6

Mechanisms of Attitude Change

Jocelyn Finlay (Harvard University) served as moderator for this session. She explained that presenters would focus on the following question: What are the mechanisms by which changes in attitudes toward and

opportunities for girls and women affect changes in attitudes toward and use of contraceptives (and vice versa)?

SKILLS THAT EMPOWER: MEASURING ATTITUDINAL CHANGE IN THE CONTEXT OF INTERVENTIONS IN BANGLADESH

While empowerment is an important goal for adolescent girls’ programming, empowerment remains elusive and unobserved in most quantitative surveys. Sajeda Amin (Population Council) presented results from a study that measured patterns of empowerment among early adolescent girls in Bangladesh, which ranks fourth-highest in the world and the highest in Asia in rates of child marriage. The study was based on data from the Bangladeshi Association for Lifeskill Income and Knowledge for Adolescents (BALIKA) Program. BALIKA aims to impart skills and build empowerment for girls with the goal of changing child marriage.

However, many of the empowerment indicators used in this analysis were based on research on women’s empowerment, so an important goal of the analysis was to determine whether these same constructs could also be adapted to adolescent girls. Specifically, the analysis focused on three central questions:

1. Do distinct classes of empowerment emerge among girls in early adolescence, and if so, how are they characterized?
2. Are background characteristics of wealth, parental education, religion, and marital status predictive of latent class membership?
3. Did the BALIKA project’s different skill-building strategies influence girls’ probability of belonging to a certain class of empowerment?

BALIKA is an intervention-based research project that tested three intervention strategies—education support, gender awareness, and livelihood skills—in a cluster-randomized controlled trial within 96 Bangladeshi villages. BALIKA leveraged considerable community engagement in all intervention arms, though the skills emphasized were different. Qualitative and quantitative data were collected from 9,000 girls (ages 12–19) from 2013 to 2015 (with a 14.1% attrition from baseline to endline). Amin said that BALIKA had a striking impact on child marriage, with a 30 percent reduction in all intervention arms compared to controls.

Outcome measures from BALIKA included workforce participation, time spent in school, learning outcomes (numeracy and English), school retention, child marriage, and empowerment. Amin highlighted that the gender awareness intervention arm appeared to have the most consistently positive impact on these outcome measures compared to the other two interventions, particularly on time spent in school, workforce participation, and learning outcomes.

To assess empowerment, the project used multiple indicators of mobility (e.g., access to school, adolescent center or NGO, library, playground), social support (e.g., networks of friends), participation in society (e.g., affiliation with community groups or sports), freedom from family domination (e.g., ability to socialize with people from other religions, ability to disagree with parents), and access to information (e.g., recent TV watching, mobile phone ownership, basic computer skills). Two attitudinal measures—attitude toward gender roles and attitude toward violence—were also included. Latent class analysis was used to identify relationships between these seven classes of empowerment indicators. A latent class regression analysis was used to assess predictors of class membership and included four background characteristics for each girl: household wealth, marital status, religion, and parents’ level of education. The BALIKA study arm (i.e., gender, education, livelihood, control) was also considered.

The analysis revealed three distinct classes of girls at baseline. One class (about one-third) provided relatively unempowered responses to the survey. The second class of girls was very mobile and socially active but also very accepting of traditional gender roles and violence. The third class of girls—while less mobile and social than the second class—had more progressive attitudes toward gender roles and violence. At endline, a fourth class of girls emerged; nearly one-third of the girls shifted toward more empowered responses on both the mobility and social indicators as well as the attitudinal indicators. Household wealth was associated with class membership at baseline but not at endline, whereas marital status, religious affiliation, and parents’ education level were predictive of class membership throughout; more empowered girls tended to be unmarried and non-Muslim and to have parents with more education. The BALIKA study arm was also predictive of class membership; girls who participated in the gender awareness and education support arms were significantly more likely to belong to the emergent fourth, most empowered class.

Overall, BALIKA increased empowerment among girls: the proportion of girls belonging to the least empowered class diminished from 32 percent to 17 percent after the BALIKA intervention. Furthermore, Amin described BALIKA as an equalizer that promoted empowerment, in terms of both wealth and other differences, across girls.

