Research Concepts

Census or Sample

If you collect data on every member of your population, then you have gathered census data.

If you collect data on only some members of your population, then you have gathered sample data.

Since census data involves complete information, there is no uncertainty involved: you have a complete description of your population.

Sample data, on the other hand, always involves incomplete information and hence always involves uncertainty.

Since census data eliminates uncertainty, why would one ever use sample data? The most obvious reason is that sample data is both cheaper and easier to collect. In other cases, a census may not be possible. For example, a white blood cell count is always based on sample data: census data would exsanguinate the subject, a highly destructive oucome!

In other cases it may not be ethical to gather census data. In drug trials, for example, the experimental drug may not work or may have harmful side effects. It is unethical to expose the entire population to such risks unless there is evidence to support the change.

Since samples represent incomplete information, the selection of the sample must be done with care to avoid bias.

Finally, even with a census there may be some uncertainty from measurement error. All measures have some limitations to their accuracy, for example. Thus while a census has "complete information" in the sense that data has been collected on every possible member of the population, there may be residual errors from other sources. Finally, even the US census missed some members of the population and only recorded data on seven questons for all members of every household that it did survey.