Is news consumption related to abortion attitudes? An exploratory study with a nationally representative sample of US adults

Abortion remains a contentious social issue, with sharply divided ideological camps: “pro-life” (broadly conceptualized as anti-abortion) and “pro-choice” (broadly conceptualized as pro-abortion) (Adams, 1997; Cates, 2012; Solon et al., 2022). Increased polarization regarding abortion has happened through decades of partisan rhetoric and divisions across party lines (Evans, 2003; Vox, 2019), which has been historically attributed to traditional media (e.g., newspapers, magazines, radio, television) (Carmines et al., 2010; Condit, 1990), and exacerbated by the mass proliferation of modern media (e.g., Internet, digital media, and social media) (Mitchell et al., 2014; Mouw and Sobel, 2001). Indeed, some argue that abortion attitudes have become divided through the reliance on partisan news—news that covers stories that are “framed, spun, and slanted so that certain political agendas are advanced” (Jamieson et al., 2007) and that offer opinionated and distinct point-of-view coverage (Levendusky, 2013)—to spread ideological views on abortion (Prior, 2013; Tucker et al., 2018).

Both content and sources of news have an impact on political polarization, and partisan news is especially polarizing (Kubin and Von Sikorski, 2021). People who report higher news consumption also prefer attitude-consistent information (Garrett, 2009a; Knobloch-Westerwick and Meng, 2009), or content that reflects personal views (Hart et al., 2009). This “selective exposure” to news (Garrett, 2009b; Goldman and Mutz, 2011) may result in people’s abortion attitudes becoming more extreme and cohesive, reflecting the news discourse they consume. Exposure to news that reflects only one perspective may also result in a process of attitude polarization (Gvirsman, 2014; Iyengar and Hahn, 2009; Stroud, 2010) leading to “echo chamber” biases that may reinforce incorrect, misleading, or negative beliefs about an issue (Tucker et al., 2018; Garrett et al., 2016; Nguyen and Vu, 2019). That is, if people only engage with news depicting one perspective, they are more likely to believe, think, and act based on this information.

Cultivation theory posits that as people consume news—especially for prolonged periods of time—their worldviews are more likely to reflect the messages and framing they see portrayed, influencing their attitudes and behaviors (Gerbner, 1969; Gerbner et al., 2002). Research on other social issues, such as policing, crime, and sexual permissiveness, suggests that the frequency and format of media consumed can influence people’s beliefs (Callanan and Rosenberger, 2015; Intravia et al., 2018; Rosenberger and Callanan, 2011). Although research examining abortion attitudes and news consumption is quite limited, based on findings regarding the role of exposure to entertainment media on abortion and abortion attitudes (Sisson et al., 2021; Brooks et al., 2022; Herold et al., 2024; Mulligan and Habel, 2011), it seems possible that the framing of news content may influence abortion attitudes. For example, Simon and Jerit (2007) explored the relationship between the terminology and language used during the public debate on “partial-birth abortions” (PBA). They found that the media’s word choice influenced respondents’ support for, or opposition to, legislation restricting PBA.

Trends in US adults’ attitudes toward legal abortion have remained relatively stable since the 1970s, with the percentage of people throughout the decades divided similarly in their endorsement of legal abortion. For instance, from the 2000s and beyond, about half of adults indicated that abortion should be legal under some circumstances, about a third legal under any circumstances, and about a fifth that it should be illegal under all circumstances. These differences may be indicative of the polarization and politicization of abortion, as well as the way abortion has come to be both a partisan and debated issue (Evans, 2003; Vox, 2019; DiMaggio et al., 1996). We also observe stability, as well as polarization, in terms of identification with socio-political terms associated with abortion (e.g., pro-choice, pro-life), as US adults are relatively split between identifying as either pro-life or pro-choice, and these trends remain relatively stable over time (Gallup, 2024). So, in this sense, we observe polarization in how abortion is discussed, presented, and addressed in partisan media and political rhetoric across political ideologies, and at the same time, stability in people’s endorsement of legal abortion.

Cultivation theory also posits that media presents and reinforces the status quo rather than challenges it (Gerbner et al., 2002). The status quo regarding abortion is the perception of polarization, politicization, and contentiousness. However, some have argued that people’s attitudes are nuanced, and common measures used to assess abortion attitudes may not be adequately capturing such nuance and instead reinforce the notion of polarization (Bruce, 2020; Hans and Kimberly, 2014; Jelen and Wilcox, 2003; Jozkowski et al., 2018; Jozkowski et al., 2021). Indeed, researchers have argued that common measures inherently create a dichotomy regarding abortion attitudes, potentially overinflating the extent to which attitudes are polarized. There is a need, therefore, to concurrently use measures that may allow for more nuance or complexity in abortion attitudes (Jozkowski et al., 2021).

