scholarly journals Why and How We Should Join the Shift From Significance Testing to Estimation

2021 ◽  
Author(s):  
Daniel Berner ◽  
Valentin Amrhein
Keyword(s):  
Null Hypothesis ◽  
Evolutionary Biology ◽  
Statistical Power ◽  
Alternative Hypothesis ◽  
Effect Sizes ◽  
Significance Testing ◽  
Significance Threshold ◽  
Beta Error ◽  
Statistical Hypotheses ◽  
Zero Effect

A paradigm shift away from null hypothesis significance testing seems in progress. Based on simulations, we illustrate some of the underlying motivations. First, P-values vary strongly from study to study, hence dichotomous inference using significance thresholds is usually unjustified. Second, statistically significant results have overestimated effect sizes, a bias declining with increasing statistical power. Third, statistically non-significant results have underestimated effect sizes, and this bias gets stronger with higher statistical power. Fourth, the tested statistical hypotheses generally lack biological justification and are often uninformative. Despite these problems, a screen of 48 papers from the 2020 volume of the Journal of Evolutionary Biology exemplifies that significance testing is still used almost universally in evolutionary biology. All screened studies tested the default null hypothesis of zero effect with the default significance threshold of p = 0.05, none presented a pre-planned alternative hypothesis, and none calculated statistical power and the probability of ‘false negatives’ (beta error). The papers reported 49 significance tests on average. Of 41 papers that contained verbal descriptions of a ‘statistically non-significant’ result, 26 (63%) falsely claimed the absence of an effect. We conclude that our studies in ecology and evolutionary biology are mostly exploratory and descriptive. We should thus shift from claiming to “test” specific hypotheses statistically to describing and discussing many hypotheses (effect sizes) that are most compatible with our data, given our statistical model. We already have the means for doing so, because we routinely present compatibility (“confidence”) intervals covering these hypotheses.

scholarly journals The earth is flat (p>0.05): Significance thresholds and the crisis of unreplicable research

2017 ◽  
Author(s):  
Valentin Amrhein ◽  
Fränzi Korner-Nievergelt ◽  
Tobias Roth
Keyword(s):  
Publication Bias ◽  
Null Hypothesis ◽  
Statistical Power ◽  
Alternative Hypothesis ◽  
Statistical Significance ◽  
Practical Importance ◽  
Decision Rules ◽  
Effect Sizes ◽  
P Values ◽  
True Effect

The widespread use of 'statistical significance' as a license for making a claim of a scientific finding leads to considerable distortion of the scientific process (American Statistical Association, Wasserstein & Lazar 2016). We review why degrading p-values into 'significant' and 'nonsignificant' contributes to making studies irreproducible, or to making them seem irreproducible. A major problem is that we tend to take small p-values at face value, but mistrust results with larger p-values. In either case, p-values can tell little about reliability of research, because they are hardly replicable even if an alternative hypothesis is true. Also significance (p≤0.05) is hardly replicable: at a realistic statistical power of 40%, given that there is a true effect, only one in six studies will significantly replicate the significant result of another study. Even at a good power of 80%, results from two studies will be conflicting, in terms of significance, in one third of the cases if there is a true effect. This means that a replication cannot be interpreted as having failed only because it is nonsignificant. Many apparent replication failures may thus reflect faulty judgement based on significance thresholds rather than a crisis of unreplicable research. Reliable conclusions on replicability and practical importance of a finding can only be drawn using cumulative evidence from multiple independent studies. However, applying significance thresholds makes cumulative knowledge unreliable. One reason is that with anything but ideal statistical power, significant effect sizes will be biased upwards. Interpreting inflated significant results while ignoring nonsignificant results will thus lead to wrong conclusions. But current incentives to hunt for significance lead to publication bias against nonsignificant findings. Data dredging, p-hacking and publication bias should be addressed by removing fixed significance thresholds. Consistent with the recommendations of the late Ronald Fisher, p-values should be interpreted as graded measures of the strength of evidence against the null hypothesis. Also larger p-values offer some evidence against the null hypothesis, and they cannot be interpreted as supporting the null hypothesis, falsely concluding that 'there is no effect'. Information on possible true effect sizes that are compatible with the data must be obtained from the observed effect size, e.g., from a sample average, and from a measure of uncertainty, such as a confidence interval. We review how confusion about interpretation of larger p-values can be traced back to historical disputes among the founders of modern statistics. We further discuss potential arguments against removing significance thresholds, such as 'we need more stringent decision rules', 'sample sizes will decrease' or 'we need to get rid of p-values'.

