Liberellas Versus Konservatives: Social Status, Ideology, and Birth Names in the United States

The term “cultural capital” has a wide range of usages and meanings. We use it as defined by Lamont and Lareau (1988) as “widely shared, high status cultural signals (attitudes, preferences, formal knowledge, behaviors, goods and credentials) used for social and cultural exclusion.” (p.156).

The term ideology has a host of connotations from loose systems of belief to myth to even “consumerism” itself. For the purposes of this paper, ideology is conceptualized as set of logically coherent ideas about the proper ordering of collective activity that constrain opinion and dictate specific courses of action (Mullins 1972).

For purposes of expedience, this paper will examine only ideology as it expressed in a liberal-conservative continuum where attitudes are organized primarily by concerns with the scope of government in the distribution of resources and the advancement of traditional and moralistic concerns (Kinder 1983, Gerring 1997). Many smaller ideologies, such as environmentalism, socialism, or certain religiously inspired ideologies have explicit conceptions of appropriate consumption practices.

One can, for instance, categorize names relative to Biblical or nature references, emotions or specific values, ethno-nationalist sentiments, culture (both highbrow and low), historical figures, etc.

Because of inconsistencies and inaccuracies in the birth record data, particularly in the annotation of the addresses, we were only able to identify locations for 478,355 of the 545,018 recorded births. However, there are no significant differences in the distribution of variables between these two samples. Thus we have no reason to think that these errors are systematically distributed or that these missing cases are distorting our findings.

For a full description of the database, see http://​swdb.​berkeley.​edu/​index.​html.

Because of variations in spellings in common sounding names (e.g., Madison, Madyson, Madisun, Madysyn, etc.) these variables somewhat overstate the phonetic variations in names.

There are other vowel sounds that are more predominant in Boys’ names, like UH and AW, but these have a very low frequency in names.

White includes mothers whose race was coded as white but whose ethnicity was listed as “American,” “Caucasian,” or “White” but not any other nationality. Latina includes those mothers whose race is coded as white and who identify a Hispanic ethnicity (e.g., Mexican, Puerto Rican, Chicano, etc.). Excluded from this list are 11,199 mothers who are identified as white but provide some other ethnicity (e.g., Armenian, Russian, Swedish). This measure was meant to exclude of European extraction who may be immigrants or come from non-English languages and thus have culturally idiosyncratic naming practices. This “ethnic white” category included 7894 birth mothers or roughly 1.4 percent of the sample. Excluded from the sample are also 2972 birth mothers whose race was identified as American Indian. There were 18,485 records for which there was either no recorded race/ethnicity or that did not fit within these four categories and 15,152 cases with no education level recorded. Because the overwhelming majority of cases missing data on education were also missing data on race, this left only 19,161 cases with no data. In total, there are 26,780 cases excluded because of missing data or because of ethnicity or race that does not fit easily within the above four categories.

Proposition 61 was a statewide $750 million bond measure to provide funding for children’s hospitals that passed with 58 percent of the vote; Proposition 63 increased the tax rate on residents making over $1 million a year by 1 percent in order to fund county mental health services and it passed with 53 percent of the vote; Proposition 71 made stem cell research a constitutional right and authorized the sale of $3 billion worth of general obligation bonds to subsidize stem cell research and passed with 59 percent of the vote.

Precincts with ideology scores above .8 or below .25 typically have less than 30 voters.

The data provided by the state mistakenly identified mothers with Chinese names as Vietnamese and grouped these two nationalities together. In addition, the tremendous differences between Korean, Japanese, Filipino, and other nationalities cast doubt on whether Asian is a meaningful unifying category. Because the relative proportions of each of these nationalities is so much smaller than other groups and because each is subject to distinct cultural and linguistic idiosyncrasies in naming practices, they are excluded from further analysis.

The dividing points for median household income, percent with a college degree, and ideology scales conform largely to the quintile and sextile distributions in the entire sample. This leaves a few skewed distributions, i.e., there are a very small number of blacks in the most conservative voting precincts.

