Slide 1: Introduction & Professor Datum Teaser

Good evening Dr. Wu and everyone! Welcome to today's presentation: "Beyond Neutral Numbers: Who Controls the Evidence in Great Data Journalism?"

When we open a news article filled with percentages, interactive maps, and official statistics, our natural instinct is to relax our guard. We tend to treat words as someone's opinion, and numbers as hard, indisputable reality. After all, two plus two equals four—how could a chart possibly have an agenda?

Today, I want to challenge that instinct. Over the next 30 minutes, we are going to look behind the curtain of both award-winning data journalism and everyday newsrooms. We will see that numbers do not gather, clean, or visualize themselves. Every dataset in the news is the end product of dozens of human and institutional choices—decisions about who gets counted, who gets left out, and how much you, the reader, are allowed to verify.

Before we dive in, please take out your phones and scan the QR code on screen (Press Q to enlarge QR) and enter your name. We have 8 live quizzes, a live Leaderboard, and a Top Scorer Crown running throughout today's session!

To kick things off and give you a quick movie-trailer preview of the five clues we are investigating today, let's hand the screen over for 60 seconds to our animated guest, Professor Datum. (Press V to play 60s Teaser Video)

Slide 2: Warm-Up Poll — Same Evidence, Different Emphasis

Now that Professor Datum has warmed us up, let's test how easily the exact same statistical evidence can tell two completely opposite stories. Look at your phones right now—this opening poll is unscored, so go with your gut instinct!

Here is a real empirical finding from Loosen, Reimer, and De Silva-Schmidt (2020), who analyzed 225 global Data Journalism Award nominees across a four-year period from 2013 to 2016. They discovered that 68.4% of the world's best data journalism projects relied on official government or institutional data, while 31.6% mobilized non-official data—such as leaked records, NGO datasets, or data gathered by journalists themselves.

• Frame A spotlights the 68.4% and concludes: "Global award contenders still lean heavily on official institutions." It makes data journalists sound dependent on the state.
• Frame B spotlights the 31.6% and concludes: "Nearly one-third of elite projects mobilize non-official evidence." Suddenly, data journalism sounds like a bold independent watchdog force!

Cast your vote on your phone now: Which frame feels more persuasive to you—Frame A or Frame B? (Pause 10s as live bars shift)

Look at the live bars moving on screen. Notice what just happened: neither headline lied, and neither made up a number. Yet simply by choosing which part of the ratio to emphasize, the journalist changes your conclusion. That brings us directly to our central argument today.

Slide 3: Our 3 Big Questions — What Makes a Claim Defensible?

If the exact same numbers can be framed in opposite ways, what actually makes a data journalism story trustworthy?

Our central argument today is straightforward: Numbers do not speak for themselves. A data story is only as strong as the human choices behind it, and whether the public is given enough visibility to question those choices.

To evaluate any piece of data journalism across our readings today, we ask three connected questions across three layers:

• 1. Published Numbers (What is shown on the surface): What do the charts, rankings, and interactive maps appear to prove? And what happens when the interface looks sophisticated, yet hides the underlying evidence? We will examine this through Young et al. (2018) and Appelgren (2018).
• 2. Production Choices (What happens backstage in the newsroom): Before a journalist even draws a chart, who collected the data? Who defined the categories? How did institutional access, deadline pressure, and free software tools shape the story? We will unpack this using Dr. Shan Wu's (2021) study of Singapore newsrooms and Loosen et al. (2020).
• 3. Public Justification (What the audience is given to verify): Finally, can readers actually inspect the method and challenge the conclusion, or are we just told to "trust the math"? We will evaluate this through Tong and Zuo (2021) and Arias Robles et al. (2023).

Slide 4: Three Claims That Must Not Be Confused

Before we examine our first case study, we need to clear up a major conceptual confusion. Both journalists and readers often lump Accuracy, Procedural Objectivity, and Value-Free Neutrality together as if they mean the same thing. As Tong and Zuo (2021) argue, confusing these three terms is how subjective human choices get disguised as "pure science."

