Newquay Lap Dance Claims Wrong

Via Lori Smith, I was alerted to this claim last week by police in Cornwall that a lap dance venue license application should be rejected because such clubs 'might' cause sexual violence. As Lori points out over on BitchBuzz, this is territory I've covered before: the widely-publicised claims that lap dance clubs in Camden caused more rape turned out to be false.

Of course the statistics for a specific area of London over a certain number of years are only that: specific to London and those years. It's dangerous to take a trend for one area, at one point in time, and generalise it to all places at all times. In order to claim that "Factor X causes Outcome Y" you need a lot more data. In the book I set out some comparisons, then, with London and other locations summarising what we know from the scientific literature, national statistics, and so on.

So what's interesting is that The Sex Myth discusses not only the situation in cities like London but also specifically, as coincidence would have it, Newquay.

Guess what? The link between lap dance and sexual violence that the police claim 'might' exist? Not only does it not exist, local media in the Southwest have already reported on this.

In 2010, the Newquay Voice obtained Devon and Cornwall Constabulary’s figures of sexual assaults. They found that the total number of recorded sexual assaults (including rapes) in and around Newquay peaked at 71 in 2005, the year before Newquay's first lap dance club opened. In 2006 however,  following its opening, the number fell to 51.

In 2007, when the town’s second lap dancing venue opened, the total number of recorded sexual assaults fell again to 41, then dropped to 27 in 2008 when a third lap dancing club opened. In 2009 the number rose slightly, but with a total of 33 offences, it is still less than half the total than before the clubs appeared.

Using publically available population data, I took these figures and calculated the incidence rate (since population varies from year to year as crime stats do, if you don't calculate a rate, the numbers are not very informative). Here are the incidence rate calculations using midyear population levels for the council of Restormel where Newquay is located:

Looking at these numbers, you'd be tempted to think that lap dancing actually reduces sexual assault. In other words the opposite of what the BBC article claims.

This like the Camden data is only a single example. Making such a broad conclusion would be rash – to conclusively demonstrate that an increase in lap dancing corresponds with a decrease in rape and sexual assault, there would have to many more such results, over longer time periods, from many places. What it does do is reinforce the same thing the statistics from Camden show: lap dancing does not correlate with higher occurrence of rape. And if there is no rise in rape, then it is impossible to claim that lap dancing “causes” rape.

Unfortunately, the myth that sex work causes violence has become so deeply embedded in media and criminology storytelling that one only needs to raise that dread spectre for the city council to take such claims seriously. In spite of the fact that the real data are easy to find and analyse, and the local papers in Cornwall have already suggested the opposite to what the police claim is true, the BBC and other media outlets don't seem to notice or care.

In the end it looks as if the council rejected the application. St Austell  and Cornwall MP Stephen Gilbert tweeted that this was "a victory for people power". And indeed if the rejection was made because the majority of residents decided they did not want it, then so be it. Nothing wrong with not liking things for the simple reason that you don't like them.

But consider that the information put about by police and reported by the BBC is misleading and poorly researched. What if, instead of the council's main criterion being what residents preferred, the decision was made because of police and media scaring people with potential crimes that turn out not to be true at all?  I don't know about the good folks of Cornwall, but where I come from, that's called lying.

Lap dancing and rape in Camden: part 5

Read all the posts in this
series: [1] [2] [3] [4] [5]

The causes of rape are not well understood. If they were, it would be easier to fight them, since we would know how to apply resources. The fact that rape is such a difficult, under-reported, under-investigated, and under-prosecuted problem indicates that we really don't know all that much about its causes.

Because the numbers involved are relatively small, the fluctuations in rate could be influenced by any number of things not actually to do with rape per se. After all, the number of reports could change even when the rate of rape stays the same. There could be subtle reasons why. A sympathetic and approachable officer in a particular area, for instance. Availability of crisis support and hotlines.  Changes in, or absence of, these things. It's not only hard to say - it's impossible. Far easier, and what this series of posts has sought to do, is to sieve out what does not cause rape, so as better to focus on the real job at hand.

And of course, being small numbers... sometimes a fluctuation is just a fluctuation.

