It’s been a while since I’ve got my teeth into a close reading of a paper, and this week has gifted me with a doozy: Pornography actresses: An assessment of the damaged goods hypothesis. The study was authored by psychology academics and former porn performer turned founder of a healthcare programme for porn performers.
The paper aimed to test the veracity of a set of beliefs surrounding women in porn. These attitudes were gleaned from a studies into attitudes towards porngraphy, finding that those with a negative attitude towards porn tended to believe that porn performers had low self-esteem, were drug addicts and had experienced sexual abuse in childhood. These attitudes, the authors point out, are also apparent in anti-porn feminist writing, which is backed up with little evidence. The authors also point out the distinct lack of quantitative research into the women in porn themselves, drawing attention to the fact that while there’s a couple of qualitative studies about why women get into acting in porn, there’s nothing quantitative.
So they decided to examine quality of life, self-esteem, attitudes towards sex, sexual behaviour and drug use in a sample of porn actresses. The headline findings were rather interesting: it turns out that the stereotypes aren’t true. Comparing porn actresses to a sample of women matched by age, marital status and ethnicity, they found that the porn actresses actually had higher self-esteem than comparable women, were more likely to feel positive, felt they had better social support and were more spiritual. There was no difference in current drug use, apart from marijuana (porn actresses were more likely to get high), although the porn actresses reported more drug use in the past. There was also no difference in incidence of sexual abuse in childhood. And finally, the porn actresses reported greater levels of sexual satisfaction, were more likely to identify as bisexual, enjoyed sex more, were having more sex than the women who weren’t in porn (sex as part of their work was not counted: this was entirely extracurricular sex), were more likely to be concerned about catching an STI, and had started having sex a little earlier.
Does this mean that the stereotypes about women in porn coming from some feminists and the general population can finally be put to bed? I’ll get back to that after we’ve had a little look through a few criticisms of the paper.
The sample
The study used a clever sampling method for accessing porn actresses–a task which is usually rather difficult and goes some way to explaining why there is little, if any, quantitative examination of porn performers’ lives. The porn industry requires that performers have regular STI tests, particularly for HIV, so participants were recruited from a clinic where they were tested. The comparison group were recruited from a college and an airport (annoyingly, it’s not specified whether this airport and university were also in California). While it is not ideal that the porn actresses were all recruited from Los Angeles, which might not be representative and generalisable to the entire population of porn performers, it is not as bad as one might think: the authors were testing whether stereotypes about women in porn were true. The majority of porn distributed in the west comes from southern California, which shapes discussion and thought about porn by western people as about this group of porn actresses. They’re not entirely representative of everyone in porn, but they’re certainly the people about whom the stereotypes are formed, and therefore this is a reasonable sample to draw from.
Since the authors were also concerned about stereotypes about women in porn, it’s also not a problem that men were not included in the study. The “damaged goods” stereotype that was being examined exists only about women!
One commenter on Jezebel (yes, I looked at a Jez discussion thread. Yes, I’m traumatised. No, I don’t ever want to go back to Jez ever again) points out that the sample of porn actresses may differ from average in being slightly older and having worked in the industry for longer. However, this isn’t actually backed up by a link to where she got this information from, and her comment is preceded by “I believe”. This might be true, but I haven’t been able to find this information anywhere, so can’t comment on whether this is a problem for sampling. However, it is important to note that the women included in this study were those who were participating in above-board porn which was compliant with the regulations, and there might be differences in women who are working in grey or black market porn. Unfortunately, these women are even harder to access and study.
Ultimately, this was an impressively large sample for such a difficult-to-access group: data from 177 porn actresses were collected (and, of course, 177 women in the comparison group). Of course, in any quantitative study, no sample is going to be completely representative, but as far as things go, this was reasonably strong.
The design
This study used a matched pair design: data collected from each porn actress was matched with data collected from a woman the same age, ethnicity and marital status. This is a fairly robust design when comparing groups, and means that differences cannot be attributed to these variables. I am even more impressed at the sample size with the researchers using this type of design, as it’s notoriously difficult to collect data for these designs, being massively time- and resource-intensive.