Let's use a simple Chef's Recipe Analogy to separate them cleanly:

• 1. Accuracy (Correct Measurement): Think of accuracy as weighing ingredients on your kitchen scale without making a math error. If the scale says 500 grams of flour, it really is 500 grams. In data journalism, accuracy means the calculations are correct and the spreadsheet has no typos.
• 2. Procedural Objectivity (Transparent Recipe): Think of this as writing down your step-by-step recipe so another chef can walk into the kitchen, follow your steps, and test how you made the dish. In journalism, this means following a transparent, repeatable method for gathering, cleaning, and checking data.
• 3. Value-Free Neutrality (The Myth of No Chef): This is the illusion that the dish cooked itself without a chef ever choosing the menu! In reality, long before you weigh the flour, a human being had to decide what dish to cook, which ingredients to leave out, and who the meal is for.

Key Takeaway: A chart can be 100% mathematically accurate, yet zero percent neutral—because a human being still chose which variables matter and how to define the problem.

Slide 5: Quiz 1 — Accuracy vs. Value-Free Neutrality

Let's put that distinction to the test right now in Quiz 1 of 8! Look at your phones and lock in your answer quickly—remember, faster correct answers earn up to 1,500 points!

Question on your phones: "A newsroom publishes a housing affordability tracker. Every price calculation is mathematically verified, and their cleaning steps are documented. Under our conceptual framework, what does this guarantee?"

• A) Value-free neutrality
• B) Accuracy and procedural discipline, NOT value-free neutrality
• C) Complete independence from institutional sources
• D) Open-ended exploratory freedom for users

(Wait 10 seconds for class votes, then Press R to Reveal Answer)

Explanation: The correct answer is B: Accuracy and procedural discipline, NOT value-free neutrality! Even when the math is spotless, human journalists still had to decide what counts as "affordable," which neighborhoods to include, and which time period to measure.

Slide 6: Young et al. (2018) — Method & Sampling Boundary

With that conceptual distinction clear, let's examine our first assigned reading: Young, Hermida, and Fulda (2018), who investigated the quality and design of Canadian data journalism.

Notice how carefully we need to examine their methodology before jumping to their findings. Young and her colleagues conducted a qualitative and quantitative content analysis of 26 Canadian data journalism projects that were nominated as award finalists between 2012 and 2015 across prestigious competitions like the Canadian Online Publishing Awards and the Digital Publishing Awards.

Why did they study award finalists instead of taking a random sample of daily articles? Because in any profession, awards define the gold standard. Award finalists show us what the journalism industry itself celebrates as its finest, most ambitious work—where newsrooms invested their best reporters, coders, and time.

However—and this is a crucial methodological boundary for our critical examination—studying 26 elite Canadian finalists tells us about the ceiling of data journalism, not the floor of everyday routine reporting. Furthermore, content analysis only lets us inspect the finished website on the screen; it cannot directly show us the invisible newsroom negotiations or budget cuts that happened behind closed doors. Even so, if we find transparency flaws at the very top of award-winning journalism, that tells us something profound about the field.

Slide 7: Young et al. (2018) — The Sports Car Paradox

And what did Young et al. (2018) discover when they looked under the hood of those 26 award finalists? Look at the three gauges on screen—they reveal a striking contradiction between surface interactivity and deep transparency.

On the surface, 96% of these award-winning projects were interactive. Readers could click buttons, hover over maps, and type in their postal codes. It looked like a golden age of user empowerment!

But when the researchers tested what that interactivity actually allowed users to do, the picture flipped completely:

• Only 15% of projects offered high or open-ended exploration, where users could actually cross-check variables or test their own questions. The vast majority were simple "click-and-look" interfaces.
• Even more striking: 69% of these celebrated projects provided zero accessible raw data for the public to inspect or download!

Think of this as the Sports Car with a Welded Hood Analogy. Imagine buying a shiny red Ferrari. You can sit in the leather seat, honk the horn, and flick the windshield wipers—that is the 96% surface interactivity. But the dealer has welded the engine hood shut—that is the 69% missing raw data. You cannot see the engine, you cannot check if parts are missing, and you can only drive down the single street the designer pre-programmed for you. Clicking a button is not the same as checking the evidence.

