Tools: Interviews and observations

Question 1

Observation

Answer 1

Anything that isn’t experimental and doesn’t involve direct manipulation of variable67

Question 2

Subjectivity and objectivity

Answer 2

Subjecivity is important for psychology as psychological events are a subjective process - we cannot completely abandon
But good research also should be somewhat objective to be scientific
We can use coding etc. to make a subjective event (e.g. experience of their child) more objective - categories that different experiences fit into, for example
- but still need to be careful to fit into their experience - cannot use a too reductionist approach, e.g. forced-choice analysis when something doesn’t have one clear code can be reductionist and make us lose some insight/complexity

Question 3

Interviews

Answer 3

Can be subjective (unstructured or open-ended such as Foley who asked to describe their infant) or objective (highly structed - e.g. checklist such as in Tamis-LeMonda for language development of their child)

Question 4

Child interviews

Answer 4

Helpful as they probe children’s actual knowledge, perspective and emotions
Need clinical training to do
Not relying on secondary information from parents which could be biased
But children can be particularly susceptible to leading questions and can interpret questions differently

Question 5

Coding scheme disadvantages

Answer 5

Difficult to code more than one child at a time - can miss behaviour if not recorded
Need training to recognise specific codes
Very subjective - we cannot directly infer the child’s intentions from their externalised behaviour
Codes developed may not cover all behaviour observed - could miss out on richness of data or new discoveries
Some codes can overlap and be interpreted similarly - subjective

Question 6

Reliability - the kappa statistic

Answer 6

A form of inter-rater reliability
The extent to which different observes or raters code the same behaviour and codes overlap/correlate
Standard in all obs studies

Question 7

Event-based coding

Answer 7

Code when an event starts and stops, which leads to coding the next event
e.g. Tamis-LeMonda defined 4 types of child events that they would code when the event started/stopped, and coded for 6 types of maternal responses that could occur within 5 seconds of one of such child events
– a molecular, specific approach

Question 8

Validity of observational

Answer 8

Observational methods are highly ecologically valid as they are less constrained and naturalistic
– but presence of researchers can pose limitation to validity as it may change responses (e.g. child unfamiliar with
Validity varies greatly across different contexts

Question 9

Reliability in observational

Answer 9

More difficult with observational data due to subjectivity
Increase using the kappa statistic
Use well-defined coding schemes that have little overlap in definitions

Question 10

Causal inferences in observational

Answer 10

Seek to describe/explain behaviour and also predict future outcomes and behaviour
Causal inferences are limited due to lack of manipulation and control (longitudinal interventions that use observational data less so as they can do pre-testing and control groups)