I have beef with the matching criteria, though. While the authors were right in selecting these, particularly as their sample of porn actresses were far more likely to be single than the general population, there’s an important thing missing that wasn’t measured at all and probably should have been controlled for. Socioeconomic status–class–was never measured, so we don’t know at all whether the porn actresses were better-off or worse-off than the comparison group, and if so, whether it was this that was the cause of their general feeling a bit better about life. Perhaps a way of establishing a better comparison group would be to compare porn actresses with TV or film actresses of the same age, ethnicity and marital status. This would likely control for a lot of the noise, although it would be an absolute arse to research.
Rather irritatingly, the authors never mentioned if they asked the women in the comparison group if they had ever worked in porn (or, indeed, if they were currently porn actresses who happened to be at college or at an airport that day). Since the likelihood of them being porn actresses is fairly low, this probably doesn’t pose much of a problem, I’m just a pedant.
The statistics
Feel free to skip this bit, as it will get a little technical, and is mostly minor statistical nitpicking. The biggest statistical elephant in the room is that this study ran a lot of statistical tests. A metric fuckton, to use the accepted statistical term. The researchers conducted an awful lot of T-tests, which is a statistical test used to check if one thing significantly differs from another: in this case, whether porn actresses differed significantly from women who weren’t porn actresses on number of sexual partners, or alcohol use, or any of the other eleventy bazillion variables which were being measured.
When one conducts a metric fuckton of statistical tests, one increases the likelihood of encountering a Type I error: a “false positive”. Purely by chance, one of the tests came up as significant, when in fact there isn’t really a difference there. This can be controlled for, although the authors didn’t. Luckily, there was enough data present for me to do this task for them. I did a Bonferroni correction, where the threshold for significance is revised based on how many tests are being performed. It’s pretty easy to do. You take the generally-accepted significance threshold, which is p=0.05 (or, a 5% probability that the results are entirely down to chance and you’re seeing an effect that isn’t really there), and divide it by the number of tests performed (in this study, 19 t-tests were performed). So, the significance threshold for the tests should actually be p=0.0026.
All of the p-values reported came up as less than 0.001, which means they’re still significant even with the Bonferroni correction, with the exception of sexual satisfaction, positive feelings and social support. However, enjoyment of sex was still significant, so it looks like our porn actress sample were enjoying sex significantly more than the non-porn actress sample anyway.
What wasn’t examined (and I wish it had been)
I’ve already mentioned how I wished the authors had controlled for class, but there’s a few more things I’d love to have seen addressed in this paper. Firstly, how the variables related to each other. Given that the porn actresses had had a lot more sex than those who weren’t in porn, could this be the reason they seem generally happier and with higher self-esteem? I have no idea, because the authors didn’t check this, and it would certainly be interesting to find out if this was the driving factor, or even just mediated the relationship.
The other thing missing, I feel, was the type of porn the women were performing in, and how that related to the variables. Were the participants who identified as bisexual more likely to be appearing in lesbian or bisexual porn? Do certain types of porn affect the self-esteem of the performers? Again, no fucking idea, I wish it had been measured, and I seriously hope future research addresses such questions.
Feel free to add more interesting questions you’d like to see addressed in the comments!
So what does it all mean?
Ultimately, from a single study, we can never conclude anything concrete, but it is a good thing to see these questions being addressed systematically, and I hope that it leads to future research. Too often, the experiences of those involved in porn or sex work are ignored, and it is genuinely refreshing to see research attempting to examine their experiences and feelings. This study provides a foundation for further examination and to build upon its flaws so we can better understand what it’s like for women in porn and replace the stereotypes with solid evidence.
Another angry woman 
There’s something nagging me about how they were matched and what they did and can conclude from that. As we know the 2 groups should be as similar as possible except for the IV – in this case being a porn actress. But being a porn actress has more characteristics than just doing sex for entertainment. Porn actresses are generally very attractive by societal standards, plus they meet a lot of men, and they meet a lot of men who like having a lot of sex with women. Perhaps especially ones who look like them. So – shock – they have high self esteem (based on looks?) and they have a lot of sex and enjoy it. Perhaps they should have been matched on self-rating of attractiveness and perhaps number of sexual partners and satusfaction with sex life should be control variables not DVs. Then we might see some more subtle effects being pulled out.
