One important caveat:
The effect that I describe below relies on having a lot of data
points close together. This creates the effect of a continuous color
bar (which, in actuality, is lots of small vertical lines). So, the
fewer the data points that you have the less spectacular will be the
effect.
Below is the description of how to
create one zone (highlighting in red the area from -3 standard
deviations to -4 standard deviations). I will leave it up to
you to do all the zones
The data set: Start
with a thousand normally distributed random numbers with a mean of
zero and a standard deviation of 1. This means the data set would
have about 2/3 of the points in the range -1 to +1, 95% in the range
-2 to +2 and 99.5% in -3 to +3. The remaining, or about 5 points
would be either < -3 or > +3.
Step 1: Create an x-y scatter
chart of the 1000 data points and set the line format to none.
Step 2: Set the series
marker to a minus sign (i.e., '-') with a size of 2 units.
Remove any extraneous formatting (legends, gridlines, borders, plot
area, etc.) that Excel creates by default. Adjust the min and
max values of the x- and y- axis to your satisfaction (I used -4 to
+4 for the y-axis and 0 to 1000 for the x-axis). Remove the
solid line used for drawing the axis. Also, change the y-axis
attribute so that the x-axis crosses at a value of -4 -- which is -4
sigma. I also changed the font size of the axes to 9 point and
turned off the auto scale option.
Step 3: Next, add 7 more series
each with a 1,000 points (the same number as the data points). The
values in the first series are all -3 (which happens to be -3 sigmas
from the mean in this data set), the second series consists of all
-2, ... the fourth is all zeros (the expected mean for this data
set), ... the last is all +3.
Step 4: Select the first new data
series (the -3s) and add it to the chart. Format the data
series as follows: Set the markers to none. Set the line style to
the last option in the drop-down list (it is a rectangle with about
25% fill). Set the line color to red. Set the line weight to dots
(the first option in the drop down list).
Step 5: Add a y-error-bar (the
minus type) with a value of 1.
The result at this step will be a
solid 'bar' of black in the chart. This is the cumulative effect of
all the error bars.
Step 6: Double-click the (error)
bars to format them. In the Patterns tab, set the line style, color,
and weight as for the series itself (see the previous paragraph). In
addition, select the Marker type to be the vertical type without the
horizontal cross line (it's the second option in the Marker section
of the Pattern tab).
You will find the result is that
the 1,000 data points show up against what seems to be a lightly
colored red 'bar.'
Step 7: Repeat the process for
each of the other series. For the mean (zero in this example)
series use a + and - error bar of with a fixed value of 1. For
the +1, +2, and +3 error bars use + error bars with a fixed with of
1. The final effect:
