Saturday, April 18, 2009

"The only thing I was fit for was to be a writer, and this notion rested solely on my suspicion that I would never be fit for real work, and that writing didn't require any." - Russell Baker

***

Does frequency of sex depend on climate? A preliminary empirical study

Introduction

Today, a friend of mine put forward the following theory:

"i realise cold ctries are sexual
i mean the ppl

tropical ctries like singapore tend to be more sexually repressed"

Possible reasons for this are that in cold countries, people have sex because they have nothing to do during winter and/or to keep warm and that in tropical countries, it's too hot and humid to have sex.

As a logical positivist, I desired to find some empirical grounding for this, and so turned to the only source I knew of to count how often people in various countries had sex - the Durex Global Sex Survey. One could quibble a lot about the survey's methodology, but it's the best (only) data source I can find, so.

Methodology

In the last few years, Durex has changed the format of their study (meanwhile repackaging it as a "Sexual Wellbeing Global Survey". While this has seen many welcome changes (a more diverse selection of countries, with more developing countries represented, more country-specific data, less fun and salacious facts like who had used vibrators more holistic data like how stressed people were and whether "feeling close to your partner, feeling loved, respected and secure" affected the achievement of sexual satisfaction et al.), the sample size has also gone down from 41 to 26 (in the 2007/8 version). Therefore, I have decided to use a parallel dataset - the one from the 2005 edition.

Having the data on frequency of intercourse, it now fell upon me to find an appropriate way to measure climate.

At first I was going to use the minimum temperatures recorded in the countries. However, I quickly realised that a country with a brief but chilly winter could skew the results, as could one with mild weather year-round. What I needed, then, was a measurement of average temperature - how hot or cold it was, on average, throughout the year.

At first I toyed with using the temperature range or standard deviation - tropical countries would have a low number and temperate ones a higher one. It wouldn't matter if it were hot during the day/summer, since when night/winter came people could be frisky. However, I felt that this would run into some of the same problems as my earlier idea. There was also the problem of some places where it's chilly all year round, like the Arctic.

In the end, I created a variable called "Climate", taking the value 0 for Tropical, 1 for Sub-Tropical, 2 for Mediterranean, 3 for Temperate and 4 for Chilly. Although the Sub-Tropical and Mediterranean climates do not differ much temperature-wise, the former is more humid, so that adds to the "too sweaty so I don't want to have sex" factor. Some countries were weird or too big, so I made a judgment call and assigned them a value between 2 whole numbers.

To check for data consistency, I also created a variable, defined as the difference between the frequency of sex in 2005 and 2007/8 (the former subtracted from the latter).

Summary statistics:

X_YEAR_2005 (Number of episodes of sexual intercourse a year, 2005 data):
Mean: 103.1
Median: 106 (Belgium, Italy and Slovakia)
Maximum: 138 (Greece)
Minimum: 45 (Japan. Singapore is a distant second at 73)
Standard Deviation: 17.7
Skewness: -0.807
Kurtosis: 4.53
Observations: 41

X_YEAR_0708 (Number of episodes of sexual intercourse a year, 2007/8 data):
Mean: 112.5
Median: 117.5 (Spain and Germany)
Maximum: 164 (Greece)
Minimum: 48 (Japan. Singapore actually ties with the US at 5th last)
Standard Deviation: 24.4
Skewness: -0.391
Kurtosis: 3.54
Observations: 26

DIFF:
Mean: 9.14
Median: 11 (Thailand and South Africa)
Maximum: 55 (India)
Minimum: -28 (US)
Standard Deviation: 19.3
Skewness: -0.024
Kurtosis: 3.37
Observations: 22

CLIMATE:
Mean: 2.32
Median: 3
Maximum: 4
Minimum: 0
Standard Deviation: 1.23
Skewness: -0.621
Kurtosis: 2.47
Observations: 45

All variables have a negative value for skewness: most people have a lot of sex, but some wet blanket countries weigh everyone else down. Also, most people got more sex after 2.5 years. The dataset is also weighted in favour of temperate/cooler countries. Finally, there are 4 new countries in the 2007/8 dataset, but only 22 countries appear in both datasets.

Results

Regressing X_YEAR_2005 on CLIMATE, I got the following results:

Free Image Hosting at www.ImageShack.us
Climate:
Coefficient - 5.33
p-value - 0.03
Adjusted R-squared - 0.094

At a 97% significance level, we conclude that climate is a significant factor affecting the frequency of sex. For every step up my climate scale a country goes, people have sex 5.3 times more a year, on average. However, climate is not a very significant factor explaining how often people have sex, explaining just under 10% of the variation here.

I would exclude Japan as an outlier (because, as usual, they are weird) but the results are clear enough.

To confirm my results, I regressed X_YEAR_0708 on CLIMATE:

Free Image Hosting at www.ImageShack.us
Climate:
Coefficient - 0.69
p-value - 0.86
Adjusted R-squared - -0.040

At any reasonable test size, we conclude that climate is NOT a significant factor affecting the frequency of sex.

Excluding Japan helped somewhat, but essentially did not change the results.

Conclusion

While the second regression rejected the hypothesis that climate affects the frequency of sex, the small sample size (less than 2/3 that of the first one) which in fact falls below the magic number 30 makes me skeptical about the results. I would thus tentatively conclude that climate does affect the frequency of sex. As always, of course, more research is needed in this area.

My dataset is on Google Documents, so anyone wishing to do more analysis (e.g. a better measure of climate, excluding outliers) or poke holes in my assumptions is free to do so.

(Damn, I should've done this for my thesis!)
blog comments powered by Disqus