Slide 8: Young et al. (2018) — Useful Tools Can Set Editorial Limits

Why did so many award finalists end up looking so similar and offering such shallow exploration? Young et al. (2018) point to the homogenizing power of free, off-the-shelf software tools.

First, let's be precise with their statistics: out of the 26 projects, 14 included maps, and 9 of those 14 mapped projects (64% of mapped stories) relied on Google Maps. When cash-strapped newsrooms rely on free commercial templates like Google Maps, the software quietly dictates the editorial format—turning complex social issues into standardized "pin-on-a-map" stories.

Look at their case study of Moving Forward, an interactive transit tax calculator. Users were invited to enter their monthly spending to see how much a proposed transit tax would cost them personally. It felt hyper-personalized! Yet the calculator quietly restricted the entire public debate to one narrow consumer question—"What does this cost my wallet today?"—while leaving out broader questions about corporate tax fairness, environmental impact, or long-term transit equity.

Now let's weigh a fair counterargument: newsrooms operate under brutal deadlines, and simple, personalized calculators help busy citizens understand complex policies quickly without drowning in spreadsheets. That is a real civic benefit! Our synthesis: user-friendly tools are great as an entry point, provided the newsroom also attaches a methodology note and a raw data link for citizens and watchdogs who want to audit the engine under the hood.

Slide 9: Quiz 2 — The Interactive Spectacle Paradox

Time for Quiz 2 of 8! Check your phones—let's see who is paying close attention to Young et al.'s (2018) empirical findings.

Question on your phones: "In Young et al.'s (2018) study of 26 Canadian award finalists, 96% featured interactive elements. However, what percentage actually gave users NO accessible raw data to verify the story?"

• A) 69% had no accessible raw data (and only 15% allowed open exploration)
• B) Only 11% had no raw data
• C) 52% had no raw data
• D) 31.6% had no raw data

(Wait 10 seconds for votes, then Press R to Reveal Answer)

Explanation: The correct answer is A: 69% had no accessible raw data, and only 15% allowed open exploration! Let's take a quick peek at the live Leaderboard (Press L) to see who is currently holding the Top Scorer Crown! (Press L to close)

Slide 10: Wu (2021) — Precision Can Carry Institutional Authority

So far, Young et al. (2018) showed us how software design and newsroom routines limit what readers see on the screen. Now let's turn to our second assigned reading—written by our own instructor, Dr. Shan Wu (2021)—which takes us even deeper upstream: What happens when a powerful state controls the data before journalists even touch it?

Dr. Wu examined COVID-19 data journalism in Singapore through in-depth semi-structured interviews with newsworkers across local and international outlets, alongside analysis of their published pandemic coverage. This qualitative interview method gives us something content analysis never could: direct insight into the lived negotiations, pressures, and constraints inside the newsroom.

Think back to the daily Ministry of Health (MOH) pandemic updates in 2020. Every single day, the government released hyper-precise numbers, detailed tables, and interactive situation reports. On the surface, this looked like total data transparency.

Yet Dr. Wu reveals how statistical precision can carry institutional authority and blind spots. Specifically, official briefings split daily infections into two separate categories: "Community Cases" versus "Migrant Worker Dormitory Cases." Notice how that single categorization choice shaped public perception! When headlines celebrated "Community cases drop to single digits!", the public felt safe and reassured—even while hundreds of infections were occurring daily inside migrant worker dormitories. The numbers were 100% mathematically accurate, yet the category boundary rendered a vulnerable population peripheral to the main "national recovery" narrative.

Slide 11: Wu (2021) — Where Can Influence Enter?

To understand how state influence enters data journalism without requiring overt censorship, Wu (2021) breaks data journalism into two main production stages: Stage 1 (Gather & Select) and Stage 2 (Write & Present).

