Color and the making of gender in early childhood

Most of today’s readers weren’t following this blog back when I started writing about color preferences. Those posts are listed under the color tag. Now there’s a new paper on the subject that helps me think about how gender works in young children.

It’s called, “Preferences for Pink and Blue: The Development of Color Preferences as a Distinct Gender-Typed Behavior in Toddlers,” by Wang Wong and Melissa Hines, in the Archives of Sexual Behavior, the same journal where I published my paper on how adult color preferences are affected by the sex of their children. (Their paper is paywalled, but since we’re personal friends feel free to ask me for a look at my licensed copy.)

The researchers studied 126 children ages 20-40 months in a UK college town. The pertinent parts of their findings, for my purposes are: girls prefer pink over blue more than boys; but the the gap starts out quite small before age two and widens to age 3; the preferences are unstable, that is, the pinker girls and bluer boys at age 24 months are not the pinker girls and bluer boys at age 36 months. (The preferences were measured by asking which color they liked better on a card, and letting them choose between pink and blue gender-neutral toys.)

Whenever there is research showing differences between the sexes, I always like to look for the overlap (see, e.g., this post). That’s because people fixate on the differences to confirm their presumption that the differences are total, fixed, and baked in or genetic. This underlies the whole fixation on the dimorphism question. So when they report girls are more likely to choose pink over blue than boys, I plug the means and standard deviations into my graphing spreadsheet to see the implied distributions (assuming normality). Here is the overall pattern:


So, you can decide whether you think that’s a big difference, but you should factor in the size of the overlap. The change over about 14 months was pretty impressive, with boys and girls pulling apart. Here are the curves at 20-26 versus 34-40 months:


One possible interpretation of this pattern is that color preference is learned rather than baked in at birth, and this is a time kids learn it. That interpretation is strengthened by the further finding that, while the gender difference increases from age 2 to age 3, it’s not stable within individuals. That is, whether a kid was pink-positive or -negative at time 1 was not a predictor of their preference at time two. That’s what this figure shows — girls are more likely to be in the top-right, but the time-1–time-2 slopes aren’t significant:


That’s more evidence against the idea that the sex difference in color preference is determined at birth, which is also consistent with the historical evidence, as Jo Paoletti’s work shows.

Children themselves have a strong motivation to perform their gender identity in ways that please adults or perhaps other children, and that tendency exacerbates early sex differences. They can anchor this performance to an arbitrary marker like color. From the paper (references removed):

Gender-related cognitive processes have been implicated in the acquisition of gender-typed color preferences. Specifically, gender-typed behaviors may be acquired through self-socialization after children have developed gender identity, and become self-motivated to adopt gender norms.

Unlike critics of this blog, I don’t fear that gender differences will be erased if we don’t continuously reinforce and celebrate them. People will figure out ways to make the “natural” differences count enough to get the job done when they need to. And reducing the pressure will help decrease both gender inequality and the stigma experienced by non-conforming people.

What was I supposed to do, not report the results?

In case you haven’t been following the research on this, my understanding is that there is some evidence that women in several cultures are more likely to wear red-related colors when they are trying to look sexually attractive. We know that from the article “Women Use Red in Order to Attract Mates” in the journal Ethos. That’s all well and good, but to make it really interesting, we’d like to know that women are especially likely to do that when they are in the most fertile time in their menstrual cycle. Because, you know:

Photo by D. Gordon E. Robertson from Wikimedia Commons
Photo by D. Gordon E. Robertson from Wikimedia Commons

Unfortunately, that paper from Ethos did not find that red-wearing was associated with menstrual cycles. But, Beall and Tracy were able to find that link. Their conclusion:

Our results thus suggest that red and pink adornment in women is reliably associated with fertility and that female ovulation, long assumed to be hidden, is associated with a salient visual cue.

As Kim Weeden pointed out when I mentioned this on Twitter, Andrew Gelman used that paper as an example of how researchers have many opportunities to slice findings before settling on those that support their hypotheses.

Fortunately, Beall and Tracy set out to replicate their finding. Unfortunately, when they attempted to replicate the results, they were not successful. Fortunately, they realized it was because they were being confounded by the weather. As they have now reported, this is important because in warm weather female humans don’t need to resort to red because they can manage their attractiveness by reducing the amount of clothing they wear (and then, who cares what color it is?). Thus:

If the red-dress effect is driven by a desire to increase one’s sexual appeal, then it should emerge most reliably when peak-fertility women have few alternative options for accomplishing this goal (e.g., wearing minimal clothing). Results from re-analyses of our previously collected data and a new experiment support this account, by demonstrating that the link between fertility and red/pink dress emerges robustly in cold, but not warm, weather.

