Why collect demographic data?

I’ve now run two large online surveys of Quakers – one about threshing and one about afterwords – and on both occasions, questions about age, gender, and education raised protests. Since this information isn’t directly related to the topic of the survey, I can understand some objections about its irrelevance, but it is important in some ways and I’d like to take a little while to explore these questions, why I ask them, and what I do with the information.

Firstly, I’d like to dispel a possible misconception. I do not use this information to judge the answers to other questions. In fact, before I analyse the data I usually detach different parts of it: when I read a detailed answer about afterwords in a particular meeting, I am not at the same time looking to see how old this respondent is or whether they have a degree. Perhaps knowing this would help the Friend who refused to tell me about their educational background because “whether academic or not you can make a spiritual decision”. I completely agree. In fact, I’d really like my data to include more responses from people with less formal education – I realise that by choosing to take written responses I’ve excluded some people who struggle with reading and writing, but even at a quick glance my respondents are much more educated than the population as a whole.

This brings me to one of the things I do do with this information: compare it with the UK population as a whole. According to our census data, even in the best educated sector of the population (as of 2011, this was women aged 25-34), only around 42.6% of people had a university degree or higher qualification. Roughly 85% of the people who answered my survey had a university degree, or equivalent or higher qualification. They aren’t in that age band, either – 40% of my respondents are aged between 61 and 70 – although 68% of them gave their gender as ‘f’, ‘female’ or ‘woman’.

You might have noticed if you did the survey that I chose to take all this information in free-text boxes, where respondents could write anything they liked, rather than restricting people to clicking a button marked ‘m’ or one marked ‘f’. Setting out the question in this way enables people to give other answers – such as ‘cis male’, ‘gender fluid’ or other, more individual responses (one respondent simply gave a personal name, perhaps to say ‘my gender is completely specific to me’). Allowing this freedom of response is an important point of principle. I did regret sticking to this principle in the case of the question about membership, though – I’d intended to people to answer the question “Are you a member or an attender of a meeting in Britain Yearly Meeting?” with something like ‘member’ or ‘attender’ or perhaps ‘member of another yearly meeting’, but a number of people answered it with ‘yes’ or ‘no’, which didn’t provide enough information for me to do anything useful with it.

This, more specifically Quaker, information leads to the other thing I do with these answers: compare with other Quaker sources to see whether I have a sample which is representative of the whole. For example, I can compare my information about gender – about 68% women – with the tabular statement for Britain Yearly Meeting for 2015, where 62% of members and attenders across England, Wales and Scotland are women, and see that I’m only a little bit out. On the other hand, I can also look at my information about membership, where of those who gave clear answers and indicated that they were within Britain Yearly Meeting, 88% are in membership, and see that they’re significantly over-represented: in the tabular statement, only about 63% of the adults reported are in membership (the other 37% being attenders).

I can think of ways to explain this. For example, people who have been involved with Quakers for longer are more likely to be in membership, more likely to be on the email lists and Facebook pages where the survey link was shared, and more likely to feel confident reporting on their experience of afterwords. However, if I hadn’t collected this information, I wouldn’t know that there was anything I needed to explain.

There are many kinds of information which I haven’t collected, too. For example, standard equal opportunities monitoring forms in the UK at the moment would also ask about sexuality and ethnicity, both questions I chose to leave out. I didn’t leave them out because I think they don’t matter – my whiteness and queerness make as much difference to my experience of Meeting for Worship as my age and gender, which is to say, they are incredibly important in some ways and irrelevant in others. I did leave them out mainly because there are not as many high-quality sources with which to compare, and because when the numbers are low (for both alternative sexualities and minority ethnicities, we are looking at around 10% in many locations in the UK, although this various considerably depending on several factors) it would be easier to have a small change look important when it was actually just chance.

By collecting four data points – age, gender, education, and membership status – I hope to have some points of comparison and get a feel for how my sample is similar or different to other Quaker and national surveys, without asking people to fill in more questions than is reasonable.

Advertisements

2 responses to “Why collect demographic data?

  1. My impression is that most Quakers have a degree, so 85% might on that basis indicate a sample of Quakers close to the Quaker average. Is there research on that? (I would leave debate on whether this is a “Good Thing”, or what might be done about it if not, for another time).

    It might have been me, giving my name as my gender. I am uncomfortable around these classifications, at a QLGF event stating “My gender is [Clare]” seemed to evoke enthusiasm, and I find [name]gender is a thing since at least 2014.

    • There is research on educational levels (which is why I include that question), and 85% is probably more than typical – the 2013 British Quaker survey, which was much more carefully structured to capture a representative sample, got 71% of respondents with a degree or above. (Initial findings in a pdf file here: https://www.woodbrooke.org.uk/data/files/CPQS/British_Quaker_survey_2013_initial_findings.pdf)

      No comment on who it was. I hadn’t encountered the format before (probably because I stepped a bit back from following the discussions in such detail before 2014), but was pleased to see that the free-text space had allowed that creativity – I can think of good reasons why people are uncomfortable around these classifications but also need to collect some of them.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s