Analyzing the Overturn of Roe v. Wade: A Term Co-Occurrence Network Analysis of YouTube Comments

Abstract

Objective: This study examines YouTube comments following the overturn of Roe v. Wade, investigating how perceptions of health implications differ based on commenters’ gender and abortion stance. Methods: Using Netlytic, 25,730 comments were extracted from YouTube videos discussing the overturn of Roe v. Wade, half of which featured physicians discussing public health implications. Manual coding of 21% of the comments identified discussions on abortion stance and medical implications, while Gender API approximated the commenters’ gender. A term co-occurrence network was generated with VOSviewer to visualize key terms and their interrelations. Custom overlays explored patterns related to gender, abortion views, and medical implications, and comparisons within these overlays intersected with the medical implications overlay to illustrate contextual differences across demographics. Results: Four clusters emerged in the network: Constitutional Law, addressing the U.S. Constitution’s interpretation and legal impacts; Reproductive Rights and Responsibility, discussing alternatives to abortion and access; Human Development, exploring the intersection of abortion laws and individual beliefs; and Religious Beliefs, linking abortion laws to faith. Prochoice users focused on medical and socioeconomic impacts on women, whereas prolife users emphasized the prevention of unwanted pregnancies and moral considerations. Gender analysis revealed males centered on constitutional issues, while females highlighted medical and personal effects. Conclusion: The findings underscore that monitoring YouTube discourse offers valuable insights into public responses to shifts in health policy.

1. Introduction

In May 2022, a draft of the United States Supreme Court’s majority opinion in Dobbs v. Jackson Women’s Health Organization (Dobbs) was leaked to the public, revealing the Supreme Court’s intention to overturn Roe v. Wade on 24 June 2022. The Supreme Court formally unveiled its ruling on Dobbs and overturned the established precedents of Roe v. Wade and Planned Parenthood v. Casey, which safeguarded a pregnant person’s right to abortion. Dobbs now permits states to independently regulate access to abortion services. The medical literature has stressed the significance of the relationship between rates of infant and maternal mortality and restrictive access to abortion care [1,2]. Understanding these connections highlights the implications of such legal decisions on public health and maternal well-being.

The theory of rational ignorance is prominent in the social sciences for examining and explaining public response to policy changes [3,4]. Rational ignorance suggests that voters refrain from gathering information when the cost of education outweighs the perceived benefits [4]. In the context of Dobbs, many Americans may not fully understand the complex legal and health implications of this decision. This deficiency in voter comprehension influences voting patterns and may not truly represent their best interests [3,4,5]. The overturn of Roe v. Wade has adverse implications for social determinants of health for women and children [6].

The Supreme Court’s decision was complex, and the general public has an incomplete understanding of the issues at stake [1,2]. Following Dobbs, numerous states enacted laws aimed at safeguarding abortion rights. However, this decision also sparked the reinforcement of existing regulations and the inception of new ones that either prohibit or limit abortion [7,8,9]. Presently, 35 states have legally restricted abortion access, resulting in a national landscape for abortion care that is uncertain, inconsistent among states, and subject to frequent changes [8].

Numerous surveys examining perceptions of Roe v. Wade have been conducted [9]. Yet, the response to Dobbs v. Jackson has predominantly surfaced across various social media platforms [7,8,10]. In the wake of the overturn of Roe v. Wade, there was a marked increase in social media engagement debating the merits of the Supreme Court’s decision to limit abortion services [7,8]. Abortion is primarily depicted as a gendered and moral issue on social media [7,8,10,11,12].

This study contributes to growing research that applies the theory of rational ignorance to explore how individuals avoid acquiring complete knowledge on complex issues, such as reproductive rights [3,4,5]. As social media increasingly serves as a platform for political expression and public discourse, particularly around health-related issues, it is crucial for health educators, policymakers, and communication professionals to understand how political learning and advocacy unfold in these digital spaces [7]. Within this context, the theory of rational ignorance social media suggests that social media users, particularly those with strong ideological views, may choose simplified information [4,5], emotionally resonant arguments, and relatable personal narratives [7,8,11], presenting limited interest in exploring the complex medical or legal implications of abortion and reproductive rights [13].

Our study repurposed a bibliometric software to study the semantic structure of YouTube commentary to investigate the aftermath of overturning Roe v. Wade in the United States. The combined use of YouTube and VOSviewer will allow us to study the association of specific variables with the discourse’s context, nature, and tone through visual overlays [14]. Due to rational ignorance, individuals might base their views more on ideological alignment [4] rather than a thorough understanding of the health implications of the overturn, prompting this exploration through social media. The primary goal was to understand the broader public discourse surrounding this event, focusing on the diversity of opinions, sentiments, and medical understanding present among members of the YouTube community.

Abortion is a multifaceted issue with legal, moral, medical, and religious considerations. Acquiring a deep understanding of these aspects, including the implications of overturning Roe v. Wade, demands significant effort and time [7,8]. Our study aims to examine how YouTube users perceive the health implications of the Roe v. Wade overturn, and whether their gender and abortion leanings are reflected in the discourse. Furthermore, our study explores whether interrelationships exist between medical implication terms above the scale midpoint and sociodemographic overlays across our map. Therefore, we ask the following: RQ1: What do the term co-occurrence network clusters derived from YouTube comments reveal about the overturn of Roe V. Wade?

Moral intuitions are a fundamental source of bias in political beliefs, shaping how individuals interpret facts, suggesting that even factual knowledge may become irrelevant if it does not align with an individual’s abortion stance [3]. RQ2: How are prochoice and prolife terms distributed across the map? Furthermore, moral intuitions and political ignorance influence attitudes toward gender policies and other social issues [15]. Findings from the literature prompt the following question: RQ3: Does the discourse differ by a commenter’s approximated gender, and if so, how?

Rational ignorance theory suggests that significant changes in law or policy, such as the overturn of Roe v. Wade, might not elicit a proportionally informed public response. Subsequent voter-driven actions, like protests, support for specific political candidates, or state-level referenda on abortion, may be driven more by general sentiments or moral perspectives than by a detailed legal understanding [4]. This theory frames further questions: RQ4: How are comments on the medical implications of this legislation distributed across the cluster map? RQ5: How do discussions of medical implications resulting from the overturn of Roe v. Wade differ by estimated gender and abortion stance, and which demographic is more likely to engage in informed discussions about these implications?

2. Methods

This study employs a mixed-methods approach to analyze YouTube comments in the context of the overturn of Roe v. Wade. Textual coding added visual layers to the map. The quantitative analysis facilitated the evaluation of the map, as well as the direct link strength between its components and the overlap between variables. Additional textual analysis aided in deciphering the true meaning of the most prominent nodes within the map and their overlap across overlays.

2.1. Data Collection

Seven YouTube videos were systematically selected from an initial pool of fifty-four, based on five criteria. First, the videos were either posted five days following the leak of the draft Dobbs decision in early May 2022, or five days after the official overturn of Roe v. Wade in late June 2022. Second, the videos were sourced from major news networks, with a mix of center-, left-, and right-leaning stances [16] (see

Table 1. YouTube videos selected and used.

