Empowering Developing Countries:
Combating Misinformation and Correcting (Mis)perceptions of Foreign Aid
Abstract:
How and under what conditions are misperceptions of foreign aid corrected or worsened? It is crucial to understand this question as misinformation and misperceptions regarding aid in recipient countries can impact voting behavior, aid effectiveness, and the perception of the state. However, our understanding of how information and misinformation influence perceptions, and how to correct misperceptions, remains limited. My dissertation argues that citizens who are highly confident in their beliefs are less likely to update, even if their initial responses are far from reality. Furthermore, I contend that accurate information and misinformation cause people to update their beliefs, but heterogeneous based on their confidence level.
My dissertation examines the intricate interplay between information, misinformation, and individuals' confidence levels in shaping aid-related perceptions in aid-recipient countries by taking a mixed-methods approach. Drawing on two rounds of field research conducted in Uganda, Kenya, and Tanzania, alongside pre-registered large-N survey experiments in the corresponding three countries, my dissertation investigates how different types of information influence belief updating processes. Particularly in the survey experiments, subjects are randomly assigned correct and incorrect information (with truthful debriefing) about aid to learn about their relative propensity to change their beliefs. The findings reveal that both accurate information and misinformation prompt symmetrical adjustments in beliefs, irrespective of their veracity. Furthermore, the degree of belief updating varies significantly based on individuals' confidence levels. Highly confident individuals exhibit a lower propensity to update their beliefs. This suggests a nuanced relationship between confidence and belief revision, particularly in contexts characterized by high aid dependence.
Combating Misinformation and Correcting (Mis)perceptions of Foreign Aid
Abstract:
How and under what conditions are misperceptions of foreign aid corrected or worsened? It is crucial to understand this question as misinformation and misperceptions regarding aid in recipient countries can impact voting behavior, aid effectiveness, and the perception of the state. However, our understanding of how information and misinformation influence perceptions, and how to correct misperceptions, remains limited. My dissertation argues that citizens who are highly confident in their beliefs are less likely to update, even if their initial responses are far from reality. Furthermore, I contend that accurate information and misinformation cause people to update their beliefs, but heterogeneous based on their confidence level.
My dissertation examines the intricate interplay between information, misinformation, and individuals' confidence levels in shaping aid-related perceptions in aid-recipient countries by taking a mixed-methods approach. Drawing on two rounds of field research conducted in Uganda, Kenya, and Tanzania, alongside pre-registered large-N survey experiments in the corresponding three countries, my dissertation investigates how different types of information influence belief updating processes. Particularly in the survey experiments, subjects are randomly assigned correct and incorrect information (with truthful debriefing) about aid to learn about their relative propensity to change their beliefs. The findings reveal that both accurate information and misinformation prompt symmetrical adjustments in beliefs, irrespective of their veracity. Furthermore, the degree of belief updating varies significantly based on individuals' confidence levels. Highly confident individuals exhibit a lower propensity to update their beliefs. This suggests a nuanced relationship between confidence and belief revision, particularly in contexts characterized by high aid dependence.
Key Results
Result 1. When subjects are exposed to accurate information, subjects shift from less accurate to more accurate.
|
|
|
Note: Observation level movement from pre-test (blue) to post-test (red) in regards to accuracy (aid-to-budget spending (%)) when respondents are exposed to accurate information. (Kenya (left), Tanzania (middle), and Uganda (right))
Result 2. When subjects are exposed with high misinformation, subjects shift from more accurate to less accurate.
|
|
|
Note: Observation level movement from pretest (blue) to posttest (red) in regards to accuracy (aid-to-budget spending (%)) when respondents are exposed to high misinformation. (Kenya (left), Tanzania (middle), and Uganda (right))
Result 3.
When individuals who are highly confident and highly accurate are exposed to high misinformation, they do not change much (Pane 1).
When individuals who are highly confident and less accurate are exposed to accurate information, they do not change much (Pane 2).
When individuals who are less confident and highly accurate are exposed to high misinformation, they become less accurate (Pane 3).
When individuals who are less confident and less accurate are exposed to accurate information, they become more accurate (Pane 4).
When individuals who are highly confident and highly accurate are exposed to high misinformation, they do not change much (Pane 1).
When individuals who are highly confident and less accurate are exposed to accurate information, they do not change much (Pane 2).
When individuals who are less confident and highly accurate are exposed to high misinformation, they become less accurate (Pane 3).
When individuals who are less confident and less accurate are exposed to accurate information, they become more accurate (Pane 4).
Pane 1. Kenya Round (Exposure to High Misinformation):
High Confidence and High Accuracy |
Pane 2. Kenya Round (Exposure to Accurate Information):
High Confidence and Low Accuracy |
Pane 3. Kenya Round (Exposure to High Misinformation):
Low Confidence and High Accuracy |
Pane 4. Kenya Round (Exposure to Accurate Information):
Low Confidence and Low Accuracy |
Note: Observation level movement from pretest (blue) to posttest (red) in regards to confidence in responses (x-axis) accuracy (y-axis) of responses
Result 4. The World Bank and local recipient government source treatments are more effective at altering prior beliefs compared to political party sources.
Note: The figure shows paired t-test results of pretest and posttest results regarding accuracy (aid-to-budget spending (%)) for each source at each accuracy level. The green arrow indicates that posterior beliefs, compared to prior beliefs, were more accurate and the red arrow indicates the posterior beliefs were less accurate compared to prior beliefs. Note that in Kenya, I tested only for the local recipient government and World Bank sources. Statistical significance is at p < 0.05. AI = Accurate Information; LM = Low Misinformation; HM = High Misinformation; WB = World Bank; LOC = Local recipient government; RP = Ruling Party; OP = Opposition Party.
Result 5. When subjects are provided with corrective information (in the between-subjects design), they are closer to accurate information.
Note: The figure shows two-sample t-test results comparing groups that received correction and those that did not receive correction regarding accuracy (aid-to-budget spending (%) for each source at each accuracy level. Green indicates that the left-side treatment has lower inaccuracies compared to the right-side treatment, and red indicates that the left-side treatment has higher inaccuracies compared to the right-side treatment. Furthermore, the Kenya experiment did not have a placebo. Statistical significance is at p < 0.05. HM = High Misinformation; LM = Low Misinformation.