M-c Shanahan

One of the first things that came across my Twitter feed yesterday morning was a press release announcing that “Women’s Quest for Romance Conflicts with Scientific Pursuits, Study Finds”.  I’m usually pretty sceptical of press releases, especially ones that include the words “study finds.” This one, instead of eliciting mild annoyance though, made me feel nauseous. Women’s quest for romance? Really? At first glance, it sounded more like something out of Mad Men than a real study.

I’m going to put my bias right out there. I am passionate about engaging people in science and, as a result, also intrigued by what keeps them from participating. I began my academic career with the explicit purpose of exploring gender issues in science education (and I have the embarrassingly naive grad school entrance essays to prove it). A year into my master’s research, I became increasingly frustrated with research that sought to categorize women and girls and definitively assign them characteristics that interfere with their interest in science. Through my own classroom teaching experiences I was well aware of the diversity of strengths, weaknesses, desires and goals that both female and male students bring to the science classroom. Essentialist gender research not only covers up this diversity, it also misses male students, many of whom are also discouraged and excluded in science. This is why I study inclusion and exclusion in science through a lens of identity – looking at patterns in the ways that individuals define themselves (including in relation to masculine and feminine gender norms) and how these definitions come together to influence the decisions that male and female students make about studying science. All that is to say, that I am generally not inclined towards approaches that homogenize women and men but at the same time am open to the important influences of stereotypes and societal expectations that can have particular influences on science participation.

So while my first thought was “ugh”, I was willing to look openly at the data to see what they’d found. It only seems fair to tell you all of that up front. So I tracked down the full study to investigate further.

The paper announced in the press release (Park, Young, Troisi & Pinkus, 2011) is based on three consecutive small studies done to explore the connections between romantic goals and science interest. The authors explain the prevalence of gender stereotypes in Western society and how girls are typically socialized to act in gender-typical ways in (presumably heterosexual) romantic situations. They also note that women who act in ways that break gender norms tend to be viewed negatively by others. The assumption is that having an interest in science is viewed as a masculine characteristic and therefore something that women will downplay in romantic situations and when they are pursuing romantic goals.

There are many (many!) unsupported assumptions built into this logic. The most obvious issue is that scientific disciplines are not nearly equal in their association with masculinity. The authors do not make a note of this but in their questionnaire they only list those areas with significant under-representation of women (“Computer Science, Technology, Engineering, Math, Chemistry, Physics, etc.” p. 3). So already, this isn’t a study about science but about particularly masculine identified areas in science. Good to know. (And actually it’s more complicated than that because different areas of engineering vary widely in gender representation and in association with masculinity and femininity)

Study 1

The first study examined two possible goals types: wanting to be intelligent and wanting to be romantically desirable. They primed the participants (all university psychology students) to think about one or the other goal type by showing them images. They prompted intelligence goals by showing participants images of books, libraries and eyeglasses (eyeglasses? That just makes me think about going to the optometrist, but okay) and romantic goals by showing them pictures of sunsets, candles and restaurants. The authors checked these images with a follow-up study to be sure that they primed the right goals (and not, say, goals of going to the optometrist) and the men and women that they asked responded that the images made them want to be intelligent or romantic in the expected ways. The main study then brought 119 students (approximately balanced between women and men and with approximately equal interest in science) into the lab. The participants were asked to rate their interest in science and in pursuing a science-related major. They then looked at one set of images (either romantic or intelligent) and answered the two science interest questions again. When they compared the two groups for women and men, there were no differences for those who had looked at the intelligence images. The women and the men in this group had about equal interest in science and in pursuing a science degree. Women who has seen the romantic images (and presumably were feeling like they wanted to be romantically desirable) reported lower interest in science than men and than the women who had seen the intelligence images.

The researchers conclude “These results support our hypothesis that women, but not men, show less interest in STEM when exposed to cues related to romantic goals versus intelligence goals. We think that women may have distanced themselves from STEM because they experience conflict between the goals to be romantically desirable and intelligent in the male-stereotypes domains of STEM” (p. 5). Sounds convincing.

Or maybe not. There are two things that I find problematic. The first is that their model showed no significant effect for prior science interest. That means that students walked into the lab, answered two questions about their interest in science, saw some pictures, then answered the same questions again and the answers they gave the first time didn’t predict the answers they gave second time. Wow, that’s one heck of an intervention! That, or the measures are problematic and don’t have great test-retest reliability (meaning that even without the pictures, people would answer the questions slightly differently from one moment to the next because they are not well designed questions). This was actually one of the first problems I noticed with the study. The researchers have taken a complex idea – interest in science (which includes interest in math, logical thinking, the natural world, abstract and explanatory thinking and more) and reduced it to a single question “How interested are you in Math and Science?” A single question like this is very prone to shifting answers. One minute I might rate my interest at a 5/7, then next maybe 6/7 because what do those ratings really mean? If we’re talking about nature maybe I’m really interested but mathematical formulas for motion, not so much. To be honest my personal answers would be the opposite – I’m all about the physics but kind of bored by taxonomy. A much more meaningful measure would have used several items, developed and tested for both validity and reliability for measuring interest in science. Krapp and Prenzel’s recent review might be a great a place to start for anyone who is interested. Without a measure that is reliable and really captures what it means to be interested in science, it’s difficult to interpret the results meaningfully.

The second issue is what is really being measured. The authors support their assertion that the effect is due to goal priming by showing that both men and women responded with the appropriate goals when they looked at the romantic and intelligence pictures. Can we know for sure that it was the goals that caused the difference in interest though? No, we can’t. That uncertainty is a normal part of all social research and for that reason alternative explanations should always be explored. My prime contender in this case is stereotype threat.