And here it is. Happy, Gelman?


Confirmatory classroom exercise

Since I am teaching love and romance in my family course this week, I thought we should add something to the conversation. I only did one exercise, and I am reporting the full results here. Nothing hidden, no tricky recodes, no other questions on the survey, no priming of the respondents (it was at the start of the lecture).

I have 80 students in the class, which means 53 were there in time for the exercise, 29 men and 24 women. I gave them this two-part question:


Because red and pink are both associated with fertility (see the baboon), I combined them in the analysis (but it works if you just use red, too). And these were the results:


The statistical test for the difference between date and family event for women is significant at the level of p<.035. This is not research, it’s just a classroom exercise (which means no IRB, no real publication). But if it were research, it would be consistent with the women-wear-reddish-to-attract-mates theory (although without the menstrual cycle question, its contribution would be limited).

Most sociologists might not go for this kind of stuff. Maybe it’s a slippery slope that leads to unattractive conclusions about gender inequality in the “natural” order. My perspective is that I don’t care. Of course this is not really evidence that evolution determines what American (or, in the case of the Ethos paper, Slovak) students wear on dates. But it doesn’t refute the theory, either.

More importantly, I am confident that we could, if desired, through concentrated social engineering, eliminate the practice of women wearing reddish on dates if we thought it was harmful — just as we have (almost) engineered away a lot of harmful behaviors that emerged from the primordial past, such as random murder, cannibalism, and hotmail. After all, they did it in China:


Sorry, wrong picture:


For previous posts in the series, follow the color tag.



Parents live in the gendered world of their children, too

I did a little research hinting at the way gendered childhood might affect parents – by looking at how the gender of their children affected their favorite colors.

Because gendering – especially around consumption – is so fierce, I figure that’s got to be the tip of the iceberg. I thought of that walking to the kids’ school the other day:

Maybe the mothers dressed to match the kids because it was the first day of school. Or maybe they have more matching clothes so that coincidences like this happen more often at random. Who knows?

Gender of kids, gender of parents (via color preference)

The Archives of Sexual Behavior has published my paper, “Children’s Gender and Parents’ Color Preferences,” as an online-first article. The abstract is up here, but the paper is paywalled — let me know if you’d like one of my personal copies and I’ll be glad to send it to you.

I did the research because I was interested in the effect that today’s strongly-gendered parenting might have on the adults who live in the color-coded worlds of their children. Also, I was looking for a simple way to test whether, or to what extent, color preferences were variable in adulthood (rather than genetic), and the gender of children seemed like a good, mostly-random experiment to test that.

Anyway, here is the basic breakdown of how my 749 parent respondents answered the online question, “Which color do you most prefer?” The colors here are the same ones they saw online (individual monitors vary, of course), ordered according to relative gender preference:

Pink, purple and red tend more female; green, blue and orange tend more male. In the statistical tests, with an age control, only blue, red, purple and pink had big enough gender differences to reach 95% confidence.

Then I compared people who had boys only and girls only to those who had a mix of boys and girls. The result was most clear for women: those with boys only had more “female” preferences — preferred pink more and blue less. For men, having either boys-only or girls-only increased their odds of preferring blue.

Here’s the pattern for women (with no controls):

The article has little in the way of discussion and speculation — the text is only six journal pages. My post-hoc interpretation is that gender-heavy environments (single-gender gaggles of children) push parents in gender-stereotypical directions. Maybe. I would be happy to hear your thoughts.

Start to finish: 750 days

This is the only time I’ve ever collected my own data for a project and seen it all the way to publication. For those interested in the process, this is how it went:

  • March 17, 2010: Working on the chapter about gender for my family sociology textbook, I blogged about the research I was reading regarding color preferences.
  • April 18, 2010: More blogging, this time about the far-fetched evolutionary psychology I was reading on the reasons for gendered color preferences.
  • April 21, 2010: Posted the survey online at, launched with a blog post and other social media.
  • April 26, 2010: Submitted an application to my local Institutional Review Board for permission to do the survey.
  • May 3, 2010: After one minor revision, the IRB approved it, and that gave me the green light for a UNC email blast.
  • May 4, 2010: “What’s Your Favorite Color?” email blast to UNC students, staff and faculty. More than 1,000 responses on the first day.
  • September 16, 2010: Stopped collecting responses. (The survey is still up there, though, drifting unmanned around the Internet, gathering data like so many comments on a dead blog.)
  • February 4, 2011: Submitted manuscript to Archives of Sexual Behavior.
  • June 21, 2011: Nudged the editor to see what was up with the review.
  • July 16, 2011: Received a revise-and-resubmit decision.
  • September 28, 2011: Resubmitted.
  • February 11, 2012: Received provisional acceptance.
  • March 7, 2012: Resubmitted.
  • March 19, 2012: Received proofs from the journal to review.
  • April 5, 2012: Article published online.

