When to use observational methods
When you study:
Non-verbal populations like infants
Interaction(how behaviors unfold in time)
Drawback of observational methods
Time consuming
Unsystematic observations
Good for:
Hypothesis generation
Amount of info obtained
Minimize risk of missing data
Drawback:
Time consuming
Cannot study more than one person at a time
Observer must stay alert
Systematic observations
1.Decide what specific behaviors to study
2.operationalize behaviors. Decide between:
Physically based OR socially based behaviors
Molar(combined) OR molecular(parts) behaviors
Continuous OR intermittent recording
Coding onset and offset
3.name target behaviors
4.create a behavior catalogue/coding scheme
Automated measurements
Advanced in computer vision and AI make advanced motion tracking possible.
Currently, manual measurements can still be more complete than computer coding
Event coding
Manual measurements
Event triggered
Observe all the time, only record when target behavior occurs
Pros and cons Event Coding
Pros:
Not as time consuming
Useful when observing more than one person
Cons:
No time info unless onset and offset are noted
No information about pre-and-post-target behavior behaviors
Interval coding
Time triggered
Don’t observe or record all the time
A.whole/full/complete interval
B.partial interval
C. Momentary sampling
Pros and cons interval coding
Pros:
Objectivity
Good if observing for prolonged periods
Good when recording frequently occurring behaviors
Cons:
Not good if behaviors seldom occur
No info on the connection between behavior in observation intervals
Intervals need to be small enough to avoid losing information
situation/episode coding
“situation triggered”
Don’t observe or record all the time
+ easy to find the target behavior
Time sampling
“time triggered”
Don’t observe or record all the time
- instantaneous sampling
just record at a particular moment and
record what happens
+ useful when recording long-lasting behaviours
- doesn’t give information about behaviour sequences
naturalistic observation
Observation without influence
Context makes actions meaningful –> it is important to study natural environments
Nominal scale:
label variables without any quantitative value or order. AKA Categorical
Ordinal Scales:
represent variables with a natural order or rank, but the intervals between them are not equal or meaningful.
Interval Scales:
have a defined order, and the intervals between values are equal. However, there is no true zero point.
Ratio Scales:
possess a true zero point and equal intervals
Experimental Observations
Characteristic: test a hypothesis
Many factors are kept constant
Manipulate one factor to see what happens
Laboratories typically use this
Field Experiments
Combines naturalistic observation and
experimental control
Introduces a systematic change in a
natural environment
Quantitative studies
Categorisation and quantification
Estimate differences in distributions
Test if differences are statistically
significant
Replicability is important
Qualitative Studies
Categorisation without quantification
Reports interpretations of observations
Qualitative studies are often used to scout
the field and generate hypotheses for
quantitative studies
Observer Reliability
Observations should give the same
values for repeated observations of the
same events
Important with clearly defined categories
For qualitative method, all data should
contain exact, systematic and relevant
information.
Data must be available so that others
can make their own interpretation
How is observer reliability assessed?
Many observers evaluate the same
situation. Then calculate the inter-
observer-reliability
Observer drift
an individual observer’s coding is not consistent over time
Recommendations:
- Return periodically to the rating of training examples
- Check both intra- and inter-observer reliability
Expectancy effects
observers are not blind to hypothesis