Better evidence collection and better prosecution might help. But we also need to think hard about preventing rape, not just punishing it. When someone claims a cause that is not a real cause, this can derail the real struggle against violence. If the focus is on lap dancing, in spite of the fact that it has no connection with rape, it is potentially diverting resources from preventing and investigating the real causes of crime.

It’s because rape is such a serious crime that researchers must be at least as rigorous in their analysis as they would with other serious events like cancer. Otherwise, it’s not real analysis. It’s throwing numbers around without context. It’s producing reports that look and feel like real research without the methodology to back them up. It is cargo cult science.


To avoid becoming cargo cult scientists, Richard Feynman said researchers must be willing to question their results, and investigate possible flaws in a theory. Researchers should pursue a level of honesty that is rare in everyday life. "We've learned from experience that the truth will come out,” Feynman said. “Other experimenters will repeat your experiment and find out whether you were wrong or right. Nature's phenomena will agree or they'll disagree with your theory. And, although you may gain some temporary fame and excitement, you will not gain a good reputation as a scientist if you haven't tried to be very careful in this kind of work. And it's this type of integrity, this kind of care not to fool yourself, that is missing to a large extent in much of the research in Cargo Cult Science."

Both cargo cult science and real science look similar on the outside to a layperson, in that they contain numbers and try to come to some sort of conclusion. Even if the Lilith report had managed to get its calculations correct in the first place, there still would have been plenty of clues that while the look and feel of real research was being imitated, the content wasn’t up to scratch.

When looking closely, it’s pretty easy to pick out cargo cult statistics, because usually:

1. It doesn’t calculate a rate. Rates are the bread and butter of incidence statistics, and a written-in-stone requirement of any report dealing with a population group. How do I know? Because I used to write papers reporting children's cancer rates. No rate = no paper. If one year’s incidence is being compared to another, expect to see rates, not raw numbers.

2. It doesn’t show a long-term trend. In the Lilith report, a small number of years were reported. Rapes before the lap dancing clubs weren’t shown, so they couldn’t be compared. Rapes more than two years after weren’t shown, so it was impossible to see if the trend was real.

3. It doesn’t use a control group. Control groups, when it comes to population statistics like these, are hard. I get it. There's no Truman Show bubble world kept somewhere for us to compare everything to. But as we say where I come from, hard cheese. You make do. Mention was made in the report of other boroughs (such as Islington) which have lap dancing, but crimes in areas of London without lap dancing were not even mentioned so no comparison could be made. The rest of the country was not considered.

4. It makes a causal connection without direct evidence for a cause, and doesn’t consider other factors. Statisticians talk about “confounders” – the other factors that can affect your results. On the basis of a short-term miscalculated trend, a cause and effect relationship is claimed between lap dancing and rape. However, this does not take into account the types of rapes reported, any possible correlation with crime hotspots within the borough, or any other possible contributing factors. Again, I know from personal experience this shit is hard. But that's no reason not to make an effort.

The lack of statistical rigour in the report is far from the Lilith paper's only problem, however.

A pervasive feature of poor research is that it often starts from an assumed position, and any data falling outside of that position are ignored. The writers come to the study with a bias and look to find ways for the numbers to fit with their preconceived notions of what the truth should be rather than what it actually is.

We can see this on the very first page of the Lilith report with statements like:

“This ‘fast fantasy’ approach is demeaning and insulting to women…”

“Lap dance… [is] not going away without a fight.”

“[I]f Camden were to change its policy on lap dance and striptease establishments, then this good practice could spread through London.”

It’s clear from the outset that the writers of the report have a particular agenda – prohibiting adult entertainment. Which is fine, since everyone's entitled to a say in what happens in their communities.

I don't object to opinions. Think lap dancing is a sin? Great, that's fine for you. Think it's oppressing women? Great, I look forward to your paper. What gets my goat is invoking a semblance of statistical analysis. I'm a (former) statistician, yo. You're on my turf now. Everyone is entitled to an opinion and also entitled to express it. But if the writer of any scientific research were so openly biased from the beginning, there is no chance the report would be accepted by a reputable journal.

Claiming the methods of science, without buying in to the philosophy of how and why they work, is unethical. If you don’t play by the same rules, you can’t use the same tools.

The tone of the report is so attached to its assumptions that it does not address several other theoretical problems.