Pink and blue kid(s)

Jo Paoletti got a nice cover write-up from the UMD magazine Terp, for her new book, Pink and Blue: Telling the Boys from the Girls in America. I especially love the cover photo:

The photo(s) is (are) by John T. Consoli, linked without permission. This is a great exercise for noticing how we internalize gender norms. If you thought even for a moment that one of these kids is a boy and the other is a girl, you’re busted as a product of socialization. (I won’t reveal the “true” answer.)

This is much better than my own crude Photoshop version:

I’ve done a series of posts on color and children, which you can find here. That led me to do a survey on color preferences, which got close to 2,000 responses to the question, “What’s your favorite color.” Based on a simple color chart provided, these were the responses:

More than the gender breakdown, I wanted to see if the nurture-versus-nature element of color preference could be teased out in a single survey. So I asked people if they had children, and the gender of their children — figuring that might reflect some “adult socialization.” And that modest result is what I found. From the abstract:

This study asks whether the gendered social environment in adulthood affects parents’ color preferences. The analysis used the gender of children to represent one aspect of the gendered social environment. Because having male versus female children in the U.S. is generally randomly distributed, it provides something of a natural experiment, offering evidence about the social construction of gender in adulthood. The participants were 749 adults with children who responded to an online survey invitation, asking “What’s your favorite color?” Men were more likely to prefer blue, while women were more likely to prefer red, purple, and pink, consistent with long-standing U.S. patterns. The effect of having only sons was to widen the existing gender differences between men and women, increasing the odds that men prefer blue while reducing the odds that women do; and a marginally significant effect showed women having higher odds of preferring pink when they have sons only. The results suggest that, in addition to any genetic, biological or child-socialization effects shaping adults’ tendency to segregate their color preferences by gender, the gender context of adulthood matters as well.

The paper has been provisionally accepted and should be published in a peer-reviewed journal near you soon.

Jo’s excellent blog is here, where you can read about her new project, exploring the rise and fall of unisex clothing for children.

Little man meet little cupcake

This stuff is a-dime-a-dozen, but I still love it.

These are the Luvable Friends 6-Pack Flannel Receiving Blankets available from Amazon (and this is not a paid product placement). They come in two color sets. A blue theme for “little man”…

…and a pink theme for the “little cupcake.”

So what are we supposed to wrap the little women in?

Follow the Family Inequality color obsession with posts under the color tag.

Was FDR born that way?

Back when kids were kids.

I can’t wait to see historian Jo B. Paoletti’s forthcoming book Pink and Blue: Telling the Girls From the Boys in America. Her work is profiled in a new article my friend Scott showed me. I used an article of hers to write about historical changes in the social construction of gender for children, which got me into the whole gender-color thing.

The Smithsonian article has this great portrait of Franklin Roosevelt at about age 2, dressed — as children typically were at that time — in a gender-neutral outfit of frilly white dress, patent-leather shoes, feathery hat and long hair:

I recommend the article.

It reminds me of the Born This Way blog, which features pictures of kids that, in the view of their adult selves, express their true nature. The site is “a statement in sociology. As you’ll see – time after time – their sexual orientation was simply NOT a choice.” Here’s one:

The text says:

In 1968, I was “The Flying Nun” for Halloween, as I was obsessed with her TV show. Everyone had a big laugh over the boy in a dress! But being so young, I really didn’t understand what all the fuss was about. I kept wearing my magical dress for playtime, all the way through here, in the summer of ’69. I started to sense that I was different from anyone I knew. By the time I was 12, it dawned on me that I was gay.

So, what would a gay 5-year-old in 1884 have done?

Doing color with babies

Thinking about how to study color on kids.

Color is everywhere; gender is everywhere; color and gender have a varied, changing, and hard-to-pin-down relationship, which nevertheless creates powerful, explicit and implicit signals by and for both the people who apply colors (to their kids, themselves, their cars) and those who see them (including themselves). Discuss?