EXPERIMENTAL EVIDENCE ON CHANGING GENDER NORMS, TACKLING MISINFORMATION, AND DECREASING COSTS: IMPLICATIONS FOR FAMILY PLANNING

Sarah Baird (George Washington University) enumerated several demand-side drivers of family planning for women in low- and middle-income countries, including opportunities in employment and education, misinformation (e.g., inadequate family planning counseling, provider bias,

misunderstanding of risks), poverty, and gender and social norms. These drivers interact in important and complicated ways that necessitate multipronged, context-dependent approaches to identify solutions, which Baird pointed out may require long-term research given that drivers such as gender norms are slow to change. Baird shared the following definition of gender norms from a recent paper, noting that some interventions will try to improve family planning outcomes by operating within gender norms while others will attempt to change them:

Baird presented evidence from several studies of family planning intervention efforts. The first was a 2014 experiment that provided married women in Zambia with vouchers for free contraception.² The intervention attempted to work within existing gender norms by varying whether women were given access to contraceptives alone or with their husbands. Couple treatment resulted in less uptake of family planning services and higher fertility than individual treatment, but Baird noted that these improved family planning outcomes in the individual treatment arm came at a cost of lower reported subjective well-being. This implies a psychosocial cost of concealing contraceptive treatment. In a second study, adolescent development clubs in Uganda provided girls with vocational training and information on sex, reproduction, and marriage.³ Researchers surveyed and tracked a representative sample of around 5,000 adolescents. The intervention appeared to delay marriage and childbearing at a four-year follow-up, but it did not increase contraceptive use among sexually active girls. It is also important to note that, despite low levels of empowerment at baseline, only 21 percent of girls signed onto the program, and attrition was high at 35 percent. A third study gave non-financial incentives to hair dressers to increase sales of female condoms in Zambia; condom sales were rewarded with stars that indicate contribution to the health of the

___________________

¹ B. Cislaghi and L. Heise, Using social norms theory for health promotion in low-income countries, Health Promotion International 34, 616–623.

² N. Ashraf, E. Field, and J. Lee, Household bargaining and excess fertility: an experimental study in Zambia, American Economic Review 104(7), 2210–2237.

³ O. Bandiera, N. Buehren, R. Burgess, et al., Women’s empowerment in action: evidence from a randomized control trial in Africa, American Economic Journal: Applied Economics 12(1), 210–259.

community.⁴ Baird appreciated that this approach utilized small nudges within existing structures to promote improved family planning outcomes.

Baird also highlighted four ongoing studies that are in preliminary stages. The first study is a collaboration with BRAC in Tanzania to examine three interventions targeting contraceptive uptake and SRH outcomes for unmarried adolescents: providing free contraceptives to females, empowering females through a goal-setting exercise related to staying healthy and HIV/STI-free, and engaging with boyfriends, through soccer, to educate them about healthy intimate relationships.⁵ So far, the study authors have found that engaging males and goal-setting exercises has improved intimate-partner violence outcomes but has not affected uptake of contraceptives or fertility outcomes. Baird remarked that even though this supply-side intervention provides contraceptives free of charge, it has had no impact on pregnancy-related outcomes, likely because of a very low contraceptive prevalence at baseline (<1%), which did not substantially change at endline. Baird offered lack of engagement with mothers—who were generally not happy about the free contraception intervention—as a potential explanation for this result, as well as misinformation about contraceptive side effects.

A second ongoing study seeks to increase the uptake of long-acting reversible contraceptives (LARCs) among adolescents and young women in Cameroon.⁶ This pilot qualitative study is highlighting the attitudes of providers, mothers, and young women that create barriers to LARC uptake—namely, that family planning is not relevant for young, unmarried women without children. Misinformation about side effects and price of contraception in Cameroon were also identified as barriers. In response, the study is pursuing two interventions: first, a shift in family planning counseling strategy from an informed-choice model (i.e., clients are given information about all options and given a choice) to a shared decision-making model (i.e., clients relate their family planning goals and medical histories to the provider and then receive a recommendation); and second, price discounts for the LARCs. Quantitative pilot information has revealed that adolescents and adults who are already seeking family planning will pursue LARCs at any of the offered price points. When clients already had

___________________

⁴ N. Ashraf, O. Bandiera, and B.K. Jack, No margin, no mission? A field experiment on incentives for public service delivery, Journal of Public Economics 120, 1–17.

⁵ M. Shah and J. Seager, Promoting safe sex among adolescents in Tanzania (AEA RCT Registry, September 23, 2020 (doi: 10.1257/rct.1305-6.199999999999999).

⁶ B. Özler, S. Athey, and J. Jamison, Increasing the uptake of long-acting reversible contraceptives (LARCs) among adolescent females and young women in Cameroon. Ongoing; Also see: B. Özler, An adaptive experiment to improve quality in contraceptive counseling and increase the uptake of long-acting reversible contraceptive methods, Seminar Presentation, Available: https://www.worldbank.org/en/programs/sief-trust-fund/brief/seminars#07212020.

a contraceptive type in mind, they were more likely to take up a LARC when it was free or discounted than at its regular price. When clients did not have a specific contraceptive in mind, tailored counseling increased uptake of LARCs, but only at higher price points (at lower prices there was no evidence of increased uptake). Adolescents and adults who arrive at a health facility for non-family-planning reasons are considerably more likely to take up LARCs at lower prices. A large-scale randomized controlled trial is scheduled to occur in 2021.

