How to Interpret Error Bars

Error bars are often included in graphs to relay confidence in the interpreted data. When data are compiled, an average by itself is not all that meaningful, because you do not know how much variation there may be in expected measurements. Suppose you had two data points that were 10 and 10,000; the average would be 5,005, but that's a long way away from either measurement. The error bars offer a convenient graphical interpretation of where most measurements would appear, usually 95 percent of them.

Instructions

• 1

  Look at the error bars, which are smaller measurements located above and below the top of a measurement bar or line. The primary measurement is an average, with the error bars representing some percentage of measurements that are expected to appear within that range. Usually, that range is 95 percent, although some researchers may use 90 or 99 percent.

• 2

  Compare subsequent measurements to see if they fall within that range. As suggested, 95 percent, or whatever the chosen percentage, fall within this range.

• 3

  Draw a horizontal line across the graph from each average's error bar. If the ranges overlap, then the differences in the averages are considered insignificant; that is, there is no statistically measurable difference between the two averages. If the error bar ranges do not overlap, then the averages are considered significantly different, meaning there is sufficient experimental data to suggest the values differ more than would be expected by chance alone.