I am currently undertaking a demographic research project for my employer. We have the data on income, age, sex, language, transport etc and I have mapped most of it and created some thematics of said variables.

Whilst the maps are perhaps adequate from a raw data representation perspective, I am looking for a technique or thematic style which displays statistical and demographic data in a more intuitive and interpretive fashion.

Here is an example of what i would like to improve:


Is there a specific technique or method that people are using which displays thematic data which is more easily interpreted by a non GIS user?

Similar to this question but perhaps with more technical descriptions on how to achieve some of the better cartographic outputs.

@IanS suggested using rasters to generate a heat map type approach:

Heat Map of Non-English speaking families

  • I guess I'm after something innovative or different for displaying multivariate thematics. I don't think I can word it any other way really. Not sure how that constitutes a shopping list question. I deliberately kept the tech out of it as I wanted the answers to be focussed on the theory, not the practical. I know how to use ArcGIS, qgis and MiPro effectively enough that I don't need step by step instructions, more concepts and ideas. Never mind, not fussed really, just thought it was an interesting topic and my Google-fu didn't really get the results I was after.
    – user21482
    Jan 28, 2014 at 9:55
  • And to be clear, I don't want a tool, software package, plugin, script or piece of code, more a technique.
    – user21482
    Jan 28, 2014 at 10:42
  • Hmmm, okely dokely, I have read just about every inch of the meta, including the topics you mention, I will have a go at editing. I've been posting on forums, technical and non-technical, for a long time now, the foibles of SE are subtle and peculiar, to say the least!
    – user21482
    Jan 28, 2014 at 22:19
  • let us continue this discussion in chat
    – user21482
    Jan 28, 2014 at 22:30
  • Sure, I was only moving it as I was hopeful it would remove the comments, as per the meta we don't want long drawn our protracted discussions :P
    – user21482
    Jan 28, 2014 at 22:49

1 Answer 1


I have used this kind of socio-economic data for a number of projects.

It can be very helpful to break out of the district polygons by laying a square grid over the area (side length based on either metres or minutes), and then using a script, calculate a score for each grid cell (e.g. if a grid cell straddles two districts, then calculate a cell value based on area of the cell covered by each different district value). You then thematically map that. The advantage is that it gets rid of the cluttered 'district' look, and becomes a more pure visual of 'hot and cold' areas. There is a bit of an art (or experimentation) to choosing an appropriate cell size.

A step further: once you have different variables (that came in different polygon shapes, not always the same district boundaries; e.g. police districts being different from census districts) set up into the common grid, you can perform correlations and build indexes. These new calculated values (e.g. burglaries by average income) can then be mapped.

  • Ah, thanks Ian! This is the sort of answer I was after. In essence you are suggesting using a raster with cell values, rather than the abstract geographic boundary. I have done this for more uniform type data but hadn't really thought of applying it within this context. I will see how I go and post results. Thanks again.
    – user21482
    Jan 28, 2014 at 22:27
  • Yes, a raster approach. For some data there is a question as to whether it is a valid approach if it 'cheats' by making lumpy low-precision data look like it's superb continuous-variable data. The QGIS heatmap plugin looks interesting for what we are talking about. Is that the same data you've posted in the new screenshot as in the old?.
    – IanS
    Jan 29, 2014 at 6:29
  • No they are different variables, the heatmap is for Non-English speakers, and I think it works well. You can straight away tell that within our area there is a higher saturation of that demographic. It doesn't work that well for, as you mentioned, more non contiguous (?) type data. I am finding Age to be that way for example. I am now thinking about creating a z score for each of the main demographic variables we want to measure against, then including some asset condition data, some distance analysis and then doing some basic raster calculations. That should be sufficient I think.
    – user21482
    Jan 29, 2014 at 22:23
  • Also, the heatmap plugin is really, really good. easy to use and understand. But I wouldn't mind someone explaining the different Kernel shapes. Might be a question that is more suited to GIS.SE
    – user21482
    Jan 29, 2014 at 22:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.