What you describe is known in the trade as the detonator plot (or dynamite plot) and is decried by statistical editors the world over.
There are several reasons for this
1. It doesn't show the data. Ironically, it is most used by people who use small numbers of hapless rats, who could easily display the data without crowding the graph.
2. The standard error tells you nothing. It isn't a confidence interval and it isn't a test for a difference between means.
3. The only useful part of the bar is the top, which shows where the mean is - all the rest of the ink tells you nothing.
4. They do not facilitate comparisons between groups when several groups are shown, for instance, at several time points as the eye has to work to locate the bar corresponding to each group at each time point.
For simple display of means of a number of groups, consider displaying the means as dots and the confidence interval as whiskers.
For data that vary over time, consider connecting the points for each group together with a line, showing how each group varies over time.
That said, I have had PhD students whimpering to me "but no-one else graphs it like that". The desire to conform cannot be underestimated...
Ronán
________________________________________
From: Evidence based health (EBH) [[log in to unmask]] On Behalf Of Markos Kashiouris [[log in to unmask]]
Sent: 31 March 2009 16:31
To: [log in to unmask]
Subject: Question
Frequently in basic research papers, data are graphically presented in a bar-format with one standard-deviation up and down represents the error bar. I have been working in a lab in the past. The data were presented in such way that if the SD error bars didn't overlap in the comparison groups eg cell counting etc the researchers presented their data as significant.
These basic-science research papers are the basis of the pyramid which leads to animal studies and higher phase trials including human research. I was wandering if this reporting method is statistically justified and if not what the error bars shall include
: 2 standard deviations each side (up or down from the bar), one SD each side of the bar or using the Standard Error?
Thank you.
Markos Kashiouris, MD
PGY2 Resident in Internal Medicine
|