I do not have access to an ungated version of the study and so am reliant on the representation of the study in Jacobs’ article. I took an empirical look at some of the inferred predicate assumptions necessary to Jacobs’ argument that “stereotypes plague children’s picture books.” The implication is that there is something amiss.
The necessary predicate assumptions to support Jacobs’ conclusion that “If children, especially girls, continue to be exposed to portrayals that suggest opportunities for women are limited to the home, and that men provide, their aspirations and independence will be muted” seem to include:
1) Representations in children’s books have some predictive capacity of life outcomes (ex. gender roles as represented in children’s books are predictive of future gender roles).Each of these assumptions can be tested for its veracity. Two of the four are false and two are unproven.
2) The structure of the study was sufficiently robust to reveal useful information about representations in children’s picture books.
3) Gender roles as represented in children’s books are unrepresentative of contemporary reality.
4) Traditional gender roles as represented in children’s picture books are potentially detrimental, particularly to girls.
1 – Assumption that representation is predictive. As far as I am aware, there are no studies showing that representation in children’s books has any predictive capacity on future life outcomes. This makes sense when you consider the multiplicity of influences on a child (family personalities, neighborhood, school, church, friend network, TV, Radio, language, vocabulary, culture, family structure, etc.); the causal density of outcomes (many links in a chain, each link with different causative strengths) and; the difficulty of disentangling cause and effect and degree of contribution of each variable on outcome. It is a logical assumption that IF a child reads a fair amount, then that might have some contribution to their knowledge, values, assumptions and experiences, but it is also logical that the degree of that reading contribution is likely swamped by all other sources of experience and knowledge. It makes sense that a book could have some impact, and indeed is likely to have an impact where it is consonant with the individual child’s proclivities and consistent with all the other sources of experience and knowledge. I think the best formulation we can get to is that the content of books are contributive to life outcomes but there is no empirical evidence that they are determinative or even predictive.
For arguments sake, accepting that representation is prolog, there is an inherent contradiction in Jacobs' conclusion that “If children, especially girls, continue to be exposed to portrayals that suggest opportunities for women are limited to the home, and that men provide, their aspirations and independence will be muted.” Let us accept that children’s books portray men and women with a strong (85%) skew towards men as providers and a similar skew of women as nurturers, and let’s also accept that books have some sort of demonstrable impact on life outcomes. If stereotype is prolog, there could have been no revolution in gender status in the past fifty years (since the study indicates there has been no change in portrayal). However, it is incontrovertible that men and women have many more options open to them (even if they don’t pursue them in equal numbers) than pre-1965. Men can be and are nurses and care-givers, women can be and are executives and soldiers. The logical conclusion is that stereotypes are not determinative of outcomes, probably either because the stereotyping is not perceived or because it is perceived but it has no impact. Contra Jacobs, it would appear that exposing girls to stereotypical gender roles does not mute their independence.
2 – Assumption that the structure of the study was robust. This is clearly not the case. An interesting study, but not robust. There are four causes for concern. First is the tiny sample. 300 books spanning a hundred years of children’s publishing. 300 wouldn’t be sufficient to be representative of even last year’s torrent of new titles. With some 25-35,000 new titles a year, a statistical sampling of 300 couldn’t tell you much at all, particularly when you are interested in so many subordinate issues (working, breadwinner, nurturing, care-giving, etc.).
The second concern is that this isn’t a random sample at all, even though “DeWitt and her colleagues analyzed a random sample of 300”. In fact, it is a random sample of a preselected sample: ‘“selected by an advisory committee of distinguished librarians” and which is “used to aid school and community libraries in selecting quality books”’. I don’t know what criteria for quality the distinguished librarians used in selecting their 1,400 books but I can guarantee it wasn’t random, wasn’t representative of all books published, and wasn’t representative of all books read by children. This would be like going to Mortimer Adler’s The Great Books of the Western World (containing some 350 selections), randomly picking 75 of the works and then assuming those 75 were representative of what people are reading.
The third concern is that there is the Pareto distribution problem (20% of X causes 80% of Y – ex. 20% of criminals are responsible for 80% of crime). Not all titles are created equal. One title may be read by 350,000 people and another by 5,000. You can’t average the two and assume the results are meaningful. If you are interested in stereotype propagation, you have to look at what people are actually reading rather than a sample of a selection. If you want to know what stereotypes children are being exposed to, you ought to be looking at the 300 bestselling children’s books of the last century, not a selection presumably based on critical acclaim and pedagogical value (which conceivably might be read by very few children).
The fourth and final concern is the common issue in the social sciences of bias and subjectivity. The researchers had to examine the three hundred books and mark behaviors as nurturing, breadwinning, care-giving, etc. These are nuanced concepts, texts are ambiguous, and there were multiple markers. The probability of different markers repeating the exact same exercise and arriving at roughly the same database of answers is low (an example of interpretive nuance is the frequency with which a book club will read a given book and its members arrive at differing interpretations of key events and episodes). The great majority of sociology papers end up with results either being withdrawn or the results are unreplicated by other researchers. All of which is to say that extreme caution has to be exercised in placing any confidence in the results of this study. It is an interesting topic and the study gives some interesting insights but the confidence level in those insights is perforce low. It is not that the conclusions of the study are necessarily wrong, it is that we can’t have confidence that they are right. See Seven Deadly Sins of Contemporary Quantitative Political Analysis by Philip A. Schrodt for a humorous but harsh take on current analytical practices.