Why do journalists rely so heavily on official data in Stage 1? Dr. Wu highlights three structural forces:

• 1. Access: In a tightly regulated information environment, government ministries hold a near-monopoly on comprehensive national datasets.
• 2. Speed: During a fast-moving crisis, daily newsroom deadlines reward reporters who plug ready-made, clean official feeds directly into their live trackers.
• 3. Public Trust: In Singapore, high public trust in state institutions means audiences expect news outlets to anchor their reporting in official figures.

Because the state pre-packages the data in Stage 1, journalists risk becoming mere amplifiers in Stage 2—turning state-defined categories into polished charts without questioning what is missing.

However—and this is a critical nuance in Wu (2021) that we must highlight—journalists were not completely passive puppets! When official dormitory numbers failed to explain the human reality on the ground, resourceful reporters pushed back by combining data with traditional field reporting: interviewing migrant workers directly, partnering with NGOs, and mapping dormitory living conditions themselves. Even in a state-dominant context, journalistic agency survives when reporters treat official data as a starting point to interrogate, rather than the final truth.

Slide 12: Quiz 3 — State Influence & Knowledge Production

Let's test this crucial insight from Wu (2021) in Quiz 3 of 8! Look at your phones now.

Question on your phones: "According to Wu (2021), why did Singapore's COVID-19 coverage often reinforce state definitions of the pandemic?"

• A) Because journalists made mathematical calculation errors in their charts
• B) Because the state controlled primary data gathering and categorization upstream (e.g., splitting 'Community' vs. 'Dormitory' cases), which journalists then visualized under tight deadlines
• C) Because Singapore audiences disliked interactive visualizations
• D) Because NGOs refused to talk to journalists

(Wait 10 seconds for votes, then Press R to Reveal Answer)

Explanation: Spot on—the answer is B! Influence enters upstream at the data gathering and categorization stage, long before a single bar chart is drawn.

Slide 13: Loosen et al. (2020) — Official Sourcing Is Not a Capture Measure

Now, hearing Wu's (2021) study might make us wonder: "Wait—is heavy reliance on government data purely a Singapore phenomenon? And does using official data automatically mean a journalist has been captured by the state?"

To answer that, we bring in our first additional reading: Loosen, Reimer, and De Silva-Schmidt (2020), who conducted a large-scale longitudinal content analysis of N = 225 Global Data Journalism Award (DJA) nominees from 2013 to 2016.

Look at the two bars on screen:

• On the left, 68.4% of global award-nominated projects relied on official institutional datasets. So relying on government data is not unique to Singapore—it is the global norm, even among the world's most celebrated stories!
• Now look at the right-hand bar: 52.0% of those very same award nominees performed a critical watchdog or interventionist role—exposing systemic failures, holding politicians accountable, or challenging official policies!

How can 68.4% official sourcing coexist with 52.0% critical watchdog journalism? Because official sourcing is not a measure of state capture! In open or investigative contexts, journalists regularly take the government's own procurement records, court filings, or pollution databases and cross-reference them to expose government failure. Synthesizing Wu (2021) and Loosen et al. (2020): what determines whether data journalism serves power or challenges power is not where the spreadsheet came from, but whether journalists have the institutional freedom, time, and editorial autonomy to cross-examine that official dataset.

Slide 14: Appelgren (2018) — The IKEA Showroom Illusion

We just saw how source power operates upstream. Now let's connect Young et al. (2018) with our second additional reading, Appelgren (2018), to see how design power operates downstream on the user's screen.

Appelgren (2018) analyzed N = 31 data journalism projects nominated for the Nordic Data Journalism Awards in 2016 and identified a phenomenon she calls the "illusion of interactivity."

Think of this as the IKEA Showroom Analogy. When you walk into IKEA, it feels like a giant, open building where you can touch every sofa, open every kitchen drawer, and test every lamp. You feel completely free! But look down at the floor: there is a painted yellow arrow guiding your footsteps down a strict, one-way corridor. You cannot easily skip rooms or rearrange the store layout; you see every product in the exact sequence the store architect planned for you.

