todogroup / osposurvey

Open Source Programs (OSPO) Survey
https://todogroup.org
Creative Commons Attribution Share Alike 4.0 International
71 stars 25 forks source link

How were participants were selected? #44

Closed cben closed 5 years ago

cben commented 5 years ago

Hi, coming here from reading https://thenewstack.io/research-shows-open-source-program-offices-improve-software-practices/. Obviously the survey sampled only a tiny fraction of all companies in the world. Please say something about how the survey was distributed / what kinds of companies was it targetted at?

Even if a reader trusts your conclusions "the findings are not the result of self-selection bias", i.e. that it's a representative sample, it's unclear a sample of what. For example there are probably no mom-and-pop-pizza shops here, or other completely offline businesses here... Which I agree is out-of-scope to ask about OSPO and such :wink:, just making the point there some implicit scope and knowing it would be useful to the reader.

A bar chart of verticals [and company sizes] could go a long way towards answering this. chart3 has number of answers by vertical:

vertical n
Defense n=29
Education n=112
Financial services n=127
Government n=66
Healthcare n=48
Insurance n=24
Manufacturing and raw materials n=38
Retail n=57
Technology (software or IT) n=712
Telecom, communications or media n=142
Transportation and automotive n=58
Utilities n=30
Other n=180

But is that number of individual respondents, or number of companies?


BTW, are all the graphs weighted by number of respondents from same company / company size / normalized to 1 per company? "One hundred and thirty responses were excluded because they appeared to come from the same company" — so I guess 1 per company, all weighted equally?

LawrenceHecht commented 5 years ago

@cben Here are some answers:

Q. How many people work for your company or organization? 2019, n=2671; 2018, n=748. 2019
Self-employed or not working 12.3%
2-50 22.8%
51-250 15.7%
251-1,000 13.0%
1,001-10,000 14.4%
More than 10,000 20.8%
Don't know 1.1%
caniszczyk commented 5 years ago

Thanks @LawrenceHecht