Trend Analysis

Here are example teen birth data from Brown County in Wisconsin, downloaded from a Wisconsin vital statistics website.

  1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Brown County 35.5 39.6 34.3 35.8 35.3 30.7 32.7 32.6 35.9 35.9 32.2 31.7
Wisconsin 36 35.7 34.3 32.7 31.6 30.5 30.5 31.1 32.4 31.3 29.6 26.5
United States         41.1 40.54 39.73 41.09 41.46 40.24 37.93 34.25

You will notice that we do not have as many years of data for the United States as Wisconsin and Brown County. This is because the data source we used for U.S. data does not go back as far as the local source. However, this is not a problem—as long as you have at least five years of data, you should be able to detect a trend. 

Step 1: Graph data to visualize the trends  

You can do this in many different software packages, but the example below was created in Google Docs since this software is freely available to anyone with an internet connection.

 

As you can see in this graphic, the trend in both Wisconsin and the United States has been decreasing over the past decade, with some variations. The trend line for Brown County is far less smooth, but also appears to show an overall decreasing trend.
Keep in mind that depending on the measure, a downward trend may represent an improvement (e.g., for premature death) or a concern (e.g., for high school graduation rates).

Additional guidance for interpreting trend graphs is available.

Step 2: Quantifying trends

There are a few ways of quantifying these trends. The one most statisticians would recommend would be to fit a regression line to the data, and test for slope and differences between the county, state and nation. (These are the tests that we conducted for the trend graphs we make available). However, if you do not have the statistical training do this yourself and you don’t have easy access to a statistician, you can approximate your county’s trend and its differences from the state and nation, by calculating the percent change over time.

Using the example data above, we recommend first stabilizing estimates by averaging two or more years, e.g., averaging the 1999-2000 and the 2009-2010 estimates. You can then calculate the percent change as follows: 
Percent change for Brown County = (Rate time 2 – Rate time 1)/Rate time 1 = (31.9-37.6)/37.6 = 15%

  1999 2000 2009 2010 1999-2000 2009-2010 Percent change
Brown County 35.5 39.6 32.2 31.7 37.6 31.9 -15%
Wisconsin 36.0 35.7 29.6 26.5 35.9 28.0 -22%

This calculation hints that while the rate of teen births in Brown County is decreasing, it may not be decreasing as fast as the rate in Wisconsin as a whole. Note that in this example, we cannot calculate the comparable change for the U.S. since the U.S. data source only goes back to 2003.