Appelgren shows that interactive news stories use three paternalistic techniques to build this digital IKEA corridor:

• 1. Controlling Functionality: Limiting filters and zoom levels so you can only slice the data in pre-approved ways.
• 2. Illusion of Interactivity: Giving you engaging micro-actions—like scrollytelling animations or hover tooltips—that feel participatory while keeping you on rails.
• 3. Linearity: Structuring the page top-to-bottom so the journalist's framing conclusion is unavoidable.

Guided storytelling is helpful for clarity—not every reader wants a raw spreadsheet! The problem arises when a newsroom mistakes a guided corridor for open public accountability.

Slide 15: Quiz 4 — Does Official Data Equal Capture?

We have reached our halfway mark—Quiz 4 of 8! Let's see who can synthesize Loosen et al. (2020) and Wu (2021) on your phones right now.

Question on your phones: "Loosen et al. (2020) found that 68.4% of global Data Journalism Award nominees used official data, yet 52.0% expressed criticism or calls for intervention. What does this prove when compared with Wu (2021)?"

• A) Official data is always fake and should never be used
• B) Relying on official data does NOT automatically equal state capture; watchdog impact depends on whether journalists cross-examine official categories rather than passively reproducing them
• C) Western journalists never use government statistics
• D) Interactive charts automatically make a story critical of the government

(Wait 10 seconds for votes, then Press R to Reveal Answer)

Explanation: The answer is B! Let's check the halfway Leaderboard (Press L)—Look at those scores! Anyone in the top 5 is still within striking distance of the Top Scorer Crown! (Press L to close)

Slide 16: Tong & Zuo (2021) — Replace the Neutrality Test?

Now we hit the biggest philosophical debate in our presentation. We have established from Young et al. (2018), Wu (2021), and Appelgren (2018) that data journalism is never value-free: human choices enter at data collection, cleaning, categorization, and interface design.

In our third additional reading, Tong and Zuo (2021) examine this reality and make a radical normative proposal. They argue that because subjective choices are unavoidable at every step of the data pipeline, journalism should stop pretending to be "objective." Instead, they argue that data journalism should be judged by whether its choices serve the public interest and strengthen democracy.

At first glance, that sounds inspiring! Who doesn't love "the public interest"?

But here is our critical counterquestion: If we throw away procedural objectivity and say "anything goes as long as it serves the public interest," who gets to define the public interest?

• In Singapore's pandemic response (Wu, 2021), state officials could genuinely argue that separating dormitory cases from community cases served the "public interest" by preventing public panic and preserving social stability!
• Meanwhile, a partisan activist outlet could cherry-pick housing data and claim their biased chart serves the "public interest."

If "public interest" becomes an excuse to hide our methods, then powerful institutions will always win the definition game. Therefore, we argue against throwing out procedural discipline: we must drop the false claim of value-free neutrality, while doubling down on transparent, auditable procedures!

Slide 17: Different Methods Answer Different Questions

Before we unveil our solution framework, let's pause for a Methodological Synthesis Checkpoint across all six studies we are examining today. Why? Because acing critical analysis means recognizing that different research methods answer different questions—and each has blind spots.

Look at our synthesis table across three distinct methodological lenses:

• 1. Elite Award Contenders (Young et al., 2018 [n=26 Canada]; Loosen et al., 2020 [N=225 Global DJA]; Appelgren, 2018 [N=31 Nordic]): All three studies analyze award-nominated outputs. Their strength is showing us the industry's highest aspirations across global and regional awards. Their limitation is selection bias: award entries are well-funded flagship projects, and output analysis alone cannot prove what audiences actually thought when clicking them.
• 2. Backstage Production & Normative Theory (Wu, 2021 [Singapore interviews]; Tong & Zuo, 2021 [Epistemological critique]): Wu (2021) solves the "black box" of content analysis by interviewing newsworkers directly to uncover invisible state-press negotiations. Its boundary is contextual specificity—findings from Singapore's pandemic media ecosystem cannot be mechanically copy-pasted to every country without testing local press-state relations.
• 3. Everyday Routine News (Arias Robles et al., 2023 [N=62 Spain]): Finally, to make sure we aren't only looking at elite awards, our fourth additional reading tests everyday daily data journalism—which we will inspect right after our framework!

