Public
Edited
May 8, 2023
import {cea_functions,cell,stylesheet} from "@falk-lieder/cea_library"
stylesheet
viewof scalability = cell(`
//The target group of the new intervention is emerging adults.
//We assume that this corresponds to people of ages 15 to 35. We estimate that this
//demographic constitutes between 50% and 70% of all active internet users.
//source: https://www.statista.com/statistics/272365/age-distribution-of-internet-users-worldwide/
proportion_internet_users_between_15_and_35 = 0.5 to 0.7
// What is the maximal amount of money that the intervention could absorb?
scalability_in_usd = scalability_of_online_interventions(proportion_internet_users_between_15_and_35)
`, [cea_functions],{distributionChartSettings: {showSummary: true}})
import {ce_of_previous_interventions,cost_effectiveness_of_intervention_bs} from "@falk-lieder/cea-baumsteiger"
viewof benefit_of_research = cell(`
// total amount of money that will be invested into the cause area of promoting well-being according to EA principles
budget = 10M to 1B
cost_of_RCT = 330000
// How much money would be invested into the new intervention if were is superior?
usd_deployable_to_new_intervention = investment_in_new_intervention(scalability_in_usd,budget,cost_of_research+cost_of_RCT)
// What would be the research's net effect on the total well-being of humanity?
net_effect_of_research_Baumsteiger = net_effect_of_research(ce_new_bs_hh,ce_best_preexisting_intervention_hh,usd_deployable_to_new_intervention,cost_of_research+cost_of_RCT)
net_effect_in_WELLBYs = hours_of_happiness_to_WELLBYs(net_effect_of_research_Baumsteiger)
`,[cost_effectiveness_of_intervention_bs,cea_functions,ce_of_previous_interventions,scalability,cost_of_research],{distributionChartSettings: {showSummary: true}})
viewof cost_of_research = cell(`
person_years = 0.5 to 0.7
personnel_costs = person_years*postdoc_salary_costs_per_year()
nr_participants_experiment = 116
cost_per_participant_experiment = 0 //20
nr_participants_pilot = 63
cost_per_participant_pilot = 0// 5
research_expenses = nr_participants_experiment * cost_per_participant_experiment + nr_participants_pilot*cost_per_participant_pilot
cost_of_research = personnel_costs + research_expenses
`,cea_functions,{distributionChartSettings: {showSummary: true, tickFormat: "$,f"}})
viewof cost_effectiveness_of_research = cell(`
//How cost-effective would it be to conduct this research?
ce_R_and_D_in_WELLBYs_per_dollar = cost_effectiveness(net_effect_in_WELLBYs,cost_of_research)
cost_effectiveness_of_research_hh = cost_effectiveness(net_effect_of_research_Baumsteiger,cost_of_research)
`, [benefit_of_research,cost_of_research,cea_functions],{distributionChartSettings: {showSummary: true}})
viewof beneft_per_time = cell(`
workweeks_per_year = 48 to 50
hours_per_week = 30 to 60
time_worked_on_project = person_years * workweeks_per_year * hours_per_week
value_per_hour = cost_effectiveness(net_effect_of_research_Baumsteiger,time_worked_on_project)
welby_per_hour = hours_of_happiness_to_WELLBYs(value_per_hour)
`, [benefit_of_research, cost_of_research, cea_functions,scalability],{distributionChartSettings: {showSummary: true}})
viewof ce_ratio = cell(`
//How does the cost-effectiveness of this research compare to the cost-effectiveness of the best existing intervention?
cost_effectiveness_ratio = cost_effectiveness_of_research_hh / ce_best_preexisting_intervention_hh
`, [cost_effectiveness_of_research,benefit_of_research,ce_of_previous_interventions],{distributionChartSettings: {showSummary: true}})
viewof limited_uptake = cell(`
p_uptake = [.1,.2,.3,.4,.5,.6,.7,.8,.9,1]
net_benefit_full_uptake = 460M
ce_ratio_full_uptake = 135
rel_ce_by_uptake(p) = net_effect_of_research_with_limited_uptake(p,ce_new_bs_hh,ce_best_preexisting_intervention_hh,usd_deployable_to_new_intervention,cost_of_research+cost_of_RCT)/net_benefit_full_uptake*ce_ratio_full_uptake
map(p_uptake,rel_ce_by_uptake)
`, [cost_effectiveness_of_intervention_bs, cea_functions,ce_of_previous_interventions,scalability,cost_of_research,benefit_of_research],{distributionChartSettings: {showSummary: true}})
viewof effect_sizes = cell(`
//Effect size for the Incredible Years program reviewed by Menting et al.
