Patrice Salomé of Plant Editors explains what it all means!.

You are finally writing up that major study; the methods section is coming together, and you have assembled your figures, to which you need to add captions, including the level of replication in all quantitative assays.

If you are like most authors, you will be tempted to write “Data are shown as means ± standard deviation (n = 3).” But what does this mean? How are your readers (and reviewers!) meant to understand this statement?

Let’s deconstruct why the above line is problematic in figure captions:

• You must have performed multiple independent replicates of this experiment (the *biological* replicates, although the term ‘independent’ replicates is more meaningful and will be used here); each independent replicate should consist of multiple technical replicates, comprising seedlings grown and collected at the same time. 
• Is the mean for the technical replicates of one independent replicate, or the mean of the means of each independent replicate?
• Does “n = 3” refer to the number of technical replicates in one independent replicate, or the number of independent replicates?

Surely you did not perform the experiment only once with three technical replicates, but simply adding “data are shown as means ± standard deviation (n = 3)” could be understood as meaning just that. Now, it is possible that you collected samples from trees growing in an orchard, and in this case, you may very well only have collected samples once. But consider the associated risk: what was the weather like that day? Was it very hot, or did a cloud roll by in front of the Sun? How generalizable are the results you obtained then? Might they instead be reflective of this one point in time?

At the same time, you should not write “n = 9” if you performed three independent experiments, each consisting of three technical replicates. This statement would artificially inflate your level of replication.

So what can you do? Let’s assume your datasets comprise three independent experiments, each with three technical replicates. 

• You can plot one independent experiment as a bar, with the technical replicates shown as separate symbols. You should state this fact clearly in the figure caption and add the line “We obtained similar results for all independent experiments.” The error bar represents the standard deviation or standard error of the three technical replicates. However, many reviewers do not like this approach, as it minimizes error bars.
• You can plot the results from all three independent experiments in one plot, with each independent experiment as one bar (and their respective technical replicates as symbols). As above, the error bar represents the standard deviation or standard error of the three technical replicates. This option is not practical, however, when you have many treatments or genotypes to plot.
• You can plot one bar showing the mean of the independent experiments. The error bar represents the standard deviation or standard error of the means of each of the three independent experiments. The means of each set of technical replicates can be shown as separate symbols. In this case, the error bar will be larger, but that’s exactly the point of replication! If your data vary quite a bit across independent experiments, you may consider normalizing each set of values to the control sample, or to the first time point. 
• You should consider compiling all data into an Excel spreadsheet and clearly distinguishing independent and technical replicates. More journals are asking for a summary of statistical analysis, so this spreadsheet is a great starting point!

In all cases, you should clearly disclose what you did and how you assessed the statistical significance of differences. You should add a description of replicates in each figure caption and in the Statistical analysis section of methods. For example, you may consider saying “Three independent experiments were performed, each consisting of three technical replicates. Data are shown as the means ± SD of the means from three independent experiments (n = 3).” Bonus points if you add individual data points and describe what these circles mean in the figure caption

Statistical analysis done and ready to publish? Contact us at planteditors.com/sales for scientific editing and image analysis!