Discussion

This study provides the first comprehensive evaluation of both the frequency and effectiveness of preregistration in sports science across multiple journals and quartiles. Contrary to our first hypothesis, 10% of all articles were preregistered, which was notably lower than the 15–25% anticipated. Moreover, while the proportion of supported hypotheses in preregistered studies was significantly lower than those that were not preregistered, the effect size was negligible, suggesting the difference may not be meaningful. Regarding our third hypothesis, journals ranked in the top quartile of the Sports Science category (as ranked by SCImago) had a higher proportion of preregistered studies than quartile 3 and 4 journals, although they also had a higher proportion of supported hypotheses. Finally, our exploratory analysis identified a concerning trend whereby more than two-thirds of preregistered studies reported a hypothesis that was not provided in their original preregistration.

The Uptake of Preregistration in Sports Science

Despite recent calls for greater transparency in sports science research, including recommendations for study preregistration [21, 31], the proportion of preregistered studies in our sample (10%) was slightly lower than previously reported in the literature (~ 12%) [25], even though the previous review examined studies published approximately 2 years earlier. The reason for this discrepancy may be attributed to differences in the samples examined. The earlier review focused exclusively on studies published over a 6-month period in five Q1 sports science journals. In contrast, our sample included journals across all quartiles, and if our results for Q1 journals were isolated, the preregistration rate was 13.8%. Research in biomedicine suggests that randomized controlled trials are more likely to be published in high-impact journals [26], which are in turn more likely to be registered in a clinical trial registry. As such, the inclusion of lower-quartile journals in our analysis likely contributed to the lower overall proportion of preregistered studies. Further supporting this, we found that studies published in lower-quartile journals were significantly less likely to be preregistered than those in higher-quartile journals. Although our finding of a 13.8% preregistration rate in Q1 journals is similar to previous work, it may also reflect a gradual increase in preregistration uptake. However, this trend cannot be confirmed without additional future research.

Interestingly, the lack of uptake in preregistration within sports science contrasts with trends observed in other fields. For instance, the number of preregistered studies in 19 leading experimental economics journals rose from 7 in 2017 to 190 in 2023 [28]. A similar trend was observed in randomized controlled trials published in 15 general economics journals, where the rate of preregistration increased from 14% in 2018 to 40% in 2021 [23]. Additionally, recent non-peer-reviewed work examining four high-quartile psychology journals reported that 43% of studies were preregistered [22], which is notably higher than the current sample. While the reasons for low uptake in sports science remain unclear, potential explanations exist. Although researchers generally view preregistration positively, some perceive they increase overall workload and contribute to greater work-related stress [32]. Others believe it restricts their ability to explore data, but this constraint is arguably a strength as it may help prevent QRPs [33]. Field-specific factors may also play a role. For example, applied sports science studies are often conducted in collaboration with, or led by, practitioners working directly with athletic populations [34]. This can lead to the exploration of routinely collected historical data to “find a result” without a predefined research plan. And while exploratory research plays an important role in the generation of new hypotheses, it is important that it is described as such, otherwise the importance of the findings may be overstated [34]. Furthermore, based on the journals reviewed in this study (Supplementary Digital Content 1), sports science publications rarely require preregistration unless a study meets the strict definition of a clinical trial outlined by the International Committee of Medical Journal Editors [35], which is uncommon in the area. In addition to field-specific factors, cultural aspects are also likely to influence preregistration practices. For instance, psychology has been at the forefront of the open science movement, with widespread adoption of practices such as preregistration and, more recently, registered reports, both of which are notably more common in psychology than in many other disciplines [36, 37]. This may also explain why the present study identified higher preregistration rates than those previously reported in epidemiological research, where secondary analyses of large datasets appear to be less frequently preregistered [24].

The Impact of Preregistration on Supported Hypotheses

The proportion of studies with a supported hypothesis (~ 55%) was lower than previously reported in sports science, where approximately 73% of hypotheses were fully supported [14]. This discrepancy may be attributable to differences in study samples. The prior study included only 215 articles from high-impact journals, which may have biased the results. Our regression analysis supports this, indicating that studies published in higher-quartile journals were more likely to report supported hypotheses than those in lower-quartile journals. This pattern may reflect a publication bias, whereby trials with novel or positive findings are more appealing to editors and reviewers, and thus more likely to be accepted by high-impact journals, a trend that has been observed in medical research [38].

As initially hypothesized, the proportion of supported hypotheses was significantly lower in preregistered studies (~ 45%) compared with those that were not preregistered (~ 56%). While this may indicate that preregistration can reduce the likelihood of QRPs that result in significant effects, the negligible effect size associated suggests that this may not be having a substantial impact on the field more broadly. Indeed, when considering their low uptake in sports science, it makes sense that their overall effectiveness would remain limited until their use becomes more widespread. To the authors knowledge, three prior studies have explored this question in different subsets of registered and nonregistered studies in the field of psychology, of which two observed a large significant difference [39, 40] and one did not [41]. Although the precise reason for the disparity between the present study and prior research is unknown, there may be some field-related differences at play. Most notably, the replication crisis in psychology has been extensively discussed and well-publicized [6], which may have led to widespread changes (including a greater importance on preregistration and data sharing) that has increased the scientific rigor of the field [42]. Such changes are unlikely to have spread to sports science to the same extent, which may also be reflected by the quality of preregistrations within the current sample.

