The empirical record on masculinity threat is now large enough to support quantitative synthesis, but it is also heterogeneous enough to warrant careful interpretation. A recent comprehensive meta-analysis of masculinity-threat experiments in heterosexual men found an overall small-to-medium effect, d = 0.45, with effects varying across components of the threat process, including trigger type, consequence type, publicity, and outcome order. Internalized responses such as negative emotions and reduced self-perceived masculinity were stronger than externalized compensatory reactions. Yet at the same time, recent preregistered replication work has failed to reproduce some influential masculinity-threat effects in a large, nationally representative U.S. sample. In particular, a replication and extension of Willer et al.’s highly cited masculine-overcompensation experiments found no consistent evidence that masculinity threat altered men’s political attitudes, whereas the original studies had reported that threat increased support for war, homophobic attitudes, and dominance-related political positions.

This failure need not imply that PMT is false as a theory. It may instead reflect a false-positive original result, changes in historical context, weaker threat salience in an online survey, differences between student and population-based samples, or theoretically important boundary conditions. Such ambiguity motivated the authors to conduct a field-level credibility analysis: the same literature may contain genuine effects, boundary-conditioned effects, inflated estimates, and false-positive findings all at once. Conventional meta-analyses primarily address how large masculinity-threat effects are on average and across moderators. The second question—how credible is the published significant-results record from which those effect-size estimates are derived—is less straightforward. Audits of more than 1,000 meta-analyses suggest that publication-bias assessment and correction are applied unevenly, heterogeneity and dependence among effect sizes are common, and conclusions often rely on diagnostics with limited sensitivity.

To answer this second question, the researchers turned to z-curve, a selection-model-based approach for characterizing the evidentiary value of a published record under the premise that statistically significant results have a higher probability of entering and remaining in the visible literature than non-significant results. Z-curve does not estimate average effect sizes or researchers’ planned statistical power. Instead, it estimates credibility-related properties of a literature based on the observed distribution of test statistics. The method produces several indicators. The Observed Discovery Rate (ODR) is the observed proportion of extracted tests that are statistically significant at a chosen threshold, usually α = 0.05. The Expected Discovery Rate (EDR) is the model-based estimate of the proportion of significant results expected across the underlying set of conducted studies before selection for significance is taken into account. The Expected Replication Rate (ERR) is the estimated average probability of significance among the subset of studies that produced significant results, corresponding to the expected success rate of close replications using the same design, sample size, and threshold. Finally, the False-Discovery Risk (FDR) is an upper-bound estimate of the proportion of statistically significant findings that could be false discoveries, derived from the EDR using Sorić’s logic: if only a small proportion of all conducted studies are expected to produce significant results before publication selection, then even a relatively small number of false positives can constitute a substantial share of the published significant findings.

The analysis was preregistered on the Open Science Framework, and the authors searched five bibliographic databases—APA PsycINFO, ProQuest, Web of Science Core Collection, Scopus, and DOAJ—completing the search in July 2025. After deduplication, 641 unique records underwent title and abstract screening by two reviewers, with 140 proceeding to full-text screening. Citation tracing yielded additional eligible records. The final coded corpus comprised 188 unique samples contributing 452 focal hypothesis tests, drawn from 99 peer-reviewed journal articles, 13 doctoral dissertations, and 11 master’s theses. Publication years ranged from 1979 to 2025, and data were collected in 21 countries, though most samples came from the United States. Sample sizes of men only varied from N = 24 to N = 867, with a median of N = 126. The researchers coded 14 distinct manipulation types, with the most common involving false feedback: bogus gender-knowledge test feedback and false personality feedback based on instruments such as the Bem Sex Role Inventory and the Conformity to Masculine Norms Inventory. Other frequently used manipulations included writing tasks, interpersonal threat, general threat-to-men messages, and failed masculine-task procedures. Only 16 of 188 samples—8.5 percent—were formally preregistered.

Because multiple eligible tests could be extracted from the same study, the authors addressed within-article dependence using a cluster bootstrap with 10,000 iterations, sampling one test per article in each iteration. Each eligible result was represented as a two-tailed p-value converted to a z-score, the standard input for z-curve. The model treats the observed significant z-scores as a mixture of truncated folded-normal distributions, with mixture weights estimated via an expectation–maximization algorithm to best reproduce the observed distribution, after which the model extrapolates to the implied full distribution including non-significant results. Tests based on very small samples (N < 30) were excluded because conversion to z-scores would be unreliable. The authors also manually coded focal hypothesis tests rather than automatically scraping statistics, and did not code manipulation checks or auxiliary analyses.

