On Mothers and Math: Behind the Buzz

As someone who has been relatively vocal about research rigor and how findings are interpreted (and sometimes misinterpreted) in the mainstream media, it was certainly interesting for me when a small study I did as a spin-off from my dissertation picked up a bit of buzz a few weeks back. To summarize, in an analysis of number-related language (e.g., “three cats,” “how many apples?”) in everyday speech, American mothers of toddlers (mean age 22 months) spoke two to three times more often about numbers and quantities when talking to boys compared to girls. After reading some of the questions and comments (some quite angry!) about the study, I thought it might be interesting to provide a bit of a behind-the-scenes look at how this paper came to be. I’ve addressed a few issues that directly deal with gender stereotyping – and why the study did not include fathers – in a previous Q&A with Mommyish.com, so I will try my best to not be repetitive.

I was originally studying differences between Mandarin Chinese and English speaking parents, how they talk to children about number before they enter formal schooling, and how this might contribute to the well-documented cross-national differences in math scores between Mandarin speaking countries (China, Singapore, Taiwan) and their English speaking counterparts (including the United States). I never failed to get at least one question about gender differences each time I presented the data. Since my research was actually about language and concept development, it took awhile after sorting the initial study out before I took the time to carefully look at what was happening between genders.

I needed to make sure that the age of the children and the amount of overall speech generally matched between boys and girls (i.e., I controlled for age and amount of speech) in order to compare between roughly equivalent groups of mother-son and mother-daughter pairs. The fact that only mothers were left after this matching process was purely circumstantial, not intentional. (However, at least one study has found that fathers show a greater bias toward boys than mothers when explaining science museum exhibits). My colleagues and I were both shocked and fascinated to find significant differences in number speech to boys and girls… but only in the American group. Over the next two years (that’s right, years), this paper was written and revised countless times. However, it was rejected no fewer than seven times from as many journals before finally being accepted and published in the Journal of Language and Social Psychology.

This is just one example of the stark difference in pace between academia and Silicon Valley. Between authors, reviewers, and editors, iterating on a manuscript can take months, if not years, for a variety of reasons. A lot of stress can result from a sometimes incredibly long peer review process. More than once have I wondered if a reviewer actually read my manuscript, and have become incredibly frustrated when reviews that don’t seem particularly negative result in a rejection for an unclear reason. Despite all of the challenges, I firmly believe in peer review as a tenet of the scientific process. I feel that it is important for articles that go down as scientific record to be reviewed by individuals with domain expertise. In addition, my co-authors and I knew that we had an important, compelling result, and knew that the only way that it would ever have any impact in the world (even just to us in our own careers!) would be to get it published in a peer-reviewed journal. And we finally succeeded… on the eighth try.

The amount of time it takes to design, run, analyze, implement, and then publish a research study is one of the reasons why there isn’t a lot of “hard evidence” about some of the current learning technologies Silicon Valley is buzzing about. Pilot results are one thing, but full randomized controlled trials are a whole different beast. Building a bridge between rigorous scientific research and development of technology is something that I feel is incredibly important, but has quite a ways to go. It is my hope that in this next phase of my career, I can make a contribution in this area, and perhaps even inspire more girls and women to pursue STEM!


Chang, A., Sandhofer, C.M., & Brown, C.S. (2011). Gender Biases in Early Number Exposure to Preschool-Aged Children. Journal of Language and Social Psychology, 30(4), 440-450.

Crowley, K., Callanan, M. A., Tenenbaum, H. R., & Allen, E. (2001). Parents explain more often to boys than girls during shared scientific thinking. Psychological Science, 12, 258-261.

Picture by kakisky

Alicia Chang is EDUKWEST's Science & West Coast Editor. She is a cognitive and developmental psychologist (Ph.D., UCLA, 2008) with research interests in language and cognitive development, the effects of language and culture on cognition, and cognitive science applications to STEM education. She lives and works in Silicon Valley. You can follow her on Twitter @aliciac and Google+.

  • Anonymous

    My first thought was that you were doing a paper in Linguistics not education.

    In terms of the difference in math scores between Chinese and English speakers, I have seen some studies that show the difference can be partly explained by tones (and syllable length of words) which, it is suggested, make it easier to keep longer strings of numbers in short/long term memory.

    Very interesting observations about the Mommie effect by the way.