Video Date	News Outlet	Title	Video Description	Views, N	Total Comments, N
24 June 2022	ABC News	Supreme Court overturns Roe v. Wade after five decades|Nightline	This landmark 6-3 decision to overturn Roe v. Wade, led by a conservative majority, immediately activated trigger laws in several states, creating immediate and profound impacts, particularly for women in disadvantaged circumstances.	397,000	9486
24 June 2022	CBS Evening News	World reacts to U.S. overturning Roe v. Wade	The Supreme Court’s decision to overturn Roe v. Wade sparked significant international reaction, with prominent leaders and protesters globally expressing solidarity and criticism, highlighting the decision’s global impact amidst a trend of easing abortion restrictions in many countries.	397,000	6918
22 May 2022	ABC News	Medical Implications if Roe V Wade is Overturned	Dr. Bhavik Kumar, a medical director at Planned Parenthood Gulf Coast, discusses the impact of Texas’ abortion bans and the Dobbs decision, explaining how they have halted abortion care in Texas and forced patients to travel out of state, often for the first time, to access services, amidst increased demand for contraception and reconsideration of fertility plans.	65,000	1709
24 June 2022	Fox News	Today’s Supreme Court Hearing Gives States the Power to Allow, Limit or Ban Abortion Altogether	Legal scholar and professor, Jonathan Turley discusses the Supreme Court’s decision to Overturn Roe v. Wade and allowing each state to determine whether to protect or restrict access to abortion care.	110,000	1453
24 June 2022	ABC News	Supreme Court Rules to Overturn Roe v. Wade	Dr. Jen Ashton, a board-certified Ob-Gyn, discusses the complexities and options in reproductive healthcare amidst new abortion laws, emphasizing the importance of timely, individualized medical decisions for women.	56,000	2975
16 July 2022	Vice News	Why OB-GYNs Are Scared About Life After Roe	This video discusses the distress and ethical conflicts faced by OB-GYNs in Oklahoma due to restrictive abortion laws, which complicate medical decisions and may lead to a shortage of healthcare professionals in the state.	87,000	1207
4 May 2022	NBC News	Women Could Face Numerous Health Risks if Roe v. Wade Is Overturned	Medical experts warn that the possible overturning of Roe v. Wade may increase maternal mortality and long-term socioeconomic challenges, disproportionately impacting low-income women and women of color who seek abortions.	71,000	2849

). Third, only videos with a minimum of 50,000 views and 1000 comments were included. Fourth, English-language news reports were selected. Lastly, videos ranging from 3 to 10 min in length were included. The details of these videos, including titles, descriptions, and engagement levels, are provided in Table 1. We collected 26,597 comments utilizing the Netlytic YouTube Application Programming Interface (API). From this dataset, we removed 867 duplicate comments, resulting in a final count of 25,730 comments for analysis.

2.2. Map Construction Procedure

Mapping of social media reactions to Roe v. Wade’s overturn via VOSviewer involved the construction of corpus and scores files [17]. The corpus file contained textual data, such as social media comments, forming the basis for VOSviewer’s text mining and network visualizations [14]. The corpus was cleaned by normalizing contractions and possessives, replacing apostrophes with spaces for better noun extraction, and eliminating irrelevant punctuation. Additionally, manual spellchecking was conducted to address the variations in global English usage, ensuring linguistic accuracy in the dataset. Using a wildcard character allowing the search to return any comments that contained words starting with “abort”, we selected 22% of all comments (5577/25,730). We manually coded attributes such as prolife stance, prochoice stance, and discussions of medical implications, assigning a “1” when a characteristic was present and a “0” when absent. We used Gender API to automatically estimate the gender based on usernames from all comments in the corpus, generating gender codes. These manual and automated codes were converted into scores to visualize characteristics of interest across all terms on the map. To refine the term map, a thesaurus was developed [17]. Using it, we merged 73 terms through spelling corrections and synonym matching, and then eliminated 403 terms that were YouTube usernames, proper nouns, advertisements, or nouns and verbs without a clear connection to others.

2.3. Term Co-Occurrence Network Construction

To create a term co-occurrence map for social media network analysis, this study repurposed VOSviewer, a bibliometric network visualization tool developed by van Eck and Waltman at Leiden University [18]. Natural Language Processing (NLP) in VOSviewer was used to extract terms, defined as nouns and noun phrases, from the text of YouTube comments. Next, also in VOSviewer, we constructed a term co-occurrence network, also known as a map, that showed connections between terms or nodes. For terms that appeared in at least 25 comments, a term co-occurrence network was built to illustrate term co-occurrences and connections [14]. We developed a thesaurus to refine our analysis, ensuring that only terms related to abortion were present on our map. The terms were depicted as network circles or nodes, with closely related terms positioned together and connected by lines to signify their association [14]. Term or node size was determined by the frequency of the terms [15]. In our network, terms were grouped into thematic clusters [14]. A cluster is a group of closely connected terms within a group that are distantly connected with terms outside the cluster [17]. Thematic patterns were interpreted based on the term co-occurrence network analysis, followed by a deeper textual analysis of the YouTube comments.

2.4. Analysis of Network Elements and Underlying YouTube Comments

To analyze clusters and direct relationships between terms, VOSviewer employs mathematical algorithms to group terms based on their co-occurrence and relatedness [15]. VOSviewer network specifications come as two text files, a map file and a network file. A map file outlines the attributes of terms within the network—the terms, their positions, and the prominence of an item—which determine their visibility on the map’s display. A network file contains information about the links between the terms on a map, specifying which pairs of terms are connected by a link and the strength of each link. We examined VOSviewer downloads of map and network files to determine the prominence of terms and the strength of their direct associations.

2.5. Analysis of Strongest Dyadic Terms

The spatial layout of nodes depicted the interrelations of terms extracted from the social media corpus. The node size was proportional to the term occurrence [17]. The spatial proximity of nodes and links between them served as indicators of term relatedness and combined usage [15]. To understand the content and relevance of these clusters, we selected the top three nodes that appeared most frequently and had the strongest direct connections, categorizing them as intra-cluster dyads [17]. This process led to the identification of three dyads per cluster, culminating in a total of twelve dyads [17]. We searched for comments in which these dyadic terms co-occurred and conducted inductive content analysis to identify the top five recurring themes. Inductive coding guided us in understanding of the context surrounding each cluster [19], which informed the naming of each theme [17]. Terms occurring over 1000 times were excluded from dyad formulation to analyze specific discussions. When a term represents 1000 or more comments, many other mapped terms are also likely to be found within that comment collection, which might expand the qualitative comment analysis beyond the terms of interest and beyond a single cluster.

2.6. Visual and Quantitative Analysis of Overlays

To deepen the analysis of the overturn of Roe, we added visual layers to the network by recoloring each term to reflect a mean score computed for the comments that mention the term [17]. The outputs from Gender API generated two distinct visual overlays to assist in approximating perspectives from males and females. We adopted a binary definition of gender to reflect how gender was represented in our corpus of social media comments. We also created custom overlays for the three manually coded variables of prochoice, prolife, and medical implications. The analysis concentrated on examining the largest and darkest nodes to reveal the context of each respective overlay illuminated by darker shades of red.

2.7. Approximation of Gender

Gender was estimated at the aggregate term level, using Gender API as a proxy measure. In two separate overlays, we modeled the distribution of the gender characteristics across the mapped terms. Our gender analysis approximated map areas with above-mean concentrations of comments by male and female social media users. The means, one for each overlay, were calculated based on Gender API for all mapped terms, reflecting an average ratio of comments by male commenters (vs. female or unknown) in the first overlay and by female commenters (vs. male or unknown) in the second overlay.

Prior to creating the overlay, we evaluated Gender API’s potential for gender classification. Comments that contained the pronoun “I” (N = 485) were included in the analysis that compared Gender API’s gender approximations to manual coding. In 122 out of 485 comments (25%), commenters self-revealed their gender as either male or female. For these 122 comments, we calculated Cohen’s κ to assess agreement between the Gender API and manual coding. We found a better-than-chance, moderate level of agreement: κ = 0.577, 99% CI [0.379, 0.775] p < 0.001 [20]. Moreover, Gender API demonstrated 77.8% accuracy, 96.2% sensitivity, 63.2% specificity, and 67.5% precision. The F1 score of 79% suggested balanced performance in classification, combining accuracy, precision, and recall [21].

2.8. Analysis of Lowest and Highest Scoring Terms by Overlay

The map file from VOSviewer provides a score for each node, indicating the percentage of comments that contain a characteristic of interest. Using a map file, we evaluated term averages for each overlay. Calculating the percentage of terms above the mean and evaluating their position within the network helped to identify clusters with nodes that consistently scored higher or lower on the characteristics we measured. Additionally, we matched the 10 terms with the highest averages and the 10 terms with the lowest averages across the prolife, prochoice, male, female, and medical implications overlays. This approach also helped identify the terms most and least likely employed within the comments differentiated by abortion stance, gender, and engagement in discourse about medical implications. We inductively coded the comments contributing to the most used terms in each overlay to summarize the discussion behind the terms.

