Parkour still becomes more popular day by day, the image has changed from being just a trend to being a serious practice with a structured methodology. The fitness industry has taken up on the idea of parkour and slowly a broader academic interest in the practice is evolving.
Whenever academics is trying to understand and analyse new and fairly unknown fields, survey methods provide a great tool for getting the knowledge and generating new and valid information for academic analysis.
I am writing this short article because I am seeing so many qualitative bad surveys on parkour (and in general). Basically here are some of the most common mistakes you can make in a parkour related survey and in surveys in general
a) don´t lie about the survey length in your opening statement – while people are more willing to start your survey when the announced length is shorter they are more likely to drop out if the length is longer than announced. Also: respondents who took the survey and were informed correctly on the length are very much more likely to finish the survey but less likely to start it. But that´s what you want, you want data quality, so be honest and let them choose!
b) don´t reproduce data – it sounds stupid, but don´t just go out there and throw your survey in everyone’s face. First you should thoroughly check if your data is out there already. Unfortunately every bachelor project nowadays is trying to get people to finish their survey. In the time I wrote this I saw 2 surveys on injury types and frequencies being posted in the pk research facebook group and the quality was horrible. Instead of reproducing already existent data try to do either get hold of already produced data or try to check other sources. For injury related data I found an interesting data set for the US region that might provide similar insights as some of the surveys being posted out there right now. If that does not cover it, the data might help making your questionnaire shorter at least.
c) now comes the tricky part. Question quality and questionnaire quality… While many surveys are made by people from a broad variety of fields and backgrounds it is disturbing how little it seems survey methodology is something to care about. It starts with the usage of unclear terms and definitions, for example: How many hours do you spend on training a week? – Sounds clear and simple? Well what if I told you the concept of training depends on who you ask. What if I told you that if you asked me the same question 5 years ago I would have answered 4-6 hours (2hrs per outdoor training session) but now I would answer 10 hours, because the term of training has changed for me. Training might also refer to a mental side where I for myself define walking to work instead of taking the bus as training. Not that I do, but I hope you get it. So what to do in that case? Make sure to clearly define your terms and concepts, make sure every respondent understands the same. Otherwise your data quality will severely suffer, up to the point where your results will be useless. That´s quite a common mistake and you can avoid it by pretesting or peer reviews before that.
Another common mistake comes from answer categories that do not cover all the options or that overlapp. Ever saw something like this: How many hours a week do you train: i) 0 – 4, ii)4 – 8, iii) 8-12. Yeah….well…I train 4 hours a week. What do I choose? There is a ton more of these errors that can easily be avoided by the help of your fellow sociologist / survey methodologist. And these errors are really simple ones. Not mentioning question order biases or sampling errors. Just please do your research…In the worst case you draw (wrong) conclusions that get spread that are the product of a poor survey design.
If you want to know more go and get – “Dillman 2000 – The tailored design method”.
If you need help from a trained survey designer / sociologist please contact me at email@example.com
For an example dataset regarding Injury rates in the US see: https://www.cpsc.gov/vi-VN/Research–Statistics/NEISS-Injury-Data
(look out for Parkour, Freerunning or other keywords in the datasets)