The null-hypothesis significance-test procedure is still warranted

1998 ◽  
Vol 21 (2) ◽  
pp. 228-235 ◽  
Author(s):  
Siu L. Chow
Keyword(s):  
Null Hypothesis ◽  
Statistical Power ◽  
Meta Analysis ◽  
Statistical Significance ◽  
Test Procedure ◽  
Significance Test ◽  
Null Hypothesis Significance Test ◽  
Wide Range ◽  
Statistical Hypotheses ◽  
Alternative Hypotheses

Entertaining diverse assumptions about empirical research, commentators give a wide range of verdicts on the NHSTP defence in Statistical significance. The null-hypothesis significance-test procedure (NHSTP) is defended in a framework in which deductive and inductive rules are deployed in theory corroboration in the spirit of Popper's Conjectures and refutations (1968b). The defensible hypothetico-deductive structure of the framework is used to make explicit the distinctions between (1) substantive and statistical hypotheses, (2) statistical alternative and conceptual alternative hypotheses, and (3) making statistical decisions and drawing theoretical conclusions. These distinctions make it easier to show that (1) H0 can be true, (2) the effect size is irrelevant to theory corroboration, and (3) “strong” hypotheses make no difference to NHSTP. Reservations about statistical power, meta-analysis, and the Bayesian approach are still warranted.

Confidence Intervals

2009 ◽  
Vol 217 (1) ◽  
pp. 15-26 ◽  
Author(s):  
Geoff Cumming ◽  
Fiona Fidler
Keyword(s):  
Confidence Interval ◽  
Confidence Intervals ◽  
Null Hypothesis ◽  
Prediction Intervals ◽  
Effect Sizes ◽  
Significance Testing ◽  
Null Hypothesis Significance Testing ◽  
P Values ◽  
Standardized Effect Sizes ◽  
Mainstream Science

Most questions across science call for quantitative answers, ideally, a single best estimate plus information about the precision of that estimate. A confidence interval (CI) expresses both efficiently. Early experimental psychologists sought quantitative answers, but for the last half century psychology has been dominated by the nonquantitative, dichotomous thinking of null hypothesis significance testing (NHST). The authors argue that psychology should rejoin mainstream science by asking better questions – those that demand quantitative answers – and using CIs to answer them. They explain CIs and a range of ways to think about them and use them to interpret data, especially by considering CIs as prediction intervals, which provide information about replication. They explain how to calculate CIs on means, proportions, correlations, and standardized effect sizes, and illustrate symmetric and asymmetric CIs. They also argue that information provided by CIs is more useful than that provided by p values, or by values of Killeen’s prep, the probability of replication.

scholarly journals When to adjust alpha during multiple testing: A consideration of disjunction, conjunction, and individual testing.

2021 ◽  
Author(s):  
Mark Rubin
Keyword(s):  
Present Article ◽  
Null Hypothesis ◽  
Multiple Testing ◽  
Multiple Comparisons ◽  
Significance Testing ◽  
Null Hypothesis Significance Testing ◽  
Significance Threshold ◽  
Alpha Level ◽  
Replication Crisis ◽  
Test Result

Scientists often adjust their significance threshold (alpha level) during null hypothesis significance testing in order to take into account multiple testing and multiple comparisons. This alpha adjustment has become particularly relevant in the context of the replication crisis in science. The present article considers the conditions in which this alpha adjustment is appropriate and the conditions in which it is inappropriate. A distinction is drawn between three types of multiple testing: disjunction testing, conjunction testing, and individual testing. It is argued that alpha adjustment is only appropriate in the case of disjunction testing, in which at least one test result must be significant in order to reject the associated joint null hypothesis. Alpha adjustment is inappropriate in the case of conjunction testing, in which all relevant results must be significant in order to reject the joint null hypothesis. Alpha adjustment is also inappropriate in the case of individual testing, in which each individual result must be significant in order to reject each associated individual null hypothesis. The conditions under which each of these three types of multiple testing is warranted are examined. It is concluded that researchers should not automatically (mindlessly) assume that alpha adjustment is necessary during multiple testing. Illustrations are provided in relation to joint studywise hypotheses and joint multiway ANOVAwise hypotheses.

scholarly journals Haplotype-based eQTL mapping finds evidence for complex gene regulatory regions poorly tagged by marginal SNPs

2018 ◽  
Author(s):  
Robert Brown ◽  
Sriram Sankararaman ◽  
Bogdan Pasaniuc
Keyword(s):  
Gene Expression ◽  
Multiple Testing ◽  
Additive Model ◽  
Effect Sizes ◽  
Independent Effect ◽  
Nucleotide Polymorphisms ◽  
Multiple Snps ◽  
Significance Threshold ◽  
Association Data ◽  
Zero Effect

