Results

A total of 2544 articles published from 19 journals (Q1 = 4, Q2 = 4, Q3 = 6, Q4 = 5) were identified through Altmetric Explorer. Of these, 538 did not meet the inclusion criteria, leaving 2006 articles for analysis. Of these 2006 articles, 201 (10.0%) were preregistered, and 1206 provided a hypothesis (60.1%), of which 669 (55.5%) were supported by the results of the study. The observed proportion was lower than expected (P < 0.001), with a 95% confidence interval of 8.7–11.4%, indicating that the first hypothesis was not supported. Across these 1206 articles, an average of 1.5 (standard deviation [SD] = 0.8; range = 1–9) hypotheses were reported per article, of which an average of 1.0 (SD = 0.8; range = 0–5.5) were supported. Of the 800 articles that did not provide a hypothesis, 54 (6.8%) described their study as exploratory, or stated that no formal hypotheses were provided. Additional demographic information is presented in Table 2.

Quartile | Total | Preregistered (% of total) | Hypothesis provided (% of total) | Hypothesis supported (% of hypothesis provided)
Q1 | 666 | 92 (13.8%) | 408 (61.3%) | 255 (62.5%)
Q2 | 611 | 79 (12.9%) | 427 (69.9%) | 211 (49.4%)
Q3 | 588 | 26 (4.4%) | 319 (54.3%) | 173 (54.2%)
Q4 | 141 | 4 (2.8%) | 52 (36.9%) | 30 (57.7%)

Supplementary Digital Content 1 contains journal-specific information extracted from their respective author guidelines regarding requirements around hypothesis testing and preregistration. With respect to hypothesis testing, three journals (The American Journal of Sports Medicine; Journal of Strength & Conditioning Research; and Journal of Sport Rehabilitation) emphasized including a hypothesis (Q1 = 2, Q3 = 1), four journals (European Journal of Applied Physiology; Journal of Biomechanics; Human Movement Science; and Journal of Human Sport and Exercise) mention the inclusion of a hypothesis (Q2 = 1, Q3 = 1, Q4 = 2), and 12 journals (British Journal of Sports Medicine; Scandinavian Journal of Medicine & Science in Sports; European Journal of Sport Science; Journal of Applied Physiology; Research Quarterly for Exercise & Sport; International Journal of Sports Medicine; Kinesiology; Sport Sciences for Health; International Journal of Athletic Therapy & Training; Motor Control; Science and Sports; and Sports Engineering) did not mention hypothesis testing (Q1 = 2, Q2 = 3, Q3 = 3, Q4 = 4). Regarding preregistration, no journals required the prospective preregistration of all studies. However, eight journals (British Journal of Sports Medicine; The American Journal of Sports Medicine; Scandinavian Journal of Medicine & Science in Sports; Journal of Strength & Conditioning Research; European Journal of Applied Physiology; European Journal of Sport Science; Journal of Biomechanics; and Sport Sciences for Health) did require the preregistration of clinical trials if they met the definition put forward by the World Health Organization or the International Committee of Medical Journal Editors (Q1 = 4, Q2 = 2, Q3 = 2) (both of which define a clinical trial as “any research study that prospectively assigns human participants or groups of humans to one or more health-related interventions to evaluate the effects on health outcomes”) and three journals (Research Quarterly for Exercise & Sport; Human Movement Science; and Journal of Sport Rehabilitation) mentioned preregistration albeit provided no recommendations regarding their use (Q2 = 1, Q3 = 2), while the remaining eight journals (Journal of Applied Physiology; International Journal of Sports Medicine; Kinesiology; Journal of Human Sport and Exercise; International Journal of Athletic Therapy & Training; Motor Control; and Science and Sports; Sports Engineering) did not mention preregistration at all (Q2 = 1, Q3 = 2, Q4 = 5).

Table 3 outlines the number of preregistered and non-preregistered studies that had a hypothesis, and the number that were supported, partially supported, not supported, and unclear.

 | Preregistered | Non-preregistered
Total studies | 201 | 1805
Studies with hypothesis | 134 (66.7%) | 1072 (59.4%)
Supported | 61 (45.5%)a | 608 (56.7%)a
Partially supported | 35 (26.1%)a | 303 (28.3%)a
Not supported | 38 (28.4%)a | 159 (14.8%)a
Unclear if supported | 0 (0.0%)a | 2 (0.2%)a
Studies without hypothesis | 67 (33.3%) | 733 (40.6%)
Described as exploratory | 8 (11.9%)b | 46 (6.3%)b

The proportion of preregistered studies that had results that fully supported their hypothesis were significantly lower than non-preregistered studies (_X_2 (1) = 5.5978; P = 0.019), with a negligible effect size (V = 0.07, 95% CI = 0.01, 0.13) (Fig. 1), supporting the second hypothesis.

Fig. 1: Relative proportion of supported hypotheses among preregistered and non-preregistered studies

The Effect of Quartile on the Proportion of Preregistrations and Supported Hypothesis

There were no significant differences in the number of preregistrations between Q1 and Q2 journals, although Q2 journals had a significantly lower rate of supported hypotheses than Q1 journals (small effect), partially supporting the third hypothesis. Conversely, Q3 and Q4, collectively, had a significantly lower number of preregistrations (moderate effect) and supported hypotheses (small effect) than Q1 journals (Table 4), supporting the third hypothesis.

Quartile | Preregistration | Supported hypotheses
OR (95% CI) | P-value | OR (95% CI) | P-value
Q1 | 1.00 (reference) |  | 1.00 (reference) | 
Q2 | 0.93 (0.67, 1.28) | 0.643 | 0.59 (0.45, 0.77) | < 0.001*
Q3/4 | 0.27 (0.17, 0.41) | < 0.001* | 0.73 (0.55, 0.97) | 0.031*

Exploratory Analysis

Of the 134 preregistered articles that provided a hypothesis, a total of 93 (69.4%) either did not have a hypothesis in their original preregistration (n = 80, 59.7%) or presented a different hypothesis from what was reported in the original preregistration (n = 13, 9.7%). Furthermore, 12 (17.9%) of the 67 preregistered articles that did not report a hypothesis in their published manuscript did provide a hypothesis in their preregistration. Of these 12, 7 (58%) of the non-reported hypotheses were not supported by the results of the study. Finally, one study (0.7%) stated that it was preregistered in the published manuscript but did not provide a registration number or note the trial registry where it was preregistered. As such, for this one article, we were unable to access its preregistration and determine how closely it aligned with what was reported in the published manuscript.