Importance of individual data

2 posts / 0 new
Last post
Andrey173's picture
Importance of individual data

How important do you think it is to report indiviaul animal data together with group average data?

marcus muench
marcus muench's picture
Well this is a fairly broad

Well this is a fairly broad question. I assume your talking about studies performed on small animals (rodents) in which case one has a decent sized number of experimental and control animals.

Generally in my experience data are presented as median ± range or sometimes mean ± SE. The former has the advantage of suppressing the effects of outliers such as you mention. However, box plots are a great way to present data showing the median, mean, range, inter-quartile range and/or individual data (depending on your software program). Wikipedia has a good description of these graphs. The point is that you can show both the median and outliers.

The question of whether outliers should be discussed really depends on the particulars of the findings, and can lead to a fair amount of debate among co-authors. I guess the question becomes is the outlier part of a smaller subset of responses or is it just the extreme end of a normal distribution? One also has to ask if it may be the result of experimental error. Odds are that if you don't have enough evidence regarding the significance of the outlier or a solid argument to make regarding the possible meaning of the outlier, then you just ignore (show it and use the data but don't suppress!) or briefly mention the data to acknowledge its existence.