Slide 18: Accountability Must Be Proportionate (3 Rules)

How do we bring all these studies together into a practical standard that real newsrooms can actually use?

We propose a Proportionate Accountability Framework built on three actionable rules. Why "proportionate"? Because a quick daily chart made in two hours does not need a 50-page academic appendix—but the higher the public stakes, the stronger the verification duty must be:

• Rule 1: Explain (Show the Recipe): State plainly where the data came from, what time period and population it covers, and what the main cleaning or categorization rules were. Even a three-sentence "How We Did This" box transforms a black-box chart into an honest claim.
• Rule 2: Scrutinise (Question Official Categories): Never treat an official dataset as a neutral mirror of reality. Ask Wu's (2021) question: Who does this category count, and who does it push out of frame? Cross-check institutional spreadsheets with ground reporting, NGOs, or alternative metrics.
• Rule 3: Enable Challenge (Give Readers a Way to Check): Match interface design to real inspectability. Provide downloadable summary tables, links to original sources, or open filters—while respecting ethical boundaries when raw data involves vulnerable people.

Slide 19: Quiz 5 — Interaction or Guided Corridor?

Let's jump into Quiz 5 of 8 on your phones! This combines Appelgren (2018) and Tong & Zuo (2021).

Question on your phones: "Why does Appelgren (2018) call many interactive data stories an 'illusion of interactivity,' and how does our framework respond?"

• A) Because users hate clicking on charts
• B) Because interactive buttons often guide readers down a pre-scripted corridor ('IKEA showroom') without letting them inspect the underlying data or assumptions
• C) Because Nordic newsrooms never use computers
• D) Because interactivity makes stories mathematically inaccurate

(Wait 10 seconds for votes, then Press R to Reveal Answer)

Explanation: The correct answer is B! Smooth animations and pre-set filters can create an IKEA showroom corridor unless paired with real methodological transparency.

Slide 20: Arias Robles et al. (2023) — Naming a Source Is Not Enough

Earlier, we noted that Young et al. (2018), Loosen et al. (2020), and Appelgren (2018) all studied award-nominated stories. What happens when we look at ordinary, everyday data journalism?

That is exactly what our fourth additional reading—Arias Robles, Lopez Lopez, and Carvajal (2023)—tested by analyzing N = 62 everyday data journalism articles published across major Spanish news outlets between 2019 and 2022.

Look at the three bars on screen—they reveal what we call the "Supermarket Receipt vs. Recipe Gap":

• 98% of everyday articles named a data source (e.g., "Source: Ministry of Health" or "Source: National Statistics Institute").
• Yet only 24% explained how the journalists processed or calculated the numbers (methodology).
• And a tiny 11% provided a usable, downloadable dataset so readers could check the math!

Think of naming a source like handing someone an NTUC FairPrice supermarket receipt: it proves you bought groceries at FairPrice, so you didn't steal them—that is basic attribution (98%). But a supermarket receipt is NOT a kitchen recipe! It doesn't tell the reader what you threw in the bin, how you combined the ingredients, or whether you burned the sauce. In everyday journalism, newsrooms routinely confuse dropping a source name (98%) with allowing the public to verify the claim (11%).

Slide 21: Quiz 6 — Attribution vs. Usable Verification

Quick fire on your phones—Quiz 6 of 8 on Arias Robles et al. (2023)!

Question on your phones: "In Arias Robles et al.'s (2023) study of everyday Spanish data journalism, 98% named a source, but what percentage actually provided a usable dataset for readers to inspect?"

• A) 85% provided a usable dataset
• B) Only 11% provided a usable dataset (and only 24% explained their method)
• C) 68.4% provided a usable dataset
• D) 50% provided a usable dataset

(Wait 8 seconds for votes, then Press R to Reveal Answer)

Explanation: The answer is B: Only 11% provided a usable dataset, and just 24% explained their method! Let's flash the Leaderboard (Press L)—only two quizzes left after this! (Press L to close)

Slide 22: The Newsroom Decision Lab — You Are the Editor!

