How does the media shape our beliefs, attitudes, and behaviours? While many academics have investigated the impact of media on social and political outcomes, we need to learn more about the channels through which this effect operates. Two mechanisms can account for its significance. Individuals are persuaded to accept new information through the media (individual channel), but media also informs listeners about what others learn, facilitating coordination (social channel). I disentangle these effects by analyzing norms surrounding violence against women in Mexico by combining a field experiment with a plausible natural experiment. I compare the impact of a radio program when it is broadcast privately versus when it is posted publicly. There is no evidence to support the unique mechanism. However, the social channel increased the rejection of violence against women and support for gender equality while unexpectedly increasing pessimism about whether violence would decrease in the future.
ORDER WITH US AND GET FULL ASSIGNMENT HELP FOR THIS QUESTION AND ANY OTHER ASSIGNMENTS (PLAGIARISM FREE)
Political Science Research and Methods, July 2019, Volume 7, Issue 3, pp. 561-578
DOI: 10.1017/perm.2018.1 [Opens in a new window]
The 2018 European Political Science Association
Understanding the extent to which mass communication can influence social and political outcomes has been a central concern in the social sciences. Indeed, many scholars have demonstrated that media effects abound and cover a wide range of topics, from political support and electoral behaviour to violence perpetration. However, we need to learn more about the mechanisms underlying these effects. How do media influence people’s beliefs, attitudes, and behaviours? How does the media, in particular, influence social norms?
The process underlying media influence can be divided into two categories: (1) an individual or direct effect and (2) a social or indirect effect. In the former, the media informs people about new norms and persuades them to accept them (Bandura Reference Bandura1986; DellaVigna and Gentzkow Reference DellaVigna and Gentzkow2010). In the latter case, the information provided also serves as a means of coordination. Coordination is required because social norms can be thought of as coordination problems, that is, situations in which each person wants to participate only if others also participate (Mackie Reference Mackie1996; Chwe Reference Chwe1998). As a result, disseminating public information can improve coordination on that standard by fostering the development of shared knowledge (Mackie Reference Mackie1996; Chwe Reference Chwe2001).
While the individual mechanism would have an effect regardless of the method of dissemination, the social agent would be more potent when the dissemination was public. As a result, I contend that information has a different effect when it is transmitted privately and individually (e.g., through leaflets) than when it is transmitted through more social or collective channels (such as mass media or public meetings). How information is delivered is critical to fully comprehending the mechanisms underlying its influence. However, media has a general component, and media-related interventions in literature have always been public. As a result, media can induce common knowledge by design, preventing the separation of the social component from the individual part and making the task of fully understanding the micro-foundations of media influence daunting.
This paper fills that void by distinguishing between the extent to which media influence acts through the unique mechanism (via persuasion) and the extent to which it acts through the social mechanism (via higher-order beliefs). To accomplish this, I combine a plausible natural experiment with a randomized field experiment conducted in collaboration with UNESCO. I examine the effects of a norms campaign—a media (audio soap-opera) intervention—on a specific set of values and behaviours, namely attitudes and norms regarding violence against women.
The issue of violence against women is a significant and well-suited case for investigating the impact of media. First and foremost, violence against women is a worldwide issue. It is a violation of human rights with far-reaching negative consequences, ranging from direct physical and mental harm to economic losses at the individual and national levels. Second, researchers and policymakers have paid close attention to development programs aimed at improving women’s economic, political, and social status in recent years. Media and social norms marketing campaigns, with a particular emphasis on “edutainment,” have been a particularly popular type of intervention (e.g., Paluck and Green Reference Paluck and Green 2009). Improving our understanding of the mechanisms underlying these policy interventions is critical for improving their design and efficacy. Finally, the case of violence against women lends itself to research into the influence of media on social norms, as existing evidence suggests a link. According to Jensen and Oster (Reference Jensen and Oster 2009), the introduction of cable television in India exposed viewers to new information about the outside world and other ways of life, which reduced the reported acceptability of violence against women. However, this effect could also be explained by media publicity, which can influence social norms through coordination—that is, by influencing perceptions of what others consider desirable and thus promoting the rejection of violence because of the expectation that others will reject it.
The intervention manipulated the social context in which people could receive the program. To that end, from May to June 2013, I conducted research in San Bartolomé Quialana, a small rural, indigenous community in Oaxaca, Mexico, where I combined (1) a plausible natural experiment on the broadcast’s reach with (2) randomly assigned invitations to listen to the program. San Bartolomé Quialana is typical of rural communities where violence against women is prevalent (UNESCO 2012). With these elements in mind, an audio soap-opera program was broadcast over the community loudspeaker to challenge gender norms and, in particular, to discourage violence against women. However, this specific loudspeaker had a unique feature. Because of topography issues, some community members could not access the broadcast. This is significant because only the area beyond the reach of the loudspeaker provides the leverage to test the individual mechanism. Households in this area were invited randomly to listen to the program, individually and privately, on an audio CD (Audio CD treatment). Individuals were unaware that others were listening to the program, preventing the creation and coordination of common knowledge and thus isolating the individual effect.