While abortion—up to fetal viability—was constitutionally protected for nearly 50 years, the 2022 Dobbs v. Jackson Women’s Health Organization (i.e., Dobbs v. Jackson) decision returned authority over abortion to states. As a result, state lawmakers are now able to enact legislation that can further restrict abortion throughout pregnancy. With sweeping legislative changes to the abortion landscape, as well as a surge of news highlighting abortion (e.g., as a political issue; ballot initiatives), it is imperative to examine the association between news consumption—particularly through news format, leaning, and quantity—and abortion attitudes.

1.1 Current study

Understanding how abortion attitudes are formed and shaped, and specifically whether the news is an influential component in these processes, is important and has direct implications for health communication and promotion (Conti and Cahill, 2017), public and political knowledge regarding abortion (Simon and Jerit, 2007; Hunt et al., 2021), and sexual and reproductive health, more generally (Levine, 2011). Despite its relevance, limited research exists assessing the association between abortion attitudes and news, making this the first study—to our knowledge—to assess this relationship by examining frequency of news consumption by news format, news leaning, and news quantity. In this exploratory study, we hypothesize that abortion attitudes (i.e., both endorsement of legal abortion and complexity in attitudes toward legal abortion) vary depending on news consumption. We opted to measure abortion attitudes in these two ways to first capture abortion attitudes in the way most commonly assessed by social scientists (endorsement of legal abortion; see Adamczyk et al., 2020) as well as in a novel way, given calls for researchers to be more nuanced in assessing attitudes (e.g., Herold et al., 2024; Simon and Jerit, 2007). Given this context, we examined the following research questions:

1. Is frequency of engagement with different formats of news (e.g., print, radio, television, social media) associated with abortion attitudes?

2. Is greater usage of one leaning of news over another (e.g., liberal > conservative) associated with abortion attitudes?

3. Is the total sum of engagement with diverse news sources associated with abortion attitudes?

4. Is there an interaction between people’s abortion identity and (a) greater usage of one leaning of news or (b) the total sum of news consumption when measuring abortion attitudes?

Our survey was administered in September 2020 to IPSOS’ KnowledgePanel, a nationally representative panel of US adults weighted to match US benchmarks for gender, age, race/ethnicity, education, census region, household income, home ownership, metropolitan areas, and Hispanic origin (IPSOS, 2021). A total of 919 participants met eligibility and data quality criteria, including a minimum response time and successful completion of a reading check question. We excluded 33 participants from the analytic sample for not completing questions related to the variables of interest. Table 1 provides an overview of the demographic characteristics of the analytic sample (N = 886). This study was approved by the Institutional Review Board at Indiana University before data collection.

2.2 Measures

2.2.1 Abortion opinion scores

Attitudes toward abortion legality (i.e., endorsement of legal abortion) were measured via responses to six items modeled after the abortion items on the General Social Survey (GSS), a commonly used data source for assessing abortion attitudes (Adamczyk et al., 2020). The item stem reads: “Please tell me whether or not you think it should be possible for a pregnant woman to obtain a legal abortion, if…” followed by six different circumstances (see Table 2). Consistent with the GSS, response options were “Yes” (scored as 1) and “No” (scored as 0). Scores were summed to create an abortion opinions scale score ranging from 0 to 6, with higher scores indicating greater endorsement of abortion legality.

Table 2. Weighted percentage support for legal abortion by circumstance and abortion identity (N = 886).

2.2.2 Abortion complexity scores

We also measured complexity in people’s attitudes toward abortion legality (Jozkowski et al., 2018; Jozkowski et al., 2021), using a multi-step process. First, drawing on the six aforementioned GSS items, we coded “Yes” as 1 and “No” as – 1. Thus, participants’ possible sum scores ranged between negative six and positive six in increments of two. The further away from zero a score is, the more theoretically polarized and less complex the participants’ views regarding legal abortion. Thus, scores of −6 or + 6 would be the least complex, and a score of zero would indicate that a person’s view is minimally polarized and thus most complex. Second, we recoded polarized scores (+6 and − 6) to be 0, indicating participants were “not complex” regarding their stance on abortion. Alternatively, we recoded a score of 0 for abortion opinions as 3 for abortion complexity because this score represented the most conflicted set of responses available—indicating abortion should be legal and not legal in three circumstances each. Thus, abortion complexity scores ranged from 0 to 3, with higher scores suggesting more complexity or conflict in views regarding legal abortion.