Baird is personally working on the third ongoing study she presented, which targets first-time young mothers as a possible entry point for increased family planning uptake in Bangladesh and Tanzania. Baird said that this intervention strategy recognizes that the norms preventing adolescents from engaging with family planning before their first birth may be too difficult to change, and instead focuses on postpartum family planning. She said that she and her colleagues are developing program enhancements on the demand and supply sides that will work within existing platforms. Barrier and facilitator analyses have revealed similar barriers as found in other studies, including social and gender norms, harsh and judgmental attitudes, myths about family planning, and missed opportunities to integrate postpartum family planning.

Baird is also working on a fourth ongoing study in Ethiopia, which aims to conduct interventions among very young adolescents who tackle gender norms alongside their peers, families, communities, and broader institutional structures. The two-year multilevel program, known as Act With Her Ethiopia (AWH-E), will include curriculum-based programming and will operate according to the Gender and Adolescence: Global Evidence (GAGE) framework.⁷ Baird explained that the study will be a cluster-randomized trial in which intervention arms are defined by increasing levels of engagement: first for girls only, then with the addition of boys and parents, then the addition of community and systems strengthening, and finally with the addition of asset transfers. Preliminary findings have shown strong short-term impacts on voice and agency, knowledge, and nutrition, but less evidence of changing gender attitudes and norms. Baird noted that the adolescents in this study are particularly young and may not be sexually active or menstruating, but that funding will allow the study to track the influence of early intervention as the adolescents enter those life stages.

In the discussion, Ann Blanc (Steering Committee Chair, Population Council) asked how long the AWH-E Program will follow the girls and whether the timeline will encompass multiple “transitions” to adulthood, such as leaving school, marriage, and first birth. Baird replied that the project has funding through 2024 (for a total of nine years), at which point

___________________

⁷ See https://www.gage.odi.org.

the girls will be ages 17–19. The project could be funded for an additional five years, at which point the girls will be ages 22–24.

Jocelyn Finlay inquired about the role of stigma in accessing family planning interventions. Baird responded that the BRAC collaboration in Tanzania addressed stigma in the goal-setting and male engagement intervention arms. She noted that the LARC study in Cameroon may already be working with a more empowered group of women because they have already visited the health facility in the first place. Stigma will be addressed once the supply-side intervention is working robustly and the focus shifts slightly toward bringing in women who have not yet visited the health facility.

SOCIAL NORMS AS DRIVERS OF MODERN CONTRACEPTIVE USE: WHAT WE KNOW AND HOW WE USE THIS INFORMATION

Rajiv N. Rimal (Johns Hopkins University) defined three varieties of social norms: descriptive norms (beliefs about what most others do), injunctive norms (pressures to conform), and collective norms (what others actually do). Rimal commented that more research currently exists on descriptive and injunctive norms—which are alike in that they are both based on perceptions—than on collective norms.

Research on social norms investigates the normative influence that these norms—whether perceived or objective—have on behavior. The power of normative influence varies based on individual-level factors (e.g., self-efficacy, perceived benefits), behavioral attributes (e.g., public vs. private behaviors, addictiveness of behaviors), and contextual characteristics (e.g., time constraints, external monitoring). Consequently, norms are more important in some contexts than others and to some people more than to others. Rimal explained that these elements are captured by the Theory of Normative Social Behavior, in which these three classes of moderating factors are unpacked in an effort to understand the conditions in which social norms are more or less influential than others.⁸

Social norms also interact to modulate their relative influence. Injunctive norms, for example, may amplify the effects of descriptive norms. Rimal illustrated this dynamic in a scenario of college drinking behavior: students are more likely to consume alcohol when they believe that many others drink (i.e., high descriptive norm) and that there is pressure on themselves to conform (i.e., strong injunctive norm); when the pressures are not

___________________

⁸ R.N. Rimal and K. Real, How behaviors are influenced by perceived norms: a test of the Theory of Normative Social Behavior, Communication Research 32(3), 389–414 (https://doi.org/10.1177/0093650205275385); A. Chung, and R.N. Rimal, Social norms: A review, Review of Communication Research 4, 1–29.

felt, the influence of believing that everyone else drinks is less powerful.⁹ Similar patterns exist with regard to perceived benefits of a behavior. For example, college students who perceive a large benefit to consuming alcohol, such as a social benefit of conforming, are more likely to be influenced by the perception that many others drink, while those who do not perceive a benefit in the activity are not especially subjected to this normative influence.¹⁰ In this case, perceived benefits have both a main effect and interaction with descriptive norms.

