Scientific Advice under Contentious Political Conditions

What is the role of a ‘good’ policy advisor?1

Tony Chen, Elise Baumann-Coblentz, Tommaso Cappato, and May Gezaw

2021-02-03

Concensus or Contentious Politics?

According to Roger Pielke (2012), tornado politics are areas where information matters and most actors want to achieve a similar, broad goal - they just disagree on how to get there. Abortion politics are areas where goals are at odds, so information matters less. A policy advisor works within this context, but what makes the advice good largely depends on personal opinion: is it that they advised a method that worked? That they won an ideological bid? In this post, we will go over some potential courses of action, but deciding whether they are a good advisor’s reaction will be left to the reader. The Pielke typology of the four ideal-types of advisors is explained in the graph below, and is determined by the type of political negotiation (elite decision or interest group tractations) and what he calls their “view of science” (“linear” if any opinion is evenly distributed along an axis, and “stakeholder model” is opinions are cleaved).

Tornado Politics

In a tornado politics situation, an advisor’s role is less about advertising an end goal than a course of action - in Pielke’s words, it leaves space for an advisor to be a “Science Arbiter” (someone who finds empirical solutions for a clearly stated problem). Since everyone agrees on the goal (say, reducing poverty), the issue is to get actors to agree on the methods. This idea can be represented graphically and expanded on by bringing in an economic model of the democratic decision-making. The Downsian approach to this issue would be to map it according to the policy preferred by the median stakeholder, voter or user, which is, understanding what most actors agree on and placing the alternatives on an axis. Capturing the median means winning the bid; in other words, in a tornado situation, fixing the policy around the median opinion and compromising as much as possible with the other stakeholders at the margins (according to information and pre-existing analyses) would ensure a policy that rallies as many as possible. Since the outcome matters to all, compromising should be the more sensible option for the non-median groups, and mobilizing relevant information should be a given. Even the best-educated Science Arbiter advisor should take this dynamic into consideration.

RPubs

(Image credits to https://econ243.academic.wlu.edu/2017/01/30/the-median-voter-theorem/)

In this illustration of the Downsian model, let’s replace each voter with a policy proposal in a tornado setting. Each stakeholder can choose to agree to the policy or repel it. On the vertical axis, instead of utility, let’s consider the number of stakeholders - their preferences are distributed along the horizontal axis, and they are more concentrated around the median. Following the median preference (policy proposal A) ensures the policy would be adopted, compromising at the margins ensures proposal A catches the stakeholders who are in-between proposal A and C, and proposal A and B.

This idea is directly inspired by Anthony Downs’ 1957 An Economic Theory of Democracy. This theory might be applied somewhat usefully to a tornado politics situation, but it does not have the ability to accurately describe a situation where the voters, stakeholders or users of a policy are polarized not around the median, but around the extremes. For the latter, an adaptation has been made: the Hotelling-Downs model.

Abortion Politics

In an abortion politics situation, users do not have a single-peaked preference curve, they are polarized, as in the situation below. Here, an advisor can take an “Issue advocate” approach and reduce the topic to a single issue (an example would be, instead of considering the consequences of illegal abortions done under restrictive abortion policies, addressing only whether abortion is morally acceptable). The opposite stand is to be a “Honest broker” and expand the scope of the issue to encompass new themes and solutions. In a Hotelling-Downs phrasing, an Issue advocate finds the median preference of a subset of users, sticks with it, and waits until they gain a majority in a deciding organization where they can make this preference into a policy. In this context, waiting and gaining more users to their side through direct advocacy is a solution to beat the divided curve. A Honest broker could, in return, do some political entryism: agreeing to the opposite side’s propositions only if they introduce new conditions (for example, less abortion clinics or longer mandated wait-times if it is in the case of an anti-abortion group consenting to greenlighting a pro-choice policy).

RPubs

(Image credits to https://blogs.warwick.ac.uk/dcstevens/entry/the_hotelling-downs_model/)

Policy Areas

“Tornado” Policy Areas “Abortion” Policy Areas
Improving literacy Age of consent (e.g. France)
Vaccination Multiculturalism/immigration
Covid reduction Assisted euthanasia
Organised criminality Censorship of social media
Gender equality Brexit/globalization
Poverty reduction Drug legalization

What about COVID?

