A totally non-compensatory multi-criteria method for evaluating and improving level of satisfaction (LoS): proposal and application on Airport Terminal of Passengers

Imagine you are trying to grade a school cafeteria's lunch. In the old-school way of doing things, you might take a math average: if the pizza is a 10/10 but the salad is a 1/10, the final grade becomes a "5" (average). This is called a compensatory method, where a super-good thing "compensates" for a terrible thing. But in real life, that doesn't always feel right. If the food tastes like cardboard, it doesn't matter how pretty the plate is; the meal is still a disaster. This is the heart of Multi-Criteria Decision Analysis (MCDA), a branch of science that helps us make tough choices by weighing many different factors. Usually, these tools assume everyone looks at the same list of factors (like taste, price, and speed) and that a high score in one area can fix a low score in another. But what if people have totally different priorities? What if one person cares about the music in the store while another only cares about the checkout speed? And what if a high score in one area shouldn't be allowed to hide a failure in another? This paper tackles exactly that messy, human reality.

The authors, Phelipe Medeiros da Rocha and Helder Gomes Costa, are tackling a problem that feels like a giant, chaotic puzzle: How do you measure how happy people are with a service when everyone is using a different rulebook? They applied their new method to 15 major international airports in Brazil, analyzing data from 19,240 passengers. Instead of forcing every passenger to rate the exact same 37 items (like "cleanliness of restrooms" or "speed of security"), they let each passenger pick the specific things they cared about. One person might rate the "Wi-Fi" and "food prices," while another only rates "security staff friendliness" and "baggage claim speed."

Here is the kicker: The researchers used a "totally non-compensatory" approach. Think of it like a strict bouncer at a club. If you are missing a ticket (a low score on one criterion), you don't get in just because you have a really cool hat (a high score on another). In the old math models, a great hat could "compensate" for the missing ticket. In this new model, if an airport fails at a critical point for a passenger, that passenger's experience is marked as poor, no matter how good everything else was. This prevents the "average" from hiding the fact that something is actually broken.

The team fed this massive, messy dataset into a special algorithm called ELECTRE TRI ME. This isn't just a simple calculator; it's a sorting machine that looks at the "credibility" of how well an airport fits into different satisfaction categories (like "Very Good," "Good," "Fair," "Bad," or "Very Bad"). Because the passengers used different criteria, the algorithm had to be smart enough to handle the gaps without filling them with fake "average" numbers.

The results were fascinating. The 15 airports didn't just get a single score; they were sorted into four distinct groups based on how consistently they performed.

• Group A (the top tier) included airports like Viracopos (VCP) and Afonso Pena (CWB). These places were generally "Very Good" across the board.
• Group D (the bottom tier) included Recife (REC), Salvador (SSA), and Belém (BEL). These airports had significant gaps in their performance.

The study didn't just stop at ranking them; it acted like a detective to find out why the lower groups were struggling. They found that while most airports were doing great with things like "security staff courtesy" and "waiting times," almost everyone was struggling with the same two things: the price of food and the price of shopping. Across all groups, passengers rated the "value for money" of restaurants and shops as just "Fair" (a 3 out of 5).

For the airports in the bottom group (Group D), the paper suggests they need to focus on specific "deal-breakers" to move up. These included things like directions and signage, flight information screens, internet access, and parking availability. The authors argue that if Group D airports fix these specific issues, they can jump to a higher satisfaction level. However, for the top-tier airports, the advice is to "maintain" their high standards, as they are already hitting the top marks in most areas.

The paper is careful to note that this method is a new way of looking at things. It explicitly rules out the idea that you can just average out scores to get a true picture of satisfaction. It also acknowledges a limitation: the data they used didn't ask passengers how important each item was to them (everyone's criteria were treated as equally important). Despite this, the authors suggest that their method is a powerful new tool. It allows us to see the "weak links" in a service chain without letting the strong links hide them, offering a clearer path for airport managers to actually improve the passenger experience. It's a shift from asking "What's the average score?" to asking "Where did we fail, and who noticed?"