AbstractMotivationExpression quantitative trait loci (eQTLs), variations in the genome that impact gene expression, are identified through eQTL studies that test for a relationship between single nucleotide polymorphisms (SNPs) and gene expression levels. These studies typically assume an underlying additive model. Non-additive tests have been proposed, but are limited due to the increase in the multiple testing burden and are potentially biased by filtering criteria that relies on marginal association data. Here we propose using combinations of short haplotypes instead of SNPs as predictors for gene expression. Essentially, this method looks for genomic regions where haplotypes have different effect sizes. The differences in effect can be due to multiple genetic architectures such as a single SNP, a burden of rare SNPs, multiple SNPs with independent effect or multiple SNPs with an interaction effect occurring on the same haplotype.ResultsSimulations show that when haplotypes, rather than SNPs, are assigned non-zero effect sizes, our method has increased power compared to the marginal SNP method. In the GEUVADIS gene expression data, our method finds 101 more eGenes than the marginal method (5,202 vs. 5,101). The methods do not have full overlap in the eGenes that they find. Of the 5,202 eGenes found by our method, 707 are not found by the marginal method—even though it has a lower significance threshold. This indicates that many genes have regulatory architectures that are not well tagged by marginal SNPs and demonstrates the need to better model alternative archi-tectures.

scholarly journals The earth is flat (p > 0.05): significance thresholds and the crisis of unreplicable research

PeerJ ◽  
2017 ◽  
Vol 5 ◽  
pp. e3544 ◽  
Author(s):  
Valentin Amrhein ◽  
Fränzi Korner-Nievergelt ◽  
Tobias Roth
Keyword(s):  
Publication Bias ◽  
Null Hypothesis ◽  
Statistical Power ◽  
Statistical Significance ◽  
Practical Importance ◽  
Effect Sizes ◽  
Point Estimate ◽  
Selective Reporting ◽  
Interval Estimate ◽  
True Effect

The widespread use of ‘statistical significance’ as a license for making a claim of a scientific finding leads to considerable distortion of the scientific process (according to the American Statistical Association). We review why degradingp-values into ‘significant’ and ‘nonsignificant’ contributes to making studies irreproducible, or to making them seem irreproducible. A major problem is that we tend to take smallp-values at face value, but mistrust results with largerp-values. In either case,p-values tell little about reliability of research, because they are hardly replicable even if an alternative hypothesis is true. Also significance (p ≤ 0.05) is hardly replicable: at a good statistical power of 80%, two studies will be ‘conflicting’, meaning that one is significant and the other is not, in one third of the cases if there is a true effect. A replication can therefore not be interpreted as having failed only because it is nonsignificant. Many apparent replication failures may thus reflect faulty judgment based on significance thresholds rather than a crisis of unreplicable research. Reliable conclusions on replicability and practical importance of a finding can only be drawn using cumulative evidence from multiple independent studies. However, applying significance thresholds makes cumulative knowledge unreliable. One reason is that with anything but ideal statistical power, significant effect sizes will be biased upwards. Interpreting inflated significant results while ignoring nonsignificant results will thus lead to wrong conclusions. But current incentives to hunt for significance lead to selective reporting and to publication bias against nonsignificant findings. Data dredging,p-hacking, and publication bias should be addressed by removing fixed significance thresholds. Consistent with the recommendations of the late Ronald Fisher,p-values should be interpreted as graded measures of the strength of evidence against the null hypothesis. Also largerp-values offer some evidence against the null hypothesis, and they cannot be interpreted as supporting the null hypothesis, falsely concluding that ‘there is no effect’. Information on possible true effect sizes that are compatible with the data must be obtained from the point estimate, e.g., from a sample average, and from the interval estimate, such as a confidence interval. We review how confusion about interpretation of largerp-values can be traced back to historical disputes among the founders of modern statistics. We further discuss potential arguments against removing significance thresholds, for example that decision rules should rather be more stringent, that sample sizes could decrease, or thatp-values should better be completely abandoned. We conclude that whatever method of statistical inference we use, dichotomous threshold thinking must give way to non-automated informed judgment.

DETECTING DISEASE CLUSTERS: THE IMPORTANCE OF STATISTICAL POWER

1990 ◽  
Vol 132 (supp1) ◽  
pp. 156-166 ◽  
Author(s):  
DANIEL WARTENBERG ◽  
MICHAEL GREENBERG
Keyword(s):  
Null Hypothesis ◽  
Statistical Power ◽  
Alternative Hypothesis ◽  
The Other ◽  
Health Departments ◽  
Occupancy Models ◽  
False Negatives ◽  
Disease Clusters ◽  
Power Studies ◽  
Cluster Methods

Abstract A variety of methods and models have been proposed for the statistical analysis of disease excesses, yet rarely are these methods compared with respect to their ability to detect possible clusters. Evaluation of statistical power is one approach for comparing different methods. In this paper, the authors study the probability that a test will reject the null hypothesis, given that the null hypothesis is indeed false. They present a discussion of some considerations involved in power studies of cluster methods and review two methods for detecting space-time clusters of disease, one based on cell occupancy models and the other based on interevent distance comparisons. The authors compare these approaches with respect to: 1) the sensitivity to detect disease excesses (false negatives); 2) the likelihood of detecting clusters that do not exist (false positives); and 3) the structure of a cluster in a given investigation (the alternative hypothesis). The methods chosen, which are two of the most commonly used, are specific to different hypotheses. They both show low power for the small number of cases which are typical of citizen reports to health departments.

