Bilingualism, Hypothetical Norms & Misperceptions

Hill, Joshua (2025) Bilingualism, Hypothetical Norms & Misperceptions. Doctoral thesis, University of East Anglia.

Preview

PDF Download (6MB) | Preview

Abstract

My thesis combines behavioural and experimental economics in a developmental context. Social norms, values and preferences serve as the common thematic structure throughout this thesis and Chapter 2: ‘Norms, Values & Preferences’ summarises the key theoretical and substantive literature in this area. Chapter 3: ‘Uganda - Culture & Context’ presents a brief history of Uganda, including economic background, social context and deep-dives on the three ethnic groups of interest throughout this work: the Gisu, Ganda and Acholi. The proceeding three chapters are the empirical heart of this work.

Chapter 4: ‘Bilinguals in the Lab: Why do Norms and Expectations not Predict Contributions?’explores whether the language you speak affects the decisions that you make? To answer this, this chapter randomly assigns the language of a public goods game for bilingual subjects to determine whether there is a difference, and if so, I explore why this is the case. This chapter initially replicates the largest previous result using 349 subjects in Uganda. I find that subjects playing a one shot public goods game in Luganda (the national language) contribute 28.9% more than those playing in Lugisu (their tribal language). This effect size is within 0.1 percentage points of the previous finding. I rule out several confounds, and test three mechanisms. First, I am able to reject the idea that language changes behaviour by activating different norms, as they do not differ by language. Second, language does not act as a coordination device, as expectations do not differ by language. Third, this leaves the explanation of different preferences in different languages. Rich anthropological evidence describes the Gisu as having a self-sufficient and non-cooperative culture. Speaking the associated language appears to activate this low-cooperation cultural frame, whereas speaking another leads to higher cooperation. These results show a large and robust language effect on behaviour, which in this case is best explained by language-dependent preferences for cooperation.

Chapter 5: ‘Hypothetical Norms & Cultural Exposure’ explores whether individuals house multiple sets of social norms (as in Hoff and Stiglitz (2016)) and whether these can be elicited hypothetically across different contexts (by using a modified version of the Krupka and Weber (2013) methodology). The context applied is twofold: geographical ‘place’ (Kampala / Mbale, 2019 experiment) and ‘frame’ (Community / Wall Street Game, 2024 experiment). I find that norms differ significantly by both place and frame, but that expectations do not. In the 2019 experiment, because of the bilingual design, I also find a significant interaction between place and language for norms. This means that speaking either Lugisu or Luganda in Mbale affects norms, but norms are unaffected by speaking either language in Kampala. For the 2024 experiment, I also find that contributions significantly differ by frame, and that for both contribution and norms, hypothetical responses across subject are able to predict actual responses. This chapter provides methodological enhancements which may be of use to policy makers; not only can hypothetical norms be measured out of context, but I have also shown that these are accurate in signalling actual behaviours.

Chapter 6: ‘Social Norms, Value & Misperceptions’ explores the nature of misperceptions and how these differ between in and out-groups‡. Looking at political in-groups in the USA, several studies have found Republicans tend to think other Republicans are more like them than they are, and that Democrats are less like them then they are (Bursztyn and Yang, 2022). These studies also suggest that Republicans are more accurate in their perceptions of Republicans when compared to Democrats. Building on this, I explore similar ideas but instead of focusing on political identity in the USA, I use ethnic identity in Uganda. I replicate and expand these finding and extend the analysis in a number of unique ways. My design considers five domains (politics, religion, deception, violence and adultery), across three ethnic groups (the Gisu, Ganda and Acholi) in a non-WEIRD setting (Uganda). My design also allows the underlying mechanisms to be determined, for both the in and out-group. I find that that the belief of others is significantly influenced by own beliefs, and this is more extreme for the in-group than the out-group. Considering the mechanisms, I find evidence that Motivated Reasoning and Cultural Identity are significant drivers of perceptions for both in and out-groups, though this varies by domain. For the in-group only, the Curse of Knowledge is also significant across all domains. Considering inaccuracy, participants are less accurate about their in-group despite feeling most confident about these responses. Lastly, the results suggest that inaccuracy is being driven by irrational attention, rather than deeper identity-related factors. These results will be of value to policy makers and researchers alike. Misperception corrections have become an increasingly popular policy tools, and they have been shown to effective, at least in the short term. However, this chapter highlights a number of potential situations where outcomes may not be as expected.

Item Type:	Thesis (Doctoral)
Faculty \ School:	Faculty of Social Sciences > School of Global Development (formerly School of International Development)
Depositing User:	Chris White
Date Deposited:	01 Dec 2025 11:03
Last Modified:	01 Dec 2025 11:03
URI:	https://ueaeprints.uea.ac.uk/id/eprint/101162
DOI:

Downloads

Actions (login required)

View Item