Copied from Animal Behaviour, 43, 185–188 (1992).

The use of animals in research raises important ethical issues. Studies in laboratory settings necessarily involve keeping animals in cages. Manipulative procedures and surgery may be necessary to achieve the aims of the research. Observation of free-living animals in their natural habitats may involve disruption, particularly if feeding, capture or marking is involved. While the furthering of scientific knowledge is a proper aim, and may itself advance an awareness of human responsibility towards animal life, the investigator should always weigh any potential gain in knowledge against the adverse consequences for the animals used as subjects, and also for other animals in the case of field studies.

In order to help their members make what are sometimes difficult ethical judgements, the Association for the Study of Animal Behaviour and the Animal Behavior Society have formed Ethical and Animal Care committees, respectively. These committees jointly produced the following guidelines for the use of all those who are planning and conducting studies of animal behaviour. These guidelines will be used by the Editors of Animal Behaviour. Submitted papers that appear to violate the spirit of the guidelines will be referred to one of the committees, and the evaluation of the committee will be used by the Editor in deciding whether to accept the manuscript.

Preliminary observations

Quantitative recording of behaviour should be preceded by a period of informal observation, aimed at understanding and describing both the subjects and the behaviour it is intended to measure. Preliminary observation is important for two reasons: first, because it provides the raw material for formulating questions and hypotheses; and second, because choosing the right measures and recording methods requires familiarity with the subjects and their behaviour. Preliminary observation is especially important if the problems or animals are new to the investigator.

The hypotheses that direct a study can seldom be formulated in isolation; rather, they reflect existing knowledge and theories, as well as the investigator's own interests and hunches. Effective research therefore requires the investigator to be familiar with the subjects, both by direct experience of watching them and by reading the literature about their biology and behaviour. A period of preliminary observation also provides a valuable opportunity for sharpening up questions and hypotheses, practising recording methods and generating additional or supplementary hypotheses. Most biologists would argue that an initial phase of description is an essential precursor to quantification, since before the right questions can be asked it is essential to know what there is to ask questions about.

We cannot over-state the importance of simply watching before starting to measure systematically. Beginners often falter early in a study because they rush to obtain ‘hard data’ and do not allow sufficient time to watch, think and frame interesting questions.

The response to the first edition of this book was gratifying. Reviewers seemed to agree that the need for such a source existed and that Measuring Behaviour satisfied that need. More than six years have slipped past since Measuring Behaviour first appeared, and the desirability of some changes became apparent. With the passage of time some sections of the first edition – notably those dealing with the technology for recording behaviour – had inevitably begun to show their age. In revising the book, we have taken the opportunity to make a number of changes to the order in which various topics are dealt with, thereby (we hope) making the ideas flow in a more cohesive manner. Some of the sections have been expanded and some entirely new ones added.

Despite our disclaimers in the introduction to the first edition, some reviewers said we should include detailed examples, both to illustrate the general points and to show in particular cases how research methods are selected and used. We have once again ignored their advice, because others – ourselves included – were pleased that we had kept the book concise. We remain convinced that a text book like this benefits from being as succinct as possible, and have therefore resisted the temptation to allow the new edition to burgeon in the cause of comprehensiveness.

We hope Measuring Behaviour will continue to be useful and that this new edition will help a new generation of scientists and students who wish to observe and measure the behaviour of animals and humans, both in the laboratory and in the field.

Why measure behaviour?

In addition to its intrinsic interest, the study of behaviour is both intellectually challenging and practically important. Animals use their freedom to move and interact, both with their environment and with one another, as one of the most important ways in which they adapt themselves to the conditions in which they live. These adaptations take many different forms such as finding food, avoiding being eaten, finding a suitable place to live, attracting a mate and caring for young. Each species has special requirements, and the same problem is often solved in different ways by different species.

Even though much is already known about such adaptations and the ways in which they are refined as individuals gather experience, a great deal remains to be discovered about the diversity and functions of behaviour. The principles involved in the evolution of increasingly complex behaviour and the role that behaviour itself has played in shaping the direction of evolution are still not well understood. Explanations of how behaviour patterns have arisen and what they are for will only come from the comparative study of different species and by relating behaviour to the social and ecological conditions in which an animal lives.

Animals are studied for many reasons (Driscoll & Bateson, 1988). Medical research has as its goals outcomes that are of direct benefit to humans. Such work is readily justified in the eyes of the public because its usefulness is obvious.

