The significance of Thunder's Place PE Data stats

Good day fellow penis statistics nerds.

One of my favorite hobbies is to crank out spreadsheets of penis data while I have a traction device attached to myself. As I’ve mentioned before, it was a very happy day when I first found Size’s db on Thunder’s Place. My life has never quite been the same.

This topic has been touched upon from what I can remember reading in the forums, but I don’t think anyone has really taken it into the depth that I think it deserves. If I have missed the thread, please send a link.

I have been away from the forums for awhile and missed Priapologist’s great thread on the Lifestyles penis size study. It was a good analysis and I enjoyed it a lot. Invisible has also been cranking out some fun data posts in his progress thread.

A lot of attention has been paid to the Lifestyles study on this site and elsewhere. I think it is informative and valuable because of the strategies employed by the Ansell company to collect the data. I also find it interesting that we have a study that is 10x the size of the Lifestyles study, but many people here seem to think that it is inferior quality. Just like the venerable Kinsey study, it is self-reported. It certainly would have a higher incidence of sabotage than the Lifestyles study, but is it enough to invalidate the data? Maybe it is my optimistic idealism (and I’m really much more of a cynic), but I think that there are several factors that make this data truly valuable in its own right, and perhaps a little undervalued in our community:

1) Big population sample. More than 8600 entries from over 3700 men. You can drown bad data with good in that kind of sample size.

2) Self-reported. This allows the contributor to work in private (and while sober) to produce his best result, and try as many times as he wants until he gets the best result.

3) Anonymous. “On the internet, nobody knows you’re a dog” … or a guy with a small penis.

4) Self-selected. Contributors were not publicly coerced or encouraged (prizes) to contribute, and most importantly: didn’t opt-out.

5) Good instructions. The site has pretty specific guidelines for measurements.

6) A mission. Sharing the data can improve public awareness of true penis sizes and may theoretically improve our chances of growing - lying is counter-productive.

Now, you can turn all of those points on their head and say:

1) Big population sample. Someone can register with a thousand different email addresses and maliciously flood the db with bad data.

2) Self-reported. Data not acquired by an unbiased 3rd party is always suspect. Motive and error are both factors.

3) Anonymous. There’s no accountability for contributing crap data when you’re anonymous.

4) Self-selected. A small penis is generally seen as undesirable in modern society and people still have an affinity for their online personas, even if they are completely separate from their meatspace embodiment.

5) Good instructions. People can’t even read the forum posting guidelines, much less how to measure properly.

6) A mission. Everyone has an agenda, but that doesn’t mean it is aligned with anyone else’s. Some people have intentionally entered bad data.

My perspective remains: In addition to the my feeling that most of the positive points I mentioned outweigh the corresponding negatives, I think most of the entries are made in good faith and under good conditions - possibly making the TP PE Data superior to any other penis size study available.

I expect to hear some reasonable discussion on these points and more.

Before looking at the data and specifics of the studies, I also want to mention that there isn’t a lot of detail to be seen from the Lifestyles study. The analysis that Priapologist did could only really focus on the extremes of the studies, which have the smallest amount of data and highlight their limited value when compared to the fidelity we have in Thunder’s PE data, which I’ll explore later on.

Let’s compare the representation that Priapologist provided for the Lifestyles data to the PE Data from Thunder’s Place. I have some python scripts (a PEist’s favorite programming language) that slice the data several different ways and put it into excel spreadsheets for further manipulation. I’ll use the starting measurements from Thunder’s data and pontificate later about why I also like to look at the ending data.

The first comparison is a look at the mean and standard deviation for length:

————————- Lifestyles - TP PE Data - difference
————— std dev: — 0.825 — 0.919 — 0.094

5th sd below mean: — 1.752 — 1.788 — 0.036
4th sd below mean: — 2.577 — 2.707 — 0.130
3rd sd below mean: — 3.402 — 3.627 — 0.225
2nd sd below mean: — 4.227 — 4.546 — 0.319
1st sd below mean: — 5.052 — 5.465 — 0.413
—————— Mean: — 5.877 — 6.384 — 0.507
1st sd above mean: — 6.702 — 7.304 — 0.602
2nd sd above mean: — 7.527 — 8.223 — 0.696
3rd sd above mean: — 8.352 — 9.142 — 0.790
4th sd above mean: — 9.177 - 10.062 — 0.885
5th sd above mean: - 10.002 - 10.981 — 0.979

First, the TP PE Data mean is half an inch larger than the Lifestyle’s data. That is within a standard deviation of either study, but I agree that this is substantial. Second, the standard deviations are within 1/10th of one another. This is pretty remarkable and does provide some credibility that the distributions are fairly well aligned. If the distributions are close, but the means are different, what can that say?