The report focuses on the difference in rapes between 1999 and 2002. However, in its first paragraph, the report states that lap dancing ‘arrived in Britain in 1997 with the opening of Secrets in Hammersmith’. So why pick and choose statistics starting two years later? If the opening of lap dancing clubs had an impact, wouldn’t you expect the impact to be evident reasonably soon afterwards?

Actually, you wouldn’t – not because someone has proven that the lag time between opening a strip club and increase in rape is 2 years, but because no one has conclusively proven there is any link between the two at all. So the choice of year can be completely arbitrary and it does not matter. Strip clubs are not correlated with rapes in any credible study.

In fact, the question of what effect adult services have on local crime has been studied so thoroughly that there can now be studies of the studies, or what statisticians call “meta-analysis”. A meta-analysis, or pooled analysis, combines the results of published studies by many different groups in order to arrive at an overall conclusion.

A meta-analysis examined 110 papers that claimed adult businesses increased crime rates. 73% of these were records of political discussions, not actual studies. Removing these and anecdotal reports about only one crime incident, the authors were left with 29 studies. Of the papers that did not contain flaws, there was no correlation between any adult-oriented business and any negative effect. Of the ten most frequently cited papers, not one met the minimum standards for good research – comparable controls, sufficient time, and valid data collection.

So while many people might be tempted to believe dodgy statistics because they sound like something that “should” be true, the analysis shows no demonstrable link between adult entertainment and crime. The idea that adult businesses have negative fallout for communities is a myth that should be put to bed for good.

Lap dancing and rape in Camden: part 4

Read all the posts in this
series: [1] [2] [3] [4] [5]

Another problem with the original paper is a lack of an appropriate control population, to compare the results with. Having a control population is particularly important in assessing risk. Controls are populations where the thing you want to test – strip clubs, in the case of the Lilith report – doesn’t exist, for comparison purposes.

For instance, in order to suggest a link between smoking and lung cancer, the original epidemiologists had to examine lung cancer rates not only in smokers, but also in non-smokers. You need to show that the factor being examined - smoking, or in our case, lap dance clubs - is the influencing factor.

Lack of a control group means that the numbers of rapes in Camden were not reported against the rape stats in a non-lapdancing area. It’s perfectly possible that the trends happening in Camden were happening everywhere, regardless of whether there was lap dancing or not. It’s impossible to know from the Lilith study if such other parts of London were experiencing similar trends in their crime rates.

The report makes comparisons between Camden, Westminster and Islington, all of which contain lap dancing clubs. As far as control populations go, that’s no good: you need somewhere where it doesn’t happen. Kind of like a placebo group in a medical trial.

So let’s run the statistics using Camden, one of the other areas they pick which does have strip clubs, and choose an additional one that has none at all. Because crime can be influenced by factors such as poverty, it would preferably be of a similar demographic profile. Then an assessment of the occurrence of rape in that area can be made, for comparison’s sake. Without doing this, it’s impossible to say whether any trend was locally concentrated or happening everywhere regardless of strip clubs.
Lambeth has a somewhat larger population than Camden and similar makeup in terms of ethnic origin. It contains no lap dancing clubs at all. Islington has a somewhat smaller population than the other two boroughs and has two venues licensed for fully nude lapdancing. And since these statistics are also available for the entire country, let’s throw that in too. After all the original claim was that Camden's rape stats were three times the national average.

Comprehensive statistics are available for crimes reported to police throughout England and Wales, so these are straightforward to find.

I shan't bore you with another table, though of course, those numbers are available (both from me and from the Metropolitan Police) if you're interested. It pains me to leave one out, because I love tables like a fat kid loves cake. But one woman's cakey feast is another's sugar rush nightmare, so. Let's skip straight to the graph. Again, the years covered by the Lilith paper for Camden are highlighted in red:



The graph shows that adding comparison changes the picture considerably. It no longer appears that lap-dancing clubs lead to an increase in rape, since boroughs with fewer or no clubs had consistently higher rates than Camden’s. The data from the original study is shown to be a small blip in a larger – downward – trend all over London.