2.9. Analysis of Abortion Stance, Gender, and Discussions Surrounding Medical Implications

Initially, terms scoring above average and shaded in dark red in the medical implications, abortion stance, and gender overlays were selected. The objective was to explore how perceptions of medical implications prompted by changes in health policy differed by sociodemographic characteristics like gender and abortion stance. Terms from the medical implications overlay like safe abortion, miscarriage, ectopic pregnancy, complications, medical procedures, mental health, medical reason, birth control, and health issue were established as the baseline for comparison; this overlay consisted of 53 terms in total. Terms from the abortion stance or gender overlays matching any of the 53 terms were recorded. Following the completion of this matching exercise, percentages of overlap between sociodemographic overlays and the medical implications overlay were calculated. This analysis allows for examining the utilization of specific medical terms utilized by individuals identifying as prochoice, prolife, male, or female. Subsequently, comments containing these terms were examined to deepen our understanding of the contextual usage of the terms.

3. Results

3.1. Co-Occurrence Term Map

The frequency of comments posted on the videos decreased significantly over time. The majority of the activity occurred within the first eight days, with 7249 comments posted on Day 0 and 12,627 on Day 1. However, as the days elapsed, comment activity sharply declined, with only a handful of comments posted by Day 120. This pattern highlights a rapid peak in engagement following the videos’ release, followed by a steady decline in comment activity over the subsequent weeks (see Figure 1). We created and examined a network of term co-occurrences to explore the conversation on social media about the reversal of Roe v. Wade. The terms were extracted with the NLP algorithm from 25,730 comments on seven YouTube videos. Using a binary counting option in VOSviewer, we identified 34,889 terms. Our inclusion criteria limited terms to those appearing at least 25 times. We refined the list of terms using a thesaurus to build a four-cluster network of 169 terms (see Figure 2 and

Table 2. Most frequently occurring terms by cluster.

Cluster	Terms	Ten Largest Terms	Term Occurrences	Cluster Label
1	59	law	1071	Constitutional Law
		Roe v. Wade	681
		Constitution	641
		Supreme Court	478
		democrat	444
		republican	401
		ruling	302
		power	285
		vote	211
		amendment	192
2	54	abortion	4185	Reproductive Health and Responsibility
		pregnancy	762
		sex	739
		birth control	679
		rape	605
		parenthood	321
		condom	320
		consequences	319
		action	287
		contraception	244
3	27	fetus	721	Human Development
		human being	373
		week	351
		doctor	322
		womb	295
		unborn	292
		cell	254
		conception	175
		brain	253
		science	144
4	16	God	822	Religious Beliefs
		religion	269
		scripture	262
		prolife	194
		church	157
		Christian	127
		love	110
		hell	104
		innocent child	106
		Satan	70

Note: Terms are italicized in the tables and throughout the manuscript to distinguish their presence on the map from the explanations of their meanings.

).

3.2. Cluster Analysis of YouTube Comments on the Overturn of Roe v. Wade

The first research question examining the context of the clusters is answered by Figure 2 and Table 2 and

Table 3. Comments representing term dyads with highest link strength: an inductive content analysis.

Linked Dyads	Link Strength	Comments with Both Terms	Comments Thematically Coded	Comments Assigned a Theme	Themes
Cluster 1: Constitutional Law
Constitution and Supreme Court	111	96	96	58	1. Constitutional interpretation and the role of the Supreme Court; 2. State rights vs. federal oversight; 3. Public misinterpretation of the Supreme Court’s decision; 4. Debate over Supreme Court’s legitimacy; 5. Democratic participation in legislation and elections.
Roe v. Wade and Supreme Court	90	84	84	31	1. Misinterpretation of the Supreme Court’s Role decision; 2. Debate about accountability of Supreme Court justices; 3. Call for protest and mobilization; 4. Clarification of the Supreme Court’s decision; 5. Impact on women’s rights.
Roe v. Wade and Constitution	61	101	101	101	1. Constitutional interpretation and judicial role; 2. Political and ideological divisions; 3. Public misinterpretation of the Supreme Court’s decision; 4. Impact on women’s rights; 5. State rights versus federal oversight.
Cluster 2: Reproductive Health and Responsibility
birth control and condom	115	40	40	40	1. Women’s autonomy and rights; 2. Negative outcomes of restricting abortion; 3. Moral and ethical considerations of abortion; 4. Misinformation about abortion and reproductive health; 5. Political and legal aspects of the abortion debate.
birth control and sex	86	174	174	101	1. Misconceptions about abortion and the need for better sex education; 2. Critiques of views that minimize the impact of rape on abortion decisions; 3. Contraception and responsibility; 4. Health risks and medical necessity.
pregnancy and sex	96	154	154	101	1. Defining life and personhood; 2. Ethical considerations of abortion; 3. Legal and societal implications; 4. Medical facts about pregnancy and contraception; 5. Challenges in accessing abortion services.
Cluster 3: Human Development
fetus and week	66	148	148	148	1. Personhood and scientific definitions; 2. Moral and ethical considerations of abortion; 3. The rights of the fetus versus the mother; 4. Social and psychological impacts; 4. Misinformation regarding abortion; 5. Fetal development
fetus and human being	58	60	60	60	1. Redemption and forgiveness for abortion; 2. Consequences of abortion; 3. The sanctity of life; 4. Salvation and repentance; 5. Morality of aborting unborn children.
fetus and womb	48	65	65	17	1. Markers of fetal viability; 2. Scientific and medical definitions; 3. Fetal awareness and pain; 4. Societal and moral decay.
Cluster 4: Religious Beliefs
God and scripture	68	61	61	61	1. Personal testimonies of faith and transformation; 2. Divine judgement; 3. Interpretation of scripture; 4. Salvation through Christ; 5. Moral and ethical consideration of abortion.
God and religion	32	41	41	20	1. Interpretation of religious texts; 2. Role of religion in governance; 3. Religious identity; 4. Critique of religious institutions; 5. Spirituality and religious beliefs.
God and love	24	45	45	40	1. Repentance; 2. Divine judgment; 3. Misinterpretation of scripture; 4. Salvation through Christ; 5. Role of religion in governance.

. VOSviewer defined each cluster by its color, indicating distinct thematic groupings in the network (see Figure 2). To identify the context of each cluster, we analyzed the ten most prominent terms (see Table 2). We also identified three dyads per cluster for a total of 12 dyads (see Table 3). We conducted an inductive content analysis of comments containing a co-occurrence of terms, formulating dyads for all four clusters [19]. The organization of the clusters was based on their size, beginning with the largest. Cluster 1 (red) focuses on the relationship between the U.S. Constitution’s interpretation and the societal and legal impacts of the Supreme Court’s decision to overturn Roe v. Wade. The cluster was labeled as the Constitutional Law cluster. Cluster 2 (green) contains themes addressing reproductive health and responsibility, approaches to abortion and reproductive rights, and the medical, legal, ethical, and societal consequences of the overturn. The cluster was labeled as the Reproductive Health and Responsibilities cluster. Cluster 3 (blue) contained themes exploring the intersection of abortion laws, policies, and individual beliefs with concepts of human development. This cluster was labeled as the Human Development cluster. Cluster 4 (yellow) contains themes exploring the intersection of abortion laws and religious beliefs. This cluster was labeled as the Religious Beliefs cluster.

Table 3 outlines the criteria we used to select comments for thematic coding. We focused on dyadic terms that exhibited the highest frequencies of co-occurrence and the strongest connections. By examining the comments that included both terms, we aimed to delve deeper into the primary themes present in the discourse. Additionally, we identified and listed the five predominant themes derived from an in-depth analysis of the comments to name each cluster.