On the other hand, the area within the range of the loudspeaker allows us to put the social mechanism to the test. In this area, the program was broadcast only once so that households could listen to it (Village Loudspeaker treatment). Furthermore, families were randomly invited to attend the program in a community meeting-style setting. That is, they were asked to listen to the same program at the same time as other members of the community, but to do so in person (Community Meeting treatment). This treatment aims to match the invitation component of the Audio CD treatment, which may aid in the generation of common knowledge. Table 1 shows how the design resulted in four groups.
Table 1: Research Design-Created Groups
I discovered that social effects, rather than individual persuasion, drive media influence by surveying 340 people in 200 households and measuring norms, attitudes, and behaviour. Social interactions, such as community meetings, are not always required for such effects. The evidence suggests that the social channel decreased personal and perceived social acceptance of violence against women, increased support for gender equality roles, and decreased optimism about the future of violence. The results, however, show that the individual channel had no effect.
A competing explanation is that systematic differences may exist between areas within and outside the reach of the loudspeaker, which could potentially affect beliefs and behaviours related to violence against women. This is not the case, and I demonstrate that a set of individual and household characteristics is balanced between the two areas. Another concern is that the design could have been vulnerable to spill-overs due to the town’s small size and the nature of the treatment conditions. However, as I will explain further below, the experiment was designed to address this issue to the greatest extent possible, and the presence of spill-overs would bias against the paper’s findings.
This study adds to the growing body of evidence that exposure to information provided by the media can influence a wide range of attitudes and behaviours. This paper adds to the body of knowledge by empirically distinguishing between media influence’s personal and social effects. This is significant for several reasons. First, it advances our understanding of the mechanisms by which media influence attitudes and social norms; these estimates resolve an ongoing puzzle in the empirical literature on media influence. Second, such estimates are critical when considering policy interventions. Third, it sheds light on how media interventions can have negative or unintended consequences.
CHANGE IN THE MEDIA AND THE MICRO FOUNDATIONS OF SOCIAL NORMS
Norms are important because they are behavioural standards based on widely held beliefs about how individual group members should behave in a given situation. As a result, these customary rules of behaviour coordinate individuals’ interactions with others (Young Reference Young2008). As a result, they have a significant influence on shaping individual behaviour, including discrimination and violence against a specific group, such as women. Norms can protect against violence, but they can also encourage and support its use. Acceptance of violence, for example, is a risk factor for all types of interpersonal violence (Krug et al. Reference Krug, Dahlberg, Mercy, Zwi and Lozano2002). Indeed, social norms, gender stereotypes and societal expectations shape and reinforce behaviour and attitudes toward violence against women. These norms persist in society because people prefer to conform to the expectation that others will also work (Lewis Reference Lewis1969; Mackie Reference Mackie1996). Participation in such norms and behaviours (or the spread of new ones) is thus a coordination issue. This is because people are motivated to coordinate with one another when there are strategic complementarities: An individual only gains social approval if a sufficient number of people express their attitudes and behave similarly. Social sanctions can be imposed if others do not join those with different expressed attitudes and behaviours (Coleman Reference Coleman1990). For example, these sanctions could include shaming, shunning, or any other form of social ostracization (Paluck and Ball Reference Paluck and Ball 2010). Other scholars contend that norms can exist without the threat of punishment. Other mechanisms that support criteria include negative emotions such as guilt or shame triggered when norms are internalized and (ii) the desire to avoid intrinsic costs resulting from coordination failure (Young Reference Young2008). In short, beliefs about the acceptability of a particular behaviour, such as violence against women, play an essential role in explaining its occurrence (Mackie Reference Mackie1996).
One might argue that violence against women is motivated by different forces because it is often a private interaction in the home. People are unlikely to engage in violent acts simply because they believe others will. However, a person who engages in violence may frequently consider the larger social context. For example, whether people who learn about these actions will recognize it as a crime and report it. When discussing the psychology of abusive men, Bancroft makes the following point: “While a man is on an abusive rampage, verbally or physically, his mind maintains awareness of several questions: ‘Am I doing something that other people could find out about, so it could make me look bad? Is there anything I’m doing that could land me in legal trouble?'” (Reference Bancroft2003: 34). (Reference Bancroft2003: 34). Furthermore, while the physical consequences of domestic violence can be concealed publicly, other gender inequality-related behaviours, such as early marriage or a lack of financial independence, are more visible.
