Observation Flashcards
What is observation?
Watching people’s behaviour to investigate it
Could be used as a research method in of itself
Or a way to gather data in the context of an experiment
Research aim
The concept researchers want to investigate: specific to reflect what they want to find out about
Begins with ‘to see if…’
Research question
A question about the concept we are investigating, asking ‘do x do this when y?’
Question mark at the end
5 things to plan before an investigation
Who to observe
What acts should be observed
Where do we observe people
When do we observe
How do we carry it out
Unstructured observation
Non focused: recording every behaviour of who we are observing in the form of qualitative data
No coding frame is imposed to narrow down what data we collect
What type of data does unstructured observation give?
Qualitative data eg written descriptions of everything we see
Structured observation
The observations of behaviour are guided by behavioural categories (defined prior)
Using a coding frame
To focus on specific behavioural events + frequency each occurs
Coding frame
A tally chart and list of behaviours (known as behavioural categories) that we look out for and note down frequency of this
Behavioural categories
The DV: what behaviours we have operationalised in order to look out for to make a tally of
In the coding frame in a controlled observation
When developing a coding frame with behavioural categories: FOCUS
Focus on behaviours relevant to research aim/question or the hypothesis so unnecessary data is not collected
When designing coding frames for behavioural categories: operationalise
Fully Operationalise all categories to ensure we know what behaviours fall into the category and what don’t: reliable observation as coding frame is standardised and applied consistently by observers
When designing behavioural categories for a coding frame: objective
Objective so observers don’t have to make inferences on if the behaviour falls in the category or not: collecting data on outside observations not guessing someone’s thoughts/motivations/feelings that may be subjective
When designing behavioural categories for a coding frame: context independent
The behaviours should be consistent over different contexts and can apply to different situations
When designing behavioural categories for a coding frame: exhaustive
All possible behaviours from the research aim need to be covered + include an ‘other’ option to include unexpected behaviours
When designing behaviour categories for a coding frame: mutually exclusive
Try create the categories/codes so at any time recording one code doesn’t mean we have to record another code at the same time
Raw data tables in structured observation
Set up a tally chart with a heading for each behavioural code
What are advantages of structured observation?
-Narrows down behaviour we are recording so it’s more systematic and focused
-Collects quantitative data which is easier to compare frequency of different codes/to different situations
-Lessens observer bias as a coding frame was imposed with operationalised codes
-Increases inter-rater reliability because using a coding frame ensures consistency across recorded behaviours
What are disadvantages of structured observation?
-May neglect unexpected behaviours because they weren’t accounted for in coding frame
-Some codes may be up for interpretation
What are advantages of unstructured observation?
-Collects very detailed qualitative data which can inform later research
-No behaviours eg unexpected behaviours are missed
-Can be replicated anywhere no matter the context
What are disadvantages of unstructured observation?
-Might be too difficult to record everything so may miss behaviours
-Collects data irrelevant to aim so hard to condense all data
Naturalistic observation
Observing participants in their natural environment: little to no manipulation done to environment
Advantages of naturalistic observation
Obtains accurate data on how people behave in real life scenarios (high ecological validity)
More ethical as it does not involve manipulating anything
Disadvantages of naturalistic observation
Does not control for Extraneous variables present which may affect the participant outside of the independent variable thus decreases the validity
Environment may not be standardised for all participants so the behaviour codes collected aren’t consistent thus replicable across different scenarios
Controlled observation
When an experimenter sets up a controlled environment to observe participants in by manipulating some variables to ensure extraneous variables don’t affect how a participant acts
Can be done with or without an IV
Strengths of controlled observation
Increases validity as extraneous variables controlled for
May increase reliability as this is consistent
Weaknesses of controlled observation
Lacks ecological validity as the environment is artificial so may not be accurate reflection of behaviour irl
Less ethical as it may involve manipulation of environment
Participant observation
When the researcher is involved with the participants they are observing: they are in the environment
What are the advantages of participant observation?
