Comparing policy options with a criteria matrix

A recommendation hides the trade-offs inside one answer. A matrix lays them on the table so everyone can see which value is doing the deciding.

The idea

The core policy-shop move is to resist jumping to a single recommendation. Instead you generate genuinely different options — the status quo, a tax, a ban, a subsidy, a mandate, disclosure — and score each against explicit criteria: effectiveness, cost, equity, feasibility, political viability.

Then comes the honest part. You attach weights to the criteria, and the weights are values, not facts. Below, one goal — cutting sugary-drink consumption — and six instruments. Slide the weights and watch the ranking reshuffle. There is no value-free “best.”

Goal

Reduce sugary-drink consumption — compare six policy instruments

Under the default weights, the reformulation mandate leads. Push equity up or effectiveness down and a different option takes the top — the highlighted cell shows what’s deciding it.

The 1–5 cell scores are illustrative analytic judgments for teaching, not empirical findings. In real work each score must be defended with evidence, and reasonable analysts will disagree.

How it works

Score every option on every criterion, weight the criteria by your declared values, and take a weighted average.

fit = Σ (weight_c × score_oc)  ÷  (Σ weight_c × 5)  × 100

  scores are 1–5   (higher = more favourable on that criterion)
  weights are 0–5  (YOUR declared values, stated openly)

worked — reformulation mandate, default weights E4 C2 Q3 A2 P2:
  4×4 + 2×4 + 3×4 + 2×3 + 2×3            = 48
  denominator = (4 + 2 + 3 + 2 + 2) × 5 = 65
  fit = 48 / 65                          = 74%

reweight and the winner moves:
  weight only politics + feasibility  ->  status quo climbs to the top
  weight equity heavily               ->  labels and subsidy rise
  weight effectiveness heavily        ->  tax and reformulation lead

When to use it

Situation	What the matrix buys you
A decision with several real options and no obvious winner	The trade-offs become visible instead of buried in prose
Stakeholders who disagree on values	You argue about weights openly, not about a hidden recommendation
A memo or briefing to a decision-maker	They can see how their priorities change the answer
Trade-off / limit	It looks objective but the scores and weights are judgments — garbage in, garbage out. It structures the argument; it doesn’t settle it.

Watch out for

Straw-man options. Padding the list with obvious losers so your favourite wins. The alternatives must be genuinely viable, or the matrix just launders a predetermined choice.
Forgetting the status quo. “Do nothing” is always an option and often the incumbent. Leave it out and you overstate the case for acting.
False precision. A 1–5 score looks hard, but a 3-vs-4 is a judgment call. Don’t let two decimal places of “fit” imply certainty you don’t have.
Double-counting criteria. If “cost” and “feasibility” largely measure the same thing, you’ve secretly doubled that dimension’s weight.
Reverse-engineering the weights. Tuning weights until the answer you already wanted wins. The weights should come from stated values first, and be held even when the ranking surprises you.

Worked example

A policy-analysis interviewer asks: “How would you advise a city weighing options to cut sugary-drink consumption?” A strong answer refuses to lead with one instrument. You’d lay out genuinely different options — status quo, an excise tax, a portion cap, subsidies for healthier drinks, a reformulation mandate, front-of-pack labels — and score each on effectiveness, cost, equity, feasibility, and political viability. Then you make the values explicit: “If the mayor’s priority is measurable health impact, the tax and the mandate lead. If protecting low-income households is paramount, the regressive tax falls and subsidies rise. If the council can’t spend political capital this year, labels or a modest subsidy may be the only movable option.” You’re showing the decision-maker how their priorities drive the answer — which is exactly what a good analyst delivers instead of a single number.

Check yourself

Your matrix ranks the tax first. A colleague raises the weight on equity and now subsidies win. What does that flip tell you?

You have four options, but three are plainly unworkable and one is your preferred plan. What’s the problem?