There are many potential Traps with Census maps.
Below, we talk about the most common you’ll likely find:
Small populations must be disguised for privacy reasons. This gives people the confidence to answer census questions more truthfully. There are two different ways of doing this…
- Small geographies: Where you have small geographies, when there’s only a very small number of houses or people in them, often the Census Organisation only reports basic statistics such as the number of people and dwellings there.
- Large geographies: For large geographies, the approach is a bit different. Where there are a small number of Cases, these get rounded-up. For example there might be just one person who’s income is more than $10 billion a year. Their details would be hidden by rounding-them-up to 3 people or they might not be reported at all.
Temporal analyses are difficult if data definitions or geographies have changed. So if you’re comparing the level of unemployment between two census periods, just be careful that the “definition” of what it is to be unemployed is the same in each census period.
For example, being unemployed might be defined in one census as working less than 10 hours a week, and in a different census, that definition might be 15 hours per week.
There are two important parts to this:
- Redefinition of boundaries due to population changes: Boundaries might be changed between the census periods because populations have increased or decreased in a way that makes it easier or more difficult for a census collector to visit each household.
- Land use change: For example, farm land might become urban between two census periods.
The census is a snapshot
It’s taken on a particular day. Particularly in countries where the census is taken once-a-decade, inter-census economic booms and busts can be missed. For example, banks might have folded and people lost their houses and jobs soon after a census was taken, but by the next census the economy might have recovered.
Temporal incompatibility with your datasets
The Census can be temporally (time) incompatible with your own data sets. For example a study on fruit pickers is not likely to be very well represented by the census because fruit pickers are a mobile workforce that are rarely active at the time the census is taken.
Spatial incompatibility with your datasets
An understanding that the census is about where people live, not where people work can help you overcome some of the things you’re finding difficult to explain.
For example, the census might not be a good tool for understanding problems in business centres because business centres often don’t have a lot of residences in them. In some places census bureaus will, for a price, reconfigure their data to represent where people work and not where they live. They can do that if one of the questions is “What’s the address of your workplace?”.
Census Bureaus rarely publish all themes at all map scales. Some detailed cross-tabulations are only made available at small scales (big areas), and can hide the detail that exists at larger scales such as Suburbs. This can lead to misrepresentation of some themes.
For example, an analysis of the census at a national scale might revealed that a country is predominantly middle class. However, if you were to repeat that analysis at suburb scale, you might find that there’s very poor areas in that country and very wealthy areas, and only on “average” is it middle class.
Figure: Geographies can be out-of-focus
Census geographies can be out-of-focus with other geographies. Geographies can be changed by both census and non-census organisations.
For example the Post Office uses zip / post code boundaries to assist it with postal delivery. For this reason, their boundaries do not always align with road centrelines, and may even align with the back fences of houses on “some” street frontages.
This means that the same house can be in two postcodes – one for the census, and one for the post office. This can cause problems when you’re comparing non-census postcode datasets (eg. retailers asking customers “what postcode do you live in?”) to census postcode datasets.
Census geographies can introduce misleading statistics
In many towns on the urban fringe, many people who live in the town work outside the town. There’s a couple of ways that this can lead to misleading statistics.
- The income level for this area is going to be based on people working in the city and returning to this rural environment to live. So, if you’re a farmer on a low income, the urban population is probably propping up the “average” income of this area.
- Boundaries that include both broad-acreage and town areas can skew population density measures. On the one hand the township will have a lower population density than if the census boundary coincided with the township boundary. And on the other hand, the rural area will be seen to have a higher population density because of the town.
Not all census geographies are compatible with each other
In order to protect privacy, and to make it difficult for people to decompile the dataset often census bureaus ensure that some of the boundaries are incompatible with each other. For example, electoral district boundaries might be deliberately incompatible with zip code boundaries.
There are many other examples of traps with census maps. None are issues that should stop you from using the census for your studies, but they are issues that you to be aware of when you’re designing your studies.
Learn How To Build A Map And Avoid These 10 Cruical Traps In Our New 5 Step Video Challenge, Click Here