The headline finding is a marked gap between the observed discovery rate and the expected discovery rate at α = 0.05: ODR ≈ 0.51 versus EDR ≈ 0.10, with non-overlapping confidence intervals. In other words, roughly half of the extracted focal tests were statistically significant, but the fitted selection model implies that only about 10 percent of all conducted studies would have been expected to yield significant results before selection for significance is taken into account. This discrepancy points to substantial selection for statistical significance across the experimental PMT literature. The clustered z-curve estimate of the expected replication rate was approximately 0.38, suggesting that exact replications using similar designs and sample sizes would be expected to succeed less than half the time at the field level. The Sorić FDR at α = 0.05 was non-trivially large, indicating that a sizeable share of statistically significant findings could be false positives, although the confidence interval was imprecisely estimated.

Subgroup analyses revealed informative patterns. Main-effect tests showed a lower ODR than interaction tests but relatively higher EDR and ERR and a lower FDR, whereas interaction tests had the highest ODR in the table but comparatively low EDR and ERR and a high FDR. This pattern suggests that interaction findings in this literature are especially prone to selection bias and less likely to replicate than main-effect findings, consistent with the experimental design reality that interactions typically require larger samples than main effects to be estimated with adequate precision. Preregistered tests showed a higher EDR, higher ERR, and a much lower Sorić FDR than non-preregistered tests, even though confidence intervals were wide due to the smaller number of preregistered studies. Comparing pre-2015 and post-2015 studies pointed to a modest improvement in evidential value over time: earlier studies showed high ODR combined with very low EDR and ERR and a high FDR, while more recent studies displayed a somewhat lower ODR but clearly higher EDR and ERR and a lower FDR, although selection bias remained visible. An exploratory analysis by outcome domain suggested that affective outcomes showed the most successful published record but did not exhibit a clearly stronger underlying evidential profile, whereas behavioral outcomes appeared most vulnerable to false discoveries, with the highest estimated FDR and lowest ERR. Manipulation types also differed: general threat-to-men and interpersonal threat manipulations showed comparatively higher EDR and ERR alongside lower FDR, whereas writing tasks and heterogeneous other-threat categories combined relatively high ODRs with very low EDRs, low ERRs, and high FDRs.

A practically important implication concerns the significance threshold itself. As a retrospective sensitivity analysis, the authors examined how z-curve indicators change when more stringent thresholds are applied to a literature that was overwhelmingly conducted and published under the conventional α = 0.05. Within the existing literature, lowering the analysis threshold from α = 0.05 to α = 0.01 substantially reduced the estimated upper-bound FDR to 0.028 while maintaining a non-trivial ODR of 0.243 and an ERR of 0.433 consistent with meaningful expected replication. In contrast, moving to α = 0.005 or 0.001 restricted the analysis to the extreme tail of the z-distribution, leaving too few studies for stable estimation. The authors emphasize that these retrospective threshold analyses should not be read as a proposal that changing α alone would control false discoveries prospectively, since such a change would likely alter study design, sample-size planning, publication incentives, and researcher behavior. Readers are encouraged to interpret existing single-study PMT claims supported primarily by p-values between 0.01 and 0.05 more cautiously unless they are accompanied by strong design features or successful direct replication.

The authors stress what z-curve does and does not identify. It evaluates whether the distribution of significant results is consistent with a mixture of probabilities of obtaining statistically significant results under selection, but it does not pinpoint which specific questionable research practices, if any, produced that distribution. The observed pattern may reflect publication bias, selective outcome reporting, undisclosed analytic flexibility, incomplete reporting of non-significant results, or some combination of these processes. A further implication is that published effect-size estimates in the PMT literature may be upwardly biased, and this concern extends to quantitative synthesis, because meta-analyses that rely primarily on the published record can inherit inflation when non-significant or ambiguous results are selectively missing. The authors caution against using the pooled estimate from the recent meta-analysis as a benchmark for planning future PMT experiments without accounting for selection-related effect size inflation in the underlying literature. At a theoretical level, the results do not imply that PMT is false or that masculinity threat never produces compensatory behavior. Instead, they suggest that the published experimental record most likely overstates evidential strength, and that a non-trivial share of statistically significant claims—particularly those supported primarily by p-values in the 0.05–0.01 range and interaction tests based on small sample sizes—should be treated as uncertain. Because PMT is used not only to explain laboratory outcomes but also to inform interventions and public-facing claims about men’s health behavior, violence, and responses to gender equality initiatives, a credibility review is a theory-maintenance task: it clarifies which parts of the evidential base can carry interpretive weight and where the field most needs high-power replication.

