Data reporting for smaller teams

bit.ly/small-data-j

Kate Martin, APM Reports

Cody Lillich, Arizona's Family (KTVK/KPHO-TV)

Junyao Yang, Mission Local

Justin Myers, Chicago Sun-Times

Cody Lillich

Senior Investigative Producer

KTVK 3TV/KPHO CBS 5 (Phoenix)

Email: cody.lillich@graymedia.com

X: @CodyLillich | Bluesky | LinkedIn

​

Data Reporting in Television

• “Data Team” may only be you

• What can you get – and get quick?
• Think about what you might need in a moment’s notice
• City salaries, including OT and “other pay”
• Vehicle crash data
• Assessor data

• Be a resource
• You’ll quickly be the go-to for all things data (that may and may not exist)
• Hold mini-sessions with staff to spread the knowledge (and so it’s not all on you!)
• One Sheets on topics that are guaranteed to come up:
• Crime
• Wildfires
• Extreme Heat

State Crash Data

• Most states require a form to be filled out for every crash.
• State DOTs typically compile all of this info into a database.

• This information is a goldmine.
• In Arizona we get data on:
• # deaths, # injuries
• Pedestrian-involved?
• Seatbelt use
• Violation(s)
• Distracted?

This is what the state puts out…

• Lengthy PDF
• Good for quick data
• But you want what feeds this!

State Crash Data

Assessor Data

• Assessor Data can tell lots of �information:
• Average price
• Rental Indicators
• People who own multiple �Properties

Investors buying thousands of Phoenix-area homes as rent prices spike

Immigration Data

• Problem: CBP data portal only goes back to FY2022.

Immigration Data

• Problem: CBP data portal only goes back to FY2022.
• Solution: Scrub monthly pdfs to get archival data to give a more accurate timeline along our sectors.
• Provides our producers with quick data and perspective that is updated once a month.

Junyao Yang

Data reporter, Mission Local

Email: junyao@missionlocal.com

Bluesky: @junyao-yang.bsky.social

Instagram: @junyaophotos

Data reporting in a small newsroom

• Mission Local: 7 reporters, 3 editors
• 3 data reporters — but we are also beat reporters

The work

• Simple graphics for quick turnaround stories
• Datawrapper

Who were the most successful endorsers? Mayor Breed, her allies — and Republicans.

The work

• Simple graphics for quick turnaround stories
• Datawrapper
• Flourish → SVG → Illustrator
• QGIS → SVG → Illustrator

SF Police Commission critics fail to understand how it works, promoting misinformation

The work

• Simple graphics for quick turnaround stories
• Datawrapper
• Flourish → SVG → Illustrator
• QGIS → SVG → Illustrator
• Enterprise stories
• Mapbox
• Javascript

The work

• Simple graphics for quick turnaround stories
• Datawrapper
• Flourish → SVG → Illustrator
• QGIS → SVG → Illustrator
• Enterprise stories
• Mapbox
• Javascript
• Python

The work

• Design
• Making sure things look pretty
• Training for colleagues: walk through easy-to-use tools
• Datawrapper and OpenRefine or even Google Sheets are a good start

Lessons

• Check your work
• Always ask sources for datasets!
• Befriend the data stewards in your city
• Design consistency
• Communicate often, show drafts — Save yourself from unnecessary revisions
• Keep the data team (no matter how small) organized

​

Kate Martin

Email: kmartin@apmreports.org

Bluesky/Twitter: @katereports

Correspondent, APM Reports

Data reporting at small newspapers

I was the only data reporter in my newsroom for most of my career.

• Started with Excel, moved to SQL, then R and dabble in Python
• Constantly pushed myself to learn new skills.
• Evangelized to colleagues about data reporting.
• Learned strategies to get editor buy in.

Workplaces: Loveland Reporter Herald, Skagit Valley Herald, The News Tribune, Carolina Public Press, NBC News, APM Reports

Anyone can become a data reporter

I had no idea how to use Excel.

​

Yes, I added these up by hand. With a pen.

​

So, I learned!

​

​

When you get data

• Make a copy
• Create two folders: Original and working
• You want to keep your original data pristine in case you mess up

Name files and versions in a reasonable way:

• Schools.final = not good
• Schools.1.0 = better
• Schools.2013-02-03 = best

Find a system that works for YOU!

Think of data as a source

You wouldn't run a one-source story on most things. Don't depend on data for the single source either.

• Data can:
• answer questions, raise questions, mislead you, and point you in the wrong direction
• You can:
• misinterpret data, take it out of context, write an incorrect formula
• find hidden gems, fact-check source statements, bring hidden issues to light

​

Time management in small newsrooms

There is never enough time.