These are all very good points and ultimately why I feel that film and TV actresses would probably be the best comparison group. Most of this would likely therefore be controlled for, particularly as measuring attractiveness is a notoriously difficult process!
Hiya! Thank you for a reasoned and well-thought criticism of the paper, something which has been almost entirely missing from coverage on it.
Just one thing re: matching studies, imo you don’t ask if the matched samples have worked in porn, because they represent the general public which does at some level contain ex-porn stars. If the matches purposely exclude them, that’s not ‘general public,’ that’s another not-control group. As in cancer studies, where you don’t ask the matched controls if they have had cancer, for example.
Totally agree with you that this needed socioeconomic data as well.
Cheers for this! I’ve never actually used a matched pairs design so I wasn’t sure how much it’s necessary to control for!
Really lovely stuff. Many thanks for doing this. One point that I can’t recalled being mentioned is parenthood. I thought a number of studies speak of satisfactions gained from family life.
It seems to me that if the sample consisted primarily of women without children/family life, then matching them with the comparison group might skew the comparison group towards older women without children who might be more inclined to dissatisfaction with their circumstances.
Just a thought…
This is quite an interesting point, and something which wasn’t measured at all by the researchers.
Though we can all be armchair/twitter critics. I think it’s important to recognise, as you do, that the most important fact is this research is being performed, and responsibly.
Yes, I would say that controlling for ‘attractiveness’ would be essential in measuring whether self-esteem was linked to being attractive rather than. And there’s also the possibility that they could guess from the questions what was being studied and wanted to represent the porn industry well. How careful were they about obscuring this? Also, since porn actresses presumably have to put on a brave face in their work, as well as in the face of critics who use any mention of unhappiness, however normal, as a weapon against their work, this may have conditioned them to respond to those types of questions in a more sugar-coated way,,,whereas the control may have no sense that this is being tested…
I think your “brave face” hypothesis might be a stereotype activating rather than an effect.
All participants–including the comparison group–were told about what the study was about, as that’s the decent thing to do. These weren’t questions which people are “trained to answer in a sugar-coated way”, but, rather, validated survey instruments which can pick up if someone’s answering in a certain way: if there was a bit of sprucing up the truth, we’d likely see it in all areas where there are stereotypes, e.g. drug use, and expressing concerns about catching STIs!
Interesting. I hope it leads to more being done welfare advocacy work *with* porn actresses, rather ‘for’ them without input from them and with little knowledge of their lives.
Such hopes are often futile: much that is said and done ‘for’ them, and ‘for’ sex workers in general, is driven by political and religious ideologues who are entirely impervious to factual input.
Nevertheless, this will prove useful for the reality-based community.
If I might offer a personal opinion, I would say that it’s always interesting to read research which challenges assumptions and prejudices; and this paper caught some unrecognised and unexamined prejudices of mine. I had perceived the porn industry as being especially and unusually exploitative, and ‘degrading’ in the objective sense of damaging the participants’ self-esteem, their mental health, and the satisfaction that they gain from their sexual relationships.
Happily this prejudice of mine appears to be unfounded – though I note your criticism about matching for class and privilege – though there’s an unpleasant inference to draw: women working in porn are not significantly more exploited or degraded than a matched sample from the general female population.
…That inference will be rejected, vehemently, by many women in ‘nice’ jobs who probably consider themselves feminists. I hope that this doesn’t drive them into the arms of the ideologues.
Thanks for your thoughts. In all honesty, I suspect that the porn actresses will probably be doing better financially than the college sample, which is probably something which ought to be controlled for!
Agreed.
Good post, thanks – wouldn’t have known about the paper otherwise.
I’d certainly agree with you that they should have used socioeconomic background as a control but wondering if it may have been problematic: as you’d pointed out the performers would probably be doing better financially than the control group, so it’s correlated against the very properties tested (the test group may be financially better-off than the control group by virtue of being the test group). So the economic background would likely not be an independent variable, all the other variables are likely to be independent. It’s been a while since I’ve done statistics so not sure if that would be a massive issue or not.