Looking forward, the authors recommend that researchers replace rule-of-thumb sample-size rationales with explicit justifications that link design, smallest effect size of interest, and error costs; justify α rather than defaulting to 0.05; prioritize wider uptake of preregistration and Registered Reports; report null outcomes rather than selectively emphasizing clean results; and replicate important findings using samples from multiple countries, including non-WEIRD contexts. Replication resources should prioritize widely cited claims, widely used manipulation types, and interaction-heavy hypotheses that have relied on small per-cell sample sizes. The authors also urge greater methodological attention to manipulation construction and validation: what specifically constitutes a threat to manhood, for whom, and under which cultural and situational conditions. Recent work indicates that the content of unmanliness varies across countries, so threat paradigms developed and repeatedly used in U.S. samples may not transfer easily to other cultural settings. The authors acknowledge limitations: z-curve models selection as a function of the probability of obtaining statistically significant results and cannot identify the mechanisms that produced the record; estimates may be biased if selection operates among already significant results; several subgroup subsets remain relatively small, leading to wide uncertainty intervals; and the estimates should be treated as time-bound. Still, the overall conclusion is clear: precarious manhood remains an influential framework, but the published record shows substantial selection for statistical significance and only moderate expected replicability, and improving cumulative inference will require more than increasing sample size—it will require systematic and transparent reporting of all theoretically relevant outcomes, including null findings.

Subject of Research: Psychology

Article Title: How credible is the experimental evidence on precarious manhood? A z-curve analysis

Article References: Krivoshchekov, V., & Gulevich, O. (2026). How credible is the experimental evidence on precarious manhood? A z-curve analysis. Frontiers in Social Psychology, 4, Article 1864827. https://doi.org/10.3389/frsps.2026.1864827

Image Credits: AI Generated

DOI: 10.3389/frsps.2026.1864827

Keywords: precarious manhood, masculinity threat, z-curve, replication crisis, meta-analysis, social psychology, false discovery risk, preregistration, credible, experimental, evidence, precarious

The core problem, the authors explain, is that traditional single-run clustering methods lack stability, robustness, and generalisability. Stability refers to whether results remain consistent across repeated runs; robustness describes resilience to noise; generalisability concerns whether findings hold in different datasets. All three are chronically weak in conventional approaches. This is not simply a matter of sloppy implementation. Kleinberg’s impossibility theorem shows that no clustering method can simultaneously satisfy all desirable properties, and the optimisation problem underlying clustering has no exact solution. Algorithms therefore rely on assumptions and trade-offs, and their stochastic optimisers mean that minor perturbations of the data, different starting points, or a different choice of algorithm can produce substantially different subgroups from the same dataset.

The downstream consequence is a replication crisis for subgroup research. Identified subtypes often vary dramatically across studies, and when broadly similar patterns do emerge, they frequently reflect arbitrary cut-offs along a single severity dimension rather than genuine heterogeneity. In mental health research, the stakes are particularly high: despite decades of effort, current evidence does not support reproducible, biologically valid, and clinically agreed-upon neuroimaging subtypes for mental illnesses. By contrast, fields such as oncology have made real translational progress with tumour subtyping, largely by adopting ensemble clustering methods that combine solutions from different data distributions, algorithms, and optimisation runs into a single consensus.

Ensemble clustering works much like ensemble methods in supervised machine learning, where predictions from many models are pooled to reduce the weaknesses of any single model. The clustering version follows a two-stage framework. The first stage generates many base clusterings, or BCs, by running different algorithms, random seeds, parameter settings, or resampled subsets of observations and variables. The second stage aggregates these partitions into a consensus, defined not as complete agreement but as the stable grouping pattern that recurs across solutions. The tutorial stresses that good BCs should be both diverse and high quality, a difficult balance because aggressive sub-sampling increases diversity while degrading individual partition quality, and optimising quality tends to drive all BCs toward the same answer.