Keep two running lists of story ideas:

• One list of stories you can do today to satisfy your boss
• One list of project candidates that you can chip away at
• Have five minutes? File records requests, look for documents and experts, write sections of the story
• Write a story and don’t turn it in. Work on your project story. When your boss asks for your daily story, then turn it in.

Check yourself before you wreck yourself

That story you thought was awesome, might not be

WARNING SIGN: If you are really excited about a story, double check your facts and formulas. You might have made a BIG error.

• Report it out a little before you tell your boss
• When you know the story is solid, send your boss a memo with a few paragraphs
• Underpromise and overdeliver
• Have a minimum viable product if the story falls through

This is all about managing expectations

You can do linear regressions!!

Helps predict how poverty affects test scores.

​

Other examples:

• Pollution and asthma
• Stress levels and exercise
• Height and weight

Beware of spurious correlations

​

​

​

​

​

​

​

Source: https://www.tylervigen.com/spurious-correlations

Justin Myers

Email: justin@justinmyers.net

Bluesky: @justinmyers.net

Fediverse: @myersjustinc� @mastodon.sdf.org

Interactives editor, Chicago Sun-Times

(any pronouns)

Experience with teams of all sizes

• Solo with occasional design help
• About 4 to 8, with different specialties
• 15 or so, and managed about 10

And now:

• A newsroom in transition
• Am I all alone?
• Am I on a four-person team?

And now:

• A newsroom in transition
• Chicago Public Media—produces both the Sun-Times and WBEZ
• Buyouts, reorganization
• Am I all alone?
• Am I on a four-person team?

And now:

• A newsroom in transition
• Chicago Public Media—produces both the Sun-Times and WBEZ
• Buyouts, reorganization
• Am I all alone?
• Sun-Times has one dedicated person for graphics/data/etc.
• Am I on a four-person team?

And now:

• A newsroom in transition
• Chicago Public Media—produces both the Sun-Times and WBEZ
• Buyouts, reorganization
• Am I all alone?
• Sun-Times has one dedicated person for graphics/data/etc.
• Am I on a four-person team?
• WBEZ has three dedicated people focusing more on data reporting
• Had been up to seven over the summer

“Graphics”

• Actual graphics
• Help others with heavy lifting
• Occasional technical R&D

“Graphics”

• Actual graphics
• Daily for print and online
• Big feature presentations
• Stuff in between
• Help others with heavy lifting
• Occasional technical R&D

“Graphics”

• Actual graphics
• Daily for print and online
• Big feature presentations
• Stuff in between
• Help others with heavy lifting
• Occasional technical R&D

“Graphics”

• Actual graphics
• Daily for print and online
• Big feature presentations
• Stuff in between
• Help others with heavy lifting
• Mostly investigative teams
• Occasional technical R&D

“Graphics”

• Actual graphics
• Daily for print and online
• Big feature presentations
• Stuff in between
• Help others with heavy lifting
• Mostly investigative teams
• Occasional technical R&D
• Workflows, tools, etc.

Two guiding principles

• Enable as much reuse as possible, within reason
• Remember later WTF I’ve done

Tools and techniques

• Python, in JupyterLab notebooks
• Lots of GIS-related work

Tools and techniques

• Python, in JupyterLab notebooks
• DVC for backing up large files outside of GitHub
• (But use whatever works in your specific context, seriously!)
• Lots of GIS-related work

Tools and techniques

• Python, in JupyterLab notebooks
• DVC for backing up large files outside of GitHub
• (But use whatever works in your specific context, seriously!)
• Lots of GIS-related work
• QGIS for experimenting
• GeoPandas for posterity
• Illustrator for publication

Communicate, communicate, communicate

• Know how to share intermediate results with people
• How do you provide your colleagues with…

Communicate, communicate, communicate

• Know how to share intermediate results with people
• Even simple annotated screenshots help a ton
• How do you provide your colleagues with…

Communicate, communicate, communicate

• Know how to share intermediate results with people
• Even simple annotated screenshots help a ton
• How do you provide your colleagues with:
• Clarity?
• Empowerment?
• Inspiration?

Free/cheap resources (including from IRE)

https://www.ire.org/join-ire/member-benefits/

• Thousands of tipsheets (useful for exploring new topics)
• Curated databases (free and paid): dams, disaster loans, business loans
• Tableau Public and Tableau Prep Builder: Free license
• SmartProcure/GovSpend: Discounted subscription for IRE members
• PacerMonitor: Discounted subscription for IRE members

Questions?

bit.ly/small-data-j

Data reporting for smaller teams

Kate Martin, APM Reports

Cody Lillich, Arizona's Family (KTVK/KPHO-TV)�Justin Myers, Chicago Sun-Times

Junyao Yang, Mission Local