An Empirical Analysis of the Gender Gap in News Consumption

Abstract

Survey results reveal that women consume less news than men. This article empirically analyzes this gender gap and explores several explanations. In the United States, the gender gap cannot be explained by differences in education, income, and other socio-demographics or by differences in preferences and job benefits of news consumption. However, the dual burden of paid and household work appears to be one of the drivers of the gender gap. In a cross-country comparison, the gender gap is linked to measures of gender equality in the economy and in politics.

Notes

¹For a general description of the data, see European Social Survey (2010).

²Of course, for an individual it can be completely rational to spend little time on political news consumption because the probability of being pivotal in an election and, hence, the benefits of making an informed voting decision are extremely small (see the discussion in the next section).

³For a survey, see Prat and Strömberg (2011) and, for a similar discussion, Benesch (2010).

⁴There exists a larger literature in communication studies on the motives and motivations for media consumption. Closest to the economic perspective is the uses-and-gratifications approach. The studies in this tradition analyze the different functions media consumption can fulfill under the assumption that individuals use the media in an active and goal-directed manner to satisfy their needs and desires and spend their time and mental resources on those sources of gratification (i.e., media outlets or programs) where they expect the most benefit (e.g., Rubin 2002).

⁵Here, media consumption refers to the amount and type of media people consume. A different aspect of media consumption is how consumers absorb and react to a given media content. Regarding the latter aspect, a gender gap can exist as well (e.g., Kern & Just, 1997).

⁶I reversed the original categories in order for higher values to reflect higher attention to hard news.

⁷One concern with the use of survey data is that respondents' answers may be (systematically) biased. In the framework of this study, it would be of particular concern if women underreported or men overreported their news consumption. Although such a response behavior cannot completely be ruled out, the result that women follow several news categories to a larger extent than men do indicates that the survey measures are not just the result of general differences in men's and women's response behaviors.

⁸It might be possible that women, when asked this question, automatically think of those types of news they consume more often (e.g., news related to religion, local affairs, and celebrities), and not of news related to politics.

*p < .10.

**p < .05.

***p < .01.

⁹For a similar empirical strategy, see Fryer and Levitt (2010) on the gender gap in mathematics.

*p < .10.

**p < .05.

***p < .01.

¹⁰In these regressions, the number of observations is smaller because only about one-half of the entire sample were asked the question on time for the news.

*p < .10.

**p < .05.

***p < .01.

*p < .10.

**p < .05.

***p < .01.

¹¹For the World Values Survey data, see European Values Study Group and World Values Survey Association (2006). For the European Social Survey (ESS) data, see Jowell et al. (2003, 2005, 2007) and European Social Survey (2010). The ESS data are archived and distributed by the Norwegian Social Science Data Services.

¹²Not all countries are surveyed in each wave.

¹³This assumption is arbitrary to some extent. The results remain robust with different values for the top categories.

¹⁴In both samples, as in the U.S. sample, there is a statistically significant gender gap in news consumption. This gap remains significant when controlling for a large set of socio-demographic characteristics (results not shown).

¹⁵The data on gross domestic product per capita is not available for three countries in the World Values Survey (WVS) sample (Zimbabwe, Puerto Rico, and Serbia/Montenegro), the Polity IV data is not available for six countries in the WVS sample (Hong Kong, Iceland, Luxembourg, Malta, Puerto Rico, and Serbia/Montenegro) and two countries in the European Social Survey sample (Iceland and Luxembourg), and the press freedom data is not available for one country in the WVS sample (Puerto Rico).

¹⁶Control variables are age, household income, level of education, employment status, marital status, type or size of place of living; and, in addition to these variables in the European Social Survey, working hours, household size, whether children are living in the household, and whether the respondent is a citizen of the country.

¹⁷Malta, the only country where the share of women who regularly watch the news is higher than the share of men, influences the statistical significance to some extent. Without Malta, the relation between gender equality and the gender gap in news consumption in column 2 would be statistically significant at conventional levels (result not shown).

¹⁸The regions are Western Europe, Central and Eastern Europe, Scandinavia, Africa, Middle East, Asia Pacific, North America, and Latin America.

¹⁹Malta is the country that influences the statistical significance of the results, but not so much that the size of the coefficient (see footnote 18) drops out as well.

²⁰The World Economic Forum's gender equality index is comprised of four sub-indexes: gender equality in politics, the economy, in health, and in education. The latter two sub-indexes do not vary to a large extent in the World Values Survey sample.

²¹In countries with lower economic gender equality, the share of a wife's time spent on housework as compared to a husband's time is larger than in countries with higher economic gender equality. In fact, the inclusion of the female share of housework in Regression 9 in Table 5 leads to a statistically insignificant coefficient of economic gender equality (results not shown). The data on wives' and husbands' time spent on housework is from Knudsen and Wærness (2008) and is available for 27 countries in the World Values Survey.

*p < .10.

**p < .05.

***p < .01.

*p < .10.

**p < .05.

***p < .01.

***p < .01.