Table 4. Dyads with frequently occurring terms (>1000 comments) excluded from inductive content analysis.

Excluded Dyads	Link Strength	Comments with Both Terms
abortion and pregnancy	318	400
abortion and rape	285	499
abortion and birth control	266	549
abortion and sex	194	355
abortion and incest	140	354

lists the dyads containing the terms “Law” and “abortion”. These dyads were excluded from the inductive content analysis because they represented terms that appeared in more than 1000 comments and connected to nodes in multiple clusters. Our goal was to understand the context of the terms within each cluster to assign meaningful names, rather than focusing on their connections across all four clusters. Excluding these highly frequent and connected terms ensured a more accurate and contained analysis of each cluster.

3.3. Overview of Dyadic Terms and Themes

The analysis of dyadic terms across the four clusters revealed distinct thematic patterns. In Cluster 1, Constitutional Law, dyadic combinations of the terms Constitution, Supreme Court, and Roe v. Wade highlighted discussions around constitutional interpretation and the legitimacy of the Supreme Court’s decision. Cluster 2, Reproductive Health and Responsibility, contained the dyads birth control and condom, birth control and sex, and pregnancy and sex, which address issues of women’s autonomy, personal responsibility to avoid unwanted pregnancy, and the implications of restricting abortion access. In Cluster 3, Human Development, the dyadic terms fetus and week, fetus and womb, and fetus and human being centered on debates about personhood, fetal rights, and the ethical and scientific aspects of abortion. Cluster 4, Religious Beliefs, was shaped by dyadic combinations of the terms God, scripture, love, and religion, and reflected moral and spiritual discussions on abortion, particularly from a Christian perspective. An inductive content analysis of Cluster 1 revealed discussions about the conflict between federal versus state power, personal freedoms, and judicial interpretation. Ideological differences surfaced, with Democrats advocating for abortion rights laws, while Republicans focused on state sovereignty and ethical issues. This dialogue extended to various perspectives on abortion law, including the interpretation of the constitution. The dialogue presented an ideological rift, as Democrats supported legislative measures aimed at guaranteeing abortion rights, while Republicans prioritized state sovereignty and ethical deliberations.

In Cluster 2, Reproductive Health and Responsibility, the inductive content analysis highlighted a focus on reproductive health and depicted contrasting views on fetal development and the moral implications of abortion. One perspective emphasized women’s autonomy and the need for accessible reproductive healthcare, emphasizing the challenges and health risks of restricting abortion access. Additional comments centered on the ethical considerations of fetal development stages, advocating for the rights of the unborn and proposing adoption as an alternative. Unique informants shared firsthand experiences and ethical viewpoints, used storytelling, and discussed past, current, and expected health outcomes to articulate their positions.

Moreover, comments also discussed the dangers of delays to care due to restrictive abortion laws and the requirement to travel across state lines. Some commenters cited higher rates of maternal mortality and infant mortality arising from restrictive abortion policies, especially in the case of necessary terminations. Some comments addressed increased health risks and complications during the abortion procedure due to inaction by clinicians out of fear of prosecution or uncertainty over the definition of medical necessity. Debates also emerged regarding whether abortions are a convenient way to end an unwanted pregnancy or an essential medical procedure in situations where there is a medical necessity, such as when carrying a nonviable fetus or when the mother’s life is critically at risk.

In Cluster 3, the Human Development cluster, the inductive content analysis uncovered discussions about the onset of life and debates about a fetus’s humanity and rights compared to the pregnant woman’s. The dialogue dissected fetal awareness, life viability, and ethical–medical debates on abortion. There was a divide between the scientific understanding of fetal growth and ethical views on life’s sanctity and maternal rights. The cluster also addressed the fetus’s potential suffering during abortion, with a notable prolife bias urging for stricter abortion laws to safeguard the unborn from ethical harm and physical distress.

Comment authors highlight various concerns regarding abortion, particularly emphasizing the emotional and psychological toll it can take on women. Many point out that abortion can lead to feelings of guilt, depression, and regret, impacting mental health significantly. Additionally, there are noted health risks associated with abortion procedures, such as infections, heavy bleeding, and potential damage to reproductive organs, which could compromise future fertility.

An inductive content analysis of Cluster 4, the Religious Beliefs cluster, revealed a focus on Christian principles, with discussions emphasizing the sanctity of life. Social media users cited specific verses from the Bible to guide and promote repentance and spiritual salvation for past abortions. The majority of these comments were prolife, highlighting abortion’s moral and spiritual implications. The clusters contained evidence of Christian dogma for opposing abortion and spreading Christian ideals on life and ethics. While minimal, some voices advocated for a separation of church and state.

3.4. Analysis of Overlays

Customized overlays showed the distribution of prolife, prochoice, male gender, female gender, and medical implications comments across the terms in the Figure 2 map. Term color corresponds to the proportion of comments behind the term that scored “1” on each variable of interest. Applying an overlay amplified a characteristic of interest, like gender, within the comments contributing to terms across clusters. Terms with above average concentrations of approximated female users relative to all terms are recolored in a darker red, whereas those with below-average concentrations are depicted in blue (see Figure 2, Figure 3 and Figure 4).

3.5. Distribution of Prochoice and Prolife Comments Across the Map

In our analysis of abortion stance, we coded 21% of the comments. Of these, 21% expressed a prochoice position and 33% a prolife stance. The remaining 46% of comments did not indicate a clear stance on abortion. There were 1.55 prolife comments for every 1 prochoice comment. If all comments were coded, the maximum average share of comments contributing to prochoice terms was 13% and 27% for prolife terms. We address research question 2 by creating a table depicting overlay concentrations across all four clusters. The visual overlays on the map clearly highlight differences across terms, with those in darker red indicating a higher prevalence of the targeted abortion stance within the contributing comments (see Figure 3).

3.6. The Prochoice Stance

As depicted in Figure 3, Overlay 1 provided additional visual evidence about the spread and meaning of prochoice terms across our map. The clusters focusing on Reproductive Health (Cluster 2), Human Development (Cluster 3), and Religious Beliefs (Cluster 4) had a higher proportion of nodes that scored above the prochoice scale midpoint. Cluster 2 had the greatest concentration. Through an examination of the comments derived from the darkest nodes within the overlay, we learned that prochoice users advocate strongly for individual rights and women’s control over their bodies. This focus extended to reproductive choices, including safe access to abortion. Additionally, these areas frequently discussed the broader social issues that can arise from carrying an unwanted pregnancy. Comments discussing the ethics of abortion, favoring state control over federal abortion rights, and condemning women for being promiscuous were less likely to contain prochoice perspectives.

3.7. The Prolife Stance

As depicted in Figure 3, Overlay 2 provided additional visual evidence about the spread of prolife terms across our map. Nodes scoring above the mean scale were composed of many comments, complicating observation independence. The clusters focusing on Reproductive Health (Cluster 2), Human Development (Cluster 3), and Religious Beliefs (Cluster 4) had a higher proportion of prolife views, with Cluster 4 showing the strongest concentration. An examination of the comments derived from the darkest nodes within the overlay reflected concerns over personal responsibility, the value of life from conception, and skepticism towards certain scientific practices and ideologies conflicting with prolife principles. Prolife perspectives were less prominent in comments discussing legal precedents, medical necessity, and the economic and mental health consequences of keeping an unwanted pregnancy.

Table 5. Terms contributing to abortion stance overlays.