effect_menting = 0.23
se_menting = (0.31-0.15)/(2*1.96)
//Effect sizes from Figure 21.3 in Mesurado et al., the study by Hare was excluded because the average age was < 10
effects_mesurado = [0.058, 0.039, 0.115, 0.375, -0.03, 0.679, 0.398, 0.317, 0.549]
se_mesurado = [0.278, 0.287, 0.117, 0.224, 0.100, 0.152, 0.084, 0.072, 0.313]
effects_laguna_and_shin_lee = [-0.0782, -0.3082, 0.2665, 0.2367, 0.1455, -0.0175, 0.6079, 0.3491, 4.3942, 0.7730, -0.2410, 0.1223, 0.0461, 0.2987, 0.0598, 0.0231, -0.0492, 1.2782, 1.6518, 0.2168, 0.5576]
se_laguna_and_shin_lee = [0.0509, 0.0583, 0.0922, 0.0940, 0.1212, 0.1236, 0.0665, 0.0251, 0.1284, 0.3535, 0.2814, 0.0993, 0.1018, 0.0538, 0.0601, 0.0586, 0.0665, 0.1817, 0.1970, 0.0882, 0.0888]
effects = List.concat(List.concat(effects_mesurado,effects_laguna_and_shin_lee),[effect_menting])
//[effect_menting, 0.058, 0.039, 0.115, 0.375, -0.03, -0.069, 0.679, 0.398, 0.317, 0.549]
se = List.concat(List.concat(se_mesurado,se_laguna_and_shin_lee),[se_menting]) //= [se_menting, 0.278, 0.287, 0.117, 0.224, 0.100, 0.107, 0.152, 0.084, 0.072, 0.313]
SampleSet.fromList(effects)
`, cea_functions,{distributionChartSettings: {showSummary: true}})
viewof predicted_effect_size = cell(`
//1. Calculate the optimal standard deviation h of the Gaussian kernel
weights = map(se, {|se| se^-2})
normalized_weights = map(weights,{|w| w/sum(weights)})
// based on calculations in Matlab (calculations_for_kernel_density_estimation.m)
sigma_hat = 0.1573
iqr = 0.3485
n_interventions = 11
// calculate the optimal standard deviation of the Gaussian kernel according to Silverman's rule of thumb
h = 0.9*min([sigma_hat,iqr/1.34])*pow(n_interventions,-1/5)
//2. Create the kernels and combine the kernels into the estimate of p(g)
kernels = map(effects,{|g_hat| normal(g_hat,h)})
p_g_hat = mx(kernels,normalized_weights)
p_g_hat
`,[cea_functions,effect_sizes],{distributionChartSettings: {showSummary: true, title: "Effect Size on Prosocial Behavior", tickFormat:"," }})
half_lifeMeta = FileAttachment("half_life@3.meta.json").json()
viewof predicted_duration_of_benefits = cell(`
//predicted half-life of the intervention's effect on prosocial behavior in days
half_life = SampleSet.fromList($half_life_meta)
`, undefined,{jsImports: {half_life_meta: half_lifeMeta},
distributionChartSettings: {showSummary: true, title: "Number of days until the effect of the intervention drops to 50%"}}
)
viewof predicted_effect_on_wellbeing = cell(`
effect_on_happiness(p_g) = {
initial_increase_in_freq_of_prosocial_behavior_per_day = p_g *
std_freq_of_prosocial_behavior_in_acts_per_day()
additional_prosocial_acts_in_the_first_month =
initial_increase_in_freq_of_prosocial_behavior_per_day*days_per_month()
half_life_in_months = half_life/days_per_month()
increase_in_nr_acts_of_kindness_per_person =
sum_over_time(additional_prosocial_acts_in_the_first_month,half_life_in_months)
increase_in_nr_acts_of_kindness_per_person * benefit_of_one_kind_act()
}
increase_in_happiness = effect_on_happiness(p_g_hat)
increase_in_happiness
`,[cea_functions,predicted_effect_size,predicted_duration_of_benefits],{distributionChartSettings: {showSummary: true, title: "Hours of Happiness per Person Completing the Intervention"}})
viewof predicted_cost_effectiveness = cell(`
predicted_ce_intervention(p_g) = {
increase_in_happiness = effect_on_happiness(p_g)
cost_effectiveness(increase_in_happiness,cost_of_online_intervention_per_person_reached())
}
predicted_ce = predicted_ce_intervention(p_g_hat)
predicted_ce
`,[cea_functions,predicted_effect_size,predicted_duration_of_benefits,predicted_effect_on_wellbeing],{distributionChartSettings: {showSummary: true, title: "Ex-ante cost-effectiveness of a new intervention for promoting prosocial behavior"}})
viewof predicted_scalability = cell(`
//The target group of the new intervention is emerging adults.
//We assume that this corresponds to people of ages 15 to 35. We estimate that this
//demographic constitutes between 50% and 70% of all active internet users.