This finding may also highlight the need for field-specific strategies to mitigate the impact of QRPs. In large-scale epidemiological studies, evidence suggests that when multiple research teams independently analyze the same dataset to answer a common research question, the variability in effect size estimates tends to be small, even when a wide range of statistical approaches are applied [43]. Conversely, similar investigations in the social sciences have shown the opposite pattern, with teams producing widely divergent results and reaching vastly different conclusions despite identical starting conditions [44]. Such discrepancies may partly reflect differing research cultures, particularly the degree of emphasis placed on achieving statistically significant results. For example, in epidemiology, there has been a shift away from traditional null hypothesis significance testing toward the estimation and interpretation of point estimates and their accompanying 95% confidence intervals [45].

Quality of Preregistrations in Sports Science

Concerningly, our exploratory analysis revealed that more than two-thirds of preregistered articles reported hypotheses that were not included in their original preregistration. This issue has been documented in previous research. For example, an examination of 459 registered studies published in psychology journals found that 52% omitted originally preregistered hypotheses, while 57% added new, unregistered ones [46]. Similar concerns were raised in a subset of 300 preregistered psychology studies, which frequently lacked methodological detail and included undisclosed deviations, both of which may increase the risk of false positives occurring [47].

These findings may suggest that authors publishing in sports science journals may frequently engage in QRPs, such as selective reporting and HARKing, to make their findings more appealing. They also indicate a potential lack of attention to preregistrations during the peer-review process. Supporting this concern, a study of 201 preregistered articles published in the PLOS journal family with open peer-review histories found that only 18% of reviewers or editors mentioned the preregistration in their peer review reports, and just 10% discussed the relationship between the preregistration and the final manuscript [48]. Considering these findings, we recommend that journal editors require peer reviewers to explicitly evaluate the alignment between study preregistrations and submitted manuscripts as part of the review process. This also highlights the importance of researchers clearly reporting any deviations from their preregistered protocols in their published manuscripts. Importantly, such deviations are not necessarily indicative of QRPs, and may arise from practical challenges encountered during the research process or even be implemented to strengthen the study design [49]. Moreover, deviations are not always initiated by the authors. For instance, reviewers may recommend the removal of nonsignificant results during the peer-review process.

Future Recommendations for Sports Science

Despite recommendations for sports science to adopt more transparent research practices, the present study suggests that they are not being widely adopted or appropriately used. As such, we believe the following actions would be advisable to improve research practice, and by extension, help prevent the continued infiltration of QRPs into the published literature:The transition to compulsory preregistration: As outlined earlier, when properly applied, preregistration can help prevent QRPs by limiting researchers to preplanned primary analyses. If journals required preregistration as a condition of submission, this could meaningfully improve research quality across the field.The widespread adoption of registered reports: While preregistration can reduce QRPs, there remains a risk that null findings are not published (i.e., the file drawer problem), leading to publication bias [50]. Registered reports may mitigate this risk. In a registered report, authors submit a study protocol that details the research questions, hypothesis, methodology, and analysis plan to a scientific journal for review prior to commencing the study, as an original research submission. Then, if the registered report is accepted, the journal agrees to publish the study regardless of the findings provided the protocol is followed [51]. Research in psychology has indicated that registered reports may provide a meaningful way to prevent QRPs and type 1 errors [52], although few sports science journals have adopted this practice [53, 54]. If more journals accepted registered reports, it would incentivize researchers to adopt open science practices by providing a clearer pathway to publication. Additional mechanisms to encourage their uptake could include offering expedited review processes for subsequent studies linked to an approved registered report or integrating registered report submission into grant funding requirements.The requirement to review preregistrations: On the basis of the results of the present study and prior findings [48], it appears likely that preregistrations in sports science are not routinely evaluated by journal editors or reviewers. Requiring peer reviewers to assess whether manuscripts align with their preregistered protocols would ensure deviations are acknowledged, increasing transparency and potentially reducing the incidence of type I errors.

Limitations

These findings should be interpreted considering several limitations. Firstly, owing to an unexpectedly low number of published studies in Q4 journals, we combined Q3 and Q4 journals into a single category. This decision limited our ability to assess differences in the proportion of registered studies and supported hypotheses across quartiles. Secondly, owing to the cross-sectional nature of this research, it cannot be ascertained whether the discrepancies between preregistrations and published manuscripts were the result of QRPs, or the authors feeling pressured to provide a hypothesis to ensure author guidelines were met appropriately. Moreover, while discrepancies between preregistrations and final articles were identified, the nature and extent of these deviations were not systematically described. Similarly, whether the hypotheses were tested using appropriate statistical methods was not considered. Future research should consider conducting a detailed analysis of preregistrations to identify both the most common types of deviations, and the appropriateness of the statistical approach used to answer reported research questions. Thirdly, what was considered to have “partial” support varied between studies. For instance, some studies had multiple hypotheses where only one was supported (and the remaining were not), while others had multiple hypotheses where all but one were supported. Fourth, only quantitative primary research studies were included. It is possible that systematic reviews or qualitative studies in sports science exhibit different preregistration rates than those reported here. Finally, while all extracted data were cross-checked for accuracy, the categorization was likely limited by the research team’s content-specific knowledge. For example, many studies presented multiple primary outcomes but described their hypotheses with minimal detail. In such cases, the research team had to rely on the authors’ interpretation of whether the results aligned with the stated hypotheses, which could have been made to sound more definitive and compelling to enhance the appeal of the paper to the submitted journal. Ideally authors should provide greater clarity on reported hypotheses by offering clear predictions regarding research outcomes, including the direction of relationships between variables and differences between groups [55].