The other items that you’d put on your list of things you would like to have been examined were probably just out of scope (they were testing against a control group, not within the test group really). The sample size would also have made it difficult to start splitting down into types of porn. Hopefully more research along these lines will start being published, it’s very much needed.
Again, it’s been a long time since I’ve done stats on any basis other than hurried looking things up so take my comments with pinch of salt.
I think the perception of higher drug use could be applied to all actresses, couldn’t it? And the entertainment industry in general.
On the exclusion of men being ok, I think we have to ask why were men excluded? I think the implication is that women are the victims in porn and men are just lucky! I suspect if one did the systematic research a quite different conclusion would be reached.
We always need some good systematic analysis to shoot down prejudicial pre conceptions, let us hope that the men don’t have to wait long.
Porn is a great way to express human desires and experience pleasures otherwise unattainable. Attached shows just how popular porn is.
http://unitedfamiliesinternational.wordpress.com/2010/06/02/14-shocking-pornography-statistics/
One of my friends turned me onto this article as she found the study discussed interesting, and thought I might too – thanks for blogging about it!
Having been through the paper quickly, the one thing I would pull out immediately is the average length of time in the industry of the sample – 3.5 years. This was a mean measurement, so there isn’t an easy way to see how this is skewed (for me, a median would have been more helpful).
However, from what I’ve read in this area previously, and my own understanding as an outsider to the industry, actresses tend to fall into one of two categories (massively generalising): (1) Those who are sexually confident and enjoy performing as a way of earning a living / earning some cash (2) Those who enter the industry (or are lured in by the promises of rewards, fiscal or emotional) and who find the experience damaging or unpleasant.
Those in (1) tend to stay in the industry, and integrate their work into a happy, well-rounded life. Those in (2) tend to have personal issues of one sort or the other and should be considered vulnerable. These people don’t stay in the industry for long.
Given the mean length of time in the industry of the sample, I think we can assume that the majority of the sample fall into (1). This is where the problem lies for me. Pornography can be a liberating lifestyle choice but it can also be a predatory and abusive industry. Studies like this are interesting, but it’s important that their limitations are understood, and that generalised perceptions are not spread too hastily.
I hope the majority of adult entertainers find their experience empowring, but I think in honesty, most don’t. Of course, limited research means it’s hard to back up such suggestions with quantitative figures, but I think anyone who’s spent anytime trying to understand the industry would have to admit this is the case. As in any situation where vulnerable people are potentially being taken advantage of, I think it’s important that we protect and understand wherever possible. (And incidentally, that sentiment should apply to everyone in the adult industry, male – female, chat-line-worker – porn actress)
Nice post and an interesting discussion. Thanks 🙂
Another quantitative study [that did not make the headlines]: (Conclusion: “Female adult film performers have significantly worse mental health and higher rates of depression than other California women of similar ages.”) http://www.ncbi.nlm.nih.gov/pubmed/21632733
So, how to explain the difference?
Thanks for linking me to this. I read the full text of this study, and it turns out there’s a huge gaping methodological flaw in this that you’ve linked to.
They presented the questionnaire to the porn performers online, while the control group (who weren’t comparable, and this wasn’t controlled for in analyses) were interviewed by telephone.
Now, there’s a big problem with telephone interviews, and they can’t really be compared with internet surveys, because when speaking to someone else on the phone, the participants might give more socially acceptable answers to the questions about bad things like abuse and mental illness. I’m sure you can imagine why!
There’s also a few other problems with the control group which I won’t get into here as it’s very technical, but basically it means you can’t make any comparisons.
It’s interesting to note, though, that the porn performers’ answer to the question about sexual abuse was roughly the same as the percentage who reported it in the study I blogged about above. The difference is, when comparing it to a group of people who aren’t porn performers, but the data were collected in the same way, the difference disappears.
So it looks like that’s probably what’s going on here. Hope that helps, and let me know if I got a bit technical in any places and I’ll try to explain better!
P.S. The reason I suspect it didn’t make the headlines is precisely because it fitted in with the stereotypes: a lot of things which are “everyday” findings don’t tend to get reported!