Census vs. Sample

[apdragon]

I am looking for more clarification on how a census survey is usually analyzed and reported. While there is a lot of information on how to go about applying statistical tests and estimates for sample data to make inferences about the population, there is not a lot of information on how to interpret census data. The information sources I have come across just states that it is not practical or economical to survey everyone in the entire population therefore sampling should be used. However, the population I deal with is small and it is feasible to survey everyone.

I understand that statistical tests rely on the sampling distribution to determine how certain you are that you are within a range from the population parameter. However, in a census survey wouldn’t I have the population parameter and therefore would not need tests of significance? Would I just summarize the data as found on the frequency tables (depending on my data type of course)? What about measures of associations to test relationship with subgroups? Any guidance that can be provided would greatly be appreciated.

[terzi]

Exactly, any descriptive analysis will give you the true parameters, so you won't need any inference at all.

Still, there are many descriptive multivariate analysis that may provide some cool results, but that would be based on your objectives.

[apdragon]

Terzi,
Just to clarify, are you confirming that I would simply describe the data as is and use univariate descriptive statistics? As with most surveys, there is non-response. What if I wanted to account for the uncertainty with the nonresponse?

Also, which multivariate analysis could I use for population data that wouldn't test for significance?

[terzi]

Yes, a descriptive analysis will give you the true parameters. So contingency tables, frequencies and descriptive statistics and graphics can be used. You can even obtain correlations and other bivariate measures and analysis.

Now, there are certain multivariate tools, such as Cluster Analysis, or Principal Component Analysis that give a descriptive, multivariate insight of your data. Of course, everything will depend on the results you wish to obtain.

[ndthl]

A friend made these comments about census statistics:

When you have a sample you use inferential stats to generalise to the population. When you have a census you already have data for the whole population, so there is no need to generalise.

For example, if you used sampling, and there is a 3% difference between groups, then you have to use inferential stats to decide whether that 3% difference is real, or just due to random chance when you did the sampling.

But if you did a census, and there is a 3% difference between groups, well, then there's definitely a 3% difference. That 3% difference is not due to random chance in sampling, because you have data for the whole population. However, even with a census you will still need to use your own judgement to think about why there is a 3% difference (for reasons other than random chance in sampling), and whether the 3% difference is large enough to have any practical significance for the work you are doing.

So basically, just use descriptive stats. Correlations are fine, but you only need the r value to show the strength of the correlation, not the p value which is related to random chance in sampling.

A lot of people don't get the difference between sample stats and census stats, and will complain that you didn't do the stats properly. I've had cases where I ended up having to do inferential stats on census data just because people complained so much that there were no p values on anything!

If you have a lot of missing data from a census sometimes you need some fancy inferential stats to fill it in. I doubt this will apply to you, but it does apply to the US population census because (for some bizarre libertarian reason) completing the census survey in not mandatory in the US.

[bryangoodrich]

ndthl shall henceforth be known as the resurrector.

[Dason]

(I don't think necro-ing a thread is a bad thing mind you)

[ndthl]

ndthl shall henceforth be known as the resurrector.

Thank you for the accolade! I spent about a day Googling trying to find anything useful about how to do census statistics. So I submitted the above in case someone else was in a similar position and would benefit from the added insight.

I am still interested if anyone knows any books, tutorials, presentations, videos or other publications on this as all of my academic stats training was for samples.