If there was a relationship between the number of lap dancing clubs and the occurrence of rape, you would expect Lambeth to be lowest of the three because it has no clubs. Islington would be higher because it has a couple, and Camden highest because it has more than those other boroughs. But Camden turns out to be the lowest of the three. There does not appear to be any relationship between the number of lap dancing clubs in a borough and the risk of rape.

The trend for the three London boroughs shows clearly that Lambeth (with no lap dancing) and Islington (with only 2 clubs) both have rates that are higher than Camden’s. All three have decreased over time, as well, which is why it pays to look at the longer trend rather than cherry-picking a few years in statistics. Apart from the early 2000s peak, Camden’s numbers are close to the overall rate for England and Wales, and are sometimes even below it. This is a far cry from the “three times the national average” claimed by the Lilith report.

All things considered, you might wonder why the Lilith report chose to look at Camden at all. According to the introduction, it was because “Lilith and Eaves believe that Camden’s opinion and acts carry great weight with other London boroughs.” Which from the analytical point of view (especially considering there are no references or other reasons given) doesn't make sense.

If we were to take this graph as our only evidence, we might conclude that the risk of rape goes up not because of the presence of lap dancing clubs, but by living in London, with Camden actually safer in that regard than other boroughs. We might also be tempted to conclude that the presence of lap dancing clubs in fact indicates a safer borough in terms of rape.

Naturally, that would be a very rash conclusion, something a responsible statistician would be reluctant to suggest. It would require far more data from the rest of London and the entire nation before such an idea could be suggested. But that’s the point – in order to make a conclusion about the effects of social phenomena in general, you need a huge amount of information to back it up. One limited study of a crime statistic is not enough and should never be allowed to stand on its own.

Interestingly enough, there are other places where the opening of lap dancing clubs does seem also to correspond with a reduction in rape and assaults. One of these is Newquay, in Cornwall.

In 2010, local paper Newquay Voice obtained Devon and Cornwall Constabulary’s figures of sexual assaults. They found that the total number of recorded sexual assaults (including rapes) in and around Newquay peaked at 71 in 2005, the year before Newquay's first lap dance club opened. In 2006, the year following its opening, the number fell to 51.

In 2007, when the town’s second lap dancing venue opened, the total number of recorded sexual assaults fell again to 41, then dropped to 27 in 2008 when a third lap dancing club opened. In 2009, the number rose slightly, but with a total of 33 offences, it is still less than half the total than before the clubs appeared. Here are the incidence rate calculations (using midyear population levels for the council of Restormel, where Newquay is located):



Again, this is only a single example – to conclusively demonstrate that an increase in lap dancing corresponds with a decrease in rape and sexual assault, there would have to man more such results, over longer time periods, from many places. However it does reinforce the same thing the statistics from Camden show: that lap dancing definitely does not correlate with higher occurrence of rape. And if there is no rise in rape, then it is impossible to claim that lap dancing “causes” rape.

Rape is widely thought to be a vastly under-reported crime. The calculations don’t tell us whether rapes were under-reported for the area in any particular year, nor what might cause that.

What it does tell us is that the original claim made in the Lilith report – that the number of reported rapes is rising – is not true. It was not true in 2003, it continues not to be true, yet the myth that rapes rise 50% after lap dancing clubs opened in Camden is still reported, even as recently as August 2009.

Lap dancing and rape in Camden: part 3

Read all the posts in this
series: [1] [2] [3] [4] [5]

Even more important than correcting the errors, as outlined in the previous posts, is looking at the longer-term trends. Rapes might go up one year, or two, or three… and they might fall the next. There are natural fluctuations that can mask the overall trend. The more data we have to analyse, the more accurate the results. The more accurate the results, the more informed the reporting.

A problem common to dealing with small numbers is making a hasty generalisation. This is a fallacy that happens when someone makes a large conclusion based on a small sample of evidence, such as an initial result that disappears later, when more data are collected.

Here's an example: Let’s say you’d never been to York before, and went there for a five-minute visit while changing trains. Let’s also say that while getting off the train you saw exactly three people – all of them with red hair. It would be a hasty generalisation to then go around saying that everyone in York is ginger. And yet, given the very small number of observations, saying so (while obviously not true) would not be a contradiction of the evidence you collected.