Things We Still Haven’t Learned (So Far)

2015 ◽  
Vol 37 (4) ◽  
pp. 449-461 ◽  
Author(s):  
Andreas Ivarsson ◽  
Mark B. Andersen ◽  
Andreas Stenling ◽  
Urban Johnson ◽  
Magnus Lindwall
Keyword(s):  
Bayesian Statistics ◽  
Null Hypothesis ◽  
Historical Background ◽  
Effect Sizes ◽  
Significance Testing ◽  
Exercise Psychology ◽  
Null Hypothesis Significance Testing ◽  
The Past ◽  
The Common ◽  
Sport And Exercise

Null hypothesis significance testing (NHST) is like an immortal horse that some researchers have been trying to beat to death for over 50 years but without any success. In this article we discuss the flaws in NHST, the historical background in relation to both Fisher’s and Neyman and Pearson’s statistical ideas, the common misunderstandings of what p < 05 actually means, and the 2010 APA publication manual’s clear, but most often ignored, instructions to report effect sizes and to interpret what they all mean in the real world. In addition, we discuss how Bayesian statistics can be used to overcome some of the problems with NHST. We then analyze quantitative articles published over the past three years (2012–2014) in two top-rated sport and exercise psychology journals to determine whether we have learned what we should have learned decades ago about our use and meaningful interpretations of statistics.

scholarly journals The earth is flat (p>0.05): Significance thresholds and the crisis of unreplicable research

2017 ◽  
Author(s):  
Valentin Amrhein ◽  
Fränzi Korner-Nievergelt ◽  
Tobias Roth
Keyword(s):  
Publication Bias ◽  
Null Hypothesis ◽  
Statistical Power ◽  
Statistical Significance ◽  
Practical Importance ◽  
Effect Sizes ◽  
Selective Reporting ◽  
Interval Estimate ◽  
P Values ◽  
True Effect

The widespread use of 'statistical significance' as a license for making a claim of a scientific finding leads to considerable distortion of the scientific process (according to the American Statistical Association). We review why degrading p-values into 'significant' and 'nonsignificant' contributes to making studies irreproducible, or to making them seem irreproducible. A major problem is that we tend to take small p-values at face value, but mistrust results with larger p-values. In either case, p-values tell little about reliability of research, because they are hardly replicable even if an alternative hypothesis is true. Also significance (p≤0.05) is hardly replicable: at a good statistical power of 80%, two studies will be 'conflicting', meaning that one is significant and the other is not, in one third of the cases if there is a true effect. A replication can therefore not be interpreted as having failed only because it is nonsignificant. Many apparent replication failures may thus reflect faulty judgment based on significance thresholds rather than a crisis of unreplicable research. Reliable conclusions on replicability and practical importance of a finding can only be drawn using cumulative evidence from multiple independent studies. However, applying significance thresholds makes cumulative knowledge unreliable. One reason is that with anything but ideal statistical power, significant effect sizes will be biased upwards. Interpreting inflated significant results while ignoring nonsignificant results will thus lead to wrong conclusions. But current incentives to hunt for significance lead to selective reporting and to publication bias against nonsignificant findings. Data dredging, p-hacking, and publication bias should be addressed by removing fixed significance thresholds. Consistent with the recommendations of the late Ronald Fisher, p-values should be interpreted as graded measures of the strength of evidence against the null hypothesis. Also larger p-values offer some evidence against the null hypothesis, and they cannot be interpreted as supporting the null hypothesis, falsely concluding that 'there is no effect'. Information on possible true effect sizes that are compatible with the data must be obtained from the point estimate, e.g., from a sample average, and from the interval estimate, such as a confidence interval. We review how confusion about interpretation of larger p-values can be traced back to historical disputes among the founders of modern statistics. We further discuss potential arguments against removing significance thresholds, for example that decision rules should rather be more stringent, that sample sizes could decrease, or that p-values should better be completely abandoned. We conclude that whatever method of statistical inference we use, dichotomous threshold thinking must give way to non-automated informed judgment.

Sign in / Sign up

Export Citation Format

Share Document