This page talks about the main "study" used here, the calcSD Average. It was calculated using two studies, Veale et al, 2015 and Habous et al, 2015.
Before, the 'Veale et al, 2015' study was used as the default one, but that raised some concerns since despite it being a combination of various different studies which add up to 15000 people spread across the globe, only a select few of those studies define erect measurements while most only define the flaccid sizes. Even though the study calculated a low probability of each particular data set skewing the data, it still could induce selection bias by the low amount of people involved.
The long-term solution to that is having a more reliable study made...which unfortunately might take a while due to various reasons. The short-term solution is to merge it with another reliable study, in this case, 'Habous et al, 2015' proves to be a great choice as it includes 778 people for the erect measurements from Saudi Arabia, a country not covered by Veale's study. This diversifies the data's source and reduces selection bias.
The last step is to add them up fairly. Just adding each one up directly and dividing by two (the most common solution) wouldn't be the best idea since the data from 'Veale et al, 2015' would have a greater impact on the average than normal as there's less people involved, so instead of each study being worth 50% of 100%, each is worth the percentage relative to the amount of people in it. An example is, if there's 700 people in study A and 300 in study B, A is worth 70% and B is worth 30% since A is more reliable, then just add them up and you get the average relative to the amount of people in the study. The data is as follows:
|Veale et al, 2015||Habous et al, 2015||Total Amount|
And that's everything.
Well, someone said to me that in the Veale study, there's also the "stretched flaccid length" measurements, those are in way greater number and are also relatively close (and directly correlated) to the "erect length" ones, so I went to check and indeed they are. After some small observations, I saw that that would bring the average up slightly on the Veale dataset used on this website, but would have a much bigger impact on the calcSD study. Until I went in and tested the numbers and realized that both are actually very close together. Long story short, we now have new length (and consequently volume) averages. Yay!
However: There was also quite a bit of unreliability on measuring stretched flaccid lengths and a higher risk of bias on the studies. That could be a problem using this data. I'm implementing it either way because it seems reliable enough, for now.
|Veale Dataset||calcSD Dataset|
|Veale - Erect Length||692||4.66%||692||4.43%|
|Veale - Stretched Flaccid Length||14160||95.34%||14160||90.59%|
|Habous et al, 2015||-||-||778||4.98%|