I want to come back to the maths and underlying params behind the multiplier_for_resource_sharing_within_households to get a better understanding of where there might be uncertainty but it time consuming and not a massive priority right now since GiveWell's working looks reasonable. For now, I'll just estimate that the discounted number of people benefiting will have slightly greater uncertainty than the average household size and proportion of Proportion_of_household_income/consumption_brought_in_by_the_individual_who_is_dewormed could realistically be 1, but not higher, and could realistically be 0, but not lower. In this case income is continuous so this isn't relevant but I'm thinking about what this would look like if all wage earners earned the same income. I feel like Proportion_of_household_income_brought_in_by_the_individual_who_is_dewormed is most likely to be (roughly in order) 0, 0.5, 1, 0.33. It would be very unlikely to be 0.8 for example, or even 0.1, because most households are families with a discrete number of wage earners in the minority (roughly 0 to 4), and a discrete number of people between 0 and 20ish. I might be misunderstanding something about statistics but is there a way to represent this difference between discrete and continuous parameters in a distribution?

Hey Hannah! Can you find an average of each country and store it as a variable?