Perhaps what I wrote wasn't clear. I suggested that new immigrants are more likely than others to have a religion.…
Dear Mette Thanks very much for your observation, which I have posted online. Unfortunately, we have had no national censuses…
Net migration to the UK has added at least two million people over the past decade, most of whom will…
I have read a 10 year old article about the positive effect of immigrant communities on Church attendance in the…
Eddie You can email them to c.d.field@bham.ac.uk Best wishes. Clive
-
13 responses to “Visualising Religious Switching, Sticking and Leaving”
-
The nice thing about the InternetMonk chart for the US is that you can more easily visualise the (approximate) percent of people who stay in the faith tradition in which they were raised.
-
[…] & Science Sunday, August 15th, 2010 reddit_url=http:TweetShareThe folks at BRIN have been busy putting together this fab chart showing generational changes in religion in Britain. The data are from the 2008 British Social […]
-
Agree with Tom. The use of pies obfuscates the proportion of people flowing from a group, or making up the proportion flowing into a group.
I don’t know the research well, but is it not the case that people have more trouble comparing areas than lengths? If that’s the case, you might actually have people under-estimating the (rather large) proportion of people in the “No religion” group. At the moment it’s only 2.5x as wide as “Non-Christian”, while it is 6.25x the area. The ratio 6:1 is not immediately obvious to me. Is it to you?
-
[…] 16 Aug 2010 09:47 • atheism,linklog,religion,society The folks at BRIN have been busy putting together this fab chart showing generational changes in religion in Britain. The data are from the 2008 British Social […]
-
Please fix the link to the original post at http://www.internetmonk.com. The link has an extra http.
-
The difficulty I had reading the chart at first was that, looking at the chart before reading the accompanying article, I assumed both of the top circles represented “No Religion”, the second circle from the top on each side represented “Church of England”, etc.
This made it appear that, for instance, nearly half of respondents who were raised with no religion eventually joined the Church of England – and that virtually everyone raised as Roman Catholics lost their religion!
After reading the article, I understood that it was the colour of the leftmost circles which indicated which groups they represented, not the vertical position.
Since I have seen this graphic reproduced on other sites without the explanatory text, perhaps the circles in the leftmost column could also be labeled?
-
Thanks for the comments – I’ve been on holiday for the past fortnight, sans internet, and am sorry not to have been able to contribute sooner.
I like the Internet Monk chart too. With both charts the aim is to represent data which would otherwise require two or three bar charts, which I think both achieve.
It’s a fair point that people can’t interpret the area of circles too easily (which suggests that the default should be bars). But as a first cut I thought that bars would be less attractive. With this type of chart the important thing is to make sure the widths of the flow arrows are proportionate to the quantities – the ‘nodes’ (start and end points) don’t necessarily have to be proportionately sized (sometimes these are represented by boundary maps, or simply by text).
Ordering the religious categories by relative size also means that it’s immediately clear that the modal origin is Anglican, and the modal current status is No Religion. Once I’d left for the holiday I realised that adding the percentages would have been a good idea, so have done that now. I’ve also corrected the Anglican-to-Anglican arrow which was unfortunately slightly too narrow – which would admittedly have been caught more easily using the Internet Monk-type bars.
I also take the point that ranking each circle by size means that the arrow for those retaining affiliation is not always a straight line across, although the fact that these arrows tend to be relatively thick should help them be easily read.
I haven’t added left-hand labels because I wanted to save space – the colours are meant to identify the groups (let me know if it’s a major problem)
The link to the Internet Monk site has also been fixed – thanks for pointing it out.
Regarding the diagram type, I’ve spent a few minutes looking up resources on ‘weighted flow diagrams’ and found that these charts are known as Sankey Diagrams, or mathematically as ‘directed weighted graphs’. This paper by Mansfield and Ambachtsheer (summarised efficiently here) notes that they are difficult to draw by hand, and that there can be problems judging overlay, relative width and ordering of the arrows. (Internet Explorer can’t show the svg files for the paper – Firefox, Opera or Safari are alternatives.) This page provides a list of software for producing Sankey diagrams, some of which are free. Finally, this paper by Edwin de Jonge of Statistics Netherlands gives interesting examples of migration weighted flow maps.
If I do this again I will try using the XSLT stylesheets or packages, and see if this is more efficient and less likely to lead to tangling.
I will also upload some tables later today at http://www.brin.ac.uk/figures which provide similar data for more detailed religious categories – specifically to break down the ‘other Christian’ category.
-
More detailed categories available here (table 8.7 in brin.ac.uk/figures). My interest was specifically in the non-denominational Christians and how they differ from the Anglicans.
-
Wish I had seen this earlier! I may use your technique to improve my chart. I think I can see an “easy” way to improve the lines on your chart too!
-
Hello Michael – finally! I like your chart as it is. Thanks for the idea – these charts have made a big difference to our traffic.
Drawing the connecting lines in Excel was time-consuming to say the least – so any ideas you have would be very welcome…
-
So, here is an relatively easy way to improve your chart using Excel shapes.
First, convert your circles to rectangles. If you still want to keep your chart flowing from left to right, set a constant width to each rectangle of about two centimeters.
Next set the height of each rectangle (in centimeters) to the percentage value divided by 10. So for the Anglicans who are 39.9% as childhood faith. You would have a rectangle of 3.99 centimeters by 2 centimeters.
The line width is in points. There are 72 points in an inch. There are 2.54 centimeters in an inch. So the width of the line moving from Anglican rectangle to Anglican rectangle
would be calculated as follows:We would take the percentage of the population who were raised Anglican and are still Anglican in this case 20.8. This number is divided by 10 to get centimeters = 2.08. The
number is then divided by 2.54 to get inches = approximately .819. The number is then multiplied by 72 to get points, and then rounded to the nearest quarter point. 58.96 becomes
59 pts.So 20.8 % / 10 / 2.54 * 72 = 59 pts.
To save you some steps. You don’t have to do the extended math for each line, instead you can simply multiply the percentage by 2.835 to get the pixel width. ( 1 / 10 / 2.54 * 72)
= 2.83520.8 * 2.835 = 58.968
I did try this out and it does work.
Send me an email. If you have the time, I would be interested in publishing an updated chart at Internetmonk.com
-
[…] the numbers, the vast majority of atheists in the US come from religious backgrounds. This report seems to support my hunch that, for the UK, the trend is less pronounced, though it looks like a […]
-