The options available

Having looked at the various forms of behavioural measure, we now move on to consider the mechanical processes involved in recording them. The choice of the medium, or physical means, used to record behavioural observations has important consequences for the sorts of data that can be collected and the sampling techniques that can be used. Five basic methods of recording behaviour are available: film or video recording; written or dictated verbal descriptions; automatic recording devices; check sheets; and computer event recorders. The most flexible and commonly used methods are check sheets and computer event recorders (described separately in sections 7.B and 7.C). Using check sheets, an event recorder or any other method obviously presupposes that the observer has formulated a set of discrete behavioural categories.

1 Film or video recordings give an exact visual record of the behaviour which can subsequently be slowed down for analysis. This is useful for studying behaviour that is too fast or too complex to analyse in real time. Similarly, exact records of vocalisations can be made with an audio tape-recorder and the sound patterns analysed later using instruments such as a sound spectrogram.

The major advantage of recording behaviour in this way is that the record can be analysed repeatedly and in different ways. However, filming or videotaping is rarely a complete replacement for observation. At some stage it is always necessary to code the behaviour – that is, to transcribe the record into quantitative measurements relating to specific behavioural categories.

Types of measure: latency, frequency, duration and intensity

Behavioural observations most commonly yield four basic types of measure.

1 Latency (measured in units of time; e.g., s, min or h) is the time from some specified event (for example, the beginning of the recording session or the presentation of a stimulus) to the onset of the first occurrence of the behaviour. For example, if a rat presses a lever for the first time 6 min after it is placed in a Skinner box, the latency to press the lever is 6 min. (See Fig. 5.1.)

If, as is normally the case, the period of observation is limited and each individual is tested more than once, then the behaviour pattern may not occur at all during some tests. This makes it impossible to calculate a simple average latency for each individual, although there are methods of combining repeated latency measures into an overall score for each subject (e.g., Theobald & Goupillot, 1990).

2 Frequency (measured in reciprocal units of time; e.g., s-1, min-1 or h-1) is the number of occurrences of the behaviour pattern per unit time. Frequency is a measure of the rate of occurrence. For example, if a rat presses a lever 60 times during a 30-min recording session, the frequency of lever pressing is 2 min-1. (See Fig. 5.1.)

An alternative usage, which is perhaps more common in the behavioural literature and in statistics, is when ‘frequency’ refers to the total number of occurrences.

This book came about almost without our realising it. We needed something to give to our undergraduate students who were about to embark on projects that involved measuring behaviour. No single source was concise enough or easily accessible to them, so we prepared some notes. These grew longer and longer, they generated more and more comments and, before long, we found ourselves doing something that neither of us had planned.

As the shape of the book started to emerge, we felt that its recommendations ought to represent, as far as possible, the collective view of the practitioners of the subject, rather than merely our own opinions. Therefore, a draft version was read by a large number of friends and colleagues. The eventual book was immeasurably improved by their advice, comments, criticisms and suggestions. Inevitably, some issues still provoke substantial disagreement; we have attempted to indicate where this is the case and where we have departed from common practice.

Some of our reviewers asked us to put in detailed examples, to illustrate the general points and to show in particular cases how methods are selected and used. We decided not to do so to the extent they would have liked for several reasons. First, we wanted to keep the book as concise as possible, in order to make the information in it readily accessible. Second, we wished to draw attention to general principles of measurement and analysis.

Different approaches to studying behaviour

Not all scientists study behaviour in the same way or ask the same sorts of questions. Historically, psychology (which originally grew out of the study of the human mind) was distinguished from ethology (the biological study of behaviour) in terms of the methods, interests and origins of the two sciences. In this century, comparative and experimental psychologists have tended to focus mainly on questions about the proximate causation of behaviour (so-called ‘how’ questions), studying general processes of behaviour (notably learning) in a few species under laboratory conditions. In contrast, ethologists have their roots in biology and have asked questions not only about how behaviour is controlled but also about what behaviour is for and how it evolved (‘why’ questions).

Biologists are trained to compare and contrast diverse species. Impregnated as their thinking is with the Darwinian theory of evolution, they habitually speculate on the adaptive significance of the differences between and within species. Indeed, many ethologists are primarily interested in the biological functions of behaviour and are wary of proceeding far with laboratory experiments without first understanding the function of the behaviour in its natural context. Studies in unconstrained conditions of animals, and increasingly of humans, have been an important feature of ethology and have played a major role in developing the distinctive and powerful methods for observing and measuring behaviour. In contrast, psychologists have traditionally placed greater emphasis on experimental design and quantitative methods.

