The Impact of Coronavirus (COVID-19) Related Public-Health Measures on Training Behaviours of Individuals Previously Participating in Resistance Training: A Cross-Sectional Survey Study

Notably, our original analysis approach was incorrect as some aspects of the measures used were overlooked in the haste with which the project was set-up and launched in response to the implementation of lockdown measures. We detail these changes below, though due to the final sample size yielded we are confident that, under the assumptions of the data generating process and model used, our results offer sufficient power for any tests, and a high degree of precision for interval estimates.

Due to the known variation in both timing of the pandemic’s effects, and the implementation of containment measures, across different countries we screened prior to inclusion such that only individuals who were currently impacted by these measures were included. Based upon survey timings respondents had therefore been under lockdown conditions including closures of their usual training facility AND/OR social distancing for ~ 3–5 weeks.

The limits set can be seen in the analysis code available in supplementary file (S1).

An exception to this was for income. We asked respondents to provide their annual household incomes in the currency appropriate based on their country of residence, followed by selecting from a list of currency codes the one used. We then used currency exchange rates (as of 31st July 2020) to convert all values to EUR. However, this yielded a high proportion of very high values (860 respondents with annual household incomes above 1 million EUR) leading us to question whether these questions had been understood fully. Thus, we chose not to filter by this variable due to concerns regarding the quality of this data. We plan to attempt to address this during-follow up and obtain data that might be used for retrospective examination of income effects.

Note, the pre-registration reported that 14 questions were to be included. However, prior to launch yet after pre-registration, we added an additional question.

We opted to use MRCVs for RT variables to allow for individuals who might vary their routines from week to week (i.e. they may engage in what is referred to as ‘periodization’) to respond without feeling the need to arbitrarily select a particular category.

As noted, the analysis presented here differs from that which was pre-registered. Due to the haste of project set-up in light of the suddenness of implementation of lockdown measures we overlooked some aspects of the data that would be generated. This included the paired nature of comparisons for SRCVs, and assumptions of dependency between items underlying the MRCVs. Thus, we have ignored our simple original pre-registered analysis plan aimed at testing the hypothesis of marginal independence. Instead, in part due to the sample size achieved which we anticipated would have yielded low p values with relative lack of information regarding the meaningfulness of the findings, we have analysed and presented our data using an alternative modelling strategy. We have also opted to avoid dichotomizing the existence of an effect for the main results and therefore did not employ traditional null hypothesis significance testing, which has been extensively critiqued [30, 31]. Instead, we consider the implications of all results compatible with these data, from the lower limit to the upper limit of the interval estimates, with the greatest interpretive emphasis placed on the point estimate.

All figures include corresponding 95% compatibility (confidence) intervals for estimates from the models yet in some figures they are so precise that they appear subsumed in the point estimate on the display.

Perhaps facilitated by many individuals already having access to such home equipment; as evidenced by a greater probability of having purchased equipment prior to lockdown. Based upon the wording of this question responses may in fact reflect those purchasing equipment in preparation for the coming lockdown to facilitate continued training at home. Indeed, those with more RT experience were more likely to purchase pre- lockdown, but less likely during-lockdown. Due to the surge in purchases of home-based equipment leading into and early on in lockdown, shortages were commonly reported during-lockdown rendering it difficult to make purchases during-that period and indeed after an initial spike demand dropped [17]. Those with greater RT experience may have better anticipated this.

See supplementary file S24.

See supplementary file S25.