Does SIR require one specific aggregation formula?

No. It is a family; the additive calculation here is one explicitly chosen version.

Can it leave alternatives unranked relative to each other?

A partial-ranking approach can retain incomparability when the two flows disagree.

Are the weights objective facts?

No. They express priorities in the model and should be explained and tested rather than treated as discovered truths.

AI Glossary / Mathematics and optimization / Advanced

Superiority and inferiority ranking method

Definition

The superiority and inferiority ranking method, usually abbreviated SIR, is a family of methods for comparing alternatives across several criteria. It records how strongly each alternative performs better than others and how strongly others perform better than it. These two perspectives are aggregated into superiority and inferiority flows, which support a partial ranking or a complete ordering under a declared rule.

What it means

SIR is a decision-support method, not a machine-learning model that discovers the right preferences automatically. Its result depends on the alternatives, the criteria, the preference functions, the weights and the aggregation procedure. A high superiority flow is desirable, while a low inferiority flow is desirable. Keeping both visible can reveal a conflict that a single final score hides. The method's name refers to comparative performance within a decision model, not an absolute judgment about the value of an organization or person.

How it works

First orient the criteria so that the direction of improvement is explicit. For each criterion, compare every alternative with the others using a preference function. Sum the outgoing preferences to obtain that alternative's superiority score and the incoming preferences to obtain its inferiority score. An additive version forms weighted sums of those scores across criteria. Other SIR variants use different aggregation procedures. A partial-ranking rule can prefer A over B when A has at least as much superiority and no more inferiority, with one strict advantage. Conflicting flows can leave the pair incomparable. A declared net-flow rule can instead order alternatives using superiority minus inferiority.

Three practical examples

A procurement exercise can compare eligible suppliers using delivery reliability, technical capability and an appropriately oriented cost criterion.
A software-selection team can examine how products compare on accessibility, integration effort and service support before accepting a final ordering.
A research group can rank candidate experiments using expected information, resource requirements and feasibility, while testing how different priorities change the result.

Think of it this way

Think of a tournament report that records both advantages achieved and disadvantages conceded. Two contestants with the same final difference can have different underlying records. This helps explain why SIR retains two flows. The analogy is limited: business criteria need not be competitive matches, and the preference functions are modeling choices rather than physical wins and losses.

Worked example

Consider three alternatives with two benefit scores: A = (9, 4), B = (7, 8), and C = (5, 6). Use a simple preference of 1 for a strictly better score and 0 otherwise, with criterion weights 0.6 and 0.4. On the first criterion, superiority counts are A: 2, B: 1, C: 0; inferiority counts are A: 0, B: 1, C: 2. On the second, superiority counts are A: 0, B: 2, C: 1; inferiority counts are A: 2, B: 0, C: 1. The weighted superiority flows are therefore 1.2, 1.4 and 0.4. The inferiority flows are 0.8, 0.6 and 1.6. B outranks A, and A outranks C, because each comparison agrees on both flow directions. Net flows are 0.4, 0.8 and −1.2, producing the same order. Now increase the first weight to 0.8 and reduce the second to 0.2. A's flows become 1.6 and 0.4, while B's become 1.2 and 0.8. A now ranks above B. The change demonstrates preference sensitivity, not an arithmetic contradiction. These binary comparisons deliberately ignore the size of each winning margin.

Illustrative case study

In a fictional example, Cedar Events compares three booking platforms. Its operations team favors reliability, while finance emphasizes implementation cost. Analysts publish both sets of weights and the resulting flow tables. They discover that the leading platform changes between those scenarios and ask the stakeholders to resolve their priorities before selecting a supplier. The case illustrates transparent decision support; it does not claim that an invented platform achieved measurable savings or that a ranking replaces procurement due diligence.

Common misconceptions and mistakes

Do not interpret the highest net flow as proof that an alternative is best under every reasonable preference system. A complete ordering can conceal the disagreement that a partial ranking retains. Indifference and incomparability are different: equal flows indicate a tie under the model, whereas one better and one worse flow indicate a conflict. Do not silently treat a cost as a benefit, double-count overlapping criteria or select thresholds without considering their units. Changing the comparison set can also change flow values because alternatives are evaluated against that set.

Business use

Before using SIR, separate mandatory eligibility requirements from tradeable preferences. Record how each criterion is measured, which direction is better and why its preference function is appropriate. Obtain stakeholder agreement on weights, then vary uncertain weights and thresholds across plausible scenarios. Show the superiority and inferiority flows alongside any final order. Preserve the original score table so reviewers can distinguish errors in evidence from disagreements about priorities. For consequential choices, retain an accountable human decision and document any departure from the model's recommendation.

Check your understanding

If A has superiority 1.5 and inferiority 0.7, while B has superiority 1.2 and inferiority 0.4, does the two-flow dominance rule prefer A?

Answer: No. A has greater superiority but also greater inferiority, so the pair is incomparable under that rule.

Related terms

Sensitivity analysis
multicriteria decision analysis
preference function
outranking
simple additive weighting
TOPSIS

Learn next

reproduce the worked flow tables, then vary the first weight from 0 to 1. Find where A and B exchange positions. Next replace the binary preference with a documented threshold-based function and explain which information the new comparison preserves.

Sources

Sources: Xiaozhan Xu, The SIR method: A superiority and inferiority ranking method for multiple criteria decision making, https://www.sciencedirect.com/science/article/abs/pii/S0377221700001016 ; Mohamed Marzouk, A superiority and inferiority ranking model for contractor selection, https://www.researchgate.net/publication/235275635_A_superiority_and_inferiority_ranking_model_for_contractor_selection . The original publisher abstract and the application's author-uploaded methods were reviewed; the original full paper was unavailable. Numerical examples and Cedar Events are original.

Previous: Superintelligence ban Next: Superkey