Research & PolicyFree
Comparative Policy Analysis by Jurisdiction
Send the jurisdictions and the design features, get a comparison with a source in every cell and the cells where it does not hold.
River fills the comparison grid one cell at a time from the statute or regulation itself, and each cell carries the provision it came from. Where a cell can only be filled from somebody's summary, it says so. Where it cannot be established, it stays empty rather than being guessed. Then it does the part most comparisons skip: it marks the cells where a jurisdiction-level comparison is the wrong comparison, because the operative rule is set somewhere other than the level you are reading.
Search this and you get grid templates and comparison tables somebody else built. The templates have no method for filling a cell, and the tables are usually assembled from secondary summaries whose provenance has been lost. That matters because the most useful finding in comparative work is negative. A jurisdiction that looks like a model can be structurally different in a way that makes its design untransplantable, and a table with no comparability column has no place to record that.
Written for the drafter looking for a model, the analyst asked whether another jurisdiction's approach would work here, and the researcher who needs a table where every cell can be checked. Where a design feature turns on cost, pricing the rule by affected group supplies the numbers, and who filed what on the record explains why a jurisdiction landed where it did. For enacted text, applying amendments to the statute they edit gets you the current law.
Fifty four cells, five that cannot be filled
Six jurisdictions against nine design features is 54 cells. Forty one get filled from the statute or regulation itself, 75.9 per cent. Eight can only be filled from a secondary summary, 14.8 per cent, and those are labelled rather than silently promoted. Five cannot be established at all, 9.3 per cent, and they stay empty. A grid presented as complete when a tenth of it is unverified is the standard output of this genre, and the labelling is what makes the other 41 cells worth anything.
Then comparability. Two of the six jurisdictions administer the function locally rather than at the level being compared, and four of the nine features have their operative rule set below that level in those two. That is 8 cells, 14.8 per cent, where the comparison is not wrong so much as aimed at the wrong document. Those four features compare across four jurisdictions, not six, and the analysis says four every time it reports them rather than quietly averaging over the other two.
The outcome column is thinner still. Three of the six publish the outcome metric at all. Two of those three compute it on a different denominator, so exactly one jurisdiction's figure is comparable without adjustment. Any claim that the model works therefore rests on one of six, 16.7 per cent. And on transplantability, five of the nine features could be adopted here as written while four require a structure that does not exist, so 44.4 per cent of the model needs redesign rather than translation.
How it works
Fix the features
The design dimensions the question actually turns on, written before any jurisdiction is read.
Read the primary text
Statute and regulation for each jurisdiction, with the provision recorded in the cell.
Test comparability
Which cells are set at a different level, and which outcome metrics share a denominator.
Answer the transplant
Which features port as written, and which require a structure you do not have.
What you get
- A source in every cell, naming the section it came from rather than the jurisdiction
- Cells filled from a secondary summary labelled as such, and unestablished cells left empty
- The features whose operative rule is set below the level you are comparing, marked
- How many jurisdictions each feature genuinely compares across, restated every time it appears
- Outcome metrics checked for a shared denominator before any outcome claim is made
- Which design features could be adopted as written, and which need a structure you lack
Common questions
Why is naming the non-comparable cells the point?
Because a comparative analysis gets used to justify copying something, and the failure mode is copying a design whose preconditions do not exist here. A jurisdiction with the same statute and a different administering structure produces a different result, and nobody discovers that until implementation. A cell marked not comparable is the cheapest possible version of that discovery.
What makes two jurisdictions non-comparable in practice?
Usually the level the rule operates at. Occupational safety is the clean example: a state may assume responsibility if its standards are at least as effective in providing safe and healthful employment as the standards promulgated federally, and it designates its own administering agency. Reading only federal text misses those states entirely.
Can a city ordinance be the operative rule?
Frequently, and it is a routine source of error. Federal wage law says no provision excuses noncompliance with any State law or municipal ordinance establishing a minimum wage higher than the federal one. So a state-level cell can be accurate and still not describe what an employer in that state complies with.
What if a cell simply cannot be established?
It stays empty and gets counted. An empty cell with a note on what was searched is honest and useful, because it tells the next reader where the gap is. A cell filled with a plausible value taken from a summary is worse than empty, since nobody downstream can tell which cells were verified and which were inferred.
How does it handle outcome evidence?
Cautiously, and denominators first. Before comparing any outcome it checks whether the jurisdictions compute the metric the same way, and where they do not, either adjusts with the adjustment shown or declines to compare. In the worked example only one of six figures is comparable without adjustment, which is a very different basis for a claim than six.
Will it pick better comparators than the ones I named?
It will suggest them and say why. A jurisdiction chosen for having the same policy is often a poor comparator, while one with a similar administering structure and fiscal position is a good one even if its policy differs. Both lists come back with the reasoning. A standing register of what is moving across each jurisdiction often supplies the comparator list first.
What comes back?
A Sheet holding the comparison with a provision citation in every filled cell, a provenance flag per cell, and a comparability note per feature. A Doc reading the comparison as prose, stating what the outcome evidence does and does not support, and answering which features could be adopted here as written.
Comparative Policy Analysis by Jurisdiction
Fill in the form and your workspace opens with the work already underway.