People are either Facebook users or they are not.
Facebook user data can be used to draw conclusions about general social phenomena.
According to Eric P.S. Baumer, an assistant professor of computer science and engineering who studies human-computer interaction, the simple statements above are, in fact, not so simple—nor are they true.
Baumer’s new research takes a fine-grained look at Facebook use and non-use, using a more nuanced approach than has previously been undertaken. Rather than employing a simple binary of two categories—use and non-use—Baumer’s study looks at demographic and socioeconomic factors that impact Facebook use and non-use using four categories:
● current user, who currently has and uses a Facebook account;
● deactivated, who has temporarily deactivated her/his account but could technically reactivate at any time;
● considered deactivating, who has considered deactivating her/his account but never actually done so; and
● never used, who has never had a Facebook account.
Baumer analyzed data collected by Cornell’s Survey Research Institute in 2015 for the Cornell National Social Survey. The data set includes responses from 1,000 U.S. households gleaned from a phone survey of adults 18 years or older. Through probabilistic modeling, Baumer sought to identify predictors for the four different types of Facebook use and non-use. He presented his findings at the 2018 ACM Conference on Human Factors in Computing Systems in Montreal, Canada on April 25th. The findings were also published in a paper titled “Socioeconomic Inequalities in the Non/use of Facebook.”
Of the factors he explored, eight emerged as predictors of use and non-use: age, gender, marital status, whether the respondent had looked for work in the past four weeks, household income, race, weight and social ideology (liberal to conservative). The strongest predictors, he found, were age, gender, whether the respondent had looked for work in the past four weeks and household income.
The research shows that current Facebook use is more common among respondents who are: middle aged (40 to 60), female, not seeking employment, of Asian descent, or currently married. Deactivation, either actual or considered, is more common among respondents who are younger, seeking employment, or not married. Respondents most likely to have never had an account are older, male, from a lower income household, racially of Black or African-American descent, more socially conservative, or weigh less.
Baumer says his findings suggest how socioeconomic factors might work in concert.
“My analysis reveals that individuals from lower-income households are less likely ever to have had a Facebook account,” says Baumer. “Yet, social networks have been shown to play an important role in fostering ‘social capital,’ which can be leveraged for accomplishing certain tasks, including securing employment. Also, respondents who had looked for work within the last four weeks were more likely to have deactivated their Facebook accounts—eliminating a potential resource in their job search.”
“Facebook, rather than acting as a democratizer,” writes Baumer, “may be perpetuating existing social inequalities.”
Other demographic results from the study include:
Age: Older respondents were more likely to have never had a Facebook account. Older respondents who did have an account were less likely to have deactivated or to have considered deactivating. For example, Baumer’s analysis reveals that every one-year increase in age increases the odds of never having had a Facebook account by 4.6 percent.
Younger respondents are more likely to have either deactivated or considered deactivating their Facebook account, while they are simultaneously less likely to be a current user. The probability of deactivation, considered or actual, drops as age increases, while the probability of never having had an account goes up.
“Rather than try Facebook and leave, older respondents never had an account in the first place,” writes Baumer in the paper.
Gender: Female respondents were 2.656 times more likely than male respondents to be a current user rather than never having used Facebook. However, gender did not significantly predict deactivation, either considered or actual. This result aligns with prior findings that social media use is more common among female respondents.
Employment: A respondent’s current employment status did not emerge as a significant predictor. However, the model does include whether the respondent looked for work in the past four weeks. Respondents who had looked for work were 2.030 times more likely to have deactivated their account and 2.276 times more likely to have considered deactivating.
Household Income: A respondent’s household income had no significant effect on deactivation, either considered or actual. However, respondents with lower household incomes were more likely to never have had a Facebook account.
Race: In the final model, only two racial categories have a significant impact, and each of those only significantly impacts a single type of use and/or non-use. First, respondents who identify as Asian are only 0.278 times as likely (i.e., 3.597 times less likely) to have considered deactivating their account. These respondents are also 0.238 times as likely (i.e., 4.202 times less likely) to have never had a Facebook account.
Second, respondents who identified as Black were more likely never to have had a Facebook account. This finding aligns with previous research suggesting Facebook use is relatively less common among African Americans.
Marital Status: Being married (as opposed to single, divorced, widowed, etc.) decreases the chance of considering deactivation and reduces the odds of actually deactivating almost by half.
Social Ideology: Self-identified conservative respondents were more likely never to have had a Facebook account. Each move toward the conservative end of the response scale corresponded to being 1.152 times more likely to have never had an account. Social ideology has only a slight impact on the probability of deactivation, either considered or actual.
Weight: Heavier respondents were less likely to have considered deactivating their account and to have never had an account. Lower-weight respondents are more likely only to consider deactivating, while higher-weight respondents are more likely to have actually deactivated.
His analysis, says Baumer, also provides specific details about the types of populations that researchers are, and are not, studying when analyzing data from social media.
“The analysis helps to explain the ways that Facebook, and likely all social media, are not representative of the broader population,” says Baumer. “Facebook users are more likely older, female, higher income earners, married and ideologically liberal.”