Publications

Preprints and articles.

Total 27

2026 15 items

Shaheed Azaad · Preprint

The Systematizing Confidence in Open Research and Evidence (SCORE) project recently published a large-scale replication project within the social and behavioural sciences. The authors replicated only 151 of 274 (55%) claims, a finding covered by major outlets such as the New York Times and Forbes. Here, I audited SCORE replications of six studies where at least one claim surprisingly failed to replicate. These failures were surprising because the original study produced a p < .01 while the replication yielded a significant effect in the opposite direction. I show that, of the 12 claims tested across these six replications, five were miscategorised as failed replications because of data analysis and interpretation problems, and six replications were invalid because of poor data quality and an error in the replication’s experiment code. I conclude that individual replications within replicability projects may benefit from more thorough peer review, and that a more comprehensive audit is necessary before we can be confident in the project’s overall replicability estimate.

Shaheed Azaad, Kassandra Friebe, Günther Knoblich, Natalie Sebanz · Preprint

We routinely face the decision to act either alone or with another person. Research has found that humans exhibit a joint action preference, choosing to coordinate even when it is unnecessary and inefficient. Here, we tested one mechanism that might contribute to this preference: that joint task representations are activated spontaneously and preferentially to individual task representations. To this end, we conducted a secondary (meta)analysis of seven experiments (total N = 288) in which dyads completed a stimulus-response task requiring switching between individual and joint responses. Switch-cost asymmetry between the two trial types revealed that joint task representations were represented preferentially to individual task representations. This held even though participants did not coordinate intentionally, responding together only incidentally when a stimulus cued responses from both. In other words, the task was represented jointly and preferentially activated, spontaneously. Our results add to a body of joint action research indicating that a suite of spontaneous perceptual and cognitive processes automatically prepare us to coordinate with others.

Kassandra Friebe, Shaheed Azaad, Natalie Sebanz, Günther Knoblich · Preprint

A central debate in joint action research concerns what distinguishes joint action from individual action. Recent accounts propose that joint action is supported by joint action plans that relate representations of individual contributions to a joint goal. While experimental evidence shows that co-actors can represent aspects of a joint action that go beyond planning their individual components, it remains unclear how representations of joint and individual action relate. We propose that relations between individual actions are represented at a different level than the individual actions themselves. This in turn would imply that switching between joint and individual action requires a switch between levels of action representation that creates additional processing costs. To test this proposal, we introduce a novel experimental paradigm that examines switch costs between individual and joint action. Across three experiments (N1 = 24, N2 = 24, N3 = 36), participants were cued to press a key either alone or together with a partner. Crucially, the only difference between conditions was the instruction to act alone or together, while the action participants executed remained the same. Our results revealed significantly faster response times on repetitions than switches between individual and joint action, consistent with the proposal that they use different levels of action representations. Crucially, the results of Experiments 1 and 3 also showed that switches between different individual actions are less effortful than switches between individual and joint action. These results further support the idea that joint and individual action rely on different levels of action representation.

Shaheed Azaad · Psychological Research

The embodied view of cognition holds that cognitive and emotional processes are grounded in our bodily experiences and physical interactions with the world. In line with this view, embodied motivation studies have purported to show that enacting motivation-related postures, expressions, and actions can elicit or modulate their corresponding motivational states. While theory-building reviews on the topic have accepted these studies’ conclusions at face value, a closer reading of key articles reveals several statistical issues that call their veracity into question. The present review shows that evidence for bodily feedback effects on motivation relies on: (1) studies using multiple measures of motivation, and dismissing results from those that do not support hypotheses, (2) analytical degrees of freedom inflating the false positive rate, and (3) reviews on the topic interpreting non-significant hypothesis tests as if they were significant. Supporting this analysis of the literature, a z -curve of 51 significant results reported across 18 articles indicated little evidentiary value, with an estimated replicability rate of just 0.21 (95% CI = [0.13, 0.25]). In sum, there is little credible evidence that bodily postures modulate motivational states.

