17. Experimental Designs

Question 1

What are 3 ways of describing behaviour?

Answer 1

1. Structure
   - The appearance, physical form or temporal patterning of the behaviour
   - The behaviour is described in terms of the subject’s posture and movements
2. Consequence
   - The effects of the subject’s behaviour on the environment, on other individuals, or on itself
   - Behaviour may be described without reference to how the effects are achieved
3. Spatial Relation
   - Relation to features of the environment or to other individuals
   - Subject’s position is the salient feature; not what the subject is doing (but where and with whom)
   - Eg. “approach” or “leave”

Question 2

3 guidelines for choosing categories

Answer 2

1. Enough categories to describe the behaviour in sufficient detail
   - Many err on the side of trying to record too much, but it is essential to be selective since measuring behaviour can be limitless
   - Chances are that with fewer measures, they will be more reliable
2. Define each category precisely
3. Categories should generally be independent of one another

Question 3

Posing the initial question

Answer 3

1. Behavioural questions and hypotheses tend to initially be rather broad
   - then narrow down as more is discovered
2. The more clearly and precisely the initial question has been formulated, the more obvious will be the specific behaviours you need to measure in order to answer the question
3. When choosing behaviour categories to measure:
   - Helpful to first identify the main types of behaviour patterns that typify the species

Question 4

What are the 4 basic types of measures?

Answer 4

1. Latency
2. Frequency
3. Duration
4. Intensity

Question 5

What is Latency?

Answer 5

• measured in units of time (eg. sec, min, hr)
• Is the time from some specified event to the onset of the first occurrence of the behaviour
• eg. a ball is placed in the pen at time “0” and the length of time it takes for the first animal to approach the ball is noted

Question 6

What is frequency?

Answer 6

• Measured in reciprocal units of time
• Is the number of occurrences of the behaviour pattern per unit time
• Frequency is a measure of the RATE of a behaviour’s occurrence

Question 7

What is duration?

Answer 7

• Measured in units of time
• Is the length of time for which a single occurrence of the behaviour patter lasts

Duration is also used in two other ways
1. Total duration
- refers to the total length of time for which all occurrences of the behaviour lasted over a specified period
- can also be expressed as a proportion (or percentage) of the observation period (eg. proportion of time)

2. Mean duration
   - refers to the mean length of a simple occurrence of the behaviour pattern, measured in units of time
   - obtained by recording the duration of each and calculating the mean of these durations OR by taking the total behaviour duration and dividing by the number of occurrences

Question 8

Duration vs. Frequency

Answer 8

Most common types of measures, which can give different and complementary pictures
- HOW OFTEN 2 cows allogroom each other (frequency) tells us something different about the nature of their social relationship from HOW LONG they spend doing it (duration)
- Frequency and duration measure of the same behaviour are not always highly correlated, so it is probably wise to record both, when possible

Question 9

What is intensity?

Answer 9

• Has no universal definition
• Judgement about the intensity or amplitude of a behaviour pattern
  eg. Consequences of a behaviour can be measured in terms of some physical quantity related to the behaviour (wt of food eaten, volume of water drunk, number of prey captured or the distance traveled)

Question 10

What are the two fundamental types of behaviour patterns?

Answer 10

1. Events
2. States

Question 11

What are behaviour events?

Answer 11

Behaviour patterns or relatively SHORT duration, such as discrete body movements or vocalizations, which can be approximated at points in time
- Salient features of events in their frequency of occurrence (eg number of times a dog barks in one minute)

Question 12

What are behaviour states?

Answer 12

Behaviour patterns of relatively LONG duration, such as prolonged activities, body postures or proximity
- Salient feature of states is their duration (mean, total or proportion). Eg. total time a dog spends asleep over a 24 hr period
- In the literature “state” may also refer to the animal’s motivation state, such as thirst or hunger, so be careful not to confuse the two
- NOTE: the onset or termination of a behavioural state can itself be scored as an event and measured in terms of frequency

Question 13

What will the need to identify individuals within a group depend on?

