**The analysis by pair makes it possible to evaluate different alternatives between them, to prioritize them and to decide.**

## Introduction

Pairwise analysis is part of decision support tools. Coming from functional analysis methods, it makes it possible to develop a hierarchy in customer expectations, in product alternatives…

## The method

The method is rather simple. It will consist in making a comparison 2 to 2 of the subjects of our evaluation.

### 1 – Establish the scale

The first step is to develop an evaluation scale. Most often, we will use a scale of 1 to 3 with :

- X little superior to Y
- X superior to Y
- X very superior to Y

We note that if we have more than 20 subjects to evaluate, we will prefer a broader scale of type 1, 3 and 9.

### 2 – Develop the matrix

Second step, we proceed to the construction of the matrix. We build a matrix such as the example below (example with 6 functions to compare) :

### 3 – Conduct the evaluation

We then proceed to the evaluation by applying the scale we have determined for each pair. Note that under these conditions the number of analyzes to be performed is: (n ^{ 2 } – n) / 2. Which, for our example with 6 subject , gives us ( 6 ^{ 2 } – 6) / 2 = 15 reviews.

Using the example, we get the following evaluation :

The matrix reads as follows :

- For the cell F1 / F2 : F2 is little higher than F1
- For the cell F3 / F6 : F3 is higher than F6
- …

### 4 – Identify the final scores

Once the evaluation is done, the scores are calculated. We add the scores of the corresponding columns and lines to each of our subjects. We get the following table :

### 5 – Conclusion

Last phase, we will conclude on the relative importance of each subject between them. If the subject of our study was to evaluate the relative importance of the functions with respect to customer satisfaction, we conclude that the function F4 is twice as important as the function F5 for example.

## Source

T. L. Saaty (1984) – Décider face à la complexité

H. A. David (1988) – The method of paired comparison