Naming and choosing colors has a subjective component, as Randall Munroe showed in his xkcd color survey.

I’ve written a few posts about gender and color, and conducted a simple “what’s your favorite color?” survey, which turned up these results for the first 1,695 respondents, the vast majority from a university-sponsored email blast to faculty, staff and students:

(The survey is still up if you want to take it — click the picture — but I’ve already pulled out a sample for analyzing.)

I labeled these colors blue, green, orange, yellow, red, purple and pink, but also showed the colors on the screen to try to triangulate visual and verbal definitions. Still, a few people added comments such as, “I usually like green, but not this one.”

In the old days, some surveys have simply asked people to name their favorite colors. Silver and colleagues (Perceptual and Motor Skills 1988, 66:295-99) used that method in 1988 in a survey of undergraduates, and got these results:

“Others” included pink, which for some reason they didn’t list separately, even though it was about half the “others” in the case of women. Ironically, Black and White appear as colors on the row labels, and as races on the column labels. (I didn’t include those in my survey, because when I was a kid some other kid told me they weren’t really colors. Now I capitalize them when referring to races, since those definitely aren’t really colors.)

Anyway, nowadays, you get fancy color analyses with infants’ eye movements recorded (no gender difference found), or rapid on-screen comparisons showing adults many different colors, analyzed (obviously) according to the “neural dimensions that underlie color coding in the human visual system” rather than specific color names such as “pink.”

Many people are concerned that we are creating color preferences in children by dressing them according to gender stereotypes. I also believe — and may yet (spoiler alert) show — that the colors on and around children also affect their parents’ attitudes toward color (see, e.g., Pinkalicious, a product line that I’m betting is more common in homes with daughters). How could this not do something to the parents?


To make a short story long, with the recent news of Disney moving its product lines into the delivery room, I was wondering how one could study the application of colors by parents to their children in real life.

Just to try something, I looked at the first 20 boys and 20 girls on the “most recent” list of photo albums at I used a free eye-dropper program to select a swatch from each kids outfit (first picture that showed visible color either on the clothes or the blanket), trying to pick the color that took up the greatest space (not counting white). After the eye-dropper gave me the color codes, I dropped them into this web-color utility, to get a pure color sample. I figure you could show these to people, like xkcd did, and ask them to name the colors, something like that.

Some of them were easy, like the first kid. The second one required deciding which color was most prevalent.

It was also a little tricky deciding which pixel to select, since there are shadows, spots, etc. So it’s not as easy as if you had a single-colored fabric stretched out in set light conditions. And in some cases the mix of colors was too much, and I just couldn’t pick one that was most prevalent. Here’s what it looks like:




Color gender by the numbers

Men and women weigh in on their favorite colors.

Update: I’m curious. Will you take a color preference survey here?

More on the many mysteries of pink and blue, this time from college students expressing their own preferences, rather than adults’ choices for children.

This research is from 2001, but I just stumbled on it. In a survey of 5,000 college students from several dozen universities, men and women were asked to express, on an open-ended form, their favorite color.

MEN                                                     WOMEN

Favorite colors for college students, 1990s.

Other responses not shown (4% of men, 5% of women). My chart from data in the article.

These men have a strong blue preference; the women are more diverse in their choices. Proportionally, the biggest differences are on pink (women 10.6-times more likely to choose) and blue (men 1.8-times more likely).

Here is the interpretation of the authors:

Without ruling out any possibility at this point, we are inclined to suspect the involvement of neurohormonal factors. Studies of rats have found average sex differences in the number of neurons comprising various parts of the visual cortex. Also, gender differences have been found in rat preferences for the amount of sweetness in drinking water. One experiment demonstrated that the sex differences in rat preferences for sweetness was eliminated by depriving males of male-typical testosterone levels in utero. Perhaps, prenatal exposure to testosterone and other sex hormones operates in a similar way to “bias” preferences for certain colors in humans.

You really have to love it. Although it’s not as far gone as the speculation that color preferences evolved from the gender division of labor in the hunter-gathering prehistory, it’s not a theory well suited to the rapid historical change we’ve seen in the case of dressing children, at least.

If I were making up an explanation, I’d say maybe these college students were generally pushed toward girl-pink/boy-blue from infancy, and then the girls more actively incorporated color choice into their identities (the idea of having a “favorite color”) — resulting in greater diversity of choices. On the other hand, maybe boys were more likely not to have a color affinity in their identity toolbox and thus are more likely to have a stuck-in-childhood response that matches the preference their parents had for them, or one they consider socially desirable. How’s that?