COVID-19 has highlighted significant limitations to Pielke’s “Tornado” vs. “Abortion” model of policy advice. Specifically, for the model to hold any meaningful applicability, we must assume some stability across time in the short-run and stability in societal preferences. When we look at U.S. sub-national responses to COVID-19 in the first year alone, we see that the issue moves from a “tornado” issue to an “abortion” issue extremely quickly. Most importantly, it seems that divergences in the political treatment of COVID has less to do with the pandemic itself than the general political climate.

library(tidyverse)
library(plotly)
library(RCurl)
stateparties <- read.csv("state-parties.csv")  #Data downloaded from Kaiser Family Foundation: https://www.kff.org/other/state-indicator/state-political-parties/?currentTimeframe=0&sortModel=%7B%22colId%22:%22Location%22,%22sort%22:%22asc%22%7D#notes; state names modified for compatibility, deleted meta-information
governor <- stateparties %>% select(RegionName = Location, Party = Governor.Political.Affiliation)

githubfile <- getURL("https://raw.githubusercontent.com/OxCGRT/USA-covid-policy/master/data/OxCGRT_US_latest.csv")  #Oxford CGRT USA Data
policies <- read.csv(text = githubfile)
policies <- policies %>% mutate(RegionName = as.character(RegionName %>% na_if(., 
    ""))) %>% mutate(RegionName = replace_na(RegionName, "USA")) %>% mutate(Day = as.Date(paste(policies$Date, 
    sep = ""), "%Y%m%d"))


df <- filter(policies, RegionName != "USA") %>% select(RegionName, Day, GovernmentResponseIndex, 
    ConfirmedDeaths, ConfirmedCases) %>% drop_na() %>% inner_join(governor, by = c(RegionName = "RegionName"))
df <- df %>% mutate(Date = as.character(df$Day))

fig <- df %>% plot_ly(x = ~ConfirmedDeaths, y = ~GovernmentResponseIndex, size = ~ConfirmedCases, 
    color = ~Party, colors = c("deepskyblue", "orangered"), frame = ~Date, text = ~RegionName, 
    customdata = ~ConfirmedCases, meta = ~Party, hovertemplate = paste("<b>%{text}</b>", 
        "<br>%{meta} Governor", "<br>Index: %{y}", "<br>Confirmed Cases: %{customdata}", 
        "<br>Confirmed Deaths: %{x}"), type = "scatter", mode = "markers")
fig <- fig %>% layout(title = "State Responses to the Covid Pandemic", xaxis = list(title = "Confirmed Deaths"), 
    yaxis = list(title = "Government Response Index (OxCGRT)"), margin = list(l = 80, 
        r = 80, b = 100, t = 100, pad = 4))

fig

This chart is fully interactive; simply use your mouse on the chart itself or the slider and buttons on the top-right corner.. The data here is provided by the Oxford COVID-19 Government Response Tracker. Political affiliations of state governors are provided by the Kaiser Family Foundation.

References

Daniel Corradi Stevens. 2008. The Hotelling–Downs model of Two–Party Competition and the Median Voter Theory. https://blogs.warwick.ac.uk/dcstevens/entry/the_hotelling-downs_model/.

Downs, Anthony. “An Economic Theory of Political Action in a Democracy.” Journal of Political Economy 65, no. 2 (1957): 135-50. http://www.jstor.org/stable/1827369.

Kaiser Family Foundation. 2021. State Political Parties. https://www.kff.org/other/state-indicator/state-political-parties/.

Pielke, Jr, R. (2007). The Honest Broker: Making Sense of Science in Policy and Politics. Cambridge: Cambridge University Press. doi:10.1017/CBO9780511818110

Thomas Hale, Sam Webster, Anna Petherick, Toby Phillips, and Beatriz Kira. 2020. Oxford COVID-19 Government Response Tracker. Blavatnik School of Government. https://github.com/OxCGRT/covid-policy-tracker.

tuckerj17. 2017. The Median Voter Theorem. https://econ243.academic.wlu.edu/2017/01/30/the-median-voter-theorem/.

  1. Prepared for GV4F4 at the London School of Economics and Political Science, School of Government↩︎