The consensus stage is where the methodological richness lies. The simplest approach is majority voting, but because clustering algorithms return arbitrary cluster labels, BCs must first be relabelled and aligned, and voting fails entirely when BCs are highly diverse. More sophisticated techniques summarise the BCs before running a second-stage clustering model. Cluster labels can be treated as categorical variables and clustered with finite mixture models or k-modes. They can be binary-encoded into a cluster association matrix for k-means-based consensus clustering, or converted into bipartite graphs for graph partitioning methods such as the hypergraph-partitioning algorithm. Alternatively, a co-association matrix records how often each pair of observations is grouped together across BCs, serving directly as a distance matrix for hierarchical clustering in evidence accumulation clustering, or transformed into weighted graphs for algorithms such as the cluster-based similarity partitioning algorithm. Newer options include the link-based clustering ensemble, the meta-clustering algorithm, adaptive clustering ensemble, and even deep learning approaches.

Choosing the right components depends on the data. K-means is the most common BC algorithm because it is computationally cheap, though it assumes spherical clusters and is sensitive to outliers; model-based methods such as latent class analysis may better suit psychological data where subgroups plausibly arise from latent generative processes. The co-association matrix remains the most popular consensus tool because heatmaps of it visually display cluster stability and help select the number of clusters. Ensemble size matters too: performance generally improves with more BCs, but benchmarking studies suggest quality plateaus between roughly 50 and 100 base clusterings. A sampling proportion of about 80 percent is typically recommended, capturing most data characteristics while preserving variation among BCs.

A major practical payoff of the ensemble approach is a principled way to choose the number of clusters. Instead of relying on internal validity indices such as Silhouette, Calinski-Harabasz, Dunn, or Gap statistics, which often disagree with one another, researchers can inspect the consensus matrix. When the chosen number of clusters matches the true structure, the matrix displays clear square-shaped blocks along the antidiagonal, indicating pairs of observations that are consistently grouped together or apart. This visual pattern can be quantified through the cumulative distribution function of consensus indices, the delta area plot, the cophenetic correlation coefficient, and the proportion of ambiguous clustering score, though the authors caution that PAC carries an inherent bias toward larger cluster counts because it ignores null reference distributions.

To make the method concrete, the tutorial walks through a full R implementation using packages such as diceR, ConsensusClusterPlus, and clue, with step-by-step code provided in supplementary materials. Using a simulated dataset of 600 observations with 12 variables generated from three multivariate normal clusters, the authors demonstrate the dice function, combining k-means, Gaussian mixture models, and Ward’s hierarchical clustering with the link-based consensus function across 100 BCs. They also cover pre-registration of analysis plans, data assessment, dimensionality reduction via PCA or alternatives such as UMAP and NMF, imputation for missing data introduced by resampling, trimming and reweighting of poor-quality BCs, permutation tests to confirm solutions did not arise by chance, and cross-validation frameworks for tuning hyper-parameters in the absence of ground truth.

The most striking evidence comes from simulations built on three real-world psychological datasets covering quality of life, alexithymic traits, and cannabis expectancies. The team perturbed each dataset by randomly removing between 0.25 and 20 percent of the data, refilling gaps with multiple imputation, and repeated the process 100 times per condition. Even tiny perturbations of 0.25 percent produced substantially different results under single-run methods; hierarchical clustering in one case performed barely better than random assignment. Ensemble clustering consistently showed the slowest decay in stability, measured by the Rand index, and much narrower variability across perturbed dataset pairs, particularly for higher-dimensional data.

The authors are careful about limits. Ensemble methods inherit the assumptions of their base algorithms, so stable solutions can still miss non-convex or arbitrary cluster shapes; stability and internal validity do not guarantee external validity, and even valid, generalisable solutions may lack practical utility if they merely partition severity. Ordinal Likert-scale items and latent constructs can distort distance matrices, and longitudinal data require specialised Bayesian ensemble pipelines. Computational cost is real, though manageable with binary association matrices and high-performance computing. Still, the tutorial’s message is unambiguous: minor perturbations of psychological data can upend single-run clustering, ensemble methods tame that instability, and with mature R tooling now available, the authors advocate widespread adoption to finally turn heterogeneous psychological data into subgroups researchers can trust.

