SPROBID (Simplified PROBID)

SPROBID answers the question "how do I rank alternatives like PROBID but with less computation?" PROBID extends TOPSIS's idea of two reference points (ideal, anti-ideal) into as many intermediate reference points as there are alternatives; this increases resistance to rank reversal but increases the computational burden as well. SPROBID keeps the same logic but uses only the topmost and bottommost quartile instead of the full set of intermediate references. Its output is a ratio score for every alternative and the rank that score produces. Wang, Rangaiah and Wang proposed it alongside PROBID in 2021; it is used in multi-alternative industrial engineering problems such as chemical engineering and process selection.

PROBID's philosophy is to look not only at the best and worst extremes but at every rung in between. Where TOPSIS works with two reference points (best, worst), PROBID builds as many intermediate reference points as there are alternatives, and averages each real alternative's distance to them, weighted by rank. This makes it harder for a single extreme alternative to distort the result. SPROBID keeps this philosophy but takes a shortcut: instead of every rung in between, it accounts only for the topmost and bottommost quartile.

An analogy makes this clear. Assessing a class's overall level of achievement by looking only at the average of the top-performing quarter and the bottom-performing quarter, rather than examining every student's mark one by one, illustrates the idea. The middle group barely enters the calculation, yet the overall trend is usually still captured. The philosophical consequence is this: SPROBID remains compensatory and retains much of PROBID's resistance to rank reversal, but it trades away part of that resistance in exchange for a lighter computation.

The method proceeds through five steps; the first three are shared with PROBID.

First step (shared with PROBID), scale equalisation and weighting. Criteria are equalised through vector normalisation, then multiplied by the criterion weights.

Second step (shared with PROBID), building the intermediate reference points. As many virtual reference points as there are alternatives are constructed. These form an ordered sequence of rungs running from best to worst; not just the two extreme points but every rung in between is used as a reference.

Third step (shared with PROBID), computing the distances. Every real alternative's distance to each of these reference points is measured and a rank-weighted average is taken; closer references contribute more to the result.

Fourth step (specific to SPROBID), selecting the quartiles. Where PROBID uses the upper half of the references as good references and the lower half as poor references, SPROBID takes only the topmost quartile as good references and the bottommost quartile as poor references. This reduces the number of references taken into account not to a half but to a quarter.

Fifth step, computing the ratio and ranking. For every alternative, the ratio of its distance to the poor references over its distance to the good references is computed. The larger this ratio, the better the alternative is considered; alternatives are ranked from largest to smallest.

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

The SPROBID score shows the ratio of an alternative's distance to the poor references over its distance to the good references. Unlike TOPSIS's closeness score, this ratio is not confined between 0 and 1; it can rise above 1. A large score means good, a small score means poor; the score itself is not a percentage or a probability. It is meaningful only within the same alternative set and the same weights, and cannot be compared with a score from a different analysis.

For this reason:

"The alternative with a SPROBID score of 1.40 performed at 140 per cent"

should be written instead as:

"Within this alternative set, this alternative's distance to the poor references is 1.40 times its distance to the good references; the size of the ratio shows only its relative standing within this set"

SPROBID works with crisp data: one number per cell. DecisionMind holds SPROBID only in this base form; there is no separate data-type extension. You need alternatives in rows, criteria in columns, one number per cell; direction information for every criterion; and criterion weights summing to 1. SPROBID does not produce weights, it takes them from outside. The number of alternatives must be at least four; below four the topmost and bottommost quartile collapses to a single reference, or no reference remains at all, and the quartile logic loses its meaning.

If your number of alternatives is large and the computational burden of PROBID's full set of intermediate references poses a practical problem, SPROBID is a suitable shortcut. If your number of alternatives is small (fewer than four), SPROBID's quartile logic breaks down; the full form of PROBID should be preferred in that case. If the loss of information from the middle-ranked references is unacceptable, that is, if the intermediate rungs must also contribute to the result, full PROBID should again be used.

A multi-alternative problem, computational burden is to be reduced → SPROBID

Few alternatives (fewer than four) → PROBID (full form)

Maximum resistance to rank reversal is required → PROBID (all intermediate references retained)

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

SPROBID's core advantage is that it retains much of PROBID's resistance to rank reversal while markedly reducing the computational burden. When there are many alternatives, only two quartile groups need to be taken into account, which gives a practical advantage on large data sets. A comparative study by Wang and colleagues (2023) shows that SPROBID's ranking behaviour stays close to PROBID's.

Its limitations stem from the quartile shortcut itself. First, information from the middle-ranked references is discarded entirely; in problems where this information matters, SPROBID can diverge from PROBID. Second, once the number of alternatives falls below four, the definition of the quartiles degenerates and the method collapses into a crude shortcut that looks only at the best and worst alternative. Third, a sensitivity study by Baydaş and colleagues (2024) shows that rankings in the PROBID family can be sensitive to the choice of data transformation (normalisation); this sensitivity holds for SPROBID as well.

The most common mistake is using SPROBID with four or fewer alternatives without noticing that the quartile selection has collapsed to almost a single reference. A second mistake is reading the ratio score as a percentage or a probability; the score only shows relative standing within this particular alternative set. A third mistake is comparing PROBID and SPROBID results on the same data set and concluding that one is "more correct" than the other; the two methods use a different number of references, so a direct comparison is misleading. A fourth mistake is assigning equal weights without justification.

The governing principle is this:

A SPROBID result is a shortcut through PROBID's reference logic; the report must state plainly that information from the middle-ranked alternatives has been given up, and how much that trade-off could affect the result.