Nodes Above and Below Overlay Scale M	Abortion Stance Overlays
	Prochoice Overlay M = 0.02	Node M	Prolife Overlay M = 0.04	Node M
Highest	separation of church and state	0.11	promiscuous	0.55
	college	0.10	condom	0.14
	equal right	0.10	left	0.14
	loss	0.10	development	0.12
	assault	0.09	innocent child	0.12
	gay marriage	0.08	IVF	0.12
	politics	0.08	health issue	0.10
	species	0.08	liberal	0.10
	unwanted child	0.08	science	0.10
	safe abortion	0.07	vaccine	0.10
Lowest	2nd amendment	0	burden	0
	10th amendment	0	civil war	0
	action	0	election	0
	development	0	medicine	0
	federal law	0	mental health	0
	left	0	precedent	0
	liberal	0	state legislature	0
	move	0	state line	0
	precedent	0	suffering	0
	promiscuous	0	Supreme Court justice	0

Note: Prochoice overlay scale mean is 0.02 and prolife scale mean is 0.04.

illustrates the terms most and least likely to be used in discussions about abortion stances. The analysis revealed differences in linguistic trends tied to ideological beliefs. Terms that had higher node averages in one overlay tended to have the lowest averages in the overlay representing the opposing abortion stance, indicating an inverse relationship.

3.8. Distribution of Comments Contributed by Male and Female Users Across the Map

We address research question 3 by demonstrating how discourse on the overturn of Roe v. Wade varies by gender, as evidenced in Figure 4 and

Table 6. A comparison of terms scoring highest and lowest by gender overlay (male vs. female).

Score Type	Male Overlay (M = 0.50)	Term M	Female Overlay (M = 0.20)	Term M
Highest	evolution	0.90	zygote	0.45
	equality	0.74	state legislature	0.43
	state level	0.74	14th amendment	0.40
	species	0.73	life sentence	0.38
	November	0.72	tube	0.36
	mother’s life	0.71	welfare	0.36
	liberal	0.70	assault	0.35
	medical implication	0.69	safe abortion	0.34
	Satan	0.69	sin	0.34
	state line	0.69	uterus	0.33
Lowest	safe abortion	0.24	evolution	0.01
	14th amendment	0.31	states right	0.06
	medical reason	0.31	left	0.07
	state legislature	0.34	precedent	0.07
	zygote	0.34	liberal	0.08
	burden	0.36	congress	0.09
	childbirth	0.36	DNA	0.09
	foster care	0.36	sacrifice	0.09
	safe sex	0.36	action	0.10
	organ	0.37	suffering	0.10

. In the sections below, we offer a detailed analysis of the comments, focusing on the main conversational differences associated with nodes above the mean scale. This analysis aligns with the male and female user overlays depicted in Figure 3. Gender API approximated that 15,512 comments were contributed by males and 4877 comments were contributed by females. However, it was unable to approximate the gender of 5341 comment authors. Male engagement was dominant in the comments, corresponding to a 3:1 ratio of comments contributed by male vs. female YouTube users, as estimated by Gender API.

3.9. Male User Approximation

In our network, the average share of comments from individuals with an approximated male gender was about 50% per node (see Table 6). There was a significant presence of male-contributed comments throughout all clusters’ nodes, with male contributions dominating almost entirely in both the Constitutional Law (Cluster 1) and Religious Beliefs (Cluster 4) clusters. An analysis of the comments most likely contributed by males revealed a perspective emphasizing the need for gender equity in reproductive choices and financial obligations, highlighting frustrations over perceived double standards in financial responsibilities after the abortion decision. Male users were likely to recognize the legitimacy of abortion for maternal health risks and engaged in discussions about personhood and the balance between state and federal powers. The comments contained an analytical interpretation of the constitution and Supreme Court precedents, along with current and future implications for reproductive rights. Males were less engaged in discussions on safe abortion access, bodily autonomy, early developmental stages, safe sex practices, economic and mental health impacts on the foster care system, and childbirth complications.

3.10. Female User Approximation

In our network, the average share of comments most likely contributed by females was 20% per node (see Table 6). Approximated contributions did not show any concentrations within any of our four clusters. The largest contributions of comments were found in the Reproductive Health (Cluster 2) and Human Development (Cluster 3) clusters. An examination of the ten darkest nodes within the overlay revealed how female users reacted to the overturn of Roe v. Wade. Female users voiced concerns about losing control over reproductive choices, including fears of carrying pregnancies from assault and being denied treatment when complications arise. Furthermore, they anticipate a heightened demand for welfare assistance due to a rise in poverty. They also recounted personal experiences with pregnancy to relay societal, religious, and individual perspectives on abortion and motherhood. Female users were less likely to question scientific views on human development. They also seemed less engaged in exploring the potential benefits of state-regulated abortion access post-overturn, and in contemplating the ethical implications of ending a pregnancy.

Table 6 shows the terms that were most and least likely to be used by males or females. The analysis depicts variations in linguistic patterns according to gender. Terms that register higher node averages in one overlay generally have the lowest averages in the other overlay, indicating an inverse relationship.

3.11. Distribution of Comments Addressing Medical Implications Across the Map

We addressed research question 4 by examining the concentrations of overlay terms above the scale midpoint in

Table 7. Cluster and overlays: how high-scoring terms are distributed across the map.

Visual Overlay	Overlay M	% of Cluster Terms Scoring Above All Terms’ M (Below All Terms’ M)
		Cluster 1 59 Terms	Cluster 2 54 Terms	Cluster 3 35 Terms	Cluster 4 21 Terms
Abortion Leaning
Prochoice	0.02	61% (39%)	74% (26%)	69% (31%)	76% (24%)
Prolife	0.04	42% (58%)	65% (35%)	54% (46%)	52% (48%)
User Gender Approximation
Female	0.20	32% (68%)	46% (54%)	43% (57%)	29% (71%)
Male	0.50	90% (10%)	56% (44%)	57% (43%)	95% (5%)
Focus
Medical Discussion	0.15	8% (92%)	61% (39%)	40% (60%)	5% (95%)

Note: Concentrations above 50% are bolded. All terms within each cluster represent 100%. Cluster themes were Constitutional Law (Cluster 1), Reproductive Health (Cluster 2), Human Development (Cluster 3), and Religious Beliefs (Cluster 4).

and visually inspecting nodes shaded in dark in Figure 5. We coded a subset (21%) of comments to determine whether medical implications were discussed in response to the YouTube videos. The average share of comments contributing to terms discussing the medical implications of Roe v. Wade had a scale midpoint of 0.15 (see

Table 8. Terms contributing to medical implications overlay.

Overlay	Overlay M	Ten Terms with Highest and Lowest Concentration of Medical Implications Content in Comments
		Highest	Term M	Lowest	Node M
Medical Information	0.15	late term abortion	0.99	10th amendment	0
		safe abortion	0.93	equality	0
		miscarriage	0.60	life sentence	0
		ectopic pregnancy	0.58	species	0
		complication	0.55	evolution	0.01
		medical procedure	0.52	slavery	0.01
		mental health	0.50	democracy	0.02
		medical reason	0.47	profanity	0.02
		birth control	0.38	scripture	0.02
		health issue	0.38	shame	0.02

). If all comments were coded, the maximum average share of comments contributing to medical implication terms was 71%. Evidence of medical implications discourse was concentrated in the cluster focusing on Reproductive Health (Cluster 2) (see Table 7). An examination of the comments derived from the darkest nodes within the overlay centered on health risks resulting from the overturn. Later in the document, we examined the significance of terms that exceeded the scale midpoint in this overlay, focusing on their intersection with gender and abortion stance.

Table 8 shows the terms that digital community members were most likely and least likely to use when discussing the medical implications of the overturn of Roe v. Wade and restricting access to abortion care. Terms most commonly associated with discussing medical implications pertained to women’s health and medical procedures. Conversely, terms less frequently linked to these implications were more often connected to the political dimensions of the debate.

Table 7 examines the concentration of terms above the midpoint of the scale in each overlay across the map. Of the four clusters, Cluster 2 exhibited the highest concentration of terms from the prochoice, prolife, male, and medical implications overlays. Female-authored comments were not dominant in any of the clusters.

