Numbers to Narrative

Telling stories with data and statistics

 

 

Dhrumil Mehta

Database Journalist, Politics - FiveThirtyEight

Adjunct Lecturer in Public Policy - Harvard Kennedy School

 

dhrumil.mehta@fivethirtyeight.com  

 @datadhrumil

@dmil

Unlocking Quant Skills

Data in the Classroom

Data in Journalism

Part 1:

My Journey

Path

  • Northwestern:
    • BA in Philosophy + Minor in Cognitive Science
    • MS in Computer Science
    • Knight Lab Student Fellow
       
  • Political Framing + USA Today + APSA
  • MediaCloud & Framing - Berkman Center
     
  • Software Development Engineer @ Amazon
     
  • Database Journalist @ FiveThirtyEight
  • Adjunct Lecturer in Public Policy @ Harvard

Database Journalist, Politics

Day One

Month One

Year One

More Writing / Reporting / Picking Up the Phone

Building Software

 

Lessons

Writing

Public Opinion

(and Pollapalooza)

 

 

Media Analysis

(data-driven)

Other stuff...

Visualization

Scraping

Data Editing

Open Data

Quantitative Editing

Words Editing

I edit POLL-BOT

... which is actually just 538's politics intern

 

 

 

 

 

 

 

 

 

(Fivey Fox as of 2020...RIP poll bot)

 

Bots

 

Bots let humans do what they're good at

2016

2018

Bot reports the facts, leaving time for humans to interpret them.

2018

But the bot also helps interpret facts!

2018

Lets readers see results that FiveThirtyEight deems unexpected

Expectations are calibrated before results ever start coming in.

2020

Same concept, new branding.

2020

Bot evolves into a human...

...jk

Internal Bots

Generalized Bot Architecture

C+J Conference @ Stanford (2016)

Infrastructure

Research

Part 2:

Types of Data Stories

Huge Data Dump

  • Uber
  • Election Results
  • Census

Answer a question with data

Support/Oppose a hypothesis

Identify a Phenomenon

Identify a Phenomenon

Debunk or Justify Conventional Wisdom

Data-Driven Profile

Lack of Data

Data driven investigative work

Dig for Data

Provide relevant context

Build our own dataset

  • With Code / Scrapers
  • By Hand
  • By Survey Tool

Explain Calculations

Use Innovative Methodology

Use data to inform traditional reporting

The Rare Datapoint

Part 3:

A case study

Individual Poll

Lots of Polls

Polling Average

Polling Average

Forecast (How to..)

Step 1: Collect, analyze and adjust polls

 

Step 2: Combine polls with “fundamentals,” such as demographic and economic data

 

Step 3: Account for uncertainty, and simulate the election thousands of times

 

 

 

 

 

Forecast

  • Polls
    • Horserace (national and state)
    • Approval
    • Generic Ballot
       
  • Economic Indicators
      
  • Incumbency
     
  • Scandals
     
  • Uncertainty Index
     
  • ETC...LOTS OF THINGS

 

 

 

 

Polling Averages

 

 

"modeled estimate" (Historical Vote Patterns & Demographics)

Forecast

Popular Vote

Electoral College

Individual States

Tipping-point states

etc...etc...

Past Forecasts

 

dhrumil.mehta@fivethirtyeight.com  

 @datadhrumil

@dmil

 

http://fivethirtyeight.com/contributors/dhrumil-mehta/​

Copy of Numbers to Narrative (NU Career Trek)

By Dhrumil Mehta

Copy of Numbers to Narrative (NU Career Trek)

Telling stories with data.

  • 304