Your task
Start without advance reading. Use the explanations inside the activity. You are an independent reviewer of Larkhaven's fictional transit consultation. Your job is to explain what happened between residents' submissions and a public decision, then defend what should happen next. Read two real research explanations about AI mediation and persuasion. For each, identify a tested comparison, its measured response and a limit on applying it here. The consultation records and all teaching datasets are invented; they are not observations from those articles. Preserve your own explanation, a rival explaining the same response and a prediction before opening the later file. Next, work with response tables and the council's actual powers, budgets and access requirement. Finally, investigate a new institution without assuming that Larkhaven's explanation travels unchanged. A strong submission makes a specific finding, uses both studies and explains a real tradeoff. You may retain, narrow, change or suspend an earlier account when you can give reasons. Keep the earlier account visible so the revision can be assessed.
Explain the studies using the in-app explanations
Article A · Explain the study
Use the Article A explanation here; no journal reading is required.
1. What did people compare and rate: an AI-written statement or a statement from a trained participant? How was critique used?
2. Explain the difference between liking a statement, changing a view, and having a minority view included in the wording. Give one example of wording and a separate example of decision-making power.
3. What does the study help a public institution understand? What does it leave unsettled?
Article B · Explain the study
Use the Article B explanation here; no journal reading is required.
1. What position was the AI told to support, and which advocacy conditions were compared? What response did researchers measure?
2. Describe one test that changed an advocacy strategy. What did that test help explain?
3. Name a group or real-world outcome that the study did not test.
Use both studies
Use the two article explanations here; no journal reading is required.
1. Choose one claim an institution might make about an AI conversation.
2. Use a specific finding from each explanation to assess that same claim. Explain what each adds.
3. What can you reasonably conclude, and which question about authority or fairness remains for the institution to decide?
Before you start
No advance student reading is required. Use the simplified explanations inside the activity, with time for questions, analysis and saved continuation across sessions.
Check your instructor's assignment and use Class mode when the module is assigned. Identify which tasks are required and whether the optional frontier is included. A completion mark records an attempt; a person assesses the reasoning.
Prepare to record document IDs, dates, exact words and unknowns. Use the fictional roles and records: you do not need to share your own political preferences or anything about your family.
Choose a browser and device you can return to. Locate the progress-backup and print/export controls before starting, and keep a place to store downloaded files. Device saving does not carry work automatically to another device.
Work through the inquiry
1. Frame the opening question
Read the first notice and identify the claim you will investigate. State whose response or decision you mean. An opening choice helps you start; it does not settle the later evidence inquiry. Leave a short note you can compare with your final account.
2. Reconstruct both studies
For each article, name what the researchers compared, what response they measured and what the result supports. Explain how one detail from each changes a particular conclusion about a consultation. Keep endorsement, attitude movement and institutional authority distinct. Use the explanations as arguments to explain rather than lists of conclusions to copy.
3. Commit a document-based account
Open the available records in order. Before the later file, write your actor-and-operation explanation, a credible rival for the same step or response and an observation on which they differ. Use exact document words and dates. After the reveal, identify the earlier conclusion you retain or revise and the record that warrants it. The optional trace scratchpad can organize your notes.
4. Work through counts and council choices
Complete 'Construct two possible linked-response tables' and explain what your constructions establish. Keep the original response pool, the separate linked practice question and the later consultation distinct. Complete 'Your authority, cost and deadline ledger' from the council dossier. Turn the ledger into a connected recommendation, defend its strongest lawful alternative and specify a comparison that could test your account of procedure acceptance.
5. Investigate Alder independently
For the new institution, generate a fresh explanation, same-response rival and separating prediction before opening its later archive. Revise with reasons, then complete 'Define your Alder test after the archive'. State exactly what your proposed conditions vary, what stays comparable and how you would measure each response. A planned test is not an observed result. Preserve the rights named in the dossier.