I think there may be many circumstantial reasons for this. I’m primarily thinking it has to do with erection quality (EQ) and fat pad.

Let’s use some examples. For those of you who have tried to take a “proof” picture of your fully erect penis, you know how hard it can be to get a good measurement while somewhat distracted, even though you’re in complete privacy. I can imagine that getting a little boozed up at the bar and being measured by two nurses under the observation of a doctor would not yield a good look at my true potential. You might argue to the contrary that two good looking nurses (if you want to make that assumption) handling your gear might help things a bit, but I’ve heard that isn’t always the case, and it’s likely that the doctor’s official presence (hot or not) probably balanced things out.

I also wonder if any nurse is going to be cramming a ruler into someone’s pelvis (who probably isn’t used to measuring himself all the time) nearly as hard as some or most of us do. It’s also easy to not get an optimal press into the fat pad if you misjudge the placement of the ruler. This is one aspect where it is possible to get much more out of a self-measurement. Even more so with a little practice.

When you’re at home, measuring yourself in private, you get an opportunity to measure as many times as you like until you are confident you’ve measured your best wood. You are probably in a more comfortable setting at home and the embarrassment factor is zero. The temperature is under your control and there’s probably low distractions from noise. I don’t think there is a good way to argue that the circumstances are less optimal at home than in a tent, outside of a bar, in Mexico, during spring break. So, all factors being held equal, the TP measurements should be more optimal and perhaps be partially responsible for a higher mean.

I think there is a case to be made for the smaller mean outside of circumstantial measurement factors and more about demographics. This may be a sensitive issue, but I’ll throw them out there anyway. Half of these are questions as much as assumptions.

Who is going to be the most well represented demographic at a bar in Mexico during spring break? I would guess upper income, educated (or educating) white Americans. Who is not as well represented in this same scenario? Lower income, uneducated American minorities. Other nationalities of all income levels? Gays? (maybe Miami Beach is a much more popular destination?) Athletes? (maybe not a factor, but I was a college athlete and sequestered from many activities like this). Anti-social misfits into penis stretching. (okay, that was a joke)It should be mentioned that Thunder’s PE Data mean length is closer to the Kinsey study when accounting for (speculating/assuming) BP vs. NBP corrections and identical (when rounded to the same significant decimal) as the mean reported in “The Definitive Penis Size Survey”.

There are many reasons why the TP PE Data mean would be larger, but don’t let that stop you from speculating…

Now, let’s look at the mean and standard deviation for circumference:

Circumference at each standard deviation:
————————— Lifestyles - TP PE Data - difference
—————-std dev: — 0.508 — 0.660 — 0.152

5th sd below mean: — 2.432 — 1.725 - -0.707
4th sd below mean: — 2.940 — 2.385 - -0.555
3rd sd below mean: — 3.448 — 3.045 - -0.403
2nd sd below mean: — 3.956 — 3.705 - -0.251
1st sd below mean: — 4.464 — 4.365 - -0.099
——————-Mean: — 4.972 — 5.025 — 0.053
1st sd above mean: — 5.480 — 5.685 — 0.205
2nd sd above mean: — 5.988 — 6.345 — 0.357
3rd sd above mean: — 6.496 — 7.005 — 0.509
4th sd above mean: — 7.004 — 7.665 — 0.661
5th sd above mean: — 7.512 — 8.325 — 0.813

The mean girth is 0.05” more for the TP PE Data, which I call equal to the Lifestyles study for all intents and purposes. If there is any question, I think the previous arguments about EQ in the length differences would easily account for this. Depending on the measuring tape used, that might even be within the margin of error of the study. Finally, they reported using mid-shaft girth just like we recommend, but I’m skeptical of how accurate anyone is about finding and taking that measurement.

The interesting story here is that the distribution of the TP PE Data is much wider. Again, I’m highly skeptical of the likely homogeneous demographics of the survey and the resultant smaller standard deviation. I think a broader demographic would naturally yield a larger bin.