    With the one child policy in China an intersting thing may be happening– girls may be treated more like boys in terms of early education because mothers know that this will be their only child and they need to look beyond gender in the parenting process. If that were true then the math scores of females in China (vs. other countries) might show a greater variation than the scores between boys. Do you see anything like this?

    • Well, I can’t entirely explain the differences in scores, but I did find a difference between Mandarin and English number language input. But I don’t think it’s the tones there… but rather the classifiers: 個,隻, etc. There is a book called The Learning Gap by Stigler & Stevenson that talks more about the differences in math education between Asian countries and the U.S. that might be interesting to you. There are language, culture, and classroom differences for sure.

      However, there is generally not a ton of support that suggest boys and girls differ in math achievement in general, just differences in the rate with which they pursue STEM careers due to self-concept. So, to answer your question, there don’t appear to be any gender differences in math achievement in China (or the US)!

      • Anonymous

         Thanks for the reference!

        So what you are saying is that between genders in a given country there is no difference in math achievement. However, there is a difference in boys and girls choosing to pursue careers (?) (higher study?) in Stem. How could this be directly related to the conversations mothers are having with young children? If the pro STEM choices come much later in life aren’t a whole range of other variables much more important?

        By the way I am very interested in the impact that mothers have (because they can, in effect, act as tutors) on a child’s learning. I believe if mothers take the role of tutor seriously that they can supercharge a child’s education. I was just trying to see your results in this light.


        • Studies of gender differences in math scores get complicated. Sometimes a difference will be found, but only at the tail ends of the distribution, or with a tiny effect size that basically renders any difference statistically meaningless. But interestingly, if a nation tends to be more gender equal, any gaps in achievement tend to go away.

          I cannot conclude a causal relationship between language input and math achievement based on my own data, but other researchers have found that kids who hear more math and number speech do better in math later on. See http://blogs.kqed.org/mindshift/2012/03/why-its-important-to-talk-math-with-kids/ for more!

          If you read my whole paper, you’ll see that people tend to choose careers they feel they will succeed in, and girls’ confidence in math starts to wane very early. Again, I don’t think language tells the whole story. You are right that this is a multi-faceted issue, but yes, I do believe that mothers (and fathers, and teachers) make a difference.

          • Anonymous

            I went to the URL you suggested and boy, the author doesn’t pull any punches. She clearly states, “studies show that “number talk” at home is a key predictor of
            young children’s achievement in math once they get to school.”


            And now the $64 question. Does this interaction have to be “mommie/daddy talk” or can it be some kind of other less direct exposure? I can really see where you are coming from with the educational games if indeed this kind of exposure sans parent is sufficient to give kids the kind of input they need.

            Oh, when I first read this article I tried accessing your paper. As I am sure you are aware it is behind an academic pay wall. Having quick access to this kind of article is one of the perks of being in academia.


          • Alicia

            That’s a great question. I actually know several researchers who have found that “guided play” (i.e., direct social interactions with parents) actually leads to better learning that “free play” (a kid sitting by himself with blocks).

            I can email you the .pdf of the paper if you’d like! Let me know!

          • Anonymous

             I’d love to get it!

            redsable50 at yahoo.

          • Anonymous

            You wrote:
            On average, utterance lengths of transcripts depicting mother–son interactions contained 345.32 lines (SD = 421.27); lengths of mother–daughter transcripts averaged 473.44 lines (SD = 398.28). Lengths of transcripts did not differ significantly based on gender of child.

            Does this mean that there were an average of 130 more interactions between mother and daughter but these interactions took up the same space (on the printed page) as mother-son?

          • Alicia

            An utterance was defined as a phrase or a sentence, typically. The database has strict rules about how the transcripts are formatted.

            Not exactly. There were no significant differences in length of transcripts/interactions (look how huge those SDs are).

          • Anonymous

             Did you consider that either gender (particularly females) were intellectually ahead of the other gender at the age of two and that mothers were just responding to this discrepency?

          • Anonymous

             Did you consider that either gender (particularly females) were intellectually ahead of the other gender at the age of two and that mothers were just responding to this discrepency?

          • Alicia

            There isn’t any evidence to suggest that would be the case.

          • Anonymous

             Did you account for birth order in this study? Might that have an impact?

          • Alicia

            I didn’t have that information, but no, I don’t think so.