Henrike Maria Jungeblut, Shaheed Azaad · Preprint

Liu et al. (2024) recently published a systematic review and meta-analysis of studies on vaccine promotion interventions. One key finding highlighted in the study’s abstract is that strategies that increased access to vaccines were especially effective in lower-income countries and those with limited healthcare access. Here, we raise two issues that call this finding into question. First, Liu et al.’s dataset contains no studies in which access interventions are implemented in low- or even middle-income countries, meaning that any claims about the efficacy of access interventions in these regions are extrapolated from a pattern so far observed only in high-income countries. Second, the interaction effect supporting this conclusion appears to be an artifact of the authors’ selecting a model that systematically excluded the largest non-access intervention effects in low- and middle-income countries.

Robrecht van der Wel, Shaheed Azaad, Pernille Hemmer, John Michael · Preprint

In a recent meta-analysis, Mariano et al. (2025) conclude that people experience a stronger sense of agency over positive than negative outcomes, indexed by temporal compression. In this commentary, we evaluate that claim and identify several concerns about its robustness. First, we were unable to reproduce their main valence effect: our reconstruction of the three-level meta-analytic model yielded a smaller, non-significant difference. Second, we identified data integrity issues, including implausibly small standard deviations that appear to reflect computational errors. Third, we argue that temporal compression effects should be standardized before being combined, because values in milliseconds are not necessarily comparable across different paradigms. When we analyzed standardized effects, evidence for a valence effect was weak, and absent once adjusting for publication bias. Overall, our re-analysis suggests that stronger agency for positive outcomes is not yet robustly supported.

Shaheed Azaad · Preprint

Although power analysis has become increasingly accessible over time, researchers still find it challenging to compute power for complex ANOVA interactions. anovapowersim is an R package that enables users to easily specify between, within, and mixed interactions of varying complexity and simulate power based on a target effect size. This tutorial provides a short introduction to the package's core functionality, along with worked examples and sample R code.

Shaheed Azaad · Preprint

It is common and recommended practice in psychology to report standardised effect sizes alongside significance tests. As Cohen (1988) notes, the interpretation of these effect sizes as either small, medium, or large should depend on the field of research. The present study analysed 13,194 effects from 3,302 articles to empirically derive thresholds for social psychology (SP), individual differences psychology (IDP) and cognitive psychology (CP). Thresholds (in Pearson’s r) were smallest for SP (small: .17, medium: .24, large: .34), followed by IDP (small: .19, medium: .27, large: .41), while CP produced the largest effects (small: .24, medium: .34, large: .51). Thresholds for within-subjects designs (based on bias-corrected dz, or g) followed a different pattern: effects were smallest for IDP (small: 0.27, medium: 0.44, large: 0.78), slightly larger for SP (small: 0.33, medium: 0.49, large: 0.74), and largest for CP (small: 0.45, medium: 0.64, large: 0.95). Effect sizes were smaller in all fields when studies were preregistered. Confidence intervals for these estimates and conversions to other effect size metrics (e.g., d and η2) are provided.

Shaheed Azaad, Yoshihisa Kashima, Nick Haslam · Preprint

Research has revealed considerable seasonal variation in human behaviour and emotion. However, whether cognitive performance is similarly seasonal remains an open question. While some studies have found a performance advantage in summer compared to winter, others have found either the reverse effect or no effect at all. Because demographics, locations, and cognitive domains of interest vary between studies, it is difficult to determine whether such heterogeneity reflects systematic variation due to study characteristics or whether significant findings are better explained by sampling variance and publication bias. In the present study, we tested for seasonality in 100,113 chess games played by 27,335 players. Not only did we not find a seasonal effect, but we also found no interactions with age, gender, the absolute latitude of the tournament location, and the human development index of a player’s country. Our results suggest that the cognitive capacities required by chess do not vary seasonally, and that there are unlikely to be populations or locations in which it does.

Shaheed Azaad, Simon Laham, Natalie Sebanz · Preprint

Evidence from joint action research suggests that a suite of spontaneous cognitive and perceptual processes underpins our ability to coordinate with others. We tested whether people spontaneously perceive affordances (or action possibilities) for the dyad when a co-actor is present. To test this hypothesis, we asked participants to complete an object-based response time task either alone (individual condition) or while seated next to a partner who completed the complementary task (joint condition). Participants in both conditions made unimanual go/no-go keypresses to objects that afforded either a bimanual (e.g., a laundry hamper) or unimanual (e.g., a bucket) grasp. In total, we conducted seven experiments (total N =293). An initial experiment revealed an affordance × task interaction effect on response times, such that responses to bimanually graspable objects were faster in the joint than in the individual condition. However, several follow-up experiments failed to replicate this finding, providing no overall evidence for spontaneously perceived joint affordances.

