Our trends are estimated using linear regression for all of the years of data included in the graph. This sometimes creates unusual situations, particularly when a measure both improves and worsens over the time period. For example, in many counties, unemployment increased dramatically between 2007 and 2011. Since then, in many counties, unemployment rates have been improving. However, the overall trend for the county depends on the relative magnitude of worsening to improvement. For example, even if your county has had improving rates over the past 3 years, but had higher rates of worsening in the years prior, it might show a worsening trend.