The Household Census : A Summary Post

Over the last year or so, we’ve done several separate posts on different types of deprivation (or access) using the Census 2011 data. We’ve looked at everything from phones to toilets, to drinking water to electricity. It’s time to put these together in one chart, and try and link these diverse measures together. If a district has high ownership of TV sets for example, thus implying a certain standard of living among households there, how likely is it that those households will also have access to the basics – say water or electricity?

The chart below may look a confusing mass of different coloured lines, but it’s quite simple to interpret. Each line represents one district. Each state is allocated a different colour, and states are ranked from bottom to top (red to green) according to their per capita income.

There are a bunch of different axes from left to right, measuring different parameters – phones, electricity, TV sets etc (also note the ‘rich’ and ‘poor’ measures which I’ll come to in a moment). Tracing the points at which each line (a district), hits an axis, tells us how that district scores on each parameter. So for instance, Paschim Champaran, in Bihar, which is the first entry in the table below the chart, has 52% of households with a phone, 9% with a TV set, 52% with access to water and so on. Hover your mouse over a row in the table and the corresponding district in the graphic gets highlighted.

You can go deeper. Click and drag on any axis to filter only those districts whose lines pass through that part. Now you can move the filter up and down to update the graph. The data table is updated automatically as well, enabling you to get at the actual numbers. You can put in multiple filters on different axes as the same time. Simply click on the ‘reset filters’ button to get back to the original view.

If for instance, we highlight the state of Delhi at the top left, we can see that Delhi’s districts are clustered fairly close together on all measures. Now move the filter down, and we see that other high income states such as Maharashtra, Haryana and Gujarat, are actually quite diverse at the district level, with some districts scoring high, and others scoring low. A state like Kerala may rank lower in the scale of per capita income, but on most measures, Kerala’s districts are tightly clustered together (like Delhi), showing that district-level inequality is lower in that state, than in other, richer, states.

A word about the ‘rich’ and ‘poor’ measures. ‘Rich(a)’ and ‘poor(a)’ are measures of access to a basket of consumption goods, rather than just one. ‘Rich(a)’ for each district, indicates the percentage of households who have a TV, computer, phone and a motorised vehicle (2 or 4-wheeler). ‘Poor(a)’ indicates the percentage of households in a district who have none of the consumption goods covered in the census – not even a phone.

Similarly, ‘rich(b)’ and ‘poor(b)’ measure access to a basket of household amenities like drinking water, toilets, and electricity simultaneously. ‘Rich(b)’ measures those households who have electricity, and access to drinking water and a toilet at home. ‘Poor (b)’ measures those households who have no electricity, and no toilet at home, and have to travel more than 0.5 km for drinking water. Note that unlike the other measures, the lower you are on the ‘poor’ measures, the better off you are.


The table displays data for only ten districts at a time.

As always, all data is from the census 2011.

The charts are made with the D3 javascript library ( Also used is the utterly invaluable parallel coordinates library which uses D3. The chart also uses divgrid and is based on this example here. This chart is one of the more complex ones I did, but it was actually one of the easiest to code thanks to the supporting libraries available. I’m still amazed at how easy it was.

February 22, 2014

Comments are closed.