//source: https://www.statista.com/statistics/272365/age-distribution-of-internet-users-worldwide/
proportion_internet_users_between_15_and_35 = 0.5 to 0.7
// What is the maximal amount of money that the intervention could absorb?
predicted_scalability = scalability_of_online_interventions(proportion_internet_users_between_15_and_35)
predicted_scalability
`, [cea_functions],{distributionChartSettings: {showSummary: true}})
viewof predicted_cost= cell(`
//This is a list of the amounts that the Templeton World Charity Foundation Awarded to projects on project developing interventions for character development. I identified then by going to https://www.templetonworldcharity.org/projects-database and searching for projects in the Core Funding Area "Character and Virtue Development" that were part of the priority "Global Innovations for Character and Virtue Development". I went through all projects listed in the results and selected all that involved the development of a psychological intervention.
project_costs_TWCF = [100000, 233999, 234498,233393,234000,999977,1003610,229599,230940,233859,233919,228651,228870,233400,1000000,233919,231679,233753,234054,233784]
predicted_cost_of_research = SampleSet.fromList(project_costs_TWCF)
predicted_cost_of_research
`,[cea_functions],{distributionChartSettings: {showSummary: true}})
viewof ex_ante_CE = cell(`
//ex-post cost-effectiveness given the outcomes of the R&D project
ex_post_CE(ce_new, cost_of_research,scalability)= {
cost_effectiveness(expected_benefit(ce_new, cost_of_research,scalability),cost_of_research)
}
//Predicted cost-effectiveness of Baumsteiger's R&D project in hours of happiness per dollar
ex_ante_CE_Baumsteiger=ex_post_CE(predicted_ce,predicted_cost_of_research,predicted_scalability)
//conversion to WELLBYs per dollar
ex_ante_CE_Baumsteiger_WELLBYs = hours_of_happiness_to_WELLBYs(ex_ante_CE_Baumsteiger)
ex_ante_CE_Baumsteiger_WELLBYs
//ex_ante_CE_Baumsteiger
`, [cea_functions,ce_of_previous_interventions,ex_ante_benefit,predicted_cost_effectiveness,predicted_cost,predicted_scalability],{distributionChartSettings: {showSummary: true, title: "Ex-ante cost-effectiveness of developing an intervention for promoting prosocial behavior"}})
viewof ex_ante_benefit = cell(`
//parameters
budget = 10M to 1B
cost_of_RCT = 330000
expected_benefit(ce_new, cost_of_research,scalability)= {
net_effect_of_research(ce_new,ce_best_preexisting_intervention_hh,investment_in_new_intervention(scalability,budget,cost_of_research+cost_of_RCT),cost_of_research+cost_of_RCT)
}
ex_ante_moral_value = expected_benefit(predicted_ce,predicted_cost_of_research,predicted_scalability)
ex_ante_value_in_WELLBYs = hours_of_happiness_to_WELLBYs(ex_ante_moral_value)
ex_ante_value_in_WELLBYs
`,[cea_functions,ce_of_previous_interventions,predicted_cost_effectiveness,predicted_cost,predicted_scalability],{distributionChartSettings: {showSummary: true, title: "Predicted moral value of developing an intervention for promoting prosocial behavior in WELLBYs"}})
viewof comparison_ex_ante = cell(`
//How does the ex-ante cost-effectiveness of this research compare to the cost-effectiveness of the best existing intervention?
cost_effectiveness_ratio = ex_ante_CE_Baumsteiger / ce_best_preexisting_intervention_hh
`, [ex_ante_CE,ce_of_previous_interventions],{distributionChartSettings: {showSummary: true}})
viewof ex_post_vs_ex_ante = cell(`
// How does the ex-ante prediction compare to the ex-post evaluation?
Plot.dists({dists: [{name: "ex-ante", value: ex_ante_CE_Baumsteiger_WELLBYs}, {name: "ex-post", value: ce_R_and_D_in_WELLBYs_per_dollar}]})
`, [cost_effectiveness_of_research,ex_ante_CE],{distributionChartSettings: {showSummary: true, title:"Cost-Effectiveness of Baumsteiger's R&D project in WELLBYs per dollar"}})
viewof ex_post_vs_ex_ante_CE_ratio = cell(`
ex_post_ratio = ce_R_and_D_in_WELLBYs_per_dollar/hours_of_happiness_to_WELLBYs(ce_best_preexisting_intervention_hh)
Plot.dists({dists: [{name: "ex-ante", value: cost_effectiveness_ratio}, {name: "ex-post", value: ex_post_ratio}]})
`, [cost_effectiveness_of_research,ex_ante_CE,comparison_ex_ante,ce_of_previous_interventions,cea_functions],{distributionChartSettings: {showSummary: true, title:"Cost-Effectiveness Ratio compared to the best global health and well-being charities"}})
Purpose-built for displays of data
Observable is your go-to platform for exploring data and creating expressive data visualizations. Use reactive JavaScript notebooks for prototyping and a collaborative canvas for visual data exploration and dashboard creation.
