Types of measuresLatency:
The time from a specified event to the onset of the first occurrence of the behaviour that the observer wants to study (Martin and Bateson, 2007). It is measured in units of time (e.g. seconds, minutes, hours).
-The kitten began to suckle for the first time 3 minutes after the birth. The latency of suckling was 3 minutes.
-The horse approached a novel object 932 seconds after the object was placed in the environment. The latency of approaching a novel object was 932 seconds.
-In an experimental situation, the dog pressed the lever to get the food for the first time 2 minutes and 15 seconds after the first exposure to the experimental equipment. The latency of pressing the lever was 2 minutes 15 seconds.
Tasks: Using the following video: calculate the latency to the first capture of toy. Answers can be found on Answer page.
Using the following video, calculate the latency to first “sit” cue, the latency to a full sit (hint: rump flat on ground) and the latency from the sit cue to the sit.
The number of occurrences of the behaviour per unit time (Martin and Bateson, 2007). It is measured in reciprocal units of time (e.g. sec-1, min-1, h-1)
-The kitten suckled 5 times in the first hour after the birth (unit time). The frequency of suckling is 5 times in 1 hour (5/h or 5h-1).
-The horse approached a novel object 10 times in 10 minutes. The frequency of approaching a novel object is 10 times in 10 minutes (10/10 min or 10x10min-1).
-In an experimental situation, the dog pressed the lever to get the food 20 times in 5 minutes. The frequency of pressing the lever is 20 times in 5 minutes (i.e. 20/5 min or 20x5min-1).
Task: Calculate the frequency for left front paw standing fully on the piece of paper positioned on the floor. (Answer on answer page).
The length of time a single occurrence of the behaviour lasts (Martin and Bateson, 2007). It is measured in units of time (e.g. seconds, minutes, hours). Duration can also be expressed as total duration or as a proportion. Total duration is the total length of time for which all occurrences of the behaviour lasted over some specified period, usually the whole observation session (e.g. 10 minutes in 1 hour). Proportion is the percentage of time spent performing the behaviour during the whole observation period (e.g. 10%).
-The kitten suckled for 5 minutes the first time it suckled after birth. The duration of the sucking behaviour was 5 minutes. The time spent suckling in the first hour after birth was 20 minutes. The total duration of suckling behaviour was 20 minutes per 1 hour. The proportion is 33% of the first hour of life of the kitten.
-The horse sniffed a novel object for 40 seconds on its first exposure to the object. The duration of the first sniffing behaviour was 40 seconds. The total time spent by the horse sniffing the novel object in the first 5 minutes of testing was 2 minutes. The total duration of sniffing behaviour was 2 minutes. The proportion of sniffing was 40% for the first 5 minutes after the object was presented.
-In an experimental situation, the observer is interested in the length of time a dog spends sitting next to its owner in an unknown environment. The duration of the first episode of sitting behaviour is 2 minutes. The time spent by the dog sitting near the owner (30 minutes) during the whole experiment (1hour [i.e. 60 mins]) is therefore a total duration of 30 minutes per hour. The proportion is 50% of the time of the experiment.
Tasks: Using the following video, calculate the duration of scratching as well as the proportion of time spent scratching (answers on answer page).
Using the following video, calculate the duration of kitten’s head in physical contact with the adult cat. Also calculate this as a proportion of total video time (answers on answer page).
A bout is defined as a cluster of behaviour patterns (i.e. a brief act repeated several times in succession, known as a ‘bout’ of events). It can also refer to an action which occurs continuously for a period (in this case it is called a ‘bout’ of a single behavioural state). The concept of bout becomes important whenever we need to describe behaviours which occur as patterns or are repetitive.
Graphical representation of a behaviour occurring in discrete bouts. The horizontal axis represents the time (if we are observing more than one animal, each axis correspond to one animal). The vertical lines represent each occurrence of the event that we are looking for.
Example 1. Imagine that you are observing two cats during an aggressive encounter and you observe each time they swish their tails. The graph below represents the data collected.
Each horizontal axis represents one cat and each vertical line represents one tail swishing. We have observed three bouts of tail swishing for the cat A and four bouts of tail swishing for the cat B.
Example 2. Imagine that you are observing two dogs while someone is teaching them to press a lever. The graph below represents the data collected.
Each axis represents one dog and each vertical line represents one lever pressing. How many bouts do you count for Dog A and Dog B? (answers on answers page)
Has no universal definition however it can be a very useful measure and can relate to frequency, duration and/or bout measures or the presence of specific aspects of a behaviour.
For example it may be important to measure the intensity of vocalisations, which may be measured as a simple description which is defined prior to the experiment commencing (e.g. in humans; whispering, talking, and shouting). Alternatively, the vocalisations could be measured on a decibel meter or a scoring system for intensity could be produced utilising frequency and/or duration measures, e.g. higher frequency of the behaviour is equivalent to higher intensity. The first example (whisper, talk, shout) provides qualitative data whereas the output from the decibel meter provides quantitative data. Sometimes intensity can be measured according to the presence or absence of certain components of the act, which may be present only at high intensity (Martin and Bateson, 2007). If the observer is interested in recording the intensity in the experimental design it is essential to make judgements about the intensity of the behaviour pattern, preferably before the study is started.