3.12. Who Was More Likely to Engage in Discussions About the Medical Implications of Overturning Roe v. Wade?

By visually examining the term co-occurrence map and VOSviewer downloads, we investigated the intersection of comment characteristics to comprehend how they converged to address our fifth research question of the study: How did YouTuber users perceive the health implications of the overturn of Roe v. Wade, and how were their gender and stance on abortion reflected in the discourse? We address research question 5 by depicting the overlap of terms above the scale midpoint for the abortion stance and approximated gender overlays and match them to the terms above the scale midpoint in the medical implications overlay. Table 7 illustrates the number and percentage of terms above the scale midpoint of the sociodemographic overlays matching the 53 terms of the medical implications overlay. Visual overlays lack independence of observation. If a term scored above the midpoint on both the medical implications and sociodemographic overlays, it is highly likely that the comment addressing abortion stance, contributed by males or females, also discussed medical implications.

Table 9. Relationships between overlays: terms above the scale midpoint matching with terms in medical implications overlay.

Category and Overlay Name	Overlay Terms Above the Scale Midpoint That Also Scored Above the Mean on the Medical Implications Overlay
	N	Percent
Abortion Stance
Prochoice	41 (12)	77%
Prolife	33 (20)	71%
Gender
Female	29 (24)	55%
Male	26 (27)	49%

Note: The medical implications overlay had 91 terms, 53 of which scored above the mean. Terms in parentheses were terms in overlays that did not match with terms above the scale midpoint in the medical implications overlay.

showed that terms in the prochoice overlay demonstrated the greatest overlap with the medical implications overlay, suggesting that individuals with a prochoice stance were more likely to discuss the medical implications of the overturn. Across the map, terms above the scale midpoint for the medical implications and prochoice overlays overlapped 77% of the time (see Table 9). The overlay for approximating the male gender had the lowest degree of overlap, indicating that male commenters were the least likely to discuss the medical consequences of the overturn. Terms above the scale midpoint contributed by approximated male users matched with terms above the scale midpoint in the medical implications overlay 49% of the time (see Table 9).

3.13. Prochoice Stance

Terms above the scale midpoint in both the prochoice and medical implications overlays, such as safe abortion and medical procedure, were used by users to discuss the need for safe abortion, and they also shared that abortion is a healthcare right and should be accessed safely and without barriers. The use of the term restriction was associated with a return to unregulated and unsafe methods, which could lead to increased medical risks such as infections, complications, and higher mortality rates. Digital publics using the term state law highlighted that a federal framework for abortion rights would ensure consistent medical care across the board, contrasting with state-level control, which can result in uneven access, potentially leading to poor health outcomes and increased maternal mortality.

Terms such as miscarriage and complication were used to highlight the need for abortion services in cases of health complications and nonviable pregnancies. The term miscarriage was used to differentiate between spontaneous and induced miscarriages. In comments containing the terms ectopic pregnancy and health issue, users discussed the need for critical and lifesaving medical interventions. Commenters discussed mental health and the importance of protecting it throughout pregnancy, whether it ended in termination or the continuation of an unwanted pregnancy. Birth control, Plan B, and sex education were widely discussed as options for preventing unwanted pregnancies and reducing the need for abortions. The term late-term abortion was also prevalent in the discourse, indicating that late-term abortions were only performed during medically necessary circumstances, as no doctor would perform one past 20 weeks of pregnancy unless the mother’s life was at risk or the fetus was nonviable.

3.14. Prolife Stance

There was a convergence between the prolife and medical implications overlays within Cluster 2, with 71% of key terms being pertinent to both areas of focus (see Table 7). While the key terms in Cluster 2 were relevant to the prochoice, prolife, and medical implications discussions, an examination of the comments uncovered a distinct context for the prolife overlay. Terms like birth control and condom were used to stress the importance of using contraception to prevent unwanted pregnancies and limiting the reliance on abortion to end unwanted pregnancies. The terms rape and exception were closely connected; a review of the comments indicated that some individuals with a prolife stance would only consider abortion under extreme circumstances like rape or when circumstances endangered the mother’s life.

3.15. Female Users

Alignment existed between the female and medical implications overlays; the terms matched 55% of the time in all four clusters (see Table 7). Closer analysis of matching terms above the scale midpoint revealed that comments about reproductive health contributing to the female overlay frequently also contributed to prominent terms across the prochoice and medical implications overlays. Women emphasized the need for accessible medical and psychological care in their comments, particularly using the term safe abortion to highlight how restrictive abortion laws compromise their healthcare autonomy. They used terms like miscarriage tube and ectopic pregnancy to advocate for prompt medical attention in high-risk situations. The terms medical procedure, risk, and danger were used to point out that abortions are often medically necessary, not elective. Furthermore, mental health was used to discuss the lasting emotional distress from pregnancy loss and abortion decisions. Females used the term doctor to reiterate the importance of adhering to medical advice over political opinions for reproductive health decisions. Women also refenced the terms birth control and contraception to suggest that contraceptive methods can fail and to reinforce the need for abortion services in these instances.

3.16. Male Users

Conversely, the comments most likely contributed by males had a lower degree of overlap with the terms from the medical implications overlay, with just 49% of terms matching (see Table 7). The overlap of terms above the scale midpoint primarily occurred in Clusters 2 and 3. Male-contributed comments used the term late-term abortion to discuss the ethics of abortion. The term miscarriage was connected to comments detailing the emotional impact of suffering a miscarriage on potential fathers. Some males used the terms complications, mother’s life, and health issues to emphasize the importance of safeguarding the mother’s health, with an understanding of the medical necessity for certain abortions. These terms were also used to recount personal experiences with pregnancy complications and loss. Mentions of the terms abortion clinic, Plan B, and medicine were used in discussions surrounding pregnancy prevention and the role of abortion clinics in providing reproductive care. In discussions containing the terms teen, healthcare, risk, and health risk, some male commenters discussed the potential risks to teenagers facing pregnancy.

4. Discussion

Members of the studied digital community based their opinions and actions on ideological alignment on abortion rather than a nuanced understanding of the health implications or legal changes. Prochoice terms were mainly found in Cluster 2, centered on advocacy for women’s rights and control over their bodies, extending to reproductive choices and safe abortion access, while also addressing broader social issues from unwanted pregnancies. Prolife comments, prevalent in Clusters 2, 3, and 4, addressed abortion ethics, the rights of the unborn, and favored state control over federal abortion rights. Prochoice users condemned promiscuity, advocating contraception as an alternative to abortion.

Rational ignorance theory suggests that different genders might engage with issues to varying extents based on perceived relevance or impact [3]. In our network, comments approximated to male users dominated the Constitutional Law and Religious Beliefs clusters, often discussing topics related to their limited role in reproductive decisions, along with financial implications. They engaged in debate about constitutional interpretations and the balance of state versus federal powers. Female-contributed comments were scattered across all clusters but predominantly appeared in the Reproductive Health and Human Development clusters. Females expressed concerns over reproductive autonomy, highlighting the need for safe abortion access, the risks of pregnancies from assault, and the poverty cycles that can result when abortions are denied. They shared personal stories to articulate societal and individual views on abortion and motherhood. Female users were motivated to overcome rational ignorance, leading to deeper engagement with the future consequences of restricting access to abortion care.

Comments discussing the medical implications of the overturn were primarily concentrated in Cluster 2. Prochoice community members were more likely to discuss the medical implications of limiting access to abortion care in the case of ectopic pregnancies and when the pregnancy was no longer viable. They also discussed the health risks that could result in mortality when abortion access was denied. Male users were least likely to discuss the medical implications of the overturn. Males dominated the discussion, with many comments shifting away from medical implications to focus on assigning blame. Users who blamed women for using abortion as birth control often lacked a deep understanding of the complex legal, medical, and policy issues involved. Furthermore, they did not see the value in gaining a deeper knowledge of medical implications and health risks.

4.1. Context of the Discourse and the Future of Abortion Care

The observed patterns in the social media discourse about Roe v. Wade’s overturn provided a clear illustration of how rational ignorance facilitates selective engagement with different aspects of a complex issue. The studied digital community members likely focused on aspects aligned with their respective ideologies. They may have avoided deeper explorations of the subject, particularly those requiring specialized medical or scientific knowledge.