As a result of these considerations, numerous policies and programs have launched ambitious campaigns to address social issues such as violence against women by encouraging social norm changes. Many of these strategies involve media-driven information interventions, such as soap operas on TV or radio (Paluck and Ball Reference Paluck and Ball 2010). These efforts raise fundamental questions about the extent and conditions under which media can influence social norms and the micro-foundations of such a process. Media influence can be divided into two categories: (1) individual or direct effects and (2) social or indirect effects.
Persuasion is used in the individual or direct effect of media. The emphasis is on the content’s persuasive power, which sparks an individual learning process, updating personal values and beliefs (Staub and Pearlman Reference Staub and Pearlman2009; DellaVigna and Gentzkow Reference DellaVigna and Gentzkow2010). This “individual educational process” is consistent with arguments advanced by social learning theory, according to which the educational effect of media works through educational role models (Bandura Reference Bandura 1986). These educational role models can instruct and transmit knowledge, values, and behaviours, among other things.
The Social Impact
The media can also have an impact through a social mechanism. Because media is a public medium, its influence is rooted in the fact that it can provide information in a way that improves coordination on a norm or action through the creation of shared knowledge (Chwe Reference Chwe2001). Publicly available information aids individuals in developing an understanding of their shared beliefs. Individuals are not only forced to update their personal beliefs due to public information, but they can also update their beliefs about how widely they are shared (Morris and Shin Reference Morris Shin2002). Public information is used to know that others received the data and that everyone who received the news knows that everyone else who received the statement knows that everyone else who received the information knows this, and so on, resulting in common knowledge. In this vein, some authors argue that “attempts to change public behaviours by changing private attitudes will be ineffective unless some effort is also made to bridge the public-private divide” (Miller, Monin and Prentice Reference Miller, Monin and Prentice2000: 113).
Furthermore, a social effect may exist even without an individual impact. People’s behaviour and publicly expressed attitudes may change, but not necessarily their private beliefs. This inconsistency between private and public is referred to as pluralistic ignorance, which describes situations in which the majority of members of a group privately reject group norms while believing that the majority of members accept them (Miller and McFarland Reference Miller and McFarland 1987). Without an individual effect, such erroneous social inference facilitates social development.
As a result, the method of dissemination is a significant driver of people’s beliefs (and higher-order beliefs) and, thus, their behaviour. A public transmission of information, as opposed to a private one, facilitates the creation of common knowledge, increasing its influence on social norms. Footnote 1 This is the paper’s central Hypothesis:
1st Hypothesis: (Common Knowledge). When information is delivered publicly, its impact on attitudes and norms is more significant.
A public method of dissemination facilitates but does not guarantee shared knowledge and coordinated action (Chwe Reference Chwe1998). Individuals may need to be sure that others receive the information. Thus everyone who received such information may wonder if everyone else who received the report is confident that others received the data, and so on. Public promotion may still be influenced by uncertainty about higher-order beliefs. However, the type of social interactions generated by the conditions under which norms are promoted influence this uncertainty. Face-to-face interactions, such as community meetings, can help to increase certainty (Mackie Reference Mackie1996; Chwe Reference Chwe2001).
To address this heterogeneity in public information dissemination, I investigate the extent to which different levels of uncertainty and potential social interactions moderate norm diffusion. Within the common knowledge framework, I investigate whether the availability of information is a sufficient condition for media influence and whether face-to-face interactions increase such power. In other words, I divide Hypothesis 1 into two secondary hypotheses:
2a Hypothesis: (Public Signal). A public mode of delivery is both required and sufficient for information to influence attitudes and norms (i.e., no social interaction is required).
2b Hypothesis: (Face-to-Face). A public method of information delivery with face-to-face interactions improves the impact of information on attitudes and norms.
CAMPAIGN OF UNESCO: A MEDIA INTERVENTION IN SAN BARTOLOMÉ QUIALANA
To put these hypotheses to the test, I organized a media intervention in San Bartolomé Quialana in collaboration with the UNESCO Office in Mexico. San Bartolomé Quialana (or simply Quialana) is a small rural indigenous community in Oaxaca. Rural municipalities share their main characteristics throughout Mexico. (For more information, see Section A1.) For this paper, cultural homogeneity is an essential aspect of Quialana. In 2010, for example, 2412 of the 2470 inhabitants were born and raised in Quialana. This is significant because the ability to focus on a single community while holding cultural and social aspects “constant” allows researchers to more easily isolate the individual-level informational mechanisms that drive media influence on attitudes and social norms.