6. Submit a revised finding
Review your initial accounts, workspaces, submitted checks and final brief. Reuse earlier source notes deliberately: explain their consequence for your final conclusion. State a revised conclusion or a sharper uncertainty rather than simply repeating your first claim. Check that each numerical statement carries a unit and base. Save a backup, then export or print the record for your instructor to review.
Vocabulary
Endorsement
Approval of a specified statement. Always identify the words, the people asked and the question they answered.
Representation
How a text or process includes people's submitted priorities and conditions. Textual inclusion is one form of representation.
Agenda control
The ability to determine which issues or eligible alternatives are considered before a decision is made.
Authorization
The power granted to an actor to carry out a specified action, such as making a binding vote.
Procedural legitimacy
A defended judgment that a decision process is acceptable, including when its outcome differs from someone's preference.
Causal mechanism
An account of how an actor and an operation produce a defined response through identifiable steps.
Rival explanation
Another credible process that could produce the same outcome and may predict different evidence. Some processes can coexist.
Percentage point
A unit for the difference between two percentages. It differs from a percentage change relative to a starting value.
Joint-response table
A table linking each person's responses to two questions in mutually exclusive cells, with totals for each question.
Post-treatment selection
A comparison problem created when groups are defined by something the tested intervention may itself change.
Read and explain the graph
Write the comparison question before choosing a display. Identify whether it asks about counts, shares within a group or change over time.
For every plotted value, write its numerator, denominator, population and unit. Use the base that matches the question; keep missing or self-selected observations visible.
Calculate the values using those bases, then label the axis and groups explicitly. Distinguish a percentage-point difference from a relative percentage change.
Describe the displayed comparison and one conclusion it cannot support. Name a linked, subgroup or comparison record you would need for a stronger claim.
Six assessment criteria
Your instructor reviews these six actual criteria once each on a 0 to 3 scale, for a maximum of 18. Completion and factual checks record work; they do not grade the argument. A supported retention of an earlier account can earn full credit.
1. Source contribution
2. Positive finding and exact trace
3. Same-outcome rivals and causal comparison
4. Lawful feasible decision and accountability
5. Numerical and representation reasoning
6. Unfamiliar transfer and explicit revision
The shared progression is 0 for missing or materially inaccurate work, 1 for emerging reasoning, 2 for developed reasoning, and 3 for convincing, specific and bounded reasoning. Your evidence, assumptions and feasible comparison matter; political agreement, prose polish, help use and changing your mind are not grading criteria.
Save and export your work
Watch the save-status message. Ordinary progress is saved in this browser on this device when storage is available. Keep one active tab; if a conflict appears, preserve your version before choosing which saved record to use.
Use 'Save progress backup' to download a JSON file at a stopping point. This editable backup preserves your activity record and is the way to move work between devices.
Use 'Restore backup' in this same activity to resume from a compatible backup. Check the restored chapter and notes before continuing. Keep the original downloaded file until you have confirmed the restore.
Use 'Download brief (.txt)' for a readable record and 'Print / save as PDF' for a printable copy. Inspect the exported notes, tables and revision. Keep the JSON backup as well; the readable export is for review.
Optional private online saving
Online saving is optional. Connecting to a saving service does not upload or submit the device draft. Use Save online personal draft only when you want to create or update that personal copy; keep its private recovery key and your local JSON backup.
A personal online draft is not an instructor submission. To submit through a configured hosted class, find the class with its student submission code, review the destination, apply the assignment switches if requested, complete the Class record and save the current version online. Tick the explicit consent box and use Submit this response snapshot.
Submission sends that version to the identified class. Later edits stay personal until you explicitly submit again. This hosted-class route is separate from the Canvas assignment and its upload confirmation.
Submit separately to Canvas
First use 'Print / save as PDF' or 'Download brief (.txt)' and inspect the resulting file. Then open the separate Canvas assignment and upload the format your instructor requests. Return to Canvas to check the submission confirmation. Saving in the player or downloading a file does not submit to Canvas. The progress JSON is a backup; submit it only if the assignment explicitly requests it.