Answer 13

Your experimental unit
- eg. the individual, pen, group etc.

Question 14

There are many ways to identify individuals, but what do you have to be mindful of?

Answer 14

Want to make sure that however you mark individuals for ID, it does not affect the behaviours you are trying to observe
- if you can’t say “no” definitively, then it should not be used as a marking (as you cannot rule it out as a confounding factor)

Question 15

How can groups be defined? What is the difference between a group and a party?

Answer 15

Groups can be defined as they are housed or as they associate with one another

Groups: association whose composition is known

Parties: aggregations whose membership is uncertain

Question 16

Once you’ve decided on your behaviour measures and how to record your observations what do you need to check the quality of before you collect data?

Answer 16

Check the quality of your measurements

Reliability = good measures
- Eg. Do all observers code the same behaviours as being the same?

Validity = right measures
- Eg. Are you measuring what you think you are measuring?

Question 17

Within observer vs between observer reliability

Answer 17

Within observer reliability (observer consistency)
- Describes the extent to which a singer observer obtains consistent results when measuring the same thing on different occasions

Between observer reliability
- Describes the extent to which two or more observers obtain similar results when measuring the same thing simulataneously
- Want at least 90% agreement (Kappa coefficient)

Question 18

What are 4 factors that affect reliability?

Answer 18

1. Practice and experience
2. Frequency of occurrence
   - If behaviours occur too rapidly to document, then video recording is often used to code the behaviour
   - Conversely, rare behaviour patterns may be missed altogether if observation sessions are not long enough
3. Observer fatigue
4. Adequacy of definitions
   - Definitions and criteria tend to “drift” with the passage of time, as observers become more familiar with the behaviour and unconsciously “improve” or “sharpen up” the definitions

Question 19

How independent are the measures?

Answer 19

Statistical tests generally assume that the data consists of a random sample from the population and that individual data points are statistically independent of one another
- The basic assumption of independence is sometimes violated in behaviour research

Question 20

What is the Pooling Fallacy?

Answer 20

Common error is to treat repeated measures of the same subject as though they were independent of one another
- Eg. measured 100 strides of each of 10 dairy cows
- Pooled: n = 1000 strides
- However, n = 10
- Thus, obtaining additional measurements from the same subjects is not a substitute for increasing the number of subjects in the sample
- In general, repeated measures from an individual should be averaged to give a single data point for that subject and the sample size (n) should be equal to the number of SUBJECTS, not the number of measurements

Question 21

What are “Group Effects”?

Answer 21

Eg. when studying litters or groups of animals
- Behaviour of animals may be similar amongst related individuals (due to genetics and previous experiences)
- Likewise, socially, interactions between members of the same group can lead to problems of independence, through short-term effects on their behaviour

Thus, if within-group variation is smaller (or larger) than between-group variation, measurements of group-mates cannot be considered to be statistically independent

Question 22

What are “non-independent categories”

Answer 22

Mean, duration, total number of occurrences, and total duration are not independent since:
- Mean duration = total duration/# of occurrences
- As such, only 2 measures can be considered independent, but not all 3

Non-independence can cause problems in interpreting associations between measures
- Eg. sleeping is negatively correlated with being awake

Question 23

What is pseudo-replication?

Answer 23

• When we think we have replicated something well but we didn’t and now can’t extrapolate the findings
• An otherwise well-planned experiment may still be limited in scope because only a restricted conclusion may be drawn from the results

Question 24

Ontogeny studies

Answer 24

Behavioural development can be studied by CROSS-SECTIONAL research, which involves measuring different individuals at each age, or by LONGITUDINAL research, which involves measuring the individuals repeatedly over time

Question 25

Choice Tests

Answer 25

Test of differential responsiveness or preference

Simultaneous or successive tests?
- Simultaneous presentation of test stimuli may be distracting to the subject
- Subject may be “trapped” by its first choice due to starting position or other factor
- May need to do a three-way choice which includes a “blank” choice