The Opera of Soaps
The intervention was an audio soap opera aimed at challenging gender norms and discouraging violence against women. Un nuevo amanecer en Quialana (A new dawn in Quialana) was created in collaboration with a regional partner non-governmental organization (NGO) and consisted of four 15-minute episodes for a total running time of 57 minutes. The soap opera was embedded in the local context, with common reference points such as “Tlacolula’s market,” as this can help the audience directly relate to the situations depicted, which can increase its impact (La Ferrara, Chong and Duryea Reference La Ferrara, Chong and Duryea2012). The plot revolved around a young couple in Quialana who fell in love and started a family. The story was written so that the main male character gradually evolved from a loving and caring husband to a violent and aggressive figure. According to research, the male figure should not be portrayed as an utterly violent character from the start so that listeners can establish a rapport with him and not dismiss his behaviour as an exception (Singhal et al. Reference Singhal, Cody, Rogers and Sabido2003). Furthermore, the script’s language employed injunctive norms (Paluck and Ball Reference Paluck and Ball 2010). Instead of arguing that “beating women is wrong,” the soap opera might say, “citizens of Quialana believe that beating women is wrong.” This biases against the central Hypothesis of this paper because those in the Audio CD treatment are exposed to these injunctive norms. However, one limitation of the narrative is that it needed to include channel factors to act out these norms. Footnote 2
Un nuevo amanecer en Quialana was broadcast over the community loudspeaker as a special event: the premiere of the first-ever locally produced soap opera and the first time the community loudspeaker was used for entertainment purposes.
Design of Research
The study’s design incorporates two sources of variation. The social context in which people can receive the intervention is manipulated explicitly by (1) exploiting arguably exogenous variation generated by community topography (i.e., within community variation of “broadcast access”) and (2) randomly inviting households to listen to the program. Each one is described in greater detail below.
Loudspeaker, Topography, and Sound Check Experiment
While Quialana did not have a local radio station at the time of the intervention, it did have a loudspeaker on top of the Town Hall in the community’s centre. Before the intervention, the loudspeaker primarily and infrequently announced sales of small-scale household goods such as construction materials such as bricks or other livestock. It was never used for other announcements such as news, weather, etc. For these reasons, the variation in the reach (and sharpness) of the loudspeaker described below surprised many of our local partners, who had previously assumed that nearly everyone in the community had access to the occasional announcement.
I define two areas within Quialana based on variation in the loudspeaker’s reach: (1) the area within the loudspeaker’s reach and (2) the area outside the loudspeaker’s reach. This variation within communities is primarily a result of topographic conditions. For example, there is a 500-foot altitude difference between one end of the municipality and the other. More specifically, in some areas, the slopes become steep enough that sound cannot travel with clarity. Footnote 3 This means that the source of variation is primarily a function of altitude difference rather than distance to the loudspeaker. That is, two households can be located at the same distance from the loudspeaker, and yet only one of them will be within the loudspeaker’s range. Figure 1 depicts the reach of the loudspeaker as determined by a sound-check process from the ground (further explained in Section A2).
Fig. 1 San Bartolome Quialana and the reach of its loudspeaker Note: Population (green), households (brown). The red line represents the range of the loudspeaker. The loudspeaker is represented by a red-filled circle.
A valid concern is that systematic differences between these two areas may exist related to attitudes and norms regarding violence against women. One of the benefits of conducting the study within a single, small (slightly more than a mile long) community is that it allows researchers to capitalize on cultural homogeneity and alleviate concerns about potential differences. There is no qualitative evidence of sorting into one area or another based on attitudes and behaviour related to gender inequality based on informal and formal discussions with UNESCO personnel, NGO partners, and Quialana citizens. According to qualitative analyses and focus groups conducted by UNESCO, violence was widespread throughout the community (UNESCO 2012). I supplement these firsthand accounts with quantitative data. I rely on data from the 2012 National Housing Inventory, in particular, to demonstrate that a battery of individual and household characteristics are balanced between the two areas (Table A2).
While the focus on a small community, combined with qualitative and quantitative evidence, strengthens the plausibility of the natural experiment, such interpretation may be jeopardized if unobservables associated with each of the two areas are also associated with attitudes and behaviours toward women. This should be considered when interpreting the results.
Audio CDs and Community Meetings were chosen at random.
The Community Meeting and Audio CD treatments were created by randomly inviting households in each area to listen to the soap opera via systematic sampling with a random start. In this case, the experiment maintained the content of the media program while changing the social context in which it was received. Households in the vicinity of the loudspeaker were invited to listen to the program in the cafeteria next to the Municipal building (i.e., Community Meeting). Families were invited to attend to it in their homes using an audio CD in the area beyond the reach of the loudspeaker (i.e., Audio CD treatment). The regional partner NGO served as the public face of the treatments, which were presented as part of a project to establish a local radio station, with no mention of UNESCO’s involvement.
