1. LON-CAPA Logo
  2. Help
  3. Log In
 

Resources About Us What's New Notice Board Opinion FAQ Archive Feedback Main
Curricula
Discussion Papers
Handbooks
Lessons Learned
Methods manuals
Regional Activities
Research Techniques
Videos
Working Papers
 

Exercise 7. Pairwise comparison

Name of exercise: Pairwise comparison

Tools: Pen, paper, objects (or names written on paper) to be sorted

Time needed: 1 – 2 hours

Description: Pairwise comparison is a fine-grained method of ranking that can be used for relatively small groups of items. It is a useful cross-check for methods such as simple and matrix ranking, and it provides a good opportunity for detailed discussion of the qualities of the items being ranked.

How to do it: The items to be compared should be chosen through a process of free-listing, ranking, sorting or other techniques. Using either real objects, or the local terms for those objects written on cards, prepare all possible pairs of the items. Randomize both the order within each pair, and the order in which the pairs are presented. Before beginning the task, discuss with the respondent the process to be followed, with particular emphasis on the issue that is being addressed (this may be preference, quality, price, availability or another feature). Present the randomized pairs one by one to a respondent, and record his or her response in a half matrix such as the ones shown below. Responses from various respondents can be summed, providing insight into the consensus or majority opinion in a community or social group.

Pairwise ranking is typically used to compare a few items at a time, because it is a labor and time consuming exercise when applied to a large amount of objects. The total number of pairs that must be compared one by one is defined by the formula n(n-1)/2, where n is equal to the total number of items to be prepared. The following example of six fruits requires assessment of 15 pairs [6(6-1)/2 = 15]. Pairwise comparison of ten items would require a response on 45 distinct pairs, and 15 items implies looking at 105 pairs.

References:

IIRR. 1996. Recording and Using Indigenous Knowledge: A Manual. Silang, Cavite, International Institute of Rural Reconstruction.

Martin, G.J. 1995. Ethnobotany: A Methods Manual. London, Chapman and Hall. Chapter on anthropology.

Rastogi, A. 1999. Methods in Applied Ethnobotany: Lessons from the Field. Discussion Paper Series No. MNR 99/1. Kathmnadu, International Centre for Integrated Mountain Development.

Example:

The fifth year students of the Department of Pharmacognosy and Pharmaceutical Botany of Khon Kaen University brought samples of six fruits to class: mangosteen, orange, grape, longan, rambutan, and banana. They split into small groups to use pairwise comparison to assess the following three issues: Among the fruits, which are the most expensive?, which do you like the most?, and which are most available?. Within each group, participants split into teams of two to interview each other using the fruits that were brought to class. As shown in the following tables, the responses were summed for all participants, and the overall score and rank was calculated. The group which focused on costs ranked mangosteen and orange as the most expensive fruits, followed by grapes and longans, and then rambutan and banana. These answers, which made intuitive sense to the participants, could have been checked during a market survey in Khon Kaen. The group that compared preference for the various fruits came up with the following ranking: banana > mangosteen > orange > longan > rambutan > grape. Although banana, mangosteen and orange are typically among the favorite fruits – as seen in the exercise on simple ranking – it was surprising to find that rambutan was relatively lowly ranked. The answers of the group which worked on fruit availability were intuitively sound: bananas, oranges and rambutans were considered easier to find than grapes, longans, and mangosteens (which are more seasonal).

 

Which is the most expensive ?

Grape

Longan

Mangosteen

Orange

Rambutan

 

SCORE

RANK

Gr (7) Ba (1)

Lo (7) Ba (1)

Ma (7) Ba (1)

Or (7) Ba (1)

Ra (6) Ba (2)

Banana

6

F

 

Lo (4.5) Gr (3.5)

Ma (4) Gr (4)

Or (4) Gr (4)

Ra (1) Gr (7)

Grape

25.5

C

   

Ma (5) Lo (3)

Or (5) Lo (3)

Ra (0) Lo (8)

Longan

25.5

C

     

Or (4) Ma (4)

Ra (0) Ma (8)

Mangosteen

28

A

       

Ra (0) Or (8)

Orange

28

A

         

Rambutan

7

E

Transitivity check: Mangosteen, Orange > Grape, Longan > Rambutan > Banana

Which do you like the most?

Grape

Longan

Mangosteen

Orange

Rambutan

 

SCORE

RANK

Gr (0) Ba (8)

Lo (3) Ba (5)

Ma (4) Ba (4)

Or (3) Ba (5)

Ra (2) Ba (6)

Banana

28

A

 

Lo (5) Gr (3)

Ma (6) Gr (2)

Or (7) Gr (1)

*Ra (5.5) Gr (2.5)

Grape

8

F

   

*Ma (5.5) Lo (2.5)

Or (5) Lo (3)

Ra (3) Lo (5)

Longan

18.5

D

     

Or (3) Ma (5)

Ra (2) Ma (6)

Mangosteen

26.5

B

       

Ra (1) Or (7)

Orange

25

C

         

Rambutan

14

E

*Equally liked or disliked

Transitivity check: Banana > Mangosteen > Orange > Longan > Rambutan > Grape

Which is easiest to find?

Grape

Longan

Mangosteen

Orange

Rambutan

 

SCORE

RANK

Gr (0) Ba (9)

Lo (0) Ba (9)

Ma (0) Ba (9)

Or (1) Ba (8)

Ra (0) Ba (9)

Banana

44

A

 

Lo (3) Gr (6)

Ma (3) Gr (6)

Or (9) Gr (0)

Ra (6) Gr (3)

Grape

15

D

   

Ma (2) Lo (7)

Or (9) Lo (0)

Ra (9) Lo (0)

Longan

10

E

     

Or (9) Ma (0)

Ra (9) Ma (0)

Mangosteen

5

F

       

Ra (0) Or (9)

Orange

37

B

         

Rambutan

24

C

Transitivity check: Banana > Orange > Rambutan > Grape > Longan > Mangosteen

BACK

 
| ResourcesAbout Us  |  What's New  |  Notice Board Opinion  |  FAQ   |  Archive  |  Feedback  |  Main  |
WWF Logo Unesco Logo Kew Logo
People and Plants Online website manager: Gary J. Martin,B.P. 262, 40008 Marrakech-Medina, Marrakech, Morocco;
Fax +212.4.329544, e-mail
peopleandplants@cybernet.net.ma
Website design & maintenance by
RAM Production Sdn. Bhd.
People and Plants Online © WWF, UNESCO and Royal Botanic Gardens, Kew
Disclaimer
Links to other websites cited in People and Plants Online do not imply endorsement of these sites or their content
by the People and Plants Initiative or its sponsoring institutions