3 – Assumption that the gender roles as represented in children’s books are unrepresentative of contemporary reality. Set aside all the above mentioned concerns. The single data point in the article is the finding: '“Fathers, on the other hand, were “much more likely than mothers to participate in both physical and non-physical play.” And they were much more likely to be portrayed as breadwinners: 26.6 percent of fathers worked outside the home, compared to 5.6 percent of mothers.”' This is a verifiable statement. In children’s books, up to 32.2% of parents work outside the home (26.6 + 5.6). This compares to 96.3% of parents (one or both parents) in the real world (Bureau of Labor Statistics, Table 4, Families with own children under 18). So in children’s world, most people don’t work (or aren’t seen to be working) whereas in the real world most everyone works. That’s a pretty big reality gap and can be validated as unrepresentative. If representation is prolog, we should fear for our economic future.
In terms of gender stereotype, in children’s books females are the (or a, it is unclear from the article), breadwinner 17.4% of the time (5.6/32.2) and males are the breadwinner 82.6% of the time (26.6/32.2), a wide and, according to the study, a stable and persisting gap. How does it compare to the real world? In the real world of those families where there is only one breadwinner (37% of all married families with children, BLS again), 82.7% of the time it is the male who is the sole breadwinner and 17.3% of the time the female is the sole breadwinner. Virtually identical to children’s world. What about the 59% of families where both father and mother work? According to Pew Research, in households where there are two breadwinners, females are the primary breadwinner in 15% of the cases. 15% is again pretty close to 17.4%.
The net is that if we are speaking of families with two parents and children, the picture portrayed in children’s books is apparently pretty accurate in terms of empirical observation. Whether that breakdown in gender roles is desirable or optimum is a subjective, normative and prescriptive question which can’t really be answered. Lots of legitimate argument can be made. All we can say is that what is reflected in children’s books appears to be consistent with the choices people make in real life (as measured by the BLS).
What are the empirical realities for the rest of the people not covered by the category Married-couple families (in BLS speak)? Married-couple families are 67% of all families. 33% of all families with children are headed by either a single mother (75% of the time) or a single father (25% of the time). And here’s where it gets really complicated – different groups by class and culture have dramatically different levels of single parenthood and the impact of single parenthood is dramatically different given other variables such as educational attainment. How does that figure into the analysis? There isn’t a simple fit nor is it clear from the article how these issues were dealt with in the study.
Assumption 4 - Traditional gender roles as represented in children’s picture books are potentially detrimental, particularly to girls. Again a normative, subjective and prescriptive question which can’t be answered directly. What can be affirmed is that families following a traditional life arc (education attainment, full-time employment, then marriage, then children) are disproportionately over-represented among the top two income quintiles and are virtually absent from the bottom income quintile (see BLS as well as the data in Charles Murray’s recent Coming Apart). For those following the traditional sequence of life milestones, fewer than 2% are in poverty (BLS). (I am using income quintile as a general marker for desirable life outcomes such as health, education attainment, longevity, morbidity, etc., recognizing that the correlation is very high but not perfect).
No other combination of variables is near as predictive of desirable life outcomes (variables such as single parent male, single parent female, children before marriage, children before school completion, etc.). Since we are talking about statistical averages it should be clear that there are multiple individual exceptions to every category. There are single female breadwinners with children in the top quintile of income; just not that many – it isn’t that it is impossible, it is that it is significantly less probable. So while the traditional stereotype of gender roles may be prevalent in children’s literature, it is also a stereotype associated with generally desired life outcomes (as measured by the proxy of income quintiles). It is very Anna Karenina, “All happy families are alike; each unhappy family is unhappy in its own way.”
In summary,
Assumption 1 – Representation is predictive - No empirical evidence to support the assumption. Causal density of desirable life outcomes makes it unlikely to be predictively true.Jacobs’ conclusion is “If children, especially girls, continue to be exposed to portrayals that suggest opportunities for women are limited to the home, and that men provide, their aspirations and independence will be muted.” To reach that conclusion he has to make four critical assumptions, two of these assumptions are wrong and two cannot be affirmed therefore his conclusion cannot be affirmed either.
Assumption 2 – Structure of the study was robust – False. Study was interesting but unrepresentative and therefore tenuous. High uncertainty regarding the results.
Assumption 3 – Gender roles as represented in children’s books are unrepresentative of contemporary reality – False. Gender roles appear to be reflective of reality (as measured by BLS).
Assumption 4 – Traditional gender roles are detrimental – No empirical evidence to support the assumption. Fundamentally, it is a normative position. Empirical data establishes greater association between desirable life outcomes and traditional life practices but it is a predictive relationship not a determinative relationship. In other words, if you follow traditional decision-making you are more likely to end up in higher income quintiles, but you are not guaranteed to do so. Correspondingly, if you make life decisions at great variance of the traditional norms you are more likely to end up in the lower income quintiles but are not guaranteed of doing so. The extent to which traditional gender roles are detrimental is solely based on their alignment or misalignment with individual goals and desires which cannot be determined a priori.
Posts
No comments:
Post a Comment