To put the unique mechanism to the test, the invitation to listen to the soap opera (via audio CD) had to be delivered privately to the household. Here, care was taken to avoid families believing that other homes were also receiving the program—though, as previously argued, this would bias against my hypotheses. Enumerators were trained to keep any material that might indicate that other households were being approached out of sight. Furthermore, when contacting families, enumerators emphasized that the audio CD was a pilot program, a one-time chance to preview and provide feedback. While not explicitly stating that the household was the only one chosen to receive the audio (to avoid deception), enumerators were trained to imply such a possibility and frame such an opportunity as something very novel, exclusive, and private—which could explain the perfect level of compliance. As a result, audio CDs were distributed along with a short questionnaire designed as a listening-test device: the enumerator would leave the audio CD and questionnaire sheet and then return a few hours later to pick up the sheet, and compliance was determined to be 100 percentFootnote 4. Because of this setup, and based on enumerator comments, all family members were present at the time, and reportedly all listened to it in some cases but not others.
To put the social mechanism to the test, the design created a comparable treatment group, the Community Meeting, in which the invitation to listen to the soap opera corresponds to the invitation component of the Audio CD treatment. Furthermore, the Community Meeting provides an opportunity to investigate the effects of public information. This treatment may increase the level of specific individuals about others receiving the notification by creating a particular type of social interaction (or at least knowledge about it), namely the community meeting. At the same time, other potential interactions facilitated by the forum, such as deliberation, may confound this common knowledge mechanism. To be clear, there was no deliberation at the community meeting (out of respect to other listeners, conversations were not allowed). However, a review and discussion could have taken place after the meeting. Because the creation of shared knowledge facilitates these interactions, the design can disentangle the social and individual mechanisms of media influence (however, it cannot unbundle face-to-face certainty from deliberation). Finally, roughly one in every four households invited to the Community Meeting went to the cafeteria—that is, they followed through on the Community Meeting treatment. People did stop by the Town Hall during the broadcast, just outside the restaurant where the community meeting was taking place, and listen to the soap opera (or at least parts of it) from just outside. According to other accounts, many people sat at home and watched the soap opera.
To fully comprehend the social mechanism, I investigate whether public dissemination of information is a sufficient condition for influencing norms and the extent to which face-to-face interactions can enhance the effect on standards. To address this, the design created a public treatment that did not impose such social interactions: the Village Loudspeaker treatment consisted of households that could listen to the broadcast by being within the loudspeaker’s reach but were not in the Group condition.
Finally, the baseline group consists of households not within hearing distance of the loudspeaker and did not receive the audio CD. Table 1 summarizes these four conditions.
An unbiased estimation of the mechanisms is based on two dimensions: one that facilitates the creation of shared knowledge in social conditions, and the other that prevents it in individual needs (i.e., no spill-overs). For the broadcast to facilitate the creation of common knowledge, those who listen to it should be aware that others are also hearing it. This is less of an issue in the Community Meeting treatment because the information is explicitly provided to the household, so they are aware that others are also receiving the invitation, and so on. However, a person receiving the Village Loudspeaker treatment may believe she has heard the broadcast because she lives near the Town Hall or thinks she has particularly good hearing, but only a few of her neighbours have heard it. I make two attempts to address this. First, I include distance to the Town Hall as a control covariate in the empirical analysis. This variable is also an important covariate because it serves as a proxy for population density, which could confound violence perpetration. Second, as discussed below, the empirical strategy is based on estimating intention-to-treat effects (ITT) precisely because individuals may fail to comply with the treatment—in the case of the Village Loudspeaker, individuals may not listen to the program or be aware that others are also listening to it, and so on. As a result, it is a conservative or lower-bound estimate.
The second dimension is related to the idea that those receiving individual treatment should be unaware of other therapies. The design was vulnerable to spillovers due to the small size of the town and the nature of the treatment conditions. However, such spillovers would be biased against the paper’s central Hypothesis. This is due to the possibility that those in the individual situation will discover that other people are also receiving the soap opera. Nonetheless, the Community Meeting invitations were distributed on a Friday to minimize potential spillovers. Both treatments were implemented the following day: the Audio CD treatment began early Saturday morning, and the Village Loudspeaker and Community Meeting broadcast began late Saturday evening.