General advice

Our aim in this chapter is to consider briefly some of the main issues that arise in the statistical analysis and interpretation of behavioural data. This is not a statistics textbook, however, and for an account of statistical methodology readers should refer to one of the many excellent statistics books available.

Our general advice is not to become obsessed by statistical techniques, nor too cavalier in their use. Statistical analysis is merely a tool to help answer questions, and should be the servant rather than the master of science. The great physicist Lord Rutherford was perhaps over-stating this point when he wrote: ‘If your experiment needs statistics, you ought to have done a better experiment’; biological systems have much greater variability than is encountered in physics experiments and statistical analysis is often essential for understanding what is going on. Nonetheless, excessively complicated statistics are sometimes used as a poor substitute for clarity of thought or good research design. Statistical analysis, no matter how arcane or exquisite, can never replace real data.

Besides being sometimes over-used, statistical techniques are frequently misused in the behavioural literature. We have already outlined one common error – that of pooling data in the mistaken belief that they are independent of one another (section 3.F). Other familiar pitfalls, covered in this chapter, include the inappropriate use of one-tailed tests (section 9.C.5), the conflation of effect size with statistical significance (9.D), the application of parametric tests to data that violate their underlying assumptions (9.F), the misinterpretation of multivariate statistics (9.1) and the various misuses of correlation coefficients (9.G).

Reliability versus validity

Measuring behaviour, like measuring anything else, can be done well or badly. When assessing how well behaviour is measured, two basic issues must be considered: reliability and validity (sometimes expressed as the distinction between ‘good’ measures and ‘right’ measures).

1 Reliability concerns the extent to which measurement is repeatable and consistent; that is, free from random errors. An unbiased measurement consists of two parts: a systematic component, representing the true value of the variable, and a random component due to imperfections in the measurement process (see section 9.K and Martin & Kraemer, 1987). The smaller the error component, the more reliable the measurement. Reliable measures, sometimes referred to as good measures, are those which measure a variable precisely and consistently.

At least four related factors determine how ‘good’ a measure is:

(a) Precision: How free are the measurements from random errors? The degree of precision is represented by the number of significant figures in the measurement. Note that precision and accuracy are not synonymous: accuracy concerns systematic error (bias) and can therefore be regarded as an aspect of validity. A clock may tell the time with great precision (to within a millisecond), yet be inaccurate because it is set to the wrong time.

(b) Sensitivity: Do small changes in the true value invariably lead to changes in the measured value?

(c) Resolution: What is the smallest change in the true value that can be detected?

The SI system of units (Système International d'Unités) should be used for measurements. The SI system is completely coherent, which means that all derived units are formed by simple multiplication or division of base units without the need for any numerical factors or powers of ten. This distinguishes the SI system from earlier metric systems such as the centimetre-gramme-second (CGS) system, which it superseded. The SI system comprises nine base units, each of which is independently defined, and various other units which are derived by combining two or more base units. The base units, together with some of the more common derived units, are listed in Table A2.1. Some common non-SI units and their SI equivalents are shown in Table A2.2.

Conventions. Each unit is represented by a standard unit symbol (e.g., m, s, A, kg), which may be multiplied or divided by other unit symbols or numbers (e.g., 3 m, 0.12 kg m, 16.5 m s-2). Unit symbols are algebraic symbols and follow the conventions of algebra. They are not abbreviations, and should never be followed by a full stop or an ‘s’ (to denote plural). The names of units (e.g., metre, second, ampere) are all spelt with a lower case initial letter. Symbols for units named after a person start with an upper case letter (e.g., A for ampere, K for kelvin, Pa for pascal). When a measurement is used as an adjective, the number and unit should be joined by a hyphen (e.g., three-metre tube, 5-m distance, 9-s delay, 6-m2 area, seventy-kilogram adult, 2-A current).

The beeper was designed by Chris Hereward of the Cambridge University Department of Zoology, who kindly gave permission for it to be reproduced here.

