The Man Boobz Survey is up! Go take it! [UPDATE: Survey closes Thursday at 8 PM, EST]

Fuck you excel, fuck you M$, and especially fuck you for making office for mac a watered down version of office!!

I HATE WINDOWS!

In related news, this hissy fit brought to you by Argenti attempting to remove the 5 surveys that have been mathematically determined trolls (my mathematical method shall be included in the final report, right now it’s too mathy to make sense, or maybe I just need sleep)

Does this make sense?

First, I calculated the average and standard deviation, and labelled anything more than two standard deviations from the mean to be an outlier. As the data is highly skewed, I then calculated the skew (using excel’s skew function, because I’m lazy like that) and multiplied it by 150%; except in the case of ethnicity, where I used 75% as the skew was 20~ and there were 27 options. Anything outside that range I highlighted as “an interesting outlier”. The reason for this difference being that the data was so skewed that in many cases there were mathematical outliers that where clearly honest answers — e.g. disabled (in)employed students — I’m looking for trolls, not interesting people 🙂

That done, I figured out how many boxes one would have to check to have checked more than 75% of the options, and labelled those as possible trolls.

So at this point we have three lists — mathematic outliers, interesting outliers, and possible trolls. I assigned a number of 1-3 to each option and figured out how many times each survey was an outlier (of any sort). Note that I did this for all the “check any” questions — all the religion ones, ethnicity, gender, sexuality, relationship status, employment, both disability questions and politics.

I then went ahead and assigned every survey that was any sort of outlier an “outlier score” by multiplying the number of outliers on the survey by the number assigned above. Those scores ranged from 1 (lots of ones) to 119. Then I determined which of the “outlier scores” were outliers using the average and standard deviation method, but excluding the 119 as it was more than twice the second highest “outlier score” and made only those two into true outliers (which excluded the person who checked literally every ethnicity category and every gender option). Having excluding that obvious outlier, and calculated the average and standard deviation on the rest of the “outlier scores”, we ended up with five surveys that where outliers among outliers — I excluded them from the results as they appear to be trolls.

While finding outliers in each set of data and then determining which surveys had enough outliers to be outliers is a bit weird, it seemed far better than guessing, or excluding everyone that was a mathematical outlier on any section — that would’ve meant dropping every disabled (un) employed student! So in the end, it looks like we had five trolls, which really isn’t surprising considering the number we have around here regularly; in fact, I was expecting more than that!

Talacaris — perhaps needless to say it, but me too. If I didn’t think you capable of it, I wouldn’t have asked you to do the smart ass section (and it’s totally fine that you can’t do it, no worries)