Thunder's Place

The big penis and mens' sexual health source, increasing penis size around the world.

The (statistical) Truth About Cock Size

Originally Posted by cheeva
How do you calculate frequency in population from sd? I would like to determine the frequency in population of the sixth and seventh sd.

Let’s say you want to find the frequency of the s-th standard deviation (for example, s= 6 or s = 7). You need to find the standard normal cumulative distribution function at -s (which means integrating 1/(2*pi)^(1/2)exp(-x^2/2) from -infinity to -s with respect to x). Call the number that comes out X. Then 1 / X will be the frequency you want.

For s = 6, we get 1 in 1,013,594,647. For s = 7, we get 1 in 781,364,424,265.

Originally Posted by Invisible
Let’s say you want to find the frequency of the s-th standard deviation (for example, s= 6 or s = 7). You need to find the standard normal cumulative distribution function at -s (which means integrating 1/(2*pi)^(1/2)exp(-x^2/2) from -infinity to -s with respect to x). Call the number that comes out X. Then 1 / X will be the frequency you want.

For s = 6, we get 1 in 1,013,594,647. For s = 7, we get 1 in 781,364,424,265.

Just as a warning, the above formula works only for normal (Gaussian) distribution, which might not be valid for whatever quantity (BPEL?) you are modeling. Normal distribution has very “narrow tails”, that is, very few samples have values several deviations from the mean. Another example of what can go wrong with the normality assumption is the symmetricity: For example, let’s say we have a quantity with average value 6 and std of 1. Then approximately one sample in billion (US, 10^9) will have a negative value of the quantity. Also, as many samples have negative value as there are samples above the value of 12. 6 is probably too much for average BPEL, but I hope you get my idea :)

Originally Posted by docrob
Just as a warning, the above formula works only for normal (Gaussian) distribution, which might not be valid for whatever quantity (BPEL?) you are modeling. Normal distribution has very “narrow tails”, that is, very few samples have values several deviations from the mean. Another example of what can go wrong with the normality assumption is the symmetricity: For example, let’s say we have a quantity with average value 6 and std of 1. Then approximately one sample in billion (US, 10^9) will have a negative value of the quantity. Also, as many samples have negative value as there are samples above the value of 12. 6 is probably too much for average BPEL, but I hope you get my idea :)

:rolleyes:

This entire thread is based on a normal distribution assumption. It’s probably a reasonable approximation unless you want a higher level of precision or wish to discuss extreme values, in which case paying closer attention to the skew and kurtosis issues you highlighted would be warranted. However, this is all in the spirit of fun so why bother?

Originally Posted by Invisible
:rolleyes:

This entire thread is based on a normal distribution assumption. It’s probably a reasonable approximation unless you want a higher level of precision or wish to discuss extreme values, in which case paying closer attention to the skew and kurtosis issues you highlighted would be warranted. However, this is all in the spirit of fun so why bother?

Exactly.

When I made this thread, I was most interested in demonstrating that 95% of men’s penises fell within the length range of 4.25 inches to 7.5 inches. I was trying to inject a bit of antidote to the oft occurring misconception regarding what constitutes an average-sized penis, not determine anyone’s standard deviation to the third decimal.

But, I do enjoy the act of quantifying (99.975%ile), so I understand why some here are taking this so far.

Cheers!
Pri

Seems to me we have enough people to come up with a fairly effective average ourselves. Even if we limited to Senior Members (to attempt to get honest measurements).

Of course, I think it would have to be pre-PE size. Has this been suggested before?

Seems to me we have enough people to come up with a fairly effective average ourselves. Even if we limited to Senior Members (to attempt to get honest measurements).

Of course, I think it would have to be pre-PE size. Has this been suggested before?

I thought it was supposed to be your pre-PE measurement.

Having girl problems?

I thought it was supposed to be your pre-PE measurement.

Define “it”.

Because I’m not understanding what you’re saying. My post was that I think it would be useful to have a survey of pre-PE size. Having said that, your comment doesn’t make sense to me.

*sorry, edited all to hell*

Last edited by Blackhatbrigade : 10-01-2008 at .

Define “it”.

Because I’m not understanding what you’re saying. My post was that I think it would be useful to have a survey of pre-PE size. Having said that, your comment doesn’t make sense to me.

*sorry, edited all to hell*

Having girl problems?

At the risk of sounding stupid, where’s that survey?

At the risk of sounding stupid, where’s that survey?

It’s the height and size survey.

Having girl problems?

Also, how did we come to the conclusion of:

Originally Posted by Priapologist
8.352 inches: 1 in 741
9.177 inches: 1 in 31, 574
10.00 inches: 1 in 3.5 million

From a group of 300 people? `:-)

Or is that what the talk of distribution is? That stuff kinda lost me.

Its another word for natural average, as humans we fall into a natural pattern in all our statistics including this one so once you acknowledge that you can talk about how likely is this girth, etc

Its also how they suppose those figures, you can see how the extremes become increasingly unlikely. This is because it does not fall into the natural average we know it should

Also, how did we come to the conclusion of:
Quote
Originally Posted by Priapologist
8.352 inches: 1 in 741
9.177 inches: 1 in 31, 574
10.00 inches: 1 in 3.5 million

From a group of 300 people? `:-)

Or is that what the talk of distribution is? That stuff kinda lost me.

The 300 men in the Ansell LifeStyles condom survey act as a representative sample of the whole male population. The sample size is large enough to do such basic statistics on and those data are backed up by every other clinician-measured survey that I have seen. In fact, the Ansell’s study is actually on the high side of the reported numbers, so my statistical analysis is probably over estimating the average penis size and distribution.

The basic idea behind this process: say that I have a jar containing 10,000 marbles of various colors and I want to know more or less how many of each color I have. I can either count all 10,000, which would be tedious, or I can pull out 50 marbles at random and note the colors obtained and their frequency. This is a sample (50) of a population (10,000). So Ansell pseudo-randomly* noted the erect lengths and girths (analogous to marble color), and how often those data occurred (frequency), of 300 American university students on holiday in Cancun.

Penis sizes, like most natural phenomena, adopt a normal or near-normal distribution, wherein 68% of the sample (and hence the population from which the sample is drawn) will fall within one standard deviation from the mean, or average, value. Likewise, 96% of the sample will fall within two standard deviations, and 99.6% will fall within three standard deviations (SD). Hence, 8.352 inches is three SD from the mean, 9.177 inches is four SD, et cetera. A z-score calculator determines the odds, 1 in x.

That is how 300 drunken frat boys informs the rest of us how long the average penis is.

*I said pseudo-random here because this is a biased sample, both from the design criteria and due to self-selection bias, i.e. guys with bigger penises were probably more willing to get measured than guys with shorter/thinner penises… which further suggests that this analysis is on the high side of the ‘true’ averages.

Pri

Originally Posted by marinera