Improving behaviour classification consistency: a technique from biological taxonomy



AARE2002: Paper CHO02101

Key to classes of living vertebrates (Subphylum Vertebrata)

1. Both fins and gills present in adults

(a) Yes - go to 2

(b) No - go to 4

2. Jaws present

(a) Yes - go to 3

(b) No - Class Agnatha (lampreys and hagfish)

3. Skeleton has bone

(a) Yes - Class Osteichthyes (bony fish)

(b) No (cartilage only) - Class Chondrichthyes (cartilaginous fish - sharks and rays)

4. Skin is naked, larvae have gills

(a) Yes - Class Amphibia (amphibians - frogs, toads, salamanders, etc)

(b) No (hair, feathers, or scales present) - go to 5

5. Skin has feathers, front limbs are wings

(a) Yes - Class Aves (birds)

(b) No - go to 6

6. Skin has hairs, milk glands produce milk for young

(a) Yes - Class Mammalia (mammals)

(b) No (skin has scales) - Class Reptilia (reptiles)

Figure 1: A binary key for the classification of common vertebrate animals.

cation (Payne & Preece, 1980); keys can be used advantageously in a variety of fields: medical
diagnosis, machine fault location, pattern recognition, and, as we show here, quantitative be-
haviour analysis. This use of keys is not unknown in education and behaviour research. Bekoff
(1977) notes some examples in animal behaviour research, and Gay (1996, pg 19) gives a key
for classifying types of research.

2 Construction of a binary key

While a great deal has been written on design of optimum keys, automated key construction
programs developed, etc (Dallwitz, 1974; Osborne, 1963a, 1963b; Payne & Preece, 1980), most
applications in behaviour research will deal with a relatively small number of categories. In
cases like these, it is easiest to simply construct our key by hand, rather than using mathemati-
cal or computer tools.

The basic principles of constructing a successful key are few (Metcalf, 1954): the classifica-
tion decisions must be based on observables, with the most prominent observables dealt with
earlier in the key rather than later, the choices at each point must be mutually exclusive, the
number of decisions required for any identification (especially common identifications) should
be as small as possible, and the key should be written as simply an clearly as possible. Sup-
porting illustrations or examples can be used to clarify specific questions asked in the key. Note
that mutual exclusivity is readily obtained in a binary key if the questions with yes/no answers
can be asked. The shortness of the key is important, since the likelihood of error in answering
the questions increases with the number of questions. A short key results from each decision
dividing the relevant group of categories into two subgroups of comparable size.

The process of constructing of a key from a list of definitions can be summarised as:



More intriguing information

1. The resources and strategies that 10-11 year old boys use to construct masculinities in the school setting
2. Cardiac Arrhythmia and Geomagnetic Activity
3. Ruptures in the probability scale. Calculation of ruptures’ values
4. The name is absent
5. PERFORMANCE PREMISES FOR HUMAN RESOURCES FROM PUBLIC HEALTH ORGANIZATIONS IN ROMANIA
6. Popular Conceptions of Nationhood in Old and New European
7. The name is absent
8. The name is absent
9. Behavior-Based Early Language Development on a Humanoid Robot
10. Estimation of marginal abatement costs for undesirable outputs in India's power generation sector: An output distance function approach.
11. Work Rich, Time Poor? Time-Use of Women and Men in Ireland
12. The name is absent
13. Moffett and rhetoric
14. Inhimillinen pääoma ja palkat Suomessa: Paluu perusmalliin
15. The name is absent
16. The Veblen-Gerschenkron Effect of FDI in Mezzogiorno and East Germany
17. The name is absent
18. EXPANDING HIGHER EDUCATION IN THE U.K: FROM ‘SYSTEM SLOWDOWN’ TO ‘SYSTEM ACCELERATION’
19. The name is absent
20. NEW DEVELOPMENTS IN FARM PRICE AND INCOME POLICY PROGRAMS: PART I. SITUATION AND PROBLEM