Shaheed Azaad · Behavior Research Methods

Theoretically driven meta-analyses often involve testing whether study characteristics moderate an effect in a pattern that supports one of multiple competing theoretical accounts. To conduct such analyses, researchers must manually extract, i.e., code, these characteristics from sometimes hundreds of studies. Ideally, studies should be coded by two researchers independently to prevent errors and biases from compromising the dataset’s integrity. The laborious nature of this task, however, means that meta-analysts usually settle for double-coding just a portion of the studies included in their study. Some researchers have proposed using large language models (LLMs) as a double-coder, showing that they can reliably extract explicitly stated information (e.g., publication year) from articles. However, meta-analyses in psychology generally require studies to be coded in terms of higher-level conceptual dimensions (e.g., the type of behavior rather than which behavior). The present study investigated whether two LLMs, OpenAI’s GPT-5 and Google’s Gemini 2.5 Pro, can code studies in this way. Both models replicated the study codes from three recently published meta-analyses in psychology with high accuracy (>92% overall), despite the limited information they received. LLM double-coding, therefore, offers a practical solution for meta-analysts seeking to ensure data integrity.

Shaheed Azaad, Nick Haslam, Yoshihisa Kashima · Preprint

Past research has concluded that stereotype threat effects cause female chess players to underperform against male opponents. Here, we investigated whether this effect is large enough be practically meaningful, and whether it varies in line with the stereotype threat account. We analysed moves from 118,053 tournament chess games (N = 29,864 players), to test for a player × opponent gender interaction on performance, whether mixed-gender games were played more aggressively, and whether female players performed better in female-only tournaments. We also tested for moderation by the Gender Inequality Index of a player’s country, a player’s birth year, and the year in which a game was played. Equivalence testing (bounds: β = ± 0.10) found all effects to be unsubstantial. Results suggest that female tournament chess players do not experience stereotype threat effects, and that gender disparities in chess performance might be better explained by factors such as participation rates.

Shaheed Azaad · Preprint

The typical approach to computing standardised mean differences (d) from paired-samples (or within-subjects) designs requires knowing the often-unreported repeated-measures correlation, r_"repeated" . This adjustment enables comparison with ds from independent samples. However, when meta-analysing effects that come exclusively from paired samples, an underutilised option is to compute Cohen’s d_z, which does not require r_"repeated" . Because d_z can be computed from a range of summary and inferential statistics, it enables researchers to obtain standardised effects even when articles report their results in little detail. The present article contains equations and the corresponding R code needed to compute d_z and its variance, and adjust for small-sample bias.

2025 5 items

Shaheed Azaad, Kassandra Friebe · Preprint

Meta-analysis enables researchers to summarise the empirical literature on a phenomenon. Often, meta-analysts also conduct theoretically motivated moderator analyses, or meta-regressions, to determine whether the magnitude of an effect depends on relevant study characteristics (e.g., the type of stimuli or measures used). To do so requires first manually extracting information from, or coding, the meta-analysed studies. Although the veracity of moderator analyses depends on the accuracy of the coding process, many researchers opt to double-code only a fraction of their studies, leaving open the possibility of coding errors in the remaining data. Here, we simulated the impact of seemingly low rates (.02, .04, and .06) of random study miscoding on meta-regressions with categorical predictors. Results indicated that miscategorisation had a larger effect in smaller meta-analyses and when moderator categories were imbalanced, even when the miscategorisation rate was as low as .02. We propose a set of guidelines to enhance the efficacy of partial double-coding and mitigate such effects when complete double-coding is infeasible.