A Moran’s I test for spatial autocorrelation for the models for each racial group and child gender failed to yield a single significant coefficient. The results from the Moran’s I test are listed in the Appendix.

The dependent variable is comprised of three categories: popular name, uncommon name, and all other names. The excluded category in the regression tables are other names.

A reviewer cautioned that our socioeconomics controls might be capturing the separate effects of single mothers, who are likely subject to differing naming imperatives given the absence of a father in the naming process. Accordingly, this might induce new partial correlations between ideology and baby names. To test this possibility, we modeled the probability that a child’s birth record would omit a father’s name (an indicator for single mothers) as a function socioeconomics and ideology. As expected, there were very large socioeconomic effects, but no ideological differences.

These results are largely a function of the relatively low number of popular names chosen by black mothers for their girls.

Because of differences in distributions of populations among whites, Latinos, and blacks, different model specifications were used to measure racial and ethnic composition in the logistic equations. For Latinos, a measure of English speaking among households in the census tract was included to differentiate between ethnic and linguistic effects that might be compounded in the measure of percent Latino. Because of their relatively lower population sizes and high levels of segregation, the racial composition of the percent black in the neighborhood was measured with a quadratic term.

Because there are so few observations from the small number of precincts with ideology scores below .3 or above .7, the confidence intervals at these poles from the GAM models become so large and the results are uninterpretable. Consequently, we have truncated the illustration to the parts of the ideological scale where the GAM model can derive meaningful statistics.

Given available data, we are only able to measure mothers’ socioeconomics and ideology at the district level. This obviously induces challenges of dependence between units within common districts. To address this, the GAM reported in the paper cluster standard errors at the electoral district level. We have separately estimated a redundant set of hierarchical GAM with random effects at the district level; however, the negligible differences between the estimated coefficients suggested we should retain the GAM with clustered errors, given their ease of interpretation.

We created a variable for the top twenty Biblical names for both boys and girls and used this as a dependent variable with the same model specifications used in Table 3. In no instance did we find a statistically significant relationship between ideology and the use of a Biblical name. To be more exhaustive, we then constructed a variable which includes all biblical names recorded in Wikipedia (which includes almost 2600 unique names). This test also showed no ideological relationship. Similarly, a reviewer suggested that a relationship between ideology and baby naming patterns among Hispanic mothers might reflect their desire to use an Aztec/Mayan name. After exploiting publicly available names of Aztec names, we separately modeled the choice of these names as a function of ideology, without finding any relationship.

At a reviewer’s suggestions, we tested the possibility that the effect of ideology would interact with mothers’ reported ethnicity. These interactions were consistently insignificant, and trivially small, indicating that a single ideological coefficient is an adequate representation of the population relationship.

Some of these educational differences may also be the consequence of higher illiteracy thus generating names that are spelled phonetically relative to variations in speech patterns (e.g., “Keef,” “Jazzmin” etc.).

To calculate the Male Gender score, we first estimated a proportion score: the number of times that name was given to a boy born in California during 2004, divided by the total number of times a name was awarded. Names like Isabella and Natalie, for instance, were chosen more than 2000 times each in 2004, but never awarded to boys, so their proportion is equal to 0. A popular androgynous name is Alexis, which was awarded 2751 times, with 1248 recipients being boys—giving this name a ratio of .453. Next, the regularity of the phonemes in each name is calculated to provide a gender-phoneme ratio score for that phoneme. This is similar (although oppositely scored) to the gender ratio of phonemes illustrated in Table 5 in the Appendix. For instance, AH0 (the middle phoneme in “hut”) is predominant in girls’ names and has a gender-phoneme ratio score of .28; AW0 (the middle phoneme in “cow”), a very uncommon phoneme, appears almost exclusively in boys’ names, and accordingly has a ratio score of .99. For each name, we then take the mean of all the gender-phoneme ratio scores in each name, and rescale it from 0 to 1 for a final Male Gender Score for each name.

When looking at specific phonemes, we also considered their specific position in the name.

The effects of neighborhood ideology on phoneme choice among less educated mothers are usually insignificant, as we would predict.