Similarly, the design faced a trade-off between minimizing these concerns and maximizing treatment intensity. The ideal for the former was to shorten the time between treatments and the survey. A weekly soap opera over several weeks or months was an option for the latter. Given that the primary goal of this study was to investigate the underlying mechanisms of media influence, I prioritized addressing spillover concerns over treatment intensity. Nonetheless, experiments with one-day or even one-hour interventions have found profound effects (e.g., Ravallion et al. Reference Ravallion, Walle, Dutta and Murgai2015). Given these considerations, the norm intervention was implemented as a one-day event, with surveys administered over the next few days.
Evaluation of Results
The regional partner NGO also served as the survey’s public face, presenting it as a way to gather Quialana citizens’ opinions to inform a community radio initiative. Footnote 5 Three questions in the survey measured respondents’ beliefs and estimation of others’ beliefs and actions regarding violence against women. Three other questions measured attitudes and individual activities related to it. As a result, I assess six outcomes of interest, which I describe in detail below.
The first dependent variable is a measure of Personal beliefs, which is intended to capture the extent to which people believe and are willing to state that violence against women is a recurring problem in the community. “Do you believe that violence against women occurs here in Quialana?” was the question. “It was coded from 1 (“No, this never happens here in Quialana”) to 5 (“This happens far too frequently in Quialana”). Given the qualitative evidence of widespread violence in Quialana (UNESCO 2012). This item was created to capture the respondent’s personal beliefs about the desirability of (and thus willingness to expose) specific actions rather than a factual scenario. In other words, the intuition behind this question is to capture the shift from a perception in which “husbands are never violent to their wives—they may engage in some aggressive behaviour, but that is not violence” to a situation in which “that” type of behaviour is recognized as violence. It is also socially acceptable to judge it as a severe problem.
Perceived social rejection is the second variable of interest. That is the extent to which an individual believes that violence is a problem in the community. The question was, “Do you think that that the community, the neighbours, and other families see violence against women as a serious problem here in Quialana? “with responses ranging from 1 (“No, they do not see it as a problem at all”) to 4 (“They see it as a serious problem that must be addressed”). As with the previous question, this item aims to assess the shift in norm perception from one in which violence is tolerated (e.g., the community experiences violence but sees it as routine and excusable) to one in which violence is rejected.
Expectations about the future, the third variable, assesses individuals’ beliefs that this type of violence will decrease. “Do you think the next generation of Quialana males…?” was the question. “, with responses ranging from 1 (“Will continue to abuse women”) to 4 (“Will never abuse women”). In other words, higher values indicate more optimistic views of the future.
While these three measures can retrieve individuals’ perceptions of norms relating to violence against women, they do not directly measure individual attitudes, beliefs, or actions regarding gender roles or domestic violence. This is addressed in outcomes four through six, which include behavioural development embedded in the survey.
The fourth outcome, Value Transmission, assesses how likely the respondent is to educate a child on gender equality values. This captures parents’ decisions about which weights to instil in their children, which are influenced by perceived prevailing societal values (Tabellini Reference Tabellini2008). It focuses on attitudes toward equality regarding household chores, which many see as one of the most difficult challenges to achieving gender equality (World Bank 2012). “Would you educate your child so that domestic chores, such as laundry and cooking, are as much the responsibility of men as they are of women?” the question asked. “, with one indicating support for this type of Education and 0 indicating opposition.
The fifth variable captures an individual’s reaction to a violent episode. “What would you do if you saw or heard a neighbour’s wife being beaten by her husband?” the question asked. .” Responses are collapsed into a binary variable in the following way: Reaction to violence is coded one if the respondent answers that they would interrupt the couple to stop the violence and call the police to intervene, and is coded 0 if the respondent answers that they would not take any action at the time.
The sixth variable returns a behavioural result. Survey respondents were asked if they would sign a petition in support of the formation of a support group for victims of violence against women; the variable Petition signature is coded one if they signed the petition, 0 otherwise.
To account for multiple testing, I also examine an Index variable constructed from the previous variables’ standardized inverse-covariance-weighted (ICW) averages. The resulting index is scaled in control group standard deviations, with higher values indicating higher levels of rejection and perceived rejection of violence against women and increased support for gender equality.
Gender, Age, and Education were the three critical covariates collected. A total of 200 households were surveyed, accounting for roughly one in every three homes in Quialana. When available, both male and female household heads were polled. This resulted in a total of 340 observations. Table A8 displays descriptive statistics, while Section A4 states randomization checks.
The empirical strategy is based on calculating ITT effects. However, in this case, the invitation to the Community Meeting (i.e., the assignment to treatment) corresponds to the theoretical motivation for the treatment. In other words, the invitation provides specific information about how the soap opera will be distributed (i.e., there will be a broadcast and an event where people can receive the program together), facilitating the creation of common knowledge. This has implications for estimating local average treatment effects (LATE), as it may be interpreted as a violation of the exclusion restriction, which precludes an unbiased estimation of the LATE, providing yet another reason to focus on ITT estimation.