Small numbers are a problem in statistics, because the less information we have, the less we can reliably say about it. Dealing this problem means having to collect more evidence where available, making pertinent comparisons, and applying more than just simple arithmetic. Reported rapes are relatively rare, so writing about rape statistics requires special attention.

Now, just because a crime is rare doesn’t mean it isn’t serious. Rape is extremely serious. No matter how many people are raped, it’s too many. One rape in the course of a year would be a tragedy; 72 is obviously a big problem. However, regardless of the fact that rape is a horrific crime, it’s also not very commonly documented. By comparison, the rate of breast cancer among women in the UK hovers around 120 per 100,000 per year, or more than three times higher than the rate of reported rapes in Camden in 2000.

It’s important to also find out whether the rate was a one-off, or whether the rise implied in the Lilith report was sustained. So let’s calculate rates, but this time for a longer timespan. We know that between 1999 and 2000 the rate of reported rapes in Camden rose. But did the trend continue? Have a look at the results:



The change in rates fluctuates a lot on a year-to-year basis! Surprised? Actually, that’s another feature of dealing with small numbers. Because the event is uncommon, a few incidents either way have more power to change the trend. Which is why percentage change for a couple of years, even if a lot different from what was originally reported, is not a good indication of what is really happening. (Or as I like to say, more years equals more better!)

But without the trend, the door is left open for people to misinterpret the statistics in a way that could be sensationalist and scary. As an example, let’s say there was 1 death due to vending machines falling over in Glasgow one year, and then 2 the next. Irresponsible reporting might say “Vending machine deaths double in one year!” Technically, that true - but it misses the spirit of what is really going on. It makes people think the risk of being squashed by a vending machine is going through the roof, when in fact there aren’t many at all… and there might be fewer next year.

If we graph the rates, we can see if the trend is rising, falling, or staying the same. The years covered in the Lilith study are highlighted in red:



For the ten years 1999-2008, it appears the trend for rate of reported rapes in Camden is actually falling, not rising.

Lap dancing and rape in Camden: part 2

Read all the posts in this
series: [1] [2] [3] [4] [5]

Let's look in depth at one year's change in rape statistics in Camden. In 1999, the Metropolitan police recorded 72 reports of rape. In 2000, the number was 88. The Met numbers are available to the public so can’t be disputed. And those numbers went up. This much the Lilith report got right. But is that all there is to the story?

The problem with numbers on their own is they don't say anything about context. The number may rise from year to year, but if the population is going up as well, the rate might not be changing at all.

Imagine, for instance, if a paper claimed London has 1000% more Chinese restaurants than it did 40 years ago, but didn't report the relative populations for those years. You wouldn't think much of the numbers. Of course the raw number would have gone up - the population got a lot bigger from 1970 to 2010. Without context, the numbers don't mean very much.

When the population grows, you have to take that change into account. What you need is not just the raw number of crimes reported, but also the population of the area from one year to the next. This is used to calculate not the number of crimes, but the rate. Rate and number are two different things, but many people (even those who should know better) use them interchangeably, and this creates confusion.

You don't have to be a London native (or even a Daily Mail reader) to know the population is going up. It's on the rise in Camden. But is it going up enough to make the rate of rapes look different from the number? Let's see.

Whenever numbers of incidents are reported, they should be used to calculate the rate of occurrence. This gives you an estimate of how many times the crime occurred per 100,000 population. So let’s look at those rape numbers again. For the year 1999, we have 72 rapes reported in Camden and - according to National Statistics - a population of 195,700 people.

To determine how many rapes occurred per 100,000 residents, we divide the number of rapes by the total population. Then we multiply by 100,000:

72 ÷ 195,700 × 100,000 = 36.8
This tells us that in 1999, there were 36.8 reported rapes for every 100,000 residents of Camden. Performing the same rate calculation for 2000, when the population was 202,800 and the number of rapes 88, gives us a rate of 43.4.

Mathematically calculating the change in rate from one year to the next gives us the percent change, be it a rise or a fall. The change in rate from 1999 to 2000, or the change from 36.8 to 43.4, is 17.9%.

That is considerably different from 50%. So the rate (which is what counts) of rapes in Camden did not go up by 50% after the lap-dancing clubs opened. If you include the even more modest increases in 2001 and 2002, you still come up with a result that is nowhere close to the Lilith report’s original claim. The combined change from 1999 to 2002 is a rate increase of 26.9% - in other words, about half of what was originally reported.

