Statistically Significant Trends

I work for a community indicator project that tracks various indicators of community well-being over time, and many of these measures come from the ACS. As we update our indicators, we are incorporating more tests for statistical significance. We were able to find guidance for how to calculate statistical differences between two estimates in the compass documents, but part of our work involves identifying and comparing trends. For example, we need to know whether poverty is trending up or down, and we sometimes need to be able to identify whether poverty is increasing faster or slower in two different populations.

I've identified some statistical tests for trends from other fields, but none of them seem to be intended for data points with their own margins of error associated with them.

Is anyone aware of statistical tests for identifying and comparing trends that would be appropriate for use with ACS data?
  • Elizabeth,
    You should be able to use the Census Bureau's published formulas to conduct significance tests between ACS estimates at two different time periods.
    Their Compass Handbooks include formulas for "Calculating MOEs for Estimates of Percent Change or Percent Difference”

    See Appendix (page A-17) in this report:
    www.census.gov/.../ACSGeneralHandbook.pdf