Greater level of detail could be collected as the observer is close with the participants, may not have collected data had they been observing from a distance
Less likely to have view of observer obstructed if they are there
Only possible way to observe certain behaviours eg where CCTV/cameras aren’t there
What are disadvantages of participant observations?
Presence or observer may affect participants behaviour: even if they don’t know they are being observed esp if observer is personally involved w/ them
Observer could be disrupted by someone not knowing they are observing
Time consuming to set up + conceal observer collecting data
Could be dangerous (?)
Non participant observation
When the observer works in a separate location to the participants (Ps cannot interact with observer) such as watching CCTV footage or through a one way mirror
What are advantages of non participant observation?
No danger
Increased validity = observer has no effect on participants behaviour and Less risk of bias as observer Isn’t personally involved
Less time to disguise etc
What are disadvantages of non participant observation?
Observer can’t collect as detailed data if they are not present in same environment
At risk of view being obstructed eg if camera is blocked
Covert observation
Participants are unaware they are being observed: no consent given
What are Advantages of covert observation?
Naturalistic behaviour is measured that is not affected by knowing they are observed
Lessens demand characteristics + social desirability, participants won’t change behaviour to help researcher/ appear better
=increased validity of study, accurate behaviours recorded
Lessens effort to gain informed consent
What are disadvantages of covert observation?
Less ethical: participants did not give informed consent to being observed
Overt observation
Participants are aware they are being observed
Advantages of overt observations
More ethical as participants have given informed consent
No need to disguise observer
What are disadvantages of overt observation?
Social desirability bias/ observer effects/ demand characteristics = decreased validity
Effort to gain informed consent
Biased in terms of sampling those who consented so not representative of whole population
The 3 methods of how to take down observations
Continuous observation
Event sampling
Time sampling
When do we use continuous sampling?
In unstructured observation, taking a note down of every behaviour without using a coding frame to separate categories
Why do we use event/time sampling instead of continuous observation?
If there’s too much behaviours to record
And we are using a structured observation so there are behavioural categories requiring systematic sampling
Event sampling
The observer takes record of every time the behavioural code occurs in the category
Time sampling
When a note of a behaviour is recorded periodically according to pre-determined time intervals eg every 10 mins/30 secs/1 hour
Types of time sampling
One zero sampling
Instantaneous scan sampling
Predominant activity sampling
One zero sampling (time sampling)
During the time interval, a record is made if each behavioural category occurs at all in this time interval
Does not account for how many times it occurs in this interval just whether it does at all
Instantaneous scan sampling: time sampling
At the instant the time interval ENDS, a record is made of each behaviour that is occurring at this time on the dot
Predominant activity sampling: time sampling
Observation is consistent and estimated to see, during the entire time period, which behaviour occupied most of this time period
And a record of that is made
Advantages of event sampling
More accurate representation of behaviours shown in observation period because it takes note of every time behaviour was shown
So will not skip or miss out any infrequent behaviours
Disadvantages of event sampling
No record of when each behaviour occurred in time period: eg if it was random or in intervals
Difficult to record all that occurs
Continuous observation leads to loss of focus
Time consuming
Advantages of time sampling
You can see at what point behaviour occurred in the time period
Increased focus of observer: have to pay attention to clock
Allows time to record data properly
If you have a video, you can skip it until end of time period (instantaneous scan)
Disadvantages of time sampling?
Data is not representative of the entire time period: only occurs at certain time periods so does not include behaviours if they were infrequent, did not occur at the end of the interval etc
One zero sampling + predominant activity sampling means observer fatigue
In a structured observation, what type of data and level have we collected?
Quantitative data = frequency of each behavioural code
Nominal data = a headcount of the number of times each behaviour occurred, cannot be put on rating scale
Most appropriate graph to present data for event sampling
Pie chart
Collected nominal level data of frequency of each behavioural cue
Most appropriate graph to present data from a time sampling
Line graph: y axis with frequency of 1 behaviour and x axis with time intervals to show change in frequency of behaviours
