Procedure
Ethical approval for the study was granted by Sheffield Hallam University Research Ethics Committee on 24/07/2018 (reference number: ER7034346) and approval was granted from the parkrun Research Board. The study used an online survey, incorporating wherever possible validated and robust measures used in health and wellbeing research. An advisory board, created using the parkrun Research Board, were consulted to longlist and then shortlist the questions used in the survey. For each measure, a search was undertaken by the research team to identify appropriate questionnaires or questions. Each questionnaire or question was selected using the following criteria: relevance; validity; reliability; length; previous use. If previous questionnaires or questions could not be not identified, the research team developed study-specific questions to capture the outcome. The length and literacy were tested and re-tested via members of the research team and the advisory board. The reporting adheres to established standards for reporting internet-based surveys; The Checklist for Reporting Results of Internet E-Surveys (CHERRIES) (19).
Population and participants
The sample was drawn from all parkrun registrants in the UK aged 16 or over. Registrants received an email from parkrun containing a link to the survey. Survey participants had to be aged 16 or over and the survey was only available in online format and in the English language. There were no other explicit exclusion criteria. In this manuscript, we use the data from respondents who identified in the survey as runners/walkers and runners/walkers who also volunteer at parkrun. We present the data relating to running/walking at parkrun (not volunteering).
The survey
The measures in the survey are described fully in Additional file 1 with a full copy of the survey, including wording for consent. The list below describes the sub-set of measures used in this study.
Demographics
Demographic data included date of birth, gender, ethnicity, employment, home parkrun (the parkrun event they were most closely affiliated with) and long-term health conditions. One question asked participants to state whether they most closely identified as a parkrun runner/walker, a parkrun runner/walker and volunteer or a parkrun volunteer. Respondents were asked to provide their parkrun ID number to enable their survey responses to be matched to the parkrun database that holds their parkrun registration details (e.g. postcode, activity level at registration) and participation information (e.g. number of parkruns completed). See ‘parkrun data’ section below for more details.
Life satisfaction and happiness
Two of the four personal wellbeing questions asked in the UK's Office of National Statistics Annual Population Survey (20) were used as measures of life satisfaction and happiness: 1) “Overall, how satisfied are you with your life nowadays?” and 2) “Overall, how happy did you feel yesterday?” Statements were rated on a 10-point visual analogue scale where 0 is "not at all", and 10 is "completely ".
Subjective health status
Subjective health status was measured using the EuroQoL visual analogue scale (EQ-VAS) (21) which asks: “We would like to know how good or bad your health is TODAY. This scale is numbered from 0 to 100. 100 means the best health you can imagine. 0 means the worst health you can imagine. Please enter a number in the box below to indicate how your health is TODAY.” Permission was granted by EuroQol Research Foundation for its use.
Motivation for participating in parkrun as a runner/walker
Motivation for participation in parkrun was measured with the question: “What motivated you to first participate at parkrun as a runner or walker?” Respondents were asked to select a maximum of three answers out of a possible 21 motives. The 21 choices were displayed in randomised order to help reduce response bias. The final choice was “other” and, if selected, respondents were asked to specify the motive.
Self-reported physical activity
Self-reported physical activity was measured using three different measures: 1) the a single item four week recall physical activity question that is asked at parkrun registration; 2) a single item one week recall physical activity question (22); and 3) the International Physical Activity Questionnaire Short Form (IPAQ-SF) (23).
Four week recall
This question asked: “Over the last 4 weeks, how often have you done at least 30 minutes of moderate exercise (enough to raise your breathing rate)?” Respondents could answer: less than once per week, about once per week, about twice per week, about three times per week, four or more times per week, rather not say, don’t know. This was chosen because the same question is asked at parkrun registration, allowing direct comparison.
One week recall
This single-item physical activity measure was developed by Milton, Bull (22) and asks: “In the past week, on how many days have you done a total of 30 minutes or more of physical activity, which was enough to raise your breathing rate. This may include sport, exercise, and brisk walking or cycling for recreation or to get to and from places, but should not include housework or physical activity that may be part of your job.” Respondents could answer: 0 days, 1 days, 2 days, 3 days, 4 days, 5 days, 6 days, 7 days. This has been validated against the Global Physical Activity Questionnaire in a UK sample of 240 adults (22).
IPAQ-SF
Physical activity was also measured using the International Physical Activity Questionnaire short form (IPAQ-SF) (23). The IPAQ-SF is a validated, subjective measure of physical activity (24). Respondents answered 7 questions on the frequency, intensity (moderate, vigorous, walking, sitting) and duration of physical activity participation over the past 7 days.