2.2.3 News variables

We assessed the frequency of news consumption across both format (e.g., television, radio, etc.) and leaning (e.g., liberal, neutral, etc.). For format, we asked, “How often do you use the following media outlets to seek news about current events and social issues?” (Never, Less than once a month, Monthly, Weekly, Daily), considering the following news formats: “Television, Radio, Print newspaper or magazine, News on social media, Online news sources” (Pew Research Center, 2016).

For news consumption by leaning, we asked, “How often do you seek information about current events or the news from media outlets that are considered (1) Liberal, (2) Neutral, and (3) Conservative?” with response options: Never, Less than once a month, Monthly, Weekly, Daily. We provided news outlet examples for each category (Ad Fontes Media, 2023). To assess the frequency of news consumption relative to each leaning, we created two variables (i.e., conservative-to-liberal and conservative-to-neutral) representing the frequency of consuming neutral and liberal news relative to conservative by subtracting the amount of conservative news from the amount of liberal and neutral news. This process resulted in two variables ranging from −4 to 4. For each variable, negative scores represent consuming more conservative news and positive scores represent consuming more neutral or liberal news. We compared consumption of neutral and liberal news to conservative news because we also used other conservative groups as the reference group for other variables included in the analysis (e.g., the Bible is the word of God, Republican, Pro-life). We also include a measure of the total sum of news consumption across leanings by summing the news consumption of all three leanings. Supplementary Table S1 provides an overview of the news variables.

2.2.4 Demographic variables

We explored several demographic variables, including age, gender, race/ethnicity, political affiliation, Bible literalism, education, and abortion identity (see Table 1). We selected these variables based on previous research (Adamczyk et al., 2020).

2.3 Analysis

All analyses were conducted using Stata MP versions 16 and 17. First, we examined descriptive statistics for demographics and participants’ consumption of news across format and leaning. Model diagnostics showed no issues related to multicollinearity. The residuals from OLS models were not normally distributed. To account for this, we used Poisson and negative binomial models. We then used the countfit command in Stata to identify which model better fits the data (Long and Freese, 2014). This command suggested the use of a Poisson model for abortion opinions and a negative binomial model for abortion complexity. Given previous recommendations for interpreting results for nonlinear models with interaction effects (Long and Freese, 2014; Mize, 2019), we used the margins command in Stata to examine predictions and tests of marginal effects. We present the average marginal effects (AMEs)—rather than coefficients—which summarize the effect of the independent variables in terms of the models’ predictions. To determine if there is a significant interaction effect (i.e., if the effect of news consumption differs across abortion identity labels), we examined the second differences, or the difference in AMEs for each group (e.g., ∆
_pro-choice
– ∆
_pro-life
). For ease of interpretation, we also present the effects of the interaction terms graphically. Coefficient estimates for the models can be found in Supplementary Tables S2, S3.

Table 1 provides raw frequencies and percentages for the sample demographics after applying the weights. Mean abortion opinions scores were 3.77 (SD = 2.11), meaning that, on average, participants endorsed legal abortion in nearly four out of six circumstances; 39.45% indicated abortion should be legal in all six circumstances. Mean complexity scores were 1.11 (SD = 1.23). Nearly half of the sample (49.36%) were “not complex” (score of 0), while just over a fifth (21.15%) were most “complex” (score of 3). Table 2 presents weighted percentages of endorsement of legal abortion by each abortion circumstance and abortion identity.

3.2 Abortion opinions scores

Table 3 presents the average marginal effects (AMEs) for the Poisson regression model predicting abortion opinions scores by news format, news leaning, and total sum of news consumption, after controlling for demographic characteristics. We found that participants who believed the Bible is not the word of God (AME = 0.635, p p p p p p p p p 

Table 3. Average marginal effects from Poisson regression model predicting abortion opinions scores (N = 886).

With regard to the news variables, there were no significant effects of news format and news leaning, and barely a significant effect of total sum of news consumption (AME = 0.042, p Supplementary Figures S1–S3 for visual presentations). Consuming more liberal-leaning news relative to conservative-leaning news increases the probability of endorsing legal abortion for those who identify as pro-choice (AME = 0.097, p p p p p Supplementary Table S4 presents the second differences, or the difference in AMEs for each group, which suggest that the effect of news consumption (by leaning and total consumption) on endorsing legal abortion is significantly different across some abortion identity sub-groups.

3.3 Abortion complexity scores

Table 4 presents the AMEs for the negative binomial regression model predicting abortion complexity by news format, news leaning, and total sum of news consumption, controlling for demographic characteristics. We found that participants who identified as women (AME = −0.192, p p p 

Table 4. Average marginal effects from negative binomial regression model predicting abortion complexity scores (N = 886).

