Team analysis, and the stage-3 exercise: Lakeshore Print Co.

Organising four people around one matrix

The group PSP case is 10 % of the course grade and is marked as one document. Three decisions make or break it.

Divide by criterion, not by stage. The instinct is to give one person stage 1, another stage 2, and so on. Then nobody sees the whole case, the stage-2 cause does not match the stage-3 alternatives, and the write-up reads as four documents stapled. Divide the matrix instead: each member owns one or two criteria — the row, its weight and the reason for it, the evidence for every cell in that row, and the sensitivity of the answer to that row. Everyone therefore scores every alternative, and the matrix is built by people who have each read the whole case.

Agree the decision rule before the disagreement. Seek consensus first — a weight or score everyone can live with, with the reason written down. When consensus fails, fall back to a majority rule the team agreed on in week one, and record the dissent as a sensitivity test: the dissenter's weight is exactly the "most contestable weight" of L6.4. A team that decides its rule during the argument decides it in favour of whoever is loudest.

One voice. One member edits the final document end to end so the register, the headings and the tense are the same throughout; the others review it against the rubric line by line. Groupthink's counter-check applies: the reviewer's job is to find the case against the leader, not to admire it.

What a full-marks stage 3 looks like — the clinic

Alternatives. [≥3 genuinely different, incl. status quo] (A) Keep the paper book — status quo. (B) Add a text-reminder service, book unchanged, about $60 a month. (C) Move to an online-booking system with reminders, about $150 a month. (D) Hybrid — online booking for new patients and reminders for all, phone bookings kept and keyed into the shared calendar.

Criteria, traced. [criteria named, each traceable to a stakeholder objective] [qualitative and quantitative both present] No-show reduction 0.30 (dentists — quantitative, expected fall in the 12 %); annual cost 0.20 (dentists — quantitative, dollars); scheduling accuracy 0.20 (Tomas — quantitative proxy, double-bookings per month); ease of use 0.15 (receptionists — qualitative); patient convenience 0.15 (patients — qualitative).

Weighting, justified. [weighting stated and justified] No-shows carry 0.30 because they are the revenue objective and the reason the question was asked; cost and accuracy 0.20 each because each is a hard objective of one stakeholder; the two qualitative criteria 0.15 each because they decide whether a system is used, not whether it is right. Sum 1.00.

Matrix, with reasons. [every matrix cell has a reason] One line per cell — e.g. accuracy: A 1 (one book, two hands), B 1 (the book is unchanged), C 5 (one shared calendar), D 4 (phone bookings still keyed in). Totals shown; the leader named as the leader, not as the answer.

Sensitivity. [sensitivity stated in one line] Moving 0.10 from no-shows to ease of use lets D overtake C: the C-versus-D choice turns on training burden — not robust between those two; robust against A and B.

Bias check. [biases and assumptions checked in writing] Assumptions: reminders reduce no-shows (evidence: the peer-reviewed review, cited); patients will use online booking (an opinion either way — flagged). Fallacies caught: "patients won't book online, so software is pointless" (hidden premise); "go digital and we'll be hacked" (slippery slope). Biases: status quo ("eleven years on paper"); anchoring on the first vendor's quote; groupthink at the practice meeting — a devil's advocate assigned to argue for B.

Now Lakeshore

You have every piece: the alternatives A–D and the rejected straw (L6.1), the five criteria traced and labelled (L6.2), the matrix with a reason in every cell and the totals (L6.3), the three weightings and the contestable score (L6.4), and three lessons of bias check (L6.5–6.7). Assemble them into one stage-3 response in the shape above. The matrix is the one fixed in L6.3 — reuse its weights and scores, and write the reasons in your own words. The bias check must be in writing, and must include the assumption about volumes and the tickets: an analyst who has accepted a gift from one vendor cannot score that vendor's press without saying so.