Very nice work Slipstream :)

You’re an engineer, aren’t you! ;-)

Moving on to the frequency in population statistics:

As previously discussed, Priapologist used the standard normal cumulative distribution function to estimate the frequency of the bins in the Lifestyles data. He disclaimed it and lacking the actual data, assuming a perfectly normal distribution is really the only way to speculate about this aspect of the data. It made for an enjoyable and lively thread, but I think Thunder’s Data illustrates why it is a pretty feeble concept beyond a couple standard deviations. I mean that with all due respect and I hope nobody takes that the wrong way. It is perfectly reasonable to do with no other data to go on.

To be clear, you get the EXACT same frequencies applied to the bins in Thunder’s PE data when using this function. To illustrate the statistical differences, I’ve calculated the frequency distribution of length for Thunder’s PE data and contrasted it with the Lifestyles estimate.

Length frequency:

————————-TP PE Data —- count - (TP) freq - SNCD est - Lifestyles

5th sd below mean: — 1.788 ——— 6 —— 621 — 3,488,556 —- 1.752

4th sd below mean: — 2.707 ——— 3 —- 1243 ——- 31,574 —- 2.577

3rd sd below mean: — 3.627 ——— 2 —- 1864 ———— 741 —- 3.402

2nd sd below mean: — 4.546 ——- 56 ——- 67 ————- 44 —- 4.227

1st sd below mean: — 5.465 —— 350 ——- 11 ————— 6 —- 5.052

——————-Mean: — 6.384 —- 2797 —— 1.3 ————— 2 —- 5.877

1st sd above mean: — 7.304 —— 418 ——— 9 ————— 6 —- 6.702

2nd sd above mean: — 8.223 ——- 82 ——- 46 ————- 44 —- 7.527

3rd sd above mean: — 9.142 ——- 12 —— 311 ———— 741 —- 8.352

4th sd above mean: — 10.062 ——- 0 ———- - ——- 31,574 —- 9.177

5th sd above mean: — 10.981 ——- 2 —— 1864 — 3,488,556 — 10.002

My first observation is that the frequency around the mean is substantially higher for the TP data, I’m sure partly/mostly due to the larger standard deviation. After that, I feel like there is, once again, a pretty decent correlation in the frequency through the 2nd std dev, and even the 3rd above isn’t an order of magnitude different.

However, after that, the sample size of the TP data proves too small to provide a realistic frequency into/beyond the 3rd standard deviation. To me, that means the Lifestyles data, at 1/10th the size of TP’s data, would provide an even lower confidence and more sparse distribution at the ends of the spectrum.

The girth frequencies have a higher frequency at the extremes.

Girth frequency:

————————-TP PE Data —- count - TP freq - SNCD est - Lifestyles

5th sd below mean: — 1.725 —— 15 — 249 — 3,488,556 — 2.432

4th sd below mean: — 2.385 —— 10 — 373 ——- 31,574 — 2.940

3rd sd below mean: — 3.045 —— 24 — 155 ———— 741 — 3.448

2nd sd below mean: — 3.705 —— 12 — 311 ————- 44 — 3.956

1st sd below mean: — 4.365 —- 301 —- 12 ————— 6 — 4.464

——————-Mean: — 5.025 — 2880 — 1.3 ————— 2 — 4.972

1st sd above mean: — 5.685 —- 419 —— 9 ————— 6 — 5.480

2nd sd above mean: — 6.345 —— 54 —- 69 ————- 44 — 5.988

3rd sd above mean: — 7.005 ——- 9 — 414 ———— 741 — 6.496

4th sd above mean: — 7.665 ——- 3 - 1243 ——- 31,574 — 7.004

5th sd above mean: — 8.325 ——- 1 —3728 — 3,488,556 — 7.512

The patterns for length should probably tell us that these extremes not a good representation of the true frequency in the population, but they seem to say that there are many more smaller and larger men out there than a normal distribution would portray. According to the Lifestyles study, the girth of “about 75 percent of men were between 4.5 and 5.5 inches” (equivalent to their middle three bins) and “17 percent of erections measured under 4.5 inches”. This means that 8% had girth above 5.5 inches, and would validate the trend in the TP data which shows more men in the 2nd and 3rd std dev above the mean than the same std deviations below.

If you truly believe that this data should be a beautiful bell curve, you may also speculate that extremely large men may have a tendency toward exhibitionism and congregate where they can be admired. Similarly, extremely small men may be so unhappy that they are more willing to seek out help, yielding a frequency that is out of proportion with the actual population.

The rank in population of the Lifestyles data is based on the frequency estimation and shares the many of the same criticisms.