Now, before we wrap up our 30-minute presentation and move into our interactive Q&A discussion, I want to put you in the editor's chair. Look at your phones for our Live Newsroom Decision Lab poll!

Scenario: It is 4:00 PM on a Friday. A government agency releases a brand-new, high-precision dataset showing that heatwave emergency hospital admissions in your city dropped by 18% this month. Your competitor is about to publish the chart. However, a community clinic nurse tips you off that the agency recently changed its coding rule: outdoor construction workers treated at mobile heat-relief tents are no longer counted as "hospital admissions."

As the Data Editor, which route do you choose on your phone right now?

• Route 1: Publish Now — Run the official interactive map immediately with source attribution so the public gets fast information.
• Route 2: Investigate First — Hold publication until your team interviews clinic workers and reconstructs the missing mobile-tent numbers.
• Route 3: Bounded Release — Publish the official figures today, but prominently build the coding-change caveat, methodology note, and what is missing right into the headline and chart.

Vote on your phone now—we will use your live classroom split to kick off Round 1 of our Q&A discussion in just 60 seconds!

Slide 23: Conclusion — Numbers Do Not Remove Judgement

Let's bring our 30-minute investigation home with three takeaway conclusions across our six readings:

• 1. On Excellence (Young et al., 2018; Appelgren, 2018): Great data journalism is not defined by how many buttons a user can click (96% interactive), because interactivity often masks a closed hood (69% no raw data) or an IKEA corridor.
• 2. On Authority (Wu, 2021; Loosen et al., 2020): Official numbers carry institutional power (68.4% official sourcing). They become real watchdog journalism (52.0%) only when reporters question how categories—like community versus dormitory cases—were constructed upstream.
• 3. On Accountability (Tong & Zuo, 2021; Arias Robles et al., 2023): Because data journalism is never value-free, naming a source (98%) is not enough. Credibility comes from proportionate procedural transparency: explaining the recipe, scrutinising the categories, and enabling the public to check our work.

Slide 24: Quiz 7 — Applying the 3-Pillar Framework

Let's lock in Quiz 7 of 8 as we transition into our 15-minute Q&A facilitation!

Question on your phones: "Under our 3-Pillar Proportionate Accountability Framework (Explain, Scrutinise, Enable Challenge), what should a newsroom do when publishing an official dataset that uses contested categories?"

• A) Refuse to ever use numbers in the news
• B) Explain the methodology/caveats, scrutinise who the official categories exclude via field reporting, and enable readers to inspect the evidence
• C) Add more 3D animations so readers don't notice the categories
• D) Just write 'Source: Government' in tiny 8-point font at the bottom

(Wait 8 seconds for votes, then Press R to Reveal Answer)

Explanation: The correct answer is B: Explain, Scrutinise, and Enable Challenge!

Slide 25: Round 1 — Make and Defend Your Decision

Welcome to our 15-minute interactive seminar discussion! Look at the live bar chart from our Newsroom Decision Lab on Slide 22. We have a fascinating split in the room between Route 1 (Publish Now), Route 2 (Investigate First), and Route 3 (Bounded Release).

Take 45 seconds right now to turn to the person sitting next to you: tell them which route you voted for, and give them one concrete reason why your route is the most responsible journalistic choice under deadline pressure. Go! (Wait 45 seconds for pair-share)

Let's hear from the floor:

• Prompt for Route 2 voters: "Who voted for Route 2 (Investigate First)? Why did you decide it was too dangerous to publish the official 18% drop right away, even if your rival beats you to the headline?" (Facilitator tie-back: Connect their response to Wu, 2021—how publishing unchallenged official categories during a public health crisis can actively mislead the public into thinking vulnerable outdoor workers are safe.)
• Prompt for Route 3 voters: "Now let's hear from someone who chose Route 3 (Bounded Release). How do you stop a 'caveat note' from being ignored by busy readers who only skim the headline and the big green downward arrow?" (Facilitator tie-back: Connect their response to Appelgren, 2018 and Arias Robles et al., 2023—why caveats must be built visually into the chart itself, not buried at the bottom like fine print.)