Rimal considered this framework in relation to research on normative influence on contraceptive use. A recent study in Ethiopia demonstrated the reliable association of both descriptive and injunctive norms with condom use by young men.¹¹ Another study in India examined interpersonal communication—and in particular spousal communication—as a potential amplifier of normative influence on the choices of women.¹² In this latter study, spousal communication heightened the influence of descriptive norms among women with one child. Similarly, this interpersonal communication amplified the influence of injunctive norms for all categories of women studied: women with zero, one, or two or more children.

Research on collective norms is relatively new. Rimal highlighted a study conducted in Germany in 2019 that examined the role of all three classes of norms on the texting and driving behaviors of adolescents.¹³ This study revealed that descriptive norms are more consequential among adolescents whose social groups objectively engage in more risky behavior, whereas descriptive norms have little effect in social networks that do not engage in these behaviors. Similar effects have been seen in the realm of modern contraceptive use. An analysis of DHS data from Ethiopia and Tanzania demonstrated that women who live in communities with high rates of contraceptive use are also more likely to use modern contraception themselves.¹⁴

___________________

⁹ R.N. Rimal and K. Real, Understanding the influence of perceived norms on behaviors, Communication Theory 13(2) (https://doi.org/10.1111/j.1486-2885.2003.tb00288.x).

¹⁰ R.N. Rimal, Modeling the relationship between descriptive norms and behaviors: a test and extension of the theory of Normative Social Behavior (TNSB), Health Communication 23(2), 103–116 (doi: 10.1080/10410230801967791).

¹¹ A. Jain, H. Ismail, E. Tobey, and A. Erulkar, Stigma as a barrier to family planning use among married youth in Ethiopia, Journal of Biosocial Science 51(4), 505–519.

¹² R.N. Rimal, P. Sripad, I.S. Speizer, and L.M. Calhoun, Interpersonal communication as an agent of normative influence: a mixed method study among the urban poor in India, Reproductive Health, 12, 71 (doi: 10.1186/s12978-015-0061-4).

¹³ S. Geber, E. Baumann, F. Czerwinski, and C. Klimmt, The effects of social norms among peer groups on risk behavior: a multilevel approach to differentiate perceived and collective norms, Communication Research (doi: 10.1177/0093650218824213).

¹⁴ E. Sedlander and R.N. Rimal, Beyond individual-level theorizing in social norms research: how collective norms and media access affect adolescents’ use of contraception, Journal of Adolescent Health 64(4), S31–S36.

Rimal explained that this relationship between the collective norm and individual behavior was stronger in more isolated communities, suggesting that individuals with less access to the outside world are more likely to follow community norms.

Rimal illustrated the various ways in which perceptions interact with reality in the context of modern contraceptive use, as well as potential opportunities for intervention in each case. Women who believe that most other women do not use contraception when in fact most others do are demonstrating pluralistic ignorance rather than accurate perception. In situations of pluralistic ignorance, Rimal said that descriptive norm-based interventions (in which people’s misperceptions are corrected) should be enacted. When women accurately perceive that most other women do not use contraception, interventions should highlight the aspirational few who do. On the other hand, women who believe that most other women use contraception when in fact most others do not are experiencing the false consensus effect as opposed to accurate perception. In these cases, interventions should address barriers to use of modern contraception, and reinforced social support should be provided for women whose perception of reality is accurate.

Rimal ended the presentation with three questions that should be considered when conducting social norms research and designing norms-based interventions: First, what do social norms–based interventions to improve modern contraceptive use look like? Second, how can these interventions be differentiated between communities with low and those with high collective norms? And third, How can the power of interpersonal communication be harnessed to amplify normative influences?

In the discussion, Megan O’Donnell (Center for Global Development) asked how media, such as radio and TV shows, could serve as a way to reach more isolated communities to increase exposure to more gender-equal norms. Rimal cited a study of isolated communities in Nepal that demonstrated that collective norms are strongly correlated with people’s perceptions of those norms.¹⁵ Access to media likely serves as a role model for different manners of behavior that may not have been in the cognitive repertoire of women in these communities. Given the enormous importance of norms in isolated communities for providing information, social influence, and social sanction, Rimal pointed out that media can also empower women by providing examples of how to handle normative backlash.

Nancy Birdsall noted a study of Saudi husbands who are themselves comfortable with their wives working but believe that most other husbands

___________________

¹⁵ R.N. Rimal, A. Chung, and N. Dhungana, Media as educator, media as disruptor: conceptualizing the role of social context in media effects, Journal of Communication 65, 863–887 (doi:10.1111/jcom.12175).

oppose wives in the workforce. Rimal cited a similar finding of pluralistic ignorance from the late 1960s, in which White American families reported that they were comfortable with their children playing with the children of a neighboring Black family, but that they did not believe their other neighbors would feel the same way.¹⁶ In these situations, correcting the misperception can to some extent change the behaviors.

___________________

¹⁶ Angus Campbell, White Attitudes toward Black People, Ann Arbor, Michigan, Institute for Social Research, 1971, pp. 135-136.