**Figure 3:** Overlay of profiles before averaging

**Figure 4:** Median values with percentile error bars

You can average data points over evenly
spaced *x* intervals. Error bars can show the standard deviation of the mean in both *x* and
*y* or the error estimate of the mean. To average data, set the
number of intervals you want. To stop averaging, reset the number of
intervals to zero. If you have lots of clusters of points and you
want only nearby points averaged, try choosing a very large number of
intervals even though many intervals will have no points in them. The
absence of an error bar means that only one point is in that interval.
Symbol groupings, if defined, are
retained in the averaging process. You can fit averaged data.

A median option is also available. Figure 3 is a graph of points representing several profiles and Figure 4 is a graph of the median profile where the error bars show where a specified percentage of the points lie centered about the median.

<3=A;> Analysis: Average, Frequency, ... ? <AV>: AV <4> Number of intervals for averaging? <10>: 20 \* Options are: 1) Standard deviation of the data (sd) or 2) Error on the estimate of the mean (sd/sqrt(n)) 3) Percentile error bars (medians, not means are shown) <5> Error bar represents? <S>: PE <6> Y percentile (error bar will show inner y%) <90>: 68 <7> X percentile (error bar will show inner x%) <100>: 0

Fri Jul 11 17:05:56 EDT 1997