Understanding Statistics and the Margin of Error in Surveys - How to Assess Data Accuracy
All surveys other than a complete census or complete administrative records involve various types of sampling errors, the most well-known of which is the “margin of error.”
To properly understand and use evidence, it’s important to have a basic understanding of the margin of error in surveys.
Almost all ETA programs can benefit by using demographic and socioeconomic data from the ACS, which is often the most useful data source for states, and often the sole data source for sub-state areas.
Understanding Statistics and the Margin of Error in Sampling
On April 6, 2016, the U.S. Census Bureau conducted a webinar on statistics and error rates in survey sampling. In this very practical context, the Census Bureau’s American Community Survey (ACS) was explained.
The Census Bureau webinars require no preliminary knowledge about statistics.
For a basic explanation of Census Bureau webinars, a list of future events, and the repository for past presentations, see the Census Bureau links for training events.
New Statistical Testing Tool Answers the Question “Is This Comparison Statistically Significant?”
ACS data provides estimates on a variety of demographic and economic indicators. These estimates come from a sample of a set population that provides survey responses. So, because this is not the whole population responding, there is some statistical uncertainty in the estimates.
The margin of error, for example, is one type of statistical uncertainty. If the uncertainty is too large, then two estimates, when they are compared, may appear to be far apart. However, they might really not be statistically different.
The Census Bureau carries out statistical testing with a tool behind the scenes to provide reliability in the data.
You can also download the statistical testing spreadsheet tool by clicking on the link for the webpage on determining the statistical significance of comparison estimates to the left.