Circuit description

IC1 (a PXO 1000) is an integrated circuit crystal oscillator and divider unit. The interval between beeps (i.e., the sample interval) is determined by the settings of switches A–D, as shown in Fig. A4.1. The output of IC1 is inverted by a part of IC2 (a 4049UB) and used to turn on IC3 (a TLC 555CP) for about 0.1 s, timed by a 1 MΩ resistor and a 0.1 µF capacitor. IC3 is an astable multivibrator that produces the output tone. This is buffered and amplified by IC2 and fed into the positive battery line via a 390 Ω resistor. The battery is given a degree of isolation from the circuit supply by a silicon diode. This may seem a curious way of coupling the output, but it has the advantage that the circuit is turned off by unplugging the earphone. Plugging in the earphone switches on and resets the timer.

1. Formulate an initial question and make preliminary observations.

The question(s) asked may stem from previous knowledge or observations and may initially be rather broad or quite specific. A period of preliminary observations provides an essential opportunity to become familiar with the subjects and their behaviour and to develop hypotheses and methods of measurement (Chapter 4).

2. Formulate hypotheses and make predictions. On the basis of preliminary observations and existing knowledge, formulate hypotheses: the more the better. These should make specific predictions. Choose observations or experiments that will best distinguish between the competing hypotheses (Chapters 2 and 3). Formulating hypotheses is a creative process requiring imagination.

3. Choose behavioural measures and a research design. Be reasonably selective. Choose measures that are relevant to the question and which will help distinguish between different hypotheses. Hypotheses can be tested by observing natural variation as well as by performing experiments (Chapters 3 and 5).

4. Define each measure. Definitions should be clear, comprehensive and unambiguous. Write down the definitions before starting to collect data (Chapter 4).

5. Select the appropriate recording methods. These will depend on the nature of the behaviour to be recorded (Chapters 6 and 7). In practice, steps 3, 4 and 5 are inter-related, since the recording methods available often influence the choice and definition of categories.

6. Practise the recording methods. Assess the reliability and validity of each category. Drop categories that are clearly unreliable and irrelevant.

Rules for research

Formal prescriptions about the way in which scientific research should be conducted often fail to capture the flair, intuition and idiosyncrasies of the best scientists. Therefore, advice on how to organise thought and carry out research has to be given and taken with caution. Some people start studying a particular species simply because they find the animal interesting. Other scientists are initially more interested in a theoretical problem and, if they are wise, choose an animal that is especially suitable for studying that problem. It is not obvious that one approach is better than the other and, indeed, they are usually complementary.

Empirical data that distinguish between competing hypotheses can be obtained by observing natural variation as well as by performing controlled experiments. Thus, for many purposes, it is unnecessary to draw a rigid distinction between observational and experimental research. Many questions about behaviour are most appropriately answered by observational research. Moreover, in order to be effective, experimental research almost invariably needs to be preceded by observation. Knowledge of the normal behaviour of healthy animals, preferably in their natural environment, is an essential precursor to experimental research (see section 2.E on field studies). One of the distinguishing features of the ethological approach to studying behaviour has been its emphasis on combining observation with experiment.

Laboratory and field experiments

The point of doing an experiment is to distinguish between alternative explanations or hypotheses.

Sampling rules: ad libitum, focal, scan and behaviour sampling

When deciding on systematic rules for recording behaviour, two levels of decision must be made. The first, which we refer to as sampling rules, specifies which subjects to watch and when. This covers the distinction between ad libitum sampling, focal sampling, scan sampling and behaviour sampling. The second, which we refer to as recording rules, specifies how the behaviour is recorded. This covers the distinction between continuous recording and time sampling (which, in turn, is divided into instantaneous sampling and one-zero sampling; see Fig. 6.1). In this section, we consider the four different sampling rules.

1 Ad libitum sampling means that no systematic constraints are placed on what is recorded or when. The observer simply notes down whatever is visible and seems relevant at the time.

Clearly, the problem with this method is that observations will be biased towards those behaviour patterns and individuals which happen to be most conspicuous. For example, in a study of agonistic behaviour in rhesus monkeys, Bernstein (1991) found that ad libitum sampling tended to miss brief responses and under-estimated the involvement of younger individuals in social interactions. Provided this important limitation is borne in mind, however, ad libitum sampling can be useful during preliminary observations, or for recording rare but important events.

2 Focal sampling (or ‘focal animal sampling’) means observing one individual (or one dyad, one litter or some other unit) for a specified amount of time and recording all instances of its behaviour – usually for several different categories of behaviour.