So not only did the media take six years to correct the error in the Lilith report, they didn’t even get it right the second time around. But the story doesn’t end there...

Lap dancing and rape in Camden: part 1

Read all the posts in this
series: [1] [2] [3] [4] [5]

The borough of Camden in north London is a vibrant and diverse quarter of the city. From the Bloomsbury of feminist hero Virginia Woolf to the leafy expanse of Hampstead Heath, it embraces a colourful past and present. In the modern iconography of London, Camden Lock is as famous for its nightlife as Kentish Town and Chalk Farm are for their music venues. At night the area comes alive, host to almost 2,000 pubs, 130 licensed entertainment venues, and seven lap dancing clubs.

To the uninitiated the Spearmint Rhino may look like any of the handful of similar establishments in the area, but it has been the epicentre of controversy since its opening. Not only was it one of the first clubs granted an all-nude license in the 1990s, it also paved the way for identical clubs in cities around London and the UK.

Spearmint Rhino was notable not only for full nudity but also for its style. It gained a reputation for having a less seedy atmosphere than previous clubs in Soho.

Comfortable leather chairs curl around customers who patronised Britain's first all-nude strip club. The topless dancers at Stringfellows were modest titillation by comparison. Spearmint Rhino’s arrival signalled a new era of adult entertainment in the capital.

Customers responded by making lap dancing the talk of London. 'Table dancing has moved into the mainstream,' wrote Ben Flanagan in the Observer. 'The clubs, previously perceived as sleazy and hostile, are now seen as ideal venues for a corporate night out or a bit of celebrity-spotting.'

So when a 2003 study of the impact of lap dancing clubs in London reported a 50% rise in rapes in surrounding areas, people were aghast. Even worse, the number of rapes was claimed to be three times the national average. As a statistic, it sounded shocking, but it also had the ring of truth to it. Lap dancing was as controversial as it was popular. News outlets all over the UK reported the results as evidence why the UK should not give in to the creeping infestation of ‘high-street’ lap dancing chains. But was this claim actually true?

The "Report on Lap Dancing and Striptease in the Borough of Camden" [pdf] was produced by Lilith R&D, part of the Eaves charity founded to support homeless and vulnerable women. The stated aim of Lilith, according to its website, is “to eliminate all aspects of violence against women”. A very worthy ideal, and an important issue. But the intentions of the authors doesn't make the relationship between their stated concern (violence) and the subject of the paper (lap dancing) any more reliable than anyone else's. And there are a large number of problems with the report simply from the statistical standpoint that disprove any such connection.

The first flaw in the report is the lack of connection between the outcome (rape) and the supposed cause (lap dancing). In well-conducted studies, you expect the researchers to show some connection between the thing being studied and the outcome being measured. Otherwise, what you have is a case of ‘correlation is not the same as causation’. In other words, just because two things happened at the same time doesn’t make them related.

The complete lack of cited research about stripping causing sex crimes is unsurprising, because no such results exist. A lot of reports have claimed the two are related, but repeated studies from many fields have all failed to connect them – I'll discuss this more later.

The next flaw is an evident unfamiliarity with calculating reliable statistics.

According to the Lilith report, rapes in Camden had been on the rise since 1999 and showed no signs of dropping. The number reported in 1999 in the borough was 72 rapes. By 2000 it was 88, 2001 had 91 and for 2002, the number of reported rapes in Camden was 96. As far as the report was concerned, the numbers spoke for themselves.

Only, there's a bit of a problem with their maths. And here’s where the evidence for the paper being more cargo cult than reliable research starts to show through.

If you look at only the numbers themselves, the difference from 1999 to 2002 is the difference from 72 to 96. That’s a difference of 24 rapes, which is only 33% increase – not the 50% originally claimed. A pretty basic error in mathematics, and one that was surprisingly resistant to being corrected. It was only six years later, in early 2009, the Guardian reported this elementary miscalculation. The original claim of 50% is still widely reported without being corrected, however.

But actually, it turns out the increase wasn’t even as high as 33%. In the next part, I'll discuss what it means to calculate rates using the changes in population to make valid comparisons.