Apache OpenOffice (AOO) Bugzilla – Issue 9279
Error bars need more functionality
Last modified: 2013-02-24 21:19:43 UTC
The error bar function on the chart feature is not well implemented at present. It is an important function for use in science and business and so is in my opinion worth developing. At present the error bars follow a similar structure to MS Excel (Although it should be noted that excel imported charts do not display error bars correctly). The way they work at present is actually quite useless: you can select standard deviations etc. but the error bars display a standard deviation of the plotted bar values (ie. if you have 3 bars of values 3, 5 and 7, the error bars will be the standard deviation of these three values and will be the same for each bar). What is actually needed is a standard deviation (or better a 95% confidence interval) of the values which are averaged to get the value of each bar in the first place. The error bars should be of different sizes on each of the bars depending on the SD of the data for which the value of the bar is the mean. This will require (at least) two extra features: firstly to be able to plot the mean of values as one point in a chart (rather than just having to work out the mean and then plotting that), and secondly to allow different error bar sizes depending on the SD of the data from which the mean was obtained. If this feature was implemented it would put the chart function of open office head and shoulders above Excel. Excel does allow you to specify different error bar sizes for each data point but you have to work out the SD (or SEM) within the spreadsheet and then point to these values from the error bar section. Being able to select SD, Standard Error of Mean or 95% confidence interval directly from the chart section would be a vast improvement (look at something like sigmaplot for the way it should be done). Regards James Stone
I know this has already been mentioned but it is an important feature.
duplicate of issue 8877
duplicate *** This issue has been marked as a duplicate of 8877 ***
*** Issue 10845 has been marked as a duplicate of this issue. ***
closed