Although results suggested no significant effect of news format, higher consumption of liberal compared with conservative news (AME = −0.064, p p Supplementary Figures S4–S6 for visual presentations). Consuming more liberal-leaning news relative to conservative-leaning news decreases the probability of abortion complexity for participants who identify as pro-life (AME = −0.119, p p p Supplementary Table S5 presents the second differences, which suggest that the effect of news consumption (by leaning and total consumption) on having complex views regarding legal abortion is significantly different across some abortion identity sub-groups.

The news can both shape and reflect cultural norms, which in turn may influence people’s attitudes, health beliefs, and behaviors (Collins et al., 2003; Dhar et al., 2017; Farrar, 2006; Wakefield et al., 2010). In this exploratory study, we investigated the relationship between self-reported news consumption and abortion attitudes in a nationally representative sample of US adults. Findings are consistent with previous literature documenting that certain demographic characteristics—Bible literalism, political affiliation, educational attainment, and abortion identity—were significantly associated with abortion attitudes (Adamczyk et al., 2020; Bartkowski et al., 2012). However, our study makes the novel contribution of exploring how news format, news leaning, and total news consumption may be associated with endorsement of legal abortion and complexity in attitudes toward legal abortion.

Notably, we did not find a significant effect of news format (e.g., television, print, online sources) on abortion attitudes and complexity. This finding differs from research on other social issues that finds specific news formats (i.e., the internet and traditional media) predicted attitudes (Callanan and Rosenberger, 2015; Intravia et al., 2018; Rosenberger and Callanan, 2011). We did find some significant effects of news leaning and total news consumption on abortion attitudes and complexity, particularly moderated by abortion identity (Carmines et al., 2010; Rohlinger, 2015).

Regarding abortion opinions, overall, participants identifying as pro-life had the lowest endorsement of legal abortion compared with the other three abortion identity subgroups. Higher consumption of liberal news was associated with lower endorsement of legal abortion for pro-life participants, but higher consumption of neutral news had the opposite effect—greater endorsement of legal abortion. In direct contrast, higher consumption of liberal news was associated with an increase in endorsement of legal abortion for pro-choice participants, while higher consumption of neutral news did not have a significant effect. Consistent with previous research (Carmines et al., 2010), these results may indicate that pro-life and pro-choice individuals appear reactive to news consumption about abortion, particularly depending on the partisanship of the news they consume. Indeed, when exposed to news that presumably conflicts with pro-life individuals’ attitudes, they may be prone to increase their commitment to preexisting beliefs (Bail et al., 2018), whereas neutral news may end up resulting in attitude shifts.

Turning to abortion complexity, participants identifying as equally both pro-choice and pro-life had the highest abortion complexity scores compared with the other three abortion identity subgroups. Our results also suggest that abortion complexity decreased with higher total news consumption for all groups except for pro-life participants. These results may indicate that abortion complexity is related to news consumption and that the amount of news consumed is possibly most salient. It is possible that as people are exposed to more news, disregarding leaning, they gain more information about abortion, which may, in turn, lead them to have less conflicting views on the issue.

Collectively, our findings suggest that news may have a different effect depending on the leaning of the news and the news consumer. Such findings underscore previous research demonstrating differential effects of information on people’s attitudes across abortion identity sub-groups (Crawford et al., 2021). Perhaps “selective exposure” to news (Garrett, 2009b; Goldman and Mutz, 2011), news content (Sisson et al., 2021), and news framing (Simon and Jerit, 2007) influence attitude formation, but in varied ways based on people’s abortion beliefs (Gvirsman, 2014; Iyengar and Hahn, 2009; Stroud, 2010). It may be that how abortion is framed, including the language used to describe abortion by news outlets of particular political leanings (Ball-Rokeach et al., 1990; Nixon et al., 2017; Andsager, 2000; Blaeser, 2023; Woodruff, 2019), may play an important role in attitude formation, disregarding the content or accuracy of information (Kann and Tulbert, 2019).