4.2. Clusters

The use of dyads and inductive content analysis clarified the context of each cluster by examining the relationships between key terms and underlying themes, allowing for the precise naming and interpretation of the clusters. Among the four identified clusters—Constitutional Law, Reproductive Health and Responsibility, Human Development, and Religious Beliefs—the Constitutional Law cluster was the largest. It centered on interpreting the Supreme Court’s abortion decision, engaging in debates on its constitutional basis, critiquing the Court and its justices, and considering the implications for democracy and federal–state relations. Although focusing on legal, ethical, and societal issues, it lacked discussion on medical implications like healthcare or maternal mortality. The prominence and size of this cluster confirmed that the discourse was predominantly oriented towards legal and social considerations, rather than medical and public health aspects [1,11,22].

The Reproductive Health and Responsibility cluster was the second largest cluster and contained the majority of medical implication discussions. Yet, much of the dialogue focused on the prevention of unwanted pregnancies rather than resorting to terminating a pregnancy. In half of the video stimuli analyzed, physicians warned that stricter abortion barriers could lead to higher maternal mortality and infant mortality. Nonetheless, members of the digital community eschewed the stimuli, engaged in incivility, and accused women of using abortion as a last-minute birth control method due to lack of preparation. The Public Religion Institute conducted a survey on abortion and found that 46% of respondents deemed abortion sinful, while 35% regarded a woman’s decision to terminate a pregnancy as a selfish evasion of responsibility [12]. Similar sentiments were represented in our network.

Conti and Cahill [23] highlight how media depictions of abortion are often politicized and are inappropriately sourced. In our study, we included videos featuring physicians addressing the medical consequences of abortion, recognizing the media’s tendency to focus more on political and social issues, with only 11% of media content coming from actual medical professionals. Despite the absence of terms like “maternal death” and “mortality” in our semantic analysis, a subset of comments brought attention to the severe health risks and potential fatalities stemming from restricted abortion access. These observations align with findings by [24], suggesting that, similar to the broader U.S. public, the digital community members exhibited a limited understanding of abortion’s safety and health implications.

Unique informants entered the dialogue and shared deeply personal stories about medical complications, regrets over having an abortion, or terminating an unwanted pregnancy. Previous research conducted by [23] showed that 18% of comments contributed on social media reflected personal experiences. Social media enables contributors to share their vulnerabilities in ways that might not occur during face-to-face interactions, allowing readers to inconspicuously observe a subset of digital community members whom they would not otherwise have access to [25]. The intent behind unique informants sharing such personal experiences was to change public perception and help other members of the digital community move beyond their rational ignorance to recognize that abortion is a medical issue, not simply a way to shirk responsibilities [26]. Personal stories and the exchange of information on social media can significantly shift public perceptions [23].

In the Human Development and Religious Beliefs clusters, conversations about human development and fetal viability emphasized morality over science, revealing a skepticism towards scientific explanations of fetal viability. Valdez et al. [7] stress the critical need for pinpointing factors that predict abortion stance, aiming to tailor future discussions on reproductive health more effectively to audience perspectives. By grasping these key determinants, we could craft educational content focused on health risks, fostering discussions that could ultimately shift voting patterns and influence health policy [27]. Social media discourse may be used to draft effective communication strategies to address constraints in access to reproductive health.

Some prolife users discussed benevolent reasons for terminating a pregnancy, often citing religious doctrines that support abortion under circumstances where a child would endure suffering due to uncorrectable defects. They advocate for exceptions such as life-saving abortions during obstetrical emergencies. The dialogue also includes a focus on parental love and the difficult choice to opt for an abortion to prevent their child’s impending suffering. This mirrors recent debates in abortion-restrictive states, where prolife and Christian women are advocating for access to abortion care. Additionally, terms like termination for medical reasons (TFMR) and compassionate abortion have been introduced to educate opponents of legal abortion about the critical need for such options when the health of the mother is at risk, or to spare a baby from inevitable pain [25,28].

4.3. Gender Approximation

Our study leveraged Gender API to approximate the gender of social media users, revealing discernible gender-based patterns in discourse and perceptions. This approach was valuable at the group level, allowing us to observe gender-specific trends and concentrations of comments within distinct clusters. Unlike data from Twitter and other platforms, which provide user demographic details and hashtags, YouTube data required additional processing due to the absence of such information. This limitation restricted our ability to categorize the dataset by political affiliation, abortion stance, geographical location, and gender. To compensate, we employed textual analysis and automated gender approximation tools as supplementary methods to introduce demographic characteristics and depth to our analysis. These methods not only deepened our understanding of the data, but also illuminated the significant role of gender in shaping responses and discussions related to the overturning of Roe v. Wade.

4.4. Social Media Behavior by Gender

Although social media usage and content consumption is more common among females, males exhibit a higher propensity to share their opinions on social media [29,30]. Sharing inflammatory opinions is more likely to occur when YouTube users express beliefs surrounding social norms and polarizing topics [31]. Abortion rights, particularly restrictive legislation, stands as one of the most divisive topics in the United States, prompting women to voice their protests on social media platforms [32]. Overlapping terms used by users we approximated as males and users with a prolife stance focused on blaming individuals for their actions. They celebrated the decision and used rhetoric to encourage individuals seeking to terminate a pregnancy to move across state lines. Gender API, our tool for estimating gender, approximated that there were 16,367 males and 5257 females. Overlay patterns only reflected categorizable comments and, within them, the ratio was estimated at 3:1. Given the gendered nature of the topic, it was unexpected to find that males were the dominant voices in the discourse.

4.5. Abortion Stance

Comment analysis revealed a 1.5 prolife to 1 prochoice ratio, not reflecting the U.S. voter base, where 60% support abortion access [8]. Prolife and prochoice comments shared 24 terms above the scale midpoint, suggesting overlap and challenging independent observation assumptions. Language analysis showed distinct trends, with prochoice efforts aimed at safeguarding abortion care and prolife efforts at restricting it, mirroring post-Roe v. Wade actions. At present, prochoice activists are advocating for abortion rights in state constitutions, while conservative groups continue to leverage legal avenues to limit all forms of abortion by prohibiting mifepristone distribution [28,33,34].

4.6. Medical Implications

Social media discussions about the medical implications and personal stories related to obstetrical emergencies and abortion denials have proven predictive of outcomes after the overturn of Roe v. Wade. It is crucial to investigate how these conversations have evolved and further analyze these experiences. Previous research on YouTube comments revealed a future-focused dialogue among prochoice users who discussed the medical consequences of abortion restrictions, with findings aligning with subsequent increases in abortion rates despite new restrictions [25,28,35,36,37]. Women, in particular, focused significantly on the medical implications and emphasized the ongoing necessity for safe abortion services. Future research is essential to deepen our understanding of the consequences of healthcare denials and maternal mortality.

We observed rational ignorance more frequently among users identified as males with a prolife stance. These individuals often suggested that those affected by restricted access to reproductive care should travel across state lines to obtain an abortion. However, they seemed unaware of the numerous challenges this entails, including the legal risks, medical dangers of delayed care, risks of carrying a pregnancy with underlying health conditions, and the potential for a pregnant person to be critically ill before receiving treatment [25,28,33,36,37]. Furthermore, since the overturn, the number of abortion clinics has shrunk by half [33]. According to the Guttmacher Institute, 171,000 birthing persons traveled out of state for an abortion in August of 2023 [37].

The shifts in supply and demand pose significant challenges for healthcare administrators in states where abortion is protected [38]. The expansion of abortion access in states like Illinois, which have allocated over USD 23 million post-Dobbs decision to enhance facilities, highlights a proactive approach to healthcare administration in response to shifting supply-and-demand dynamics [38]. This allocation of resources has been crucial in not only increasing physical access, but also in subsidizing procedures for out-of-state travelers, thereby reinforcing Illinois’s role as a refuge state for abortion services.