I use ordinary least squares with two empirical strategies: (1) Community Meeting versus Audio CD and (2) all four treatment conditions. Footnote 7
Individual and Social Mechanisms: Community Meeting vs Audio CD
The first empirical strategy involves comparing the Community Meeting and Audio CD treatments, as follows:
Where I denotes individuals and h households; Y I,h denotes the outcomes of the interests mentioned above (continuous variables are expressed in standard deviations of the distribution of responses in the Audio CD condition); and Community Meeting denotes whether the household was invited to the meeting. In this estimation, those receiving the Audio CD treatment—those living outside the loudspeaker range and asked to listen to the audio CD—comprise the baseline category. For efficiency, I include a vector of controls, which consists of an indicator for Female gender, respondent’s Age, and Education, which denotes a schooling indicator for whether the respondent (1) never attended school, (2) attended but did not complete primary school, or (3) completed primary school. I also account for the natural logarithm of the distance between the household and the Town Hall in meters. After assigning treatments to families, I cluster the standard errors at the household level.
In Equation 1, the coefficient of interest is; it captures the social mechanism underlying norms diffusion. Hypothesis 1 predicts a value greater than zero. Nonetheless, I put it to the test with a two-sided test.
Total Sample: All Treatment Conditions
The Community Meeting estimates can isolate the social effects caused by shared knowledge. They may, however, be influenced by the increased certainty created by the face-to-face interaction, and they may be confounded by other social interactions, such as deliberation, facilitated by the community meeting. To address this and determine the extent to which a public mode of delivery is a sufficient condition for influencing norms, I used the entire sample as follows:
Where Y I,h denotes the results (continuous variables are expressed in standard deviations of the distribution of responses in the baseline condition). Control vectors, Community Meetings, and Audio CDs are defined as before. VillageLoudspeaker indicates whether a household is within hearing distance of the loudspeaker but was not invited to the meeting. Finally, those living in the particular area but not receiving treatment constitute the baseline category.
The coefficients of interest in Equation 2 are, and. They assess the impact of the intervention and, by definition, can shed light on the various potential mechanisms. Hypothesis 1 predicts > and >, and Hypothesis 2a predicts >0 with >, while Hypothesis 2b predicts >0 with >.
Audio CD versus Community Meeting
Table 2 shows each outcome of interest based on two different specifications. The first shows a specification that only includes the Community Meeting indicator (i.e., ), whereas the second consists of the vector of control covariates.
Table 2 compares a community meeting to an audio CD.
Standard errors are shown in parentheses at the household level.
Covariates include Age, gender, Education, and distance.
+p<0.10, *p<0.05, **p<0.01. According to the influence on personal beliefs results, those invited to the community meeting were more likely to state that violence against women is a recurring problem in Quialana than those invited to the Audio CD. When controls are introduced, the parameter estimate gains precision but remains stable, ranging from 0.33 to 0.35 SD relative to the Audio CD condition (p-value=0.065 and p-value=0.052, respectively). The evidence for perceived social rejection is consistent, stable (0.66 and 0.65) and precise estimates. The effects of the community meeting on future expectations are adverse, very stable (0.48 and 0.42), and statistically significant at conventional levels, implying that those invited to the discussion became more pessimistic about reducing violence in the future. Several factors could explain this arguably perverse effect. While the community meeting may have induced coordination around a new injunctive norm (i.e., people in Quialana should reject violence), it may also have raised awareness and facilitated coordination around a more subtle descriptive example, namely that violent behaviour is prevalent in the community. This more precise belief about the community's current situation, combined with the fact that the soap opera provided no channel factors to act upon it, may have resulted in pessimistic expectations for the future extent of violence. Another explanation is that, due to the new shared knowledge, people may expect more opposition to violence against women, which could lead to a backlash. For example, if more women speak out against violence, a subset of men may respond more violently. While the data does not allow me to rule out or pinpoint a specific explanation, it does indicate that a social mechanism drives this effect. Individual action analyses also support the social mechanism. Those invited to the community meeting were 16 percentage points more likely (Model 8) to say they would educate their children on gender equality values, 20 percentage points more likely to react to a violent event (Model 10), and 20 percentage points more likely to sign the petition (Model 12). The ICW Index analysis supports these findings. Subjects invited to the community meeting have a 0.45 SD higher index of responses than those invited to the Audio CD. To address concerns about the natural experiment's plausibility, Table A11 replicates the analysis by limiting the sample to households within 300 meters of the Town Hall, yielding similar results. The evidence is overwhelming. A social channel is primarily responsible for media influence, measured by changes in beliefs, attitudes, and behaviour. However, as evidenced by thoughts about the future prevalence of violence, creating common knowledge may also facilitate a more detailed view of the status quo, thus