-The observer is interested in recording aggressive behaviour in dogs. Dogs can use different aggressive signals for threatening another individual and these signals vary in their intensity. The escalation of dog aggressive behaviour, from the mildest threat to the most severe usually follows the pattern (simplified for the purposes of this example), stare, growl, snap, bite. If the observer is interested only in the most intense aggressive behaviour he will score only when dog bites.
-The observer is interested in grooming behaviour in cats. He may record only the most intensive behaviour (e.g. if the cat pulls out his fur).
-The observer is interested in studying the intensity of drinking behaviour in horses after a working session. He can record the volume of water drunk during first water access after work.
TASK: Listen to the following audio clips of cats vocalising prior to receiving food. Decide what measure you would like to use to measure intensity (e.g. duration, volume) and rank the vocalisations from lowest intensity to highest.
Audio 1: Horace vocalising
Audio 2: Herbie vocalising
Audio 3: Cosmos vocalising
Events and states
As well as deciding what type of measure (e.g. duration, frequency, latency) to record for each behaviour, knowing how to categorise the behaviour itself is as important.
Behaviours may be recorded as either as events or as states. Events are instantaneous while states have appreciable duration (Martin and Bateson, 2007). The performance of any kind of behaviour takes some amount of time (however brief), but if we consider behaviours at the moment of their onset (or at any other single defining instant), we are recording events. States on the other hand are those behaviours that are likely to have a measurable start time and a measurable end time (e.g. eating, walking, running, drinking, and sitting). However the onset of these behaviours would be events, for example the change in posture from standing to sitting when the behaviour sitting occurs is of relatively short and fixed duration.
Example: The following videos illustrate events (head shake, yawn and jump).
Example: The following videos illustrate states (urination, resting and walking and standing [same video]).
The observer is interested in studying stress signals in dogs during a veterinary examination and record behaviours such as lip licking and yawning. These behaviours are recorded as events. Recording events is a process for documenting the number of times the studied behaviour occurs and is thus usually a measure of frequency.
Please note that even although these behaviours are classed as events, by using very accurate measuring devices it may be possible to compare duration in these behaviours which may give valuable information. However, some events occur so quickly (e.g. eye blink) that even with using the most accurate recording device, human reaction rates and error associated with this is likely to lead to inaccurate measurement of duration.
-The observer wants to study dogs’ social play behaviour and is interested in the number of play signals that a subject needs to start a play session with an unknown dog. He records each play bow as an event.
-The observer is interested in studying seasonal variations in daily activity of a group of horses. The duration of grazing, walking, and sleeping is recorded for each subject. Those behaviours are recorded as states, start and end of each behaviour are recorded.
-The observer is interested in cats’ predatory behaviour and records locating, stalking, and chasing prey as states. However, the killing bite is recorded as an event due to its very short duration.
-The observer is interested in cats’ resting behaviour and records each time that a cat assumes a sitting posture; the onset of sitting behaviour is recorded as an event. If the observer records that the cat is seated for 10 minutes, to be seated is a state.
When recording behaviours as events the observer may be interested in the frequency or the latency to a particular behaviour. Frequency measure involves counting the number of occurrences of the studied event during the observing session.
The observer is interested in the frequency of scratching behaviour in cats and records as an event the onset of this behaviour. During the 1 hour observation period the cat scratches 4 times. The frequency of the event onset of scratching is 4 times in 1 hour.
Behaviours scored as events usually have instantaneous occurrence, they happen so fast that for the observer it is often not useful or accurate to record the duration (e.g. blinking, lip licking, yawning).
The observer is interested in blinking and lying down of a cat. Lying down has a clear beginning and end point, thus the observer can easily determine when the behaviour starts and when it ends whereas it is difficult to define onset and termination for blinking due to the short duration of this behaviour, so it is better to record this behaviour as event.
Once a defining event has been chosen, that behaviour is not scored in a sample session unless the defining event occurs during the session, even though the behaviour is otherwise “in progress” during the session.
The observer wants to record as an event when the cat opens the eyes. The recording session started when the cat had already opened the eyes. The observer could not record any event opening the eyes because when he started to record cat’s eyes were already opened. The studied event occurred before the sampling period.
When recording behaviours as states the observer may want to study the duration of a particular behaviour in other words the time spent in some activity. Each occurrence of the particular behaviour can be measured directly or the transition times (onsets and terminations) can be recorded from which duration can be calculated.
The observer wants to study if there are any differences in the time spent eating by stabled horses versus horses that live on pasture. Recordings are made of when the horse begins eating and stops eating. This behaviour is recorded as a state. The observation period is 2 hours for each subject. The pastured horse spent 35+15+25+7+8 minutes eating, whereas the stabled horse spent 10+3+4+5+13+20 minutes eating. The total duration of eating behaviour for Horse (Pastured) is 90 minutes per 2 hours, while for Horse (Stabled) is 55 minutes per 2 hours.