Importantly, to measure news leaning, we asked participants how often they sought information about current events or the news from media outlets considered liberal, neutral, and conservative (with examples of outlets in each category). While some research has found that people may have skewed perceptions of media bias (Coe et al., 2008; Feldman, 2011; Perloff, 2017), particularly when they conflict with their views (Vallone et al., 1985), other research has found that people generally selectively expose themselves to belief-confirming information (Hart et al., 2009; Knobloch-Westerwick, 2014) that further reinforces their understanding of news bias. Experimental research has found that participants attach certain meanings and values to news sources and may have an accurate understanding of the political orientation of these media outlets (Iyengar and Hahn, 2009; Baum and Gussin, 2008; Turner, 2007). A few polls have also found that a plurality of people correctly associate conservative and liberal news outlets (Pew Research Center, 2009). Nevertheless, people’s ability to correctly identify the leaning of news and media is influenced by their political sophistication and media literacy (Iyengar and Hahn, 2009; Prior, 2007; Tully and Vraga, 2018). In light of this, more research and polling are needed to assess people’s awareness, knowledge, and perceptions of the political leanings of news sources to refine survey instruments to best measure news consumption by leaning.

Greater knowledge about abortion has long been associated with more favorable attitudes toward legal abortion (Crawford et al., 2021; Esposito and Basow, 1995). In our study, people who relied solely on partisan news may have seen abortion content more frequently framed along partisan lines (e.g., against abortion to protect the “sanctity of human life” (Lockhart et al., 2023; Williams, 2011; The White House, 2021)] vs. in favor of choice because it is “her body, her choice” (Wicclair, 1981; Ludlow, 2008; MSNBC, 2022)). Researchers should more deeply explore the relationship between abortion knowledge, abortion attitudes, and news consumption. Moreover, more in-depth research that explores how content and framing may work together or in opposition to influence attitudes could be warranted.

Finally, our findings regarding greater news consumption or engagement with news generally suggest that attitudes toward abortion might not be influenced solely by the leaning of news consumed but also in combination with the quantity of its consumption and mediated by abortion identity. Although our study did not assess participants’ length of engagement with different news, cultivation theory argues that prolonged engagement is associated with attitudinal differences (Gerbner et al., 2002). It is possible that participants who reported engaging almost exclusively with partisan news, either conservative or liberal, have more prolonged engagement with these sources, which may influence their attitudes to a greater extent. This could also be explained by echo chambers, where people gravitate toward news that mirrors their preferences, validating their “extreme” beliefs (Cinelli et al., 2021).

4.1 Implications for practice and/or policy

Our findings suggest that US adults’ attitudes toward abortion may be differentially influenced by partisan news and total news consumption across abortion identity. Thus, it may be useful for advocates to employ different tactics geared at different sub-groups when working to change people’s attitudes toward abortion. For example, interventions in media literacy could be instrumental in helping people critically evaluate the content and framing of news on abortion. Our findings suggest that consumption of news sources with diverse political leanings and consumption of more news generally may offset efforts by partisan media and political groups to polarize or increase the portrayal of polarized attitudes (Prior, 2013; Tucker et al., 2018; Rohlinger, 2006). Initiatives to promote media diversity and reach audiences through diverse outlets may help spread balanced and accurate information so that people can develop more informed views on abortion. Since news can influence health beliefs and behaviors (Wakefield et al., 2010; Escobar-Chaves et al., 2005), there is a need for advocates to encourage evidence-based public health efforts and provide accurate abortion information to the general public.

4.2 Limitations

Although our study provides unique insights into the relationship between news consumption and abortion attitudes, particularly through the intersection of news format, leaning, and quantity, we would like to note some important limitations. First, we relied on self-reported news consumption, subject to recall bias. Second, we did not measure participants’ length of engagement with different news, which, according to cultivation theory, is associated with attitudinal differences. Third, as discussed earlier, there is a possibility that some participants, particularly those less politically engaged and with lower media literacy, may not have understood the political leanings of specific news platforms not listed as examples in the survey questions. Finally, we collected our data in September 2020; there have been substantial changes in the abortion legislative landscape since then, which may trickle down to abortion attitudes and news consumption, particularly after the 2022 Dobbs v. Jackson decision. Thus, we caution that these results might not be generalizable given the changing context of abortion access, abortion attitudes, and abortion news. That said, some evidence suggests that the Dobbs v. Jackson decision did not significantly influence people’s attitudes toward abortion (Jozkowski et al., 2023a).

Our findings demonstrate that people’s abortion attitudes vary depending on their consumption of partisan news, total amount of news, and abortion identification. We found that news format was not associated with attitudes toward legal abortion nor complexity in abortion attitudes. However, engagement with partisan news sources, as well as with more news generally, were associated with abortion attitudes through the moderating effect of abortion identity. Given these findings, coupled with various other sources that have documented these differential effects in attitudes based on abortion identity (Crawford et al., 2021; Jozkowski et al., 2023b), there is a need for advocates to address polarization in the post-Dobbs v. Jackson context to empower people to make reproductive health decisions and access quality care.

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The Supplementary material for this article can be found online at: