Tools: Scale Construction Flashcards
Goals of Assessment
Identify risk factors or developmentally important characteristics/needs in a target population
Identify individuals on the basis of those risk factors/needs
What is a tool?
A measurement instrument - use to measure a construct
What are the features of an assessment?
A range of content for different developmental domains and needs
Standardisation - reliable
Concurrent validity - data should fit with other data with similar developmental groups/those with specific needs
Predictive validity - we want to be able to infer something about the future - establish the risk that an infant might have due their characteristics
Parent report
Very widely used approach to checklists and inventories
Do not require high staffing or training
Parents are with their children all the time - large quality and quantity of observation, in a huge range of natural contexts
BUT
parents could be unreliable - might not be consistent over time
add systematic error due to their own biases about their children - may want to
relying on only parent report can lead to shared method variance - results are partly due to the method itself
Murphy et al. - parent report
Used parent reports to rate childrens emotionality and regulation over time
Used to create trajectories of changes over time and the relationship between two variables by testing trends
Also used teacher report for same measures
– addressed shared method variance by having multiple data sources
– however teachers are slightly less reliable and see children in more limited/different contexts to each other
Cross-informant comparisons
Achenback et al.
Correlations across informants ranged from .20 to .60
Moderating effects of this variance:
- the context that teachers saw children in - e.g. different classes/subjects they struggle in could impact attention observed
- child age when assessed - stability of the child (e.g. more instable during early development)
- the observability of issues - externalising are easier to report on than internalising
Preceding scale construction
1) What are the existing measures?
- systematic reviews on existing measures related to your construct of interest or similar
2) Is a new measure really needed?
- only need when there is no other tool for our construct - strict criterion
Need to have construct validity - measure what we aim to
Do relations reflect another causal factor? e.g. due to confounds in both
Have you considered other limitations on your conclusion? other variables or factors
Reliability
The extent to which a measure is consistent and repeatable
- across time, do we get similar results?
- reliability is easier to demonstrate than validity
Validity
The extent to which a method measures what it intends to measure
7 bad practices when constructing a scale
1) presenting items in unidimensional blocks
2) presenting items in a fixed order
3) few or no opposite scored items
4) items with similar language
5) deleting items with diverse means
6) deleting items less related to the phenomena than others
7) ignoring response bias
Scale construction order
1) determine what you want to study
- after reviews
- can be one or multiple constructs
2) generate an item pool
- based on previous theory
- brief and readable
- should capture full variance of agreement/disagreement
3) determine the format for measurement
- needs to be appropriate for developmental group, e.g. asking questions to young infants who may not be able to fully understand
4) review item pool
- discussions with formal panel of parents, specialists in area of interest (e.g. intervention administrators)
5) inclusion of validation items
- e.g. checking if young infants are attending to what we want
6) administer items to pilot samples
7) evaluate items
8)produce final scale
Baby Care Questionnaire
Wanted to measure the beliefs that parents have that guide the care they give to their infants.
Justified new scale as:
- most scales were focused on very specific/narrow types of care, e.g. feeding or sleep, and produced by specialists in that particular field
- interested in cognitive and communicative outcomes in infants
- many of the at-risk infants were preterm and so longitudinal observational data was inappropriate - unable to observe infants in the unit
Baby Care Questionnaire
Wanted to measure the beliefs that parents have that guide the care they give to their infants.
Justified new scale as:
- most scales were focused on very specific/narrow types of care, e.g. feeding or sleep, and produced by specialists in that particular field
- interested in cognitive and communicative outcomes in infants
- many of the at-risk infants were preterm and so longitudinal observational data was inappropriate - unable to observe infants in the unit
