How does one determind the construct validity of a study?
By looking at the reliability and validity of a study
What kind of evidence supports the construct validity of a measured variable
Scatterplots and correlation coefficients can be used to determine reliability by seeing how similar two people measured observations and re-test reliability as well as using r to see whether two items are similar and therefore correlated.
How does one measure variables?
- Self-reports: where people report on their attitudes, beliefs or behaviours
- Observational measures: where raters record the visible behaviors of people or animals
- Physiological measures: where researchers measure biological data such as heart rate, brain activity and hormone levels
What is the difference between categorical and quantitative variables?
Variables are either operationalised to be categorical, where the cases are categories, or quantitative variables which have meaningful numbers
Quantitative variables can be further classified into..
- Ordinal scales
- Interval scales
- Ratio scales
How do researchers determine whether a measure is reliable?
Researchers collect data to see whether the measure works consistently.
What are the 3 types of measurement reliability?
- Test-retest reliability - whether a sample gives a consistent pattern of scores at more than one testing
- Interrater reliability - whether two observers give consistent ratings to a sample of targets
- Internal reliability - when people answer similarly worded items in a consistent way
*measurement reliability is necessary but not sufficient for establishing measurement validity
How can measurement validity be established?
It can be established with subjective judgements (face validity - if a measure is subjectively considered to be a plausible operationalisation of the conceptual variable in question and content validity - a measure must capture all parts of a defined conceptual definition) or with empirical data
What are the 2 kinds of measurement validities?
- Criterion validity - requires collecting data that show a measure is correlated with expected behavioral outcomes
- Convergent and discriminant validity - requires collecting data that show a measure is correlated more strongly with measures of similar constructs than with measurements of dissimilar constructs.
What is the major difference between ordinal and interval/ratio scales?
Ordinal does not have numerals that have a set distance between them, