Introduction

SICSS, 2023

Christopher Barrie

Introduction

  1. Welcome
  2. Workflow
  3. Ethics

Welcome

Housekeeping

  • Lunches 13:00-14:00 daily:

    • Vegan/vegetarian/GF options (go to front of queue!)

    • Week 2: a more diverse diet (AKA you’re on your own)

Housekeeping

  • Slack

    • Keep it on!

    • Channels for materials

    • Channels for prep.

Summer school structure

  • 5 Organizers:

    • Aybuke, Chris, Tod

    • Chris: main point of contact

  • 1 Teaching Assistant:
    • Lara

Summer school structure

Summer school structure week 1

  • 4 Days intensive instruction

    • 10:00-16:00 everyday

    • 16:00-17:00/17.30 research talk from invited speakers

  • See: https://github.com/cjbarrie/sicss_23 for full schedule + materials
    • Updated daily + linked materials
  • Thursday: Social Data Science Hub launch, see here
  • Friday: group meal 🥳

Summer school structure week 2

  • 4 (and a bit) days group work
    • Collaboration projects OR your own idea.
      • 10:00-16/17:00 everyday
  • Final Day Presentations
    • 12:00-16/17:00

Intro.: Computational Social Science

  • What is it?

Intro.: Computational Social Science

“Computational social science is an interdisciplinary field that advances theories of human behavior by applying computational techniques to large datasets from social media sites, the Internet, or other digitized archives such as administrative records.”

Intro.: Computational Social Science

“[CSS is] the development and application of computational methods to complex, typically large-scale, human (sometimes simulated) behavioral data.”

Intro.: Computational Social Science

Commonalities:

  • Computationally intensive methods

  • New types of data

  • Outcome of interest is ultimately human behaviour (?)

Intro.: Computational Social Science

Tensions:

  • Place of theory

    • Theory from data

    • New data for old theories?

Where has this all come from?





Why are these data different?

  1. Volume
  2. Velocity
  3. Producers
  4. Variety

Digital trace data

What is digital trace data?

  • Social media posts

  • Blogs and webpages

  • Call detail records (CDRs)

  • Web searches

  • Wearables

  • Internet of Things

Digital trace data

What does it look like?

Figure 1: Salganik, Matthew. 2018. Bit by Bit: Social Research in the Digital Age. Princeton: Princeton University Press. p.7.

Examples

[{'text': 'hello freak [ __ ] I would love to play',
  'start': 0.439,
  'duration': 5.351},
 {'text': 'you the dinosaurs are not real video',
  'start': 3.72,
  'duration': 4.82},
 {'text': 'just do it God', 'start': 5.79, 'duration': 5.25},
 {'text': "we probably can't play it on YouTube",
  'start': 8.54,
  'duration': 3.76},
 {'text': "where we'll get pulled but we could play",
  'start': 11.04,
  'duration': 4.44},
 {'text': 'the audio right play the I will put the',
  'start': 12.3,
  'duration': 4.59},
 {'text': 'video up on the screen and play the',
  'start': 15.48,
  'duration': 3.389},
 {'text': 'audio for you and you could just [ __ ] it',
  'start': 16.89,
  'duration': 4.29},
 {'text': 'your head could turn beet red smokes',
  'start': 18.869,
  'duration': 4.381},

Examples

Examples

<publicwhip scraperversion="b" latest="yes">
  <oral-heading id="uk.org.publicwhip/debate/2017-01-09b.1.0" nospeaker="true" colnum="1" time="" url="">Oral
Answers to
Questions</oral-heading>
  <major-heading id="uk.org.publicwhip/debate/2017-01-09b.1.1" nospeaker="true" colnum="1" time="" url="">
WORK AND PENSIONS
</major-heading>
  <speech id="uk.org.publicwhip/debate/2017-01-09b.1.2" nospeaker="true" colnum="1" time="" url="">
    <p pid="b1.2/1">The Secretary of State was asked—</p>
  </speech>
  <speech id="uk.org.publicwhip/debate/2017-01-09b.1.3" speakername="John Bercow" person_id="uk.org.publicwhip/person/10040" colnum="1" time="" url="">
    <p pid="b1.3/1">I call Mr Gerald Jones. Where is the fella? He is not here.</p>
  </speech>
  <minor-heading id="uk.org.publicwhip/debate/2017-01-09b.1.4" nospeaker="true" colnum="1" time="" url="">
Self-employment
</minor-heading>
  <speech id="uk.org.publicwhip/debate/2017-01-09b.1.5" person_id="uk.org.publicwhip/person/25309" speakername="Peter Dowd" oral-qnum="2" colnum="1" time="" url="">
    <p pid="b1.5/1" qnum="908056">What recent assessment he has made of trends in the level of self-employment. </p>

Applied examples

Figure 2: https://advances.sciencemag.org/content/7/29/eabe6534/tab-article-info

Applied examples

Figure 3: https://arxiv.org/abs/2101.03820

Applied examples

Figure 4: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4088828

Applied examples

Figure 5: https://www.science.org/doi/10.1126/science.aac4420

Applied examples

Figure 6: https://www.sciencedirect.com/science/article/pii/S0047272714000929

Applied examples

Figure 7: https://journals.sagepub.com/doi/abs/10.1177/1461444810365313

Isn’t this all just a fad?

No

Next question?

But seriously…

  • Ubiquity
  • Diversity
  • Complementarity

But seriously…

  • Ubiquity
    • Digital trace data is everywhere and “always on”

But seriously…

  • Diversity
    • of methods and data sources means no-brainer to use

But seriously…

  • Complementarity
    • of data and methods means enrichment (of the old) not substitution (by the new)

Isn’t it all just [stats./quant./data science]?

  • In some ways: yeah…

Isn’t it all just [stats./quant./data science]?

Figure 8: https://projecteuclid.org/journals/annals-of-mathematical-statistics/volume-33/issue-1/The-Future-of-Data-Analysis/10.1214/aoms/1177704711.full

Isn’t it all just [stats./quant./data science]?

  • In some ways: yeah…

  • In other ways: well, no…

    • New types of data

      • e.g., unstructured and trace data
      • associated decisions re how to restructure (compare: surveys)

Isn’t it all just [stats./quant./data science]?

  • In other ways: well, no…

    • New emphasis

      • e.g., learning from data and prediction

Isn’t it all just [stats./quant./data science]?

  • In other ways: well, no…

    • New theory

      • e.g., human-computer interaction

Isn’t it all just [stats./quant./data science]?

  • In other ways: well, no…

    • New ethical dilemmas…

Ethics

Ethical case studies

Case studies

Case studies

What is at stake?

  • Go to chalkboard

Case studies

Case studies

Ethics

  • consequentialism
  • deontology

How do we decide?

  • IRB (rules-based)
  • ethical judgment (principles-based)

…

  • and we do this openly (transparency-based accountability)

Ethics

  • respect for persons
  • beneficence
  • justice
  • respect for law and public interest

Ethics

  • respect for persons (Belmont)
  • beneficence (Belmont)
  • justice (Belmont)
  • respect for law and public interest (Menlo)

  1. https://www.hhs.gov/ohrp/regulations-and-policy/belmont-report/index.html

  2. https://www.dhs.gov/sites/default/files/publications/CSD-MenloPrinciplesCORE-20120803_1.pdf

  3. https://aoir.org/reports/ethics3.pdf

Ethics

  • respect for persons
    • participants decide–not you (informed consent)

Ethics

  • beneficence

    • minimize risk and maximiize benefits to decide if worth it

    • N.B. third-party oversight (because researcher bias)

Ethics

  • justice

    • distribution of benefits and burdens of research

      • vulnerable individuals; global South as lab

Ethics

  • respect for law and public interest

    • compliance (TOS and general law…)
      • N.B. difficulties with cross-national research
    • be open about what you’re doing (transparency-based accountability)

Group work

What is at stake?

  • Go to chalkboard

Workflow

Workflow

  • Who uses Git?
git --version

Workflow

  1. We all need to create a Github account.

  2. We get our computer talking to Git by logging our Github credentials (either SSH or HTTPS approach)

  3. We create a Github Repo

  4. We create an R “Project”

  5. We start writing our code

  6. We “commit” and then “push” our changes when ready

Setting up in R

  • Create a personal access token
install.packages("usethis") #if you haven't already installed
library(usethis)

usethis::create_github_token()

gitcreds::gitcreds_set()

Setting up in R

Setting up in R

Setting up in R

Setting up in R

Setting up in R

Setting up in R

Setting up in R

Setting up in R

Or in Terminal…

git clone https://github.com/cjbarrie/sicss_23
cd sicss_23
git remote -v
git pull origin main
git status

Committing

touch example.R
git status 
git add --all
# or git add .
git commit -m "example commit"
git push
git pull origin main

#OR
git fetch origin main
git merge

Branching

git branch "example"
git checkout example
touch example.R
git commit -m "branch example"
git push --set-upstream origin example

PRs…

Branch delete

git checkout origin
git branch -d example
git push origin --delete example

If you want to practice more…

https://learngitbranching.js.org/

.gitignore

  • Where you can specify what to remove from the version control

.gitignore

.gitignore

README

  • Where we describe what the repo contains

README

README

Reproducibility

  • Versioning your code is step 1. But there are many more steps…

  • Reproducibility also depends on…

Reproducibility

  • Version of the programming language used;

  • Versions of the packages/libraries of said programming language used;

  • Operating System, and its version;

  • Versions of the underlying system libraries (which often go hand in hand with OS version, but not necessarily).

  • And even the hardware architecture that you run all that software stack on.

Taken from: https://raps-with-r.dev/intro.html