Design and Analysis of the Randomized Response Technique (with Graeme Blair and Kosuke Imai). Journal of the American Statistical Association, Volume: 110, Issue: 511 (September 2015), pages 1304 - 1319.
The randomized response method is a survey technique that seeks to reduce potential bias due to non-response and social desirability when asking questions about sensitive behaviors and beliefs. This survey methodology asks respondents to use a randomization device, such as a coin flip, whose outcome is unobserved by the enumerator. By introducing random noise, the method conceals individual responses and consequently protects respondent privacy. While numerous methodological advances have been made, we find surprising few applications of this promising methodology. In this paper, we address this gap by (1) reviewing standard designs available to applied researchers, (2) developing various multivariate regression techniques for substantive analyses, (3) proposing power analyses to help improve research designs, (4) presenting new robust designs that are based on less stringent assumptions than those of the standard designs, and (5) making all described methods available through open-source software. We illustrate some of these methods with an original survey about militant groups in Nigeria.
Power of four standard designs across probabilities of asking for the truth (x-axis) for N=500 and proportions of .1, .2, and .3 with the sensitive trait.
Refugee Proximity and Support for Citizenship Exclusion in Africa
As forced migration reaches unprecedented levels, understanding how it impacts local host communities, especially within the global South where the majority of this migration occurs, is critical. This article examines how the presence of refugees can change local citizens' support for citizenship exclusion in Africa. Using new data on locations of refugee communities (both formal camps and informal settlements) and 35,000 geo-referenced Afrobarometer respondents across 22 countries, I find that citizens who live near refugees in their country are more likely to endorse restrictions on citizenship access, especially with respect to granting birthright citizenship. Placebo tests show null effects of future refugee sites, suggesting that there is no selection on unobserved confounders. Citizens near refugees also report lower confidence in the national economy and less trusting of others, suggesting that the threat they feel due to refugee proximity are economic and social. These findings have implications for border politics and boundary-making in African nation-states.
 Map of citizen respondents and refugee sites.  Effects of refugee site proximity on support for jus sanguinis vs. jus soli citizenship regimes.
Presented at ISA 2014, APSA 2014, CAPERS Spring 2015, WZB Conference on Migration and Diversity 2015, USC Political Science Dept. 2017, Penn State's New Faces in Political Methodology IX. Links: (Working Paper)
Efficacy and Active Citizenship in Development: Experimental Evidence on Validated Participation in Tanzania (with Evan Lieberman)
Self-efficacy beliefs routinely predict active political participation in advanced democracies, but much less is known about the role of efficacy in developing contexts. We theorize that the link from socio-economic status (SES) to efficacy may be explained by opportunities to experience successful episodes of participation. Thus, while SES is difficult to change in the short run, it may be possible to boost efficacy and active citizenship among poor citizens through exposure to such opportunities. We investigate these propositions in the context of public education in rural Tanzania. First, we detail observational evidence demonstrating a strong relationship between self-efficacy and active citizenship behaviors. Next, we show that a novel meeting-based intervention -- Validated Participation, which affords parents a unique opportunity to discuss evidence, make decisions, and receive validation -- causes a large boost in efficacy among poor parents. We also find suggestive evidence of behavior change in parents reported by teachers.
Effects of Validated Participation (VP) vs. Information only (IW) compared to Control (SO) on measurements of efficacy for poor vs. wealthy parents.
Reducing Insurgent Support Among At-Risk Populations: Experimental Evidence from Cash Transfers and Livelihood Training in Afghanistan (with Jason Lyall and Kosuke Imai)
Governments, militaries, and aid agencies use economic interventions to win "hearts and minds" among vulnerable populations in wartime. Yet rigorous evidence of whether these programs can reduce support for insurgents remains scarce. We experimentally evaluate a program of livelihood training and one-time unconditional cash transfers on combatant support among 2,579 at-risk youths in Kandahar, Afghanistan. Employing survey methodologies for sensitive questions, we find that training alone has little effect on attitudes towards combatants. Cash initially increases pro-government sentiment, but these effects dissipate by seven months. When combined with livelihood training, cash increased support for the Afghan government while marginally decreasing pro-Taliban sentiment. Our evidence suggests that combatant support is mostly driven by political dynamics, including politicians'ability to capture credit for delivering the program itself, rather than opportunity costs of rebellion.
Effects of training (TVET) and cash (UCT) on combatant support measured with endorsement and randomized response at endline and 7 months later.
Do Refugees Spread or Reduce Conflict? (with Andrew Shaver)
As the number of forcibly displaced individuals around the world reaches record highs, understanding how their presence affects conflict is a major outstanding academic and policy question. Using new geo-coded data on refugee sites and conflict data at the subnational level from 1989 to 2008, we find no support for claims that refugee settlement increases the likelihood of conflict onset or incidence. Instead, we find that, conditional on having no other refugee sites present in the same country-year, provinces hosting refugee sites experience substantively large decreases in their likelihood of both conflict onset and incidence. We confirm these findings when examining heterogeneous effects of formal refugee camps and informal settlements. To exclude the possibility of unobserved confounders, we use placebo tests to show that there are no effects of future refugee sites on past conflict outcomes. We theorize that the influx of humanitarian and government attention, assistance, infrastructure to refugee sites may increase the stability of these settlement areas; we use nighttime lights satellite data and case studies to explore this further.
Map of civil conflict onset propensity and refugee sites (both formal and informal) for the period 1989 - 2008 at the subnational province level.
Presented at ISA 2014 and APSA 2015.
How Refugees Affect Citizens' National and Ethnic Identities: Evidence from Tanzania
A common claim about contemporary African politics is that citizens feel low attachment to their nation relative to subnational identities such as ethnicity. This project offers a theory of boundary-making in which pressures generated from weak borders lead citizens in border regions to stengthen attachment to their national identity. Through elite interviews, focus groups, and an original geo-coded survey experiment of over 2000 citizen respondents living around a rapidly expanding refugee camp in a rural Tanzanian border region, this project finds that the presence of refugees substantially increases local citizens’ attachment to their national identity and decreases their attachment to ethnic identities. Whether this increase in national solidarity leads to greater public goods cooperation among citizens is unclear; only when nearby citizens are informed about the conflict that displaced refugees are they more likely to demand for or contribute to better public goods provision for their own communities.
Measuring levels of attachment to various identities (e.g. regional, ethnic, national) by asking citizen respondents to distribute 10 beads across 6 bins.
Presented at MPSA 2017 and APSA 2017. Links: (Pre-Analysis Plan)