Psychologists should be able to falsify predictions. A common prediction in psychological research is that a nonzero effect exists in the population. For example, one might predict that American Asian women primed with their Asian identity will perform better on a math test compared with women who are primed with their female identity. To be able to design a study that allows for strong inferences (Platt, 1964), it is important to specify which test result would falsify the hypothesis in question. Equivalence testing can be used to test whether an observed effect is surprisingly small, assuming that a meaningful effect exists in the population (see, e.g.,

No abstract

Psychologists must be able to test both for the presence of an effect and for the absence of an effect. In addition to testing against zero, researchers can use the Two One-Sided Tests (TOST) procedure to test for equivalence and reject the presence of a smallest effect size of interest (SESOI). TOST can be used to determine if an observed effect is surprisingly small, given that a true effect at least as large as the SESOI exists. We explain a range of approaches to determine the SESOI in psychological science, and provide detailed examples of how equivalence tests should be performed and reported. Equivalence tests are an important extension of statistical tools psychologists currently use, and enable researchers to falsify predictions about the presence, and declare the absence, of meaningful effects.

SummaryData analysis workflows in many scientific domains have become increasingly complex and flexible. To assess the impact of this flexibility on functional magnetic resonance imaging (fMRI) results, the same dataset was independently analyzed by 70 teams, testing nine ex-ante hypotheses. The flexibility of analytic approaches is exemplified by the fact that no two teams chose identical workflows to analyze the data. This flexibility resulted in sizeable variation in hypothesis test results, even for teams whose statistical maps were highly correlated at intermediate stages of their analysis pipeline. Variation in reported results was related to several aspects of analysis methodology. Importantly, meta-analytic approaches that aggregated information across teams yielded significant consensus in activated regions across teams. Furthermore, prediction markets of researchers in the field revealed an overestimation of the likelihood of significant findings, even by researchers with direct knowledge of the dataset. Our findings show that analytic flexibility can have substantial effects on scientific conclusions, and demonstrate factors related to variability in fMRI. The results emphasize the importance of validating and sharing complex analysis workflows, and demonstrate the need for multiple analyses of the same data. Potential approaches to mitigate issues related to analytical variability are discussed.

In response to recommendations to redefine statistical significance to p ≤ .005, we propose that researchers should transparently report and justify all choices they make when designing a study, including the alpha level.

For almost half a century, Paul Meehl educated psychologists about how the mindless use of null-hypothesis significance tests made research on theories in the social sciences basically uninterpretable. In response to the replication crisis, reforms in psychology have focused on formalizing procedures for testing hypotheses. These reforms were necessary and influential. However, as an unexpected consequence, psychological scientists have begun to realize that they may not be ready to test hypotheses. Forcing researchers to prematurely test hypotheses before they have established a sound “derivation chain” between test and theory is counterproductive. Instead, various nonconfirmatory research activities should be used to obtain the inputs necessary to make hypothesis tests informative. Before testing hypotheses, researchers should spend more time forming concepts, developing valid measures, establishing the causal relationships between concepts and the functional form of those relationships, and identifying boundary conditions and auxiliary assumptions. Providing these inputs should be recognized and incentivized as a crucial goal in itself. In this article, we discuss how shifting the focus to nonconfirmatory research can tie together many loose ends of psychology’s reform movement and help us to develop strong, testable theories, as Paul Meehl urged.

Researchers often conclude an effect is absent when a null-hypothesis significance test yields a non-significant p-value. However, it is neither logically nor statistically correct to conclude an effect is absent when a hypothesis test is not significant. We present two methods to evaluate the presence or absence of effects: Equivalence testing (based on frequentist statistics) and Bayes factors (based on Bayesian statistics). In four examples from the gerontology literature we illustrate different ways to specify alternative models that can be used to reject the presence of a meaningful or predicted effect in hypothesis tests. We provide detailed explanations of how to calculate, report, and interpret Bayes factors and equivalence tests. We also discuss how to design informative studies that can provide support for a null model or for the absence of a meaningful effect. The conceptual differences between Bayes factors and equivalence tests are discussed, and we also note when and why they might lead to similar or different inferences in practice. It is important that researchers are able to falsify predictions or can quantify the support for predicted null-effects. Bayes factors and equivalence tests provide useful statistical tools to improve inferences about null effects.

Concerns have been growing about the veracity of psychological research. Many findings in psychological science are based on studies with insufficient statistical power and nonrepresentative samples, or may otherwise be limited to specific, ungeneralizable settings or populations. Crowdsourced research, a type of large-scale collaboration in which one or more research projects are conducted across multiple lab sites, offers a pragmatic solution to these and other current methodological challenges. The Psychological Science Accelerator (PSA) is a distributed network of laboratories designed to enable and support crowdsourced research projects. These projects can focus on novel research questions, or attempt to replicate prior research, in large, diverse samples. The PSA’s mission is to accelerate the accumulation of reliable and generalizable evidence in psychological science. Here, we describe the background, structure, principles, procedures, benefits, and challenges of the PSA. In contrast to other crowdsourced research networks, the PSA is ongoing (as opposed to time-limited), efficient (in terms of re-using structures and principles for different projects), decentralized, diverse (in terms of participants and researchers), and inclusive (of proposals, contributions, and other relevant input from anyone inside or outside of the network). The PSA and other approaches to crowdsourced psychological science will advance our understanding of mental processes and behaviors by enabling rigorous research and systematically examining its generalizability.