REGIME (Regime Analysis Method)

REGIME is a ranking method for when you hold a decision table and want the alternatives placed in a single order. Its output is a net superiority score for every alternative; this score reflects the weighted difference between an alternative's total wins and losses against the others. REGIME does not generate weights, it takes them from outside. The method belongs to the family of "qualitative multiple-criteria choice models" that Hinloopen, Nijkamp and Rietveld proposed in 1983 for regional planning decisions, and it has since seen wide use in the urban and regional planning literature.

The idea behind REGIME is not to trust the precision of numbers. A criterion's measurement may be contested, coarse, or reliable only as a ranking; in that case, saying merely "A is better than B" is a more honest claim than saying "A is 3.2 units better than B." REGIME uses exactly this simple information in every pairwise comparison: an alternative scores +1 on a criterion where it beats a rival, -1 where it loses, and 0 where they tie. These signs are multiplied by the criterion weights and summed, giving the alternative's net superiority over that rival.

The structural consequence is that REGIME is an ordinal method. It never uses the magnitude of a difference, only its direction. This is an advantage when data quality is doubtful, because measurement error usually distorts magnitude without reversing direction. It is also a limitation, because "slightly better" and "vastly better" contribute the same amount (+1) to REGIME.

The method proceeds through two steps.

First, the pairwise regime vector. REGIME compares every pair of alternatives. On each criterion, it checks, according to the criterion's direction (higher or lower is better), which alternative comes out ahead. The better alternative scores +1 on that criterion, the worse one -1, and ties score 0. This builds a sign sequence, the regime vector, for every pair; the sequence carries only direction, never magnitude.

Second, the weighted sum and net score. REGIME multiplies these signs by the criterion weights and sums them; this is an alternative's weighted net superiority over one specific rival. An alternative's overall score is the sum of these net superiorities across all its rivals. REGIME ranks the alternatives from the highest score to the lowest; a positive score indicates overall superiority, a negative score overall inferiority.

The formulas behind each step and the intermediate tables are given on the DecisionMind method page; this card carries no formulas.

The net superiority score is a weighted summary of how many criteria an alternative leads on against its rivals and how many it trails on; it says nothing about magnitude. An alternative with a score of 0.6 is not "60 per cent good"; it means the alternative leads its rivals on most of the weighted criteria. A score of 0 means "undecided": the alternative leads on some criteria and trails on others, and these two directions cancel out. It cannot be compared with a score computed on a different alternative set, because the score arises solely from the pairwise comparisons within this set.

Thus instead of writing:

"REGIME found the best alternative"

the report should read:

"With these weights and this alternative set, the alternative leading on the most criteria is this one; the score reflects the count of directions, not their magnitude"

REGIME works with crisp data: a single number, or a rankable value, in every cell. DecisionMind currently holds no other data-type extension of this method. REGIME's ordinal nature suits data that already arrives in rank form (1st, 2nd, 3rd); it also bridges data that demands a crisp number but whose ordering alone can be trusted.

You need alternatives in rows, criteria in columns, a comparable value in every cell with no gaps; direction information for every criterion; and criterion weights that sum to 1. REGIME does not produce weights, it asks for them. A minimum of two alternatives and two criteria is required; the sweet spot lies between three and twelve criteria.

REGIME suits situations where you trust only the ordering of your criterion values rather than their exact magnitude, or where decision-makers, even with precisely measured data, want to discuss the direction of a difference rather than its size. Its typical territory is regional and urban planning, environmental impact assessment and supplier selection; this reflects the field it originated in.

It should not be used where the magnitude of differences genuinely bears on the decision. A very small gap and a very large gap on the same criterion contribute the same amount (+1) to REGIME; if you do not want to lose this information, a magnitude-sensitive method such as RAPS or TOPSIS should be chosen instead.

Only criterion order is reliable, magnitude is doubtful → REGIME

Gap magnitude also matters and is reliable → RAPS, TOPSIS

No compromise allowed on one criterion, sub-threshold alternatives must be screened out → screening first, then ranking

Not a ranking but weights are needed → AHP, BWM, SWARA (subjective); Entropy, CRITIC (objective)

REGIME's greatest strength is its resilience to measurement error. Because it uses only direction, a small measurement error on a criterion usually leaves the sign unchanged and does not corrupt the result. It can bring qualitative and quantitative data together under one roof, because the only information required is the answer to "which is better." Its computational burden is small, and it hands the decision-maker a result that is easy to explain in terms of "on how many criteria are you ahead."

Its limitations stem from its ordinal structure. First, magnitude information is lost entirely; a small difference and a large one carry the same weight, and this causes a meaningful loss of information in some decisions. Second, when many ties (a value of 0) occur, a criterion's discriminating power falls, and the result becomes excessively dependent on the remaining criteria. Third, the pairwise comparisons must be rebuilt whenever the alternative set changes; adding a new alternative can alter the ranking (Wang and Luo, 2009). Fourth, it should not be forgotten that REGIME comes from the family of qualitative MCDM methods that has been the subject of debate over measurement, manipulation and meaning; different aggregation rules applied to the same ordinal data can give different results (Voogd, 1988).

The most common mistake is marking criterion direction wrongly. If a "lower is better" criterion is marked "higher is better," the signs reverse and the ranking comes out wrong from the start. A second mistake is reading the REGIME score as a percentage or a distance; the score is only a weighted count of how many criteria an alternative leads on. A third mistake is interpreting the result as resting on a single dominant criterion when many criteria are tied; ties reduce that criterion's discriminating power, and the report must state this. A fourth mistake is assigning weights equally without justification.

The governing principle is this:

A REGIME score is the sum of how many weighted criteria an alternative leads on and how many it trails on; it carries no information about the magnitude of the difference.