Stereotype threat is the anxiety that people can feel when they are in situation in which they might confirm a negative stereotype about their gender, racial or cultural group. For example, when reminded of the stereotype that girls aren’t good at math, girls tend to do worse on math assessments than they would otherwise.  Stereotype threat doesn’t just impact achievement though. It effects motivation and interest too. Jessi Smith and her colleagues investigated interest and motivation in a computer science task under conditions where participants felt gender stereotype threats. When they suggested that women aren’t good at math and related tasks, the women in their study showed less interest for completing the tasks. Those women who were normally achievement and goal oriented also changed their motivation from striving to achieve goals to being nervous about failing, a motivation orientation that is associated with lower interest. So, when stereotypes about women are primed in science and math, it seems that interest and motivation goes down.

The central premise of this study is itself a stereotype about women: that they constantly seek romantic validation. Participating in this study itself might have primed stereotypes. I am speculating here but I would find those images to be stereotypically feminine and the whole idea of being romantically motivated to be a negatively-associated female stereotype within an academic environment (think of the derision associated with stories of women who go to university to find husbands – no such parallel exists for men). It seems at least possible to me that priming women with stereotypical romantic images in an academic setting would elicit stereotype threat, a threat that could possibly be related to decreased motivation and interest for science, another stereotypical characteristic. I’m not actually trying to make this claim, just to show that there are other possible explanations than what the authors of the study suggest.

Study 2

 This study was very similar to Study 1, except that instead of looking at pictures the participants overheard conversations that were meant to prime either romantic goals or intelligence goals. The researchers and their assistants stood in the hallway and had a short conversation that was either about doing well on a test or about a recent successful date. The study participants inside the study room could hear the conversations. The authors also did second version that used similar conversations but this time primed romantic goals and non-romantic friendship goals. The authors report that both versions replicated Study 1. Except that it isn’t quite true. In Study 1, men and women showed equal interest when they saw intelligence pictures and when men saw romantic pictures. The basic result was that women who saw romantic pictures expressed less interest than would be expected, all other things being equal. Study 2 shows something entirely different. In both versions, women and men who overheard romantic conversations expressed equal interest in science and so did the men who overheard intelligence or friendship conversations. The only difference was that women who overheard intelligence or friendship conversations showed HIGHER interest in science than would be expected. Let me say that again – women and men responded no differently to having romantic goals primed and the non-romantic goals (intelligence and friendship) led women to express stronger interest in science. When they overheard the intelligence related conversations, women also expressed stronger desire to pursue science related degrees. I don’t see how that replicates Study 1. It seems in my view to directly contradict it.

Study 3

The third study asked women (and only women) to keep track of their daily goals and their math activities (e.g., paying attention in class, doing math homework) and desirability activities (e.g., emailing/texting someone you are interested in or spending time with them). From the students’ diary entries, they found the following relationships:

• On days that women reported pursuing romantic goals, they engaged in more romantic activities and fewer math activities. They also engaged in fewer math activities (but not more romantic activities) the day after.
• On days that women reported pursuing academic goals (and the day after), they engaged in more math activities. These days had no impact on their romantic activities.

So, the participants in their study engaged in activities that met their goals for the day, whether academic or romantic. This isn’t a surprise at all and there is no evidence that this is gender specific because no men were included in this part of study. There is also no evidence that it is exclusive to romantic goals because no other goals were studied. It is just as conceivable that people who set interpersonal goals for the day will also engage in less math and more interpersonal activities. That kind of the point of having a goal for the day isn’t it?

But wait, there’s more:

• On days when the women pursued romantic goals (and the day after) they also felt more romantically desirable.
• On days when women pursued academic goals, there was no impact on their desirability (it neither increased nor decreased their feelings of desirability). Wait, what? So the whole study is built on the premise that wanting to be romantically desirable interferes with interest and participation in science but there is no evidence that feelings of desirability are in any way negatively affected by pursuing intelligence goals.

This means that the only finding from Study 3 is that activities can be predicted by daily goal setting (either romantic or intelligent).

Some final thoughts

So, what’s my overall assessment? I’m really troubled by the study, not because it disagrees with the way that I approach gender and science but because the evidence is extremely weak and seems to have been interpreted with the researchers’ own expectations heavily in play. The back and forth nature of the findings (in one study romantic goals had a negative effect and in another they didn’t) is not acknowledged and suggests to me that there is probably something else that is actually creating the effect. Not only that, it suggests that the measures (the questions themselves) are probably not stable. There is nothing in this study that convinces me that romantic goal pursuits are in any particular way responsible for women’s underrepresentation in science – not because it’s something I don’t want to be believe but because the evidence just isn’t there.

(As a side note, I’d like to add that the title of the article suggests that the authors did not spend much time with the literature related to science education persistence and participation. The title indicates that the study is about attitudes towards science, which is something entirely different from interest and not addressed at all in the study. I have also not addressed the secondary findings related to interest in English/languages because the primary claims are related to science and math.)

References and other reading

Krapp, A., & Prenzel, M. (2011). Research on interest in science: Theories, methods, and findings. International Journal of Science Education, 33, 27-50.

Park LE, Young AF, Troisi JD, & Pinkus RT (2011). Effects of everyday romantic goal pursuit on women’s attitudes toward math and science. Personality & social psychology bulletin, 37 (9), 1259-73 PMID: 21617021

Smith, J.L., Sansone, C., & White, P.H. (2007). The stereotyped task engagement process: The role of interest and achievement motivation. Journal of Educational Psychology, 99, 99-114.

Steele, C. M. (1997). A threat in the air: How stereotypes shape intellectual identity and performance. American Psychologist 52, 613–629