Cognitive Daily

i-2430e448a0eb8f6be8a3b42100bda16b-sexy.jpgi-eca0cf2af9fc3ac4445c7dff7d8aab70-research.gifA recent study about violence and sex in TV advertising got a fair amount of press. “Violence and sex don’t sell,” the headlines proclaimed. If such a claim is true, it flies in the face of conventional wisdom and the blusterings of ad agencies worldwide.

Advertisers have always used the idea that “sex sells” to generate interest in their products, and television networks have argued that they need to offer lots of violence and sex in their programming in order to attract viewers. But what if sex and violence don’t really sell products? Doesn’t that turn the whole notion of “sex sells” on its head?

We’ve taken a look at the study, by Brad Bushman and published in the prestigious journal Psychological Science, and the answer to that question seems to be “sort of.” Bushman studied 336 adults in central Iowa by showing them violent, sexy, or neutral TV programming (some of the programs included 24 and Cops [violent], Sex in the City and Will and Grace [sexy], and America’s Funniest Animals and Trading Spaces [neutral]). Each program contained the same 12 commercials. The commercials were chosen from a selection of products that were relatively unfamiliar to the study group: “Senokot Natural Vegetable Laxative,” “Nutra Nails,” and so on.

The goal of the study was to see if different types of programming correlated to the effectiveness of the advertisements. After watching the programs, participants were asked which of the advertised products they could recall, which brands they recognized from a multiple-choice quiz, which they planned to buy, and finally they were given their choice of ten $1 coupons from a set of 40, including the 12 they had seen advertised. All the coupons were actually fakes, but participants believed that they could use the coupons to get real store discounts. Here’s a summary of the results:


The chart indicates that sex and violence in the surrounding TV program do indeed affect how well viewers recall the brands in the ads. People who watched programs with sex and/or violence recognized significantly fewer brands and chose significantly fewer coupons from the advertised brands. Viewers of sexy, but not of violent, shows recalled significantly fewer brands than viewers of neutral shows.

Bushman argues that if advertisers would refuse to buy advertising on shows with sexual/violent content, they could help their own bottom lines because the ads on these shows don’t sell as well. Society would be helped because violent/sexy shows would soon be removed from the air due to financial insolvency — and violent programming is associated with aggressive behavior while sexually explicit shows can lead to anti-social sexual attitudes.

Are Bushman’s far-reaching conclusions merited? Assuming the price for advertising during violent/sexy programming is the same as for neutral programs, perhaps they are. But this single study can’t be taken as a refutation of the mantra that “sex sells.” For one thing, the study doesn’t consider sexual and/or violent content in the ads themselves. Bushman suggests that one reason ads are less effective during violent or sexy programming is that the viewer’s attention is drawn to the sexual/violent content. Wouldn’t sexy/violent ads also draw viewers’ attention? If so, would these ads be more effective in the context of sexy/violent shows, or neutral shows?

And what if advertisers did begin to flee from violent/sexual programming? Would more viewers be drawn to the few such shows that remained? Might some advertisers not still be willing to pay to reach the large audiences for these shows? If so, then Bushman’s ideal of a sex- and violence-free media would never be attained.

Nonetheless, Bushman’s research does suggest that advertisers, media outlets, and consumers should be aware that mantras such as “sex sells” almost certainly oversimplify the issue.

Bushman, B.J. (2005). Violence and sex in television programs do not sell products in advertisements. Psychological Science, 16(9), 702-708.


  1. #1 coturnix
    May 30, 2006

    More excited you are about the show, more you hate the commercial break, don’t you? It interrupts the action at the cliffhanger, or gives you a cold shower in the middle of a steamy scene – what’s not to hate?

    As you pointed out, the “sex sells” question is a question about the content of ads themselves, not about the content of the surrounding show that the ad contrasts to and competes with.

  2. #2 Harlan
    May 30, 2006

    I agree with coturnix; this doesn’t really address the real question, and methodologically, I’m not sure if you can say that the shows are equal. Did the subjects rate the shows in terms of how much they enjoyed them? Was physiological arousal measured? Were these measures entered as covariates in the the analysis? If not, I’m not sure they were comparing apples and apples, and would take the results as preliminary.

  3. #3 baldywilson
    May 30, 2006

    The big problem with Bushman’s conclusion is simply that a lot more people watch violent and/or sexy programs than watch neutral programs, so even if the advert is selling fewer products per person the advert is reaching a far greater number of people. It’s all very well having 50% of people recognising your brand if you go for program x, and only 25% for program y, but if program x only attracts 500 people, but program y attracts 2,000, I know which one I’d go for!

  4. #4 Kapitano
    May 30, 2006

    It seems to me the Bushman’s method is backwards. Instead of measuring the response to neutral ads interrupting variable programming, he should be measuring the response to variable ads interrupting neutral programming.

    Measuring the viewers response to an advert for floor tiles in the adbreak of Sex in the City tells us something about how the viewer responds to a nonsexy interruption to a sexy programme. It doesn’t tell us anything about the response to the advert itself – except that viewers don’t like having their sex interrupted ;-).

  5. #5 chezjake
    May 30, 2006

    And since when are 336 adults from central Iowa a statistically significant sample for anywhere but central Iowa? Who, besides Wal-Mart, wants to advertise selectively to that market?

  6. #6 Charles
    May 31, 2006

    “Sex Sells” only applies to the ad itself, as in blonde babes in wet Tshirts selling beer. This study seems to be conclusions on lot of nothing.

  7. #7 Dave Munger
    May 31, 2006


    I wouldn’t say it’s a whole lot of nothing. It’s saying, all things being equal, an advertiser would be better off advertising during neutral programming.

    Chez Jake,

    Certainly I wouldn’t suggest that the lack of diversity in the sample isn’t a problem, but this is a large sample of motivated viewers (they were paid $30 for participating). Many studies’ sample groups consist only of college sophomores, and this group was much more diverse than that.

  8. #8 Michael Anes
    May 31, 2006

    There’s a great deal of ecological validity to this study. Just on the face, many products cannot be sexed up or violenced up and it would be unwise to do so (maybe) for floor wax or tissues or cat litter or kids vitamins. Many products are relatively neutral (I suspect, and have no evidence) compared to the kinds of entertainment that can be shown!

    More importantly, the study matters because we want to know what effect various situations have on encoding. What good is any advertisement if you cannot remember the product? And we know a great deal about the effects of stress and elevated cortisol before, during and after encoding from the lab of Larry Cahill and others. All that elevated physiological response surrounding an ad may not help memorial processes, and this study is another in that vein.

  9. #9 n
    June 1, 2006

    Apart from everything else everyones said, I have one thing to add, where are the error bars? Those differences to me seem small enough to be obliterated by variance.

  10. #10 Dave Munger
    June 1, 2006


    We don’t put error bars on the graphs for CogDaily because of the difficulty in interpreting them (for example, sometimes they report standard error, sometimes confidence intervals). We always report in the post which results are statistically significant. In this case, the results are significant.

  11. #11 Stephen O' Farrell
    January 10, 2008

    Blaaa bLaaa Blaaa. Has any one every watched will and grace..?
    Maybe it’s me but I don’t think Jack sashaying across the room doing “Just Jack” constitutes for “sexy” or the constant banter and jokes. In fact.. Where is the sexy moments.. rarely even see people kissing in it.. Then again.. Sometimes during the ad breaks I do copy them and strip of into my underwear and try to replicate the backup dancing scene with J-Lo.. Maybe that’s what the viewers are really doing. Fun Times….. Fun Time..

New comments have been disabled.