avoided altogether by assuming that a treatment always has some effect (however small) and then computing the strength of the evidence in favour of the observed or predicted/required effect (ie 1-P, where P is the one-sided significance level of the test). With this approach one-sided and two-sided tests yield identical results, and so there is effectively only one type of test.

Evaluation of fecal hormones for noninvasive research on reproduction and stress in humpback whales (Megaptera novaeangliae)

“…Since seasonal hormone cycles are often nonlinear, this analysis began with linear regression and proceeded to higher-order polynomials (2nd through 6th order examined in sequence) to assess whether nonlinear regression produced a markedly improved fit to the data. Statistical analyses were performed with Prism 6 or SPSS 11 for Mac OSX, with alpha (significance threshold) set at 0.05, with selection of tests following recommendations for small-n experimental designs and non-normal data as in Cohen (1988), Siegel and Castellan (1988), Ruxton and Neuhäuser (2010) and Murphy (2017). One-tailed tests (see Murphy, 2017;Ruxton andNeuhäuser, 2010, Siegel andCastellan, 1988) were used for comparisons of reproductive state to reproductive steroids, and of reproductive state to fGCs, since these analyses had a strong a priori predictions regarding direction of effect (based on consistent patterns reported for fecal hormones of cetaceans, other marine mammals, and mammals generally).…”

Adding exogenous biglycan or decorin improves tendon formation for equine peritenon and tendon proper cells in vitro

“…Mean values for technical replicates within biological replicates were calculated before differences of the mean values were compared within testing categories to avoid pseudoreplicates. Statistical analyses were all performed using nonparametric Wilcoxon signed-rank tests in which each treatment (5 nM bBGN, 25 nM bBGN, 5 nM bDCN, 25 nM bDCN) was compared to its corresponding control with a one-sided test applied [ 44 – 46 ]. The H 0 was that the addition of SLRPs to the constructs would lead to no improvement or a decline in tenogenic features; the H a was that the addition of SLRPs would improve tenogenic properties, or promote tendon formation, as described in several studies: (1) increased UTS, Young’s modulus, and MTL; (2) increased collagen content; (3) increases in collagen fibril numbers with increases in fibril diameters; (4) increased relative expression of tendon markers SCX and MKX , (5) increased expression of ECM assembly markers; and (6) decreased expression of perivascular markers [ 6 – 10 , 12 , 17 – 22 , 27 ].…”

Field Experiments and Job Posting Sources: The Consequences of Job Database Selection for Estimates of Racial Discrimination

“…10. We use both one-and two-tailed tests for racial discrimination given that our hypothesis about racial discrimination is directional due to existing theory and empirical research (for a discussion, see Murphy 2017;Tinkler et al 2019).…”