There are two main part to this answer:
First, I reverse-engineered the grouped data to come up with an appropriate mean and standard deviation on this scale.
Second, I employed some chart trickery to make the normal distribution curve look right when superimposed on the column chart. I used Excel 2007 for this; hopefully you have the same options available in your version.
Part 1: Reverse-Engineer
The column B formulae are:
Last Point =MAX(A2:A6)
Mean =SUMPRODUCT(B2:B6,A2:A6)/SUM(B2:B6)
E(x^2f) =SUMPRODUCT(A2:A6^2,B2:B6)
E(xf)^2 =SUMPRODUCT(A2:A6,B2:B6)^2
E(f) =SUM(B2:B6)
Variance =B10-B11/B12
StDev =SQRT(B13/(B12-1))
Part 2: Chart Trickery
Data table:
Column D
is just an incremental counter. This will be the number of data points in the normal distribution curve.
E2 =D2/$B$8 etc.
F2 =NORMDIST(E2,$B$9,$B$14,FALSE) etc.
Chart:
Now, add Columns E:F
to the chart. You will need to massage a few things:
- Change the series to be an X-Y plot. This might require some editing of the chart series to force a single series to use your desired X and Y values.
- Change the series to use the secondary axes (both X and Y).
- Change the secondary X-axis range to 0.5-5.5 (i.e., 0.5 on either side of the column chart category values). This will effectively align the primary and secondary X-axes.
- Change the secondary Y-axis range to 0-1
- Format the X-Y series appearance to taste (I suggest removing value markers).
The result so far:
Lastly, you can remove the tick marks and labels on the secondary axes to clean up the look.
Postscript: Thanks to John Peltier for innumerable charting inspirations over the years.