Subject of Research: Ensemble clustering methods for improving the stability and reproducibility of subgroup identification in psychological and psychiatric research

Article Title: Ensemble clustering: A practical tutorial

Article References: Gao, C. X., Wang, S., Zhu, Y., Ziou, M., Teo, S. M., Smith, C. L., Chiu, D., Talhouk, A., Wang, M., Yu, W., Cotton, S. M., & Dwyer, D. (2026). Ensemble clustering: A practical tutorial. Behavior Research Methods, 58(10), Article 286. https://doi.org/10.3758/s13428-026-03158-y

Image Credits: AI Generated

DOI: 10.3758/s13428-026-03158-y

Keywords: ensemble clustering, cluster analysis, consensus clustering, psychiatry, psychology, machine learning, unsupervised learning, R programming, replication crisis, latent class analysis, stability, Behavior Research Methods

The heart of the problem lies in the distinction between two psychological quantities that are easily confused. In an old–new recognition task, participants judge whether each test item was previously studied. Researchers usually want to know about sensitivity, the participant’s true ability to discriminate studied targets from unstated lures. But performance is also shaped by bias, the participant’s tendency to say “old” regardless of the evidence. A liberal responder says “old” often and racks up both hits and false alarms; a conservative responder says “old” rarely. A valid sensitivity measure must ignore this bias entirely. If two experimental conditions genuinely have identical memory accuracy but differ in how willing participants are to respond, the correct conclusion is that no sensitivity difference exists, and a statistical test should produce a significant result only about 5 percent of the time, matching the conventional alpha level.

Previous simulation work had already raised alarms. Rotello and colleagues showed in 2008 that common single-point measures, which compute sensitivity from a single pair of hit rate and false alarm rate, systematically confuse bias with sensitivity. Yet those findings, cited only around 110 times, failed to change practice. In a review of 253 recognition memory papers published across 40 Springer Nature journals between 2018 and 2024, the new study found that percent correct and corrected recognition together accounted for nearly a quarter of reported measures, with d′ close behind. Not one of the dominant measures had ever been validated in a way that escaped circularity, because in real experiments the true sensitivity is unknowable except through the very measures being tested.

The theoretical foundation of the crisis is a mismatch between two models. Every dataset is generated by some underlying population of memory signals, described by a data-generating model. Three major families compete to describe recognition memory: continuous Gaussian signal detection models, including the unequal-variance signal detection model, or UVSD; discrete threshold models such as the double-high-threshold model, which underlie corrected recognition; and mixture models like the dual-process signal detection model, which combines an all-or-none recollection process with a continuous familiarity process. Meanwhile, every sensitivity measure embodies a measurement model, a set of assumptions about the relationship between hits and false alarms that can be visualized as a receiver operating characteristic curve. When the measurement model does not match the data-generating model, two operating points that truly reflect identical sensitivity can appear to fall on different curves, and bias masquerades as a genuine memory difference.

The simulations confronted this problem directly. Across 1,962 parameter combinations, each replicated 10,000 times, the researchers sampled memory signals from distributions corresponding to four data-generating models, created two conditions that were identical in sensitivity but differed in decision criteria, and then tested each condition-pair with t-tests for five different measures: corrected recognition, A′, d′, the geometric area under the ROC curve, and da. Whenever a measure was valid, false positive rates should hover near 5 percent. The single-point measures failed almost everywhere outside the narrow case where their own assumptions happened to hold: d′ survived only under equal-variance Gaussian distributions, and corrected recognition only under rectangular distributions. Crucially, under unequal-variance Gaussians, which most experts consider the most plausible account of recognition data, all of these measures produced false discovery rates far above 5 percent, climbing steadily as sample size grew and reaching a staggering 100 percent in the largest, longest simulated experiments.

That last number deserves a pause. A 100 percent Type I error rate means that every single experiment would find a significant sensitivity difference where none exists, and every replication would confirm it. This inverts the usual hope that large samples and many trials protect science from error. With invalid measures, more data simply amplify the confound, producing results of enormous psychometric reliability and near-zero validity. The authors draw a pointed parallel to the replication crisis: highly replicable findings can still be false discoveries, and decades of research effort have been spent chasing effects, from the “revelation effect” to the apparent memory advantage of emotional words, that later analyses attributed to bias rather than sensitivity. Similar measurement failures have been documented in eyewitness lineup research, reasoning, social psychology, and child welfare.

The solution the researchers champion is da, a signal detection measure that, like d′, transforms hit and false alarm rates into z-scores and takes their difference, but that abandons the indefensible equal-variance assumption. Instead, da measures the distance between target and lure distributions in units of their root-mean-square standard deviation, using the zROC slope, denoted S, to estimate the actual variance ratio for each participant. Empirical studies consistently find that target distributions are roughly 25 percent wider than lure distributions, with slopes averaging about 0.8 but varying widely across individuals, from around 0.5 to above 1. The new version of da exploits this by collecting confidence ratings, which yield multiple operating points per participant, and estimating S by linear regression on each participant’s zROC curve, requiring at least three usable points per condition. Only one parametric assumption remains: that the underlying distributions are Gaussian.

The simulations vindicated this choice. For both equal- and unequal-variance Gaussian data, da held Type I error rates at approximately 5 percent across every sample size, number of trials, criterion placement, and distributional separation tested. Under the dual-process mixture model, da also stayed near 5 percent in typical scenarios, because the ROC curve implied by that model closely resembles the UVSD curve, though errors rose toward 16 percent in extreme parameterizations with very large recollection contributions. Only for rectangular threshold distributions, where the Gaussian assumption is simply wrong, did da break down, as theory predicts. The one caveat concerned precision: with few trials and many participants, noisy estimates of S could push error rates to about 10 percent, and experiments that excluded many participants for having too few operating points showed similar inflation. The remedy, the authors argue, is straightforward: run enough trials per participant, closer to 128 than to 64, so that the slope estimate converges on the true variance ratio.

Even the seemingly assumption-free alternative fared poorly. The geometric area under the empirical ROC curve, popular in machine learning and often promoted as non-parametric, was computed by connecting adjacent operating points with straight lines and summing triangles and trapezoids, which omits area whose size depends on where the points fall. Because that omitted area shifts with bias, the measure was biased too, reaching near-total false positive rates under large criterion shifts with big samples. The researchers also decline to endorse full mathematical modeling as a universal fix, noting that in a blinded expert challenge, experienced modelers reached strikingly inconsistent conclusions about which experiments had manipulated sensitivity, bias, or neither, and that modelers in the literature overwhelmingly choose threshold models whose fit is inferior to continuous alternatives. A simple scalar measure available to every researcher, they argue, beats a fragile modeling pipeline.

The implications reach beyond recognition memory to the entire family of single-interval tasks, from perceptual discrimination and attentional cueing to lexical decision and metacognitive judgment, wherever binary judgments against a criterion are made. For the existing recognition literature, the news is sobering: a substantial share of published sensitivity effects based on binary responses and standard indices may be uninterpretable, though the authors offer limited escape routes, such as mirror effects, in which higher hit rates coexist with lower false alarm rates in a way bias cannot explain, and designs where within-list criterion shifts are implausible. For future work, the prescription is sharper. Collect confidence ratings, compute da with an individually estimated variance ratio, and let reviewers and editors retire measures whose false positives replicate perfectly forever. In the ongoing struggle to make psychology’s discoveries both replicable and true, the study suggests, the battle begins with choosing the right ruler.

Subject of Research: Validation of bias-independent sensitivity measures for recognition memory using Monte Carlo simulations

Article Title: Reject common measures like d′ and Corrected Recognition, embrace da: Simulation explorations of single- and multi-point recognition measures of sensitivity

Article References: Levi, A., Danino, R., Rotello, C. M., & Goshen-Gottstein, Y. (2026). Reject common measures like d′ and Corrected Recognition, embrace da: Simulation explorations of single- and multi-point recognition measures of sensitivity. Behavior Research Methods, 58(10), Article 287. https://doi.org/10.3758/s13428-026-03130-w

Image Credits: AI Generated

DOI: 10.3758/s13428-026-03130-w

Keywords: recognition memory, sensitivity, response bias, signal detection theory, Monte Carlo simulations, measurement crisis, da measure, ROC curves, unequal variance model, Type I error, Behavior Research Methods, replication crisis