Shaheed Azaad, Natalie Sebanz, Kassandra Friebe, Vjeran Keric, Simon Laham · Preprint

For two decades, the joint Simon effect (JSE) has been a cornerstone of joint action research. Initially presented as evidence that humans spontaneously represent others’ actions in a functionally equivalent way to their own, the JSE now serves as an interdisciplinary measure of self-other integration across research on social cognition, human-robot interaction, and animal cognition. However, whether the JSE is better explained by non-social mechanisms remains an open question. The present meta-analytic review sought to provide an overview of the JSE literature and to evaluate its suitability as a measure of self-other integration. Our meta-analysis of 274 JSEs from 80 studies (7042 participants across 186 samples) revealed a significant overall effect of gD = .60. Analysis of 52 canonical JSEs (reflecting the seminal study) revealed a larger JSE of gD =.72. Finally, moderation by study characteristics suggested that JSEs at least partly reflect social representations of one’s co-actor. Importantly, our findings lend support to the continued use of the JSE as a measure of self-other integration.

Shaheed Azaad, Gerald Echterhoff · Preprint

Many experiments in psychology involve deceiving participants about aspects of the study. Researchers who employ deception sometimes administer post-experiment suspicion probes and exclude participants from analysis if their responses indicate that they were not successfully deceived. However, several studies have found that their suspicious participants behave as if they were deceived as intended. One explanation is that participants infer that they were deceived upon being presented with a suspicion probe, which biases them to overreport being suspicious. We tested whether post-experiment beliefs about deception can indeed induce such biases by manipulating whether we admitted to using deception before administering suspicion probes. Admitting to deception did not increase the amount of suspicion reported (Experiment 1), nor did the content of participants’ suspicions reflect the content of our admission (Experiment 2). Results suggest that it is unlikely that suspicion probes ‘tip off’ participants in a way that causes them to overreport suspicion.

Shaheed Azaad, Natalie Sebanz · Scientific Reports

Research on joint action planning has demonstrated that, when acting with others, individuals will choose more effortful actions if it improves co-efficiency by reducing the overall effort exerted by the group. However, because these studies use actions for which time and effort costs are confounded, it is unclear which costs participants sought to minimise and what processes underlie decision-making about individual contributions to joint actions. Across three experiments, we tested (1) whether dyads aim to minimise effort vs. time costs (Experiments 1–2) and (2) whether individuals choose actions based on a rational, deliberative process or a heuristic (Experiment 3). Data from a joint object-dragging computer task revealed that participants chose to drag objects to the closer of two goals even when it required more effort from the dyad (Experiments 1–2). Participants preferred closer goals when time and effort costs were equal (Experiment 3a), but only when the objects’ locations were salient (Experiment 3b). Together, these results suggest that: (1) individuals minimise time, rather than effort, when sharing object-moving tasks, and (2) they do so – at least in part - by using distance as a heuristic. Our findings also replicate and extend earlier work showing that people prefer acting together versus alone, even when this is less efficient.

2023 2 items

Shaheed Azaad, Natalie Sebanz · Scientific Reports

Contextual cues have been shown to inform our understanding and predictions of others’ actions. In this study, we tested whether observers’ predictions about unfolding actions depend upon the social context in which they occur. Across five experiments, we showed participants videos of an actor walking toward a piece of furniture either with (joint context) or without (solo context) a partner standing by it. We found greater predictive bias, indicative of stronger action expectations when videos contained a second actor (Experiment 1), even when the solo condition had a perceptually-matched control object in place of the actor (Experiment 2). Critically, belief manipulations about the actions the walking actor would perform suppressed the difference between social context conditions when the manipulation specified an action possible in both contexts (Experiment 5) but not when the action was one that would be difficult without a partner (Experiment 4). Interestingly, the social context effect persisted when the belief manipulation specified an unlikely action given the depicted scene (Experiment 3). These findings provide novel evidence that kinematically-identical actions can elicit different predictions depending on the social context in which they occur.

Shaheed Azaad, Natalie Sebanz · Psychological Research

In a recent Psychological Research article, Eaves et al. (2022) review the literature on how motor imagery (MI) practice combined with action observation (AO) enhances motor performance. The authors propose that the synchronous form of AO and MI (AOMI) affords unique benefits to performance that are not possible when the two interventions are performed asynchronously. We discuss three questions raised by Eaves et al.’s review: (1) are there any clear advantages to synchronous AOMI? (2) Are there super-additive benefits to AOMI, and if so, are they unique to synchronous AOMI? (3) How might coordinative AOMI, in which people imagine complementary actions, facilitate joint actions?