setting negative expectations about future change. Every Treatment Condition Table 3 shows the results for the entire sample, both without and with controls. Table 3 displays all treatment conditions. Standard errors are shown in parentheses at the household level. Covariates include Age, gender, Education, and distance. ICW=inverse-covariance-weighted. +p<0.10, *p<0.05, **p<0.01. Personal and perceived social rejection analyses show that social mechanisms drive the informational effects on beliefs and norms. When looking at future expectations, the estimated parameters for social treatments are similar in size, ranging from 0.20 to 0.24 and displaying a negative sign. The Audio CD parameters, on the other hand, are positive but far from statistically significant. These preliminary findings support both the community meeting and the Village Loudspeaker treatments. While individual attitudes and actions support the social mechanism, the evidence for the community meeting is more substantial, supporting Hypothesis 2b. When looking at the ICW Index, a similar pattern emerges. In addition, I calculated several F-tests for coefficient inequality. When either of the social conditions is compared to the Audio CD (), there is a statistically significant difference at conventional levels, supporting Hypothesis 1. When the social mechanism is examined further, the evidence shows that publicness can be a sufficient condition for norm diffusion, helping Hypothesis 2a. Some evidence suggests that face-to-face interactions can enhance such effects, supporting Hypothesis 2b. As before, I replicated the analysis by looking at households within 300 meters of the Town Hall and found the same results (Table A12). These findings support the notion that social mechanisms are the primary drivers of media influence on attitudes and norms. DISCUSSION The extent to which the results represent a one-off case in a unique setting is a valid concern when interpreting the results. As previously stated, Quialana, like many other municipalities in Mexico, has a high level of media consumption and issues with gender inequality and violence against women. Similarly, Mexico, as a large and diverse society aiming to empower women to overcome social challenges, shares many similarities with other developing and even developed countries. (For a more detailed discussion, see Section A6.) However, to what extent are the findings of this study externally valid in the sense that they generalize beyond Quialana? While numerous variations in context or treatment design could change the estimates presented here, the results still speak to a plausible phenomenon; the notion that public information, through shared knowledge and coordination, can induce differences in norms and behaviour is frequently stated as a general proposition rather than as one that applies to a specific context (Chwe Reference Chwe2001). Three specific findings merit further investigation. First, the negative results on future expectations were arguably unexpected. Understanding the conditions that cause and can prevent these types of backlashes (e.g., emphasizing channel factors) is theoretically and policy-relevant. Second, the mixed results of the Village Loudspeaker indicate the need for further investigation into the conditions under which public information is sufficient to influence norms and the conditions under which securing common knowledge through social interactions is necessary. Third, the Audio CD results indicate that private persuasion was ineffective. From the perspective of a community where social norms are deeply embedded, this result is not surprising precisely because it lacks such a link with the community. On the other hand, individual persuasion may be more effective in less sensitive issue areas, where social pressures may carry less weight. Finally, there may be questions about whether the changes in reported attitudes reflect differences in behaviour or simply in reporting. Despite the behavioural evidence on the petition signature, one might be concerned that public treatments only change what respondents believe other people want to hear and see about the acceptability of violence but do not change the incidence of abuse. However, directly observing people in their homes makes it easier to distinguish between changes in reporting and changes in behaviour. Even if media interventions only affect what is reported, it still represents a shift in social norms and progress. Changing social norms is a necessary (Jensen and Oster Reference Jensen and Oster 2009) step toward changing the desired outcomes (Mackie Reference Mackie1996). CONCLUSION Exposure to information provided by media outlets is known to influence a wide range of attitudes and behaviour. However, little is known about the precise mechanisms underlying such influence. Such effects can be explained by two broad means: a unique agency based on persuasion and a social tool based on higher-order beliefs and coordination. This paper investigates these mechanisms and deciphers their effects at the individual level by examining attitudes and norms toward violence against women. The evidence presented here supports a consistent narrative: media influence on attitudes and social norms is primarily driven by social effects rather than individual persuasion. First, I demonstrate that a public method of delivery reduced personal and perceived social acceptance of violence against women while increasing support for gender equality roles, whereas a private practice, had no discernible effects. I also demonstrate that public information is not a panacea, as it has increased scepticism about whether violence will decrease in the future. Second, I show how a pure general delivery method (i.e., without social interactions) can be necessary and sufficient for influencing attitudes and norms. Overall, understanding the interaction between individual beliefs and various types and sources of information can shed light on the social mechanism proposed here.