Length rank in population by percentile:

————————- Lifestyles ——— rank —- TP PE Data - rank

5th sd below mean: — 1.752 —- 0.0000003% — 1.788 —- 0.16%

4th sd below mean: — 2.577 —— 0.000032% — 2.707 —- 0.24%

3rd sd below mean: — 3.402 ———— 0.14% — 3.627 —- 0.30%

2nd sd below mean: — 4.227 ————- 2.3% — 4.546 —- 1.80%

1st sd below mean: — 5.052 ———— 15.9% — 5.465 — 11.19%

——————-Mean: — 5.877 ————— 50% — 6.384 — 51.48%

1st sd above mean: — 6.702 ———— 84.1% — 7.304 — 86.21%

2nd sd above mean: — 7.527 ———— 97.7% — 8.223 — 97.42%

3rd sd above mean: — 8.352 ———- 99.86% — 9.142 — 99.62%

4th sd above mean: — 9.177 ——- 99.9968% - 10.062 — 99.95%

5th sd above mean: - 10.002 —— 99.99997% - 10.981 — 99.95%

Circumference rank in population by percentile:

————————— Lifestyles ——- rank —- TP PE Data - rank

5th sd below mean: — 2.432 —- 0.0000003% — 1.725 —- 0.40%

4th sd below mean: — 2.940 —— 0.000032% — 2.385 —- 0.67%

3rd sd below mean: — 3.448 ———— 0.14% — 3.045 —- 1.31%

2nd sd below mean: — 3.956 ————- 2.3% — 3.705 —- 1.64%

1st sd below mean: — 4.464 ———— 15.9% — 4.365 —- 9.71%

——————-Mean: — 4.972 ————— 50% — 5.025 — 58.13%

1st sd above mean: — 5.480 ———— 84.1% — 5.685 — 86.96%

2nd sd above mean: — 5.988 ———— 97.7% — 6.345 — 98.20%

3rd sd above mean: — 6.496 ———- 99.86% — 7.005 — 99.65%

4th sd above mean: — 7.004 ——- 99.9968% — 7.665 — 99.89%

5th sd above mean: — 7.512 —— 99.99997% — 8.325 — 99.97%

Again, there is a very similar distribution through the 3rd standard deviations, but the fringes don’t seem so extreme in Thunder’s data.

As I said in the opening post of the thread, there is a lot of detail in Thunder’s PE data and I’m able to find a significant value in that detail. So far, this might have seemed like an attack on the Lifestyles study or its analysis, but it isn’t. I think it is a great study and Priapologist a man after my own heart in his examination. The point I’d like to make is that Thunder’s PE Data is every bit as substantial and more, for many purposes. I’d like this data set to get its fair shake in the world of penis studies and make sure our members don’t dismiss it or take it for granted.

I’ll post more about some of the insight I’ve gotten from Thunder’s PE Data, soon.

Excellent work!!! Thanks for sharing!

BHB and Iguana - Thanks!

Thank you, Lampwick for putting my data “tables” in fixed-width format. Cheers!

Slipstream,

I show at least some entries in the PE Data that look bogus. Are you seeing the same thing?

Well, just because some guys have a 18” dick and you are envious, right?

Seriously, couldn’t you cut out some very improbable measurements, correcting data with common sense?

I have no clue if Slipsteam’s model shows any of it or not.

Mine certainly is, you can go download the .zip and look at the .html in it. There’s some pretty big and low numbers.

I’ve actually added configurable filters to pull some of those bastards out. I can exclude them if their growth is beyond (above or below) a certain threshold for the duration of their entries, and a few other easy ones. There are still some others that I know are bogus that I pull out manually. It’s easy to do when you can sort all the columns. The other thing is that some of the dates are inconsistently formatted, so I have to fix them, but it was easy to do once I found the exception and saw it was formatted either one way or the other.

I can also turn off the filters to get the total data. I try to minimize my manual manipulations too much, but some are too egregious not to strike.

I noticed the date formatting too. Haven’t done anything about it yet though.

How do you choose which ones to pull out? I mean, 1” EL is possible. I think maybe 1” EG may be a bit far fetched… but taking those out causes some issues with the reliability of the data doesn’t it?

I mean, I’m not even sure some of the larger measurements should be taken out. Maybe just use the data from people with more than one entry? That might get rid of most of the bogus data I would think. I don’t know, I’m looking for sane ways to do this too :)