2022 1 item

Michael R. Donner, Shaheed Azaad, Garth A. Warren, Simon M. Laham · Emotion Review

Disgust seems to play an important role in moral judgment. However, it is unclear whether the role of disgust in moral judgment is limited to certain kinds of moral domains (versus many) and/or certain types of disgust (versus many). To clarify these questions, we conducted a multilevel meta-analysis ( k = 512; N = 72,443) on relations between trait disgust sensitivity and moral judgment (disgust-immorality association). Main analyses revealed a significant overall mean disgust-immorality association ( r = .23). Additionally, moderator analyses revealed significant specificity in disgust type and moral domain (grounded in Moral Foundations Theory): effects were stronger for (a) sexual disgust compared to pathogen disgust, (b) sanctity moral judgments compared to other domains of moral judgments, and (c) sexual-sanctity associations compared to other disgust type-moral domain pairings.

2021 1 item

Shaheed Azaad, Günther Knoblich, Natalie Sebanz · Cambridge University Press

Even the simplest social interactions require us to gather, integrate, and act upon, multiple streams of information about others and our surroundings. In this Element, we discuss how perceptual processes provide us with an accurate account of action-relevant information in social contexts. We overview contemporary theories and research that explores how: (1) individuals perceive others' mental states and actions, (2) individuals perceive affordances for themselves, others, and the dyad, and (3) how social contexts guide our attention to modulate what we perceive. Finally, we review work on the cognitive mechanisms that make joint action possible and discuss their links to perception.

2020 1 item

Shaheed Azaad, Simon M Laham · Quarterly Journal of Experimental Psychology

When participants make left/right responses to unimanually graspable objects, response times (RTs) are faster when the responding hand is aligned with the viewed object’s handle. This object-based compatibility effect (CE) is often attributed to motor activation elicited by the object’s afforded grasp. However, some evidence suggests that the object-based CE is an example of spatial CEs, or Simon effects, elicited by the protruding nature of objects’ handles. Moreover, recent work shows that the way in which objects are centred on-screen might attenuate or reverse CEs, perhaps due to differences in pixel asymmetry (the proportion of pixels either side of fixation) between centralities. In this study, we tested whether pixel asymmetry also contributes to between-object variation in object-based CEs. In experiment 1 ( N = 34), we found that between-object differences in asymmetry predicted object-based CEs, such that objects with a greater proportion of pixels to the handle-congruent side of fixation produced larger CEs. In experiment 2 ( N = 35), we presented participants with mug (low asymmetry) and frying pan (high asymmetry) images and found that between-object and within-object (due to stimulus centrality) differences in pixel asymmetry interact to moderate CEs. Base-centred stimuli (centred according to the width of the object’s base) produced conventional CEs, whereas object-centred (centred according to the object’s total width) stimuli produced negative CEs (NCEs). Furthermore, the effect of centrality was smaller for mugs than pans, indicating an interaction between within-object and between-object differences in pixel asymmetry.

2019 2 items

Shaheed Azaad, Simon M Laham · Quarterly Journal of Experimental Psychology

Tucker and Ellis found that when participants made left/right button-presses to indicate whether objects were upright or inverted, responses were faster when the response hand aligned with the task-irrelevant handle orientation of the object. The effect of handle orientation on response times has been interpreted as evidence that individuals perceive grasp affordances when viewing briefly presented objects, which in turn activate grasp-related motor systems. Although the effect of handle alignment has since been replicated, there remains doubt regarding the extent to which the effect is indeed driven by affordance perception. Objects that feature in affordance-compatibility paradigms are asymmetrical and have laterally protruding handles (e.g., mugs) and thus confound spatial and affordance properties. Research has attempted to disentangle spatial compatibility and affordance effects with varying results. In this study, we present a novel paradigm with which to study affordance perception while sidestepping spatial confounds. We use the Bimanual Affordance Task (BMAT) to test whether object affordances in symmetrical objects facilitate response times. Participants ( N = 36) used one of three (left unimanual/right unimanual/bimanual) responses to indicate the colour of presented objects. Objects afforded either a unimanual (e.g., handbag) or a bimanual (e.g., laundry hamper) grasp. Responses were faster when the afforded grasp corresponded with the response type (unimanual vs. bimanual), suggesting that affordance effects exist independent of spatial compatibility.