Perceived impact of running/walking at parkrun
The perceived impact of parkrun was measured using the following question: “Thinking about the impact of parkrun on your health and wellbeing, to what extent has running or walking at parkrun changed:” Respondents were presented with a list of 15 potential impacts and asked to rate each one on the following 5-point scale: much worse, worse, no impact, better, much better. The answer choices were displayed in randomised order to help reduce response bias. The final choice was “other” and, if selected, respondents were asked to specify the impact.
parkrun data
Additional data was exported from the parkrun database wherever enough personal details were provided to enable data matching. Additional data matched to responses included the following: postcode provided at parkrun registration; date of parkrun registration; self-reported physical activity level at registration using the four-week recall question; and total number of parkruns completed since registration.
Data collection
Pilot testing was carried out on a randomly selected sample of 200 UK participants (aged 16 or over). Subsequent power calculations suggested that the survey would have to be sent to the full parkrun population to allow segmentation to a sub-sample from socioeconomically deprived areas (derived by postcode) and who were previously inactive at registration. The survey was distributed between 29th October and 3rd December 2018.
The survey used Qualtrics online survey software (25). The web link contained an introductory page with a participation information sheet and a confirmation box to indicate it had been read, understood and consent given to be part of the take part. Only people emailed the web link could access the survey. View rate of the survey was not captured. The survey was open for five weeks from 29th October 2018 with staggered sending of emails due to email server limitations. Reminders were emailed after one week. There were no incentives offered for taking part in the survey.
Questions were asked in the order presented in Additional file 1, with the exception of the International Physical Activity Questionnaire Short Form (IPAQ-SF), which was asked as a final, optional question due to its length and to keep it apart from the other physical activity measures used earlier in the survey. Questions were not randomised, but response choices within some questions were (see Additional file 1).
Adaptive questioning was utilised, such that certain questions were displayed based on answers to previous questions. For example, people who reported being walkers/runners did not see questions about volunteering at parkrun. There was a maximum of 47 questions, with an average of 4.3 questions per page and a maximum number of 11 screens (pages) of questions (total question number and page number were shorter depending on how respondents answered questions).
Questions were optional (i.e. non-compulsory) with the exception of the question about parkrun participation type (to enable the appropriate questions to be presented to the respondent), one question about long-term health conditions and two questions about life satisfaction and happiness. Respondents could go back and forth within the survey to review or change answers. Upon clicking 'submit', answers could not be changed. With consent, partially completed survey responses were saved and data kept for analysis unless the respondent requested removal by contacting the research team.
Data handling
Survey returns that included identifiers (parkrun ID number, name, date of birth, home parkrun) were matched, with consent, to parkrun registration data for 74% of survey respondents (the remaining 26% did not contain enough information to allow the match). All data was anonymised after matching with parkrun registration data. Data was handled in accordance with the Data Protection Act 2018 and the General Data Protection Regulation 2018. Raw survey data was stored on centrally encrypted hard drives in password protected files. Raw data was exported from Qualtrics to Microsoft Excel (Microsoft Office 2011, Microsoft Corporation, USA) for data validation, cleaning and the removal of personal identifiers to create a pseudo-anonymised data set. Analysis was carried out in Microsoft Excel, SPSS (IBM SPSS Statistics 24.0) and MATLAB (version 13.0b, MathWorks, USA).
Duplicate responses were identified by their unique Qualtrics code assigned during the survey and only the latest time-stamped response retained. Responses were excluded if they consented and filled out some or all demographic data but did not fill out any other survey questions to enable analysis. Six respondents were removed either due to abusive comments in free text, because of nonsensical responses, or both. Some respondents were found to give responses between 0 and 10 for the EQ-VAS which had a scale of 0 to 100; any data from 1 and 10 inclusive were removed with zeros retained as a valid minimum. This removed 1,270 responses for this question. Respondents were not obliged to answer all questions and partially completed surveys were included in the analysis.
Respondents were allocated an Index of Multiple Deprivation (IMD) for Lower Level Super Output Areas (LSOA) derived from the postcode they provided at parkrun registration. LSOAs are the smallest units from which Population Census data is compiled and onto which official data on socio-economic context is mapped by the Office of National Statistics (26). The participants IMD scores were grouped into quartiles where quartile 1 was the most deprived.
A detailed protocol and full description of the dataset is available in the survey report published on the parkrun Research Board website (18).
Data analysis
Descriptive statistics were used to characterise the respondents and compare them to the total population of parkrun registrants from which they were drawn. Stratified analyses where then undertaken to compare health and wellbeing, motivation for participation and self-reported benefits of participation between groups defined by socioeconomic deprivation status as well as their self-reported activity level at registration.
For descriptive statistics, we report percent, mean, median and interquartile range (IQR). Data such as age, happiness, life satisfaction, health today, parkruns per year, years registered and the single activity question were non-parametric. Group comparisons were carried out using the Mann-Whitney U test.