Slide 26: Round 2 — Challenge the Weakest Assumption

Now let's push deeper in Round 2 by stress-testing the hardest edge cases in our readings. Look at the three provocation cards on screen—I want to challenge the room with Question 2: When is it actually ETHICAL for a journalist to REFUSE to publish raw data?

Earlier, we criticized the 69% of Canadian award finalists (Young et al., 2018) and the 89% of Spanish daily stories (Arias Robles et al., 2023) that failed to share downloadable raw data.

Ask the class: "Can you think of a real-world situation where publishing the full raw spreadsheet would actually be irresponsible or harmful? And if you can't publish the raw data for ethical reasons, how do you still stay accountable to your readers?"

(Take 1-2 student responses, then synthesize:)

• Rubric Synthesis Point 1: Exactly! Think about migrant worker health records (Wu, 2021), domestic violence shelter locations, confidential whistleblowers, or small-sample neighborhood data where individuals could be re-identified (doxxing).
• Rubric Synthesis Point 2: Open data is never an excuse to violate privacy! When raw row-level data cannot be published for ethical or legal reasons, a newsroom fulfills Proportionate Accountability by publishing aggregated summary tables alongside a detailed methodology and schema note—explaining how the records were verified and why individual rows are protected.

Slide 27: Round 3 — Revise, Then Leave a Question Open

For our final discussion round before the Championship Quiz, let's settle the big normative debate triggered by Tong and Zuo (2021):

Ask the class: "Knowing that human judgement shapes every step of data journalism, should journalism schools and newsrooms stop using the word 'objectivity' altogether—or should we redefine objectivity as 'transparent, auditable procedure'? By a show of hands or a quick shout-out, where do you stand after today's discussion?"

(Take 1-2 final comments from the class, then deliver the synthesis wrap-up:)

Final Facilitator Synthesis: Notice how our discussion today shows why procedural objectivity still matters. If we abandon disciplined verification procedures completely, then in moments of crisis—whether a pandemic or a contested election—the public has no way to distinguish rigorous journalism from state propaganda or partisan spin. We don't trust data journalism because the numbers fell from the sky untouched by human hands; we trust great data journalism when reporters show us their recipe, question official categories, and invite us to check the work.

Slide 28: Quiz 8 — Ethical Boundaries of Open Data

Everyone grab your phones for our Final Capstone Quiz—Quiz 8 of 8! This decides who walks away with today's Top Scorer Crown!

Question on your phones: "A data journalist investigates workplace injuries among undocumented migrant workers using confidential clinic logs. Publishing the raw spreadsheet could expose workers to deportation or retaliation. Under our Proportionate Accountability framework, what is the best course of action?"

• A) Protect individual privacy by withholding identifiable raw rows, while publishing anonymized aggregate tables and a full methodology note explaining how the records were verified
• B) Publish all workers' names and addresses anyway because 100% raw data transparency is an absolute rule
• C) Kill the entire story and never report on workplace injuries
• D) Publish a flashy interactive map with zero explanation of where the numbers came from

(Wait 10 seconds for final votes, then Press R to Reveal Answer)

Explanation: It is A! Ethical accountability means balancing public inspectability with harm minimization!

Slide 29: Live Leaderboard & 8-Quiz Championship Podium

And now, the moment of truth—look at the big screen and your phones for our Final 8-Quiz Championship Podium!

Let's give a huge round of applause to our 3rd Place, 2nd Place, and our reigning NMC5364 Data Journalism Grand Champion wearing the Top Scorer Crown at the top of the screen!

Thank you so much Dr. Wu and everyone for jumping into the polls, the Decision Lab, and the debate today! You can also grab the full one-page study handout at /handout.html on the same link. I'm happy to take any final questions or feedback from Dr. Wu!