The establishment of the Complex Abortion Regional Line (CARLA) in states like Massachusetts and Illinois demonstrates the anticipation for growing needs for out-of-state abortions and strategies implemented to streamline the coordination of care for abortion seekers [37]. By managing referrals, supporting emotional needs, and coordinating logistics, CARLA addresses several layers of barriers that might otherwise delay or prevent access to abortion services. This comprehensive approach not only enhances the efficiency of healthcare delivery, but also emphasizes the critical role of emotional and logistical support in abortion care—areas that have traditionally been underexplored in health policy initiatives.

However, the reliance on grants to subsidize abortion costs raises questions about the sustainability of such funding models [33]. While effective in the short term, these initiatives require ongoing financial support to remain viable. This dependency underscores the need for permanent policy solutions that ensure long-term sustainability without sacrificing the quality of care. Moreover, while initiatives like CARLA improve access and quality of care for individuals traveling out of state, they also reflect a broader systemic issue regarding unequal access to abortion services across the United States [25,28,35,36,37,38]. The necessity for such services highlights ongoing disparities that could be exacerbated by future political and legal challenges.

There is an exigent need to improve care and decision-making in obstetrical emergencies; particularly in states with restrictive abortion laws, it is crucial to address both the systemic and practical challenges faced by healthcare providers. The need for further safeguards and training is evident, as physicians often encounter systemic barriers and personal reluctance when required to provide lifesaving care under restrictive conditions [25,28,35,37,38]. This hesitance is not only a reflection of the legal risks involved, but also of the unclear guidelines under high-pressure situations. Healthcare leaders must prioritize ensuring that their institutions comply with the Emergency Medical Treatment and Labor Act (EMTALA), which mandates that in life-threatening conditions where an abortion is necessary, federal law supersedes state laws that restrict such procedures [34].

Despite this federal protection, there are reported instances where restrictive states not only overlook, but actively counteract EMTALA, leading to legal actions against physicians who perform abortions to save a patient’s life [39]. This legal contradiction not only puts physicians in a precarious position, but also jeopardizes patient care. Notably, the Supreme Court is on the verge of reinforcing that EMTALA supersedes state abortion bans, which would clarify the obligation of physicians to proceed with abortions when the life of the birthing person is at risk [34]. This forthcoming decision is expected to impact the landscape of reproductive healthcare by affirming federal protection and potentially reducing the legal uncertainties that currently hinder timely and life-saving medical interventions.

To navigate these complexities, obstetric care teams in abortion-hostile states need clearly defined protocols that delineate when and how they can legally and safely proceed with emergency procedures. Additionally, the introduction of training programs on national referral networks such as CARLA is imperative. These networks assist in coordinating care for patients who need to travel out of state to access necessary medical interventions [37,38]. Such training should include effective communication strategies to ensure that patients are fully informed about potential complications and the logistical aspects of obtaining care across state lines, which is vital in reducing delays and improving outcomes [37]. These teams must also ensure timely discussions with patients about potential complications and facilitate referrals to states where necessary care is accessible, considering travel time across state lines [37].

4.7. Limitations

Our analysis faced limitations due to the temporal restriction of the social media engagement studied, which spanned only eight days following the announcement of Roe v. Wade’s overturn. Despite sourcing from predominantly left-leaning news outlets, the discourse leaned conservatively. While we identified numerous themes, parts of our term co-occurrence network were obscured by repetitive, well-known abortion rhetoric from both sides, overshadowing smaller dialogues on other issues. Social media plays a key role in spreading misinformation within an echo chamber, making it harder to correct misinformation due to advanced technologies and algorithms [40]. The inclusion of posts with minimal character counts and those containing scripture may have skewed our dataset. Additionally, automating the coding of the abortion stance proved challenging, as users often employed the same term and often used sarcasm to provide an opposing viewpoint, complicating interpretation. Only 2–4% of the manually coded comments revealed the user’s abortion stance. The absence of independent observations further restricted our capacity for statistical analysis, impacting the breadth and depth of our findings. Gender API has inherent limitations and can only approximate gender, as it does not account for misgendering. The tool was developed primarily from a binary perspective on gender, and its accuracy is affected by users’ tendencies to mask their gender identities through ambiguous usernames. Gender, however, is a much more complex and nuanced concept. Additionally, the findings of this study, based on English-language comments in response to videos from U.S. news outlets, are not generalizable to non-Western social media users.

4.8. Future Research

We recommend conducting future research to explore YouTube videos that share narratives about maternal mortality and infant loss due complications, fetal abnormalities, and neonatal death to provide insight into the devastating impacts of abortion restrictions, delays to care, and subsequent losses on families. This research should also extend to examining how these restrictions contribute to socioeconomic and racial disparities and affect marginalized communities. Additionally, it is crucial to study the broader effects of delayed healthcare on pregnant individuals’ health and well-being. By analyzing personal narratives and testimonies, policymakers and healthcare practitioners can gain insights into the long-term adverse effects that these restrictions have on individuals [25]. Future studies in these areas are essential for informing effective health policies and interventions.

4.9. Advancing a Novel Research Approach

There were very limited examples of co-occurrence networks constructed using YouTube [16]. For researchers studying social movements like the overturn of Roe, YouTube is significant not only as a platform widely used by U.S. residents and activists for disseminating information and messages via videos, but also as a critical arena for political education, dissecting reasons for polarization, and incubating cultural conflicts [41]. This methodology presents a novel approach for analyzing social media data, with significant implications for public policy, regulation, and public health communication. Studying how social media influences public opinion through network visualization offers a way to educate voters and encourage informed decision-making on critical issues. This flexibility makes it applicable to a wide range of topics, particularly large-text corpora containing sensitive disclosures from unique informants.

Using dominant dyadic nodes to evaluate the discourse allowed us to assess both prominent and less prevalent reactions to phenomena [17]. This method, similar to purposeful sampling, involved custom overlays and the strategic selection of comments with nodes above the scale midpoint. It provided access to comments containing experiences and perspectives from unique informants, which would not be accessible in traditional research settings [30]. Our methodology extends beyond specific topics or populations, facilitating the study of marginalized communities and informing the development of multidisciplinary healthcare and educational strategies tailored to specific needs.

This mixed-methods approach transformed network visualization from a two-dimensional to a multidimensional perspective. By overlaying various variables, we examined the intersectionality of characteristics within our population and their manifestation in discussions. These overlapping variables can serve as predictors of behavior in specific populations and demonstrate associations between viewpoints, behaviors, and factors. This is significant because it enables us to tailor health-related strategies to improve health literacy, influence voter patterns, and modify behaviors in public health, politics, and beyond.

Additionally, this approach extends research reach to an international audience, enabling the analysis of complex global issues like international conflicts, healthcare challenges in developing nations, and geopolitical controversies. Combining social media data with term co-occurrence network visualization can illuminate underreported or censored aspects of international conflicts and political divides that spark social movements. By analyzing public sentiment and reactions, this method can help pressure governments to reconsider their positions. This approach not only showcases the transformative potential of social media in public engagement, but also amplifies the impact of research by fostering deeper, proactive involvement with societal challenges.

5. Conclusions

In conclusion, YouTube’s varied commentaries reflect a spectrum of perspectives that blend ethical and medical considerations. These discussions not only engage future voters effectively, but also provide crucial insights for policymakers and health educators, illuminating influential aspects of the abortion debate within U.S. society. By visualizing this discourse, educational content can be developed that addresses health risks and influences voting behaviors, thereby shaping health policy. YouTube plays an essential role in developing communication strategies that simplify complex medical and legal information, making it more accessible and impactful. Furthermore, social media discourse has played a pivotal role in transforming the abortion debate by shifting the narrative from blaming women to highlighting the potentially fatal consequences of restrictive abortion policies. One approach involves framing discussions around the ethical considerations of terminating pregnancies in cases where the fetus cannot survive post-delivery due to genetic conditions [28]. This method not only addresses medical and ethical aspects, but also centers on the compassion needed to support expectant parents facing difficult decisions and tragic outcomes [28]. Ultimately, this approach enhances public understanding and support for reproductive health services, leading to better access and informed decision-making.