# Should mentioning "potential" ethical issues with data analysis/display be part of an answer?

On another StackExchange site (https://mathematica.stackexchange.com/questions/270097/remove-outliers-in-boxwhiskerchart?noredirect=1#comment674078_270097), I've run into resistance because of my mentioning about potential ethical issues with removing the display of "outliers" from a standard box-and-whisker plot.

Are ethical concerns considered "off topic" here?

I have not noticed any questions with potential ethical issues at this site but I certainly don't read all of the questions.

• We see comments about the inadvisability or potential unforeseen or bad consequences of proposed procedures all the time. However, comments that cross the line over to insinuations or accusations of ethical impropriety would generate the same kinds of reactions here as they did on the Mathematica site.
– whuber Mod
Jun 29 at 19:17
• @whuber It was not the removal of outliers but rather the erasing of inconvenient data points from a display created with those outliers. That "defacing" of the original graphic caused me to make the comment about data tampering.
– JimB
Jun 29 at 20:02
• The pushback appears to be not against the message (which is correct and needed to be articulated) but against the way in which it seemed to accuse the OP of intending something nefarious. One way to get your message across would be to ask it as a question rather than employing the imperative, categorical "don't do it." For instance, "Because one point of the boxplot is to highlight such 'outlying' data and their automatic removal could be misinterpreted, maybe there are other options. Could you explain what you hope this will accomplish?"
– whuber Mod
Jun 29 at 20:26
• @whuber Good point. I still have a ways to go on my diplomacy skills.
– JimB
Jun 29 at 20:42

• The question's perhaps ambiguous, but I'd say the same thing anyway. (Plotting individual outliers isn't so standard - the boxplot function in R has an argument controlling whether they're plotted or not, & I often see boxplots in which they're not, especially for large data-sets. There's certainly no sin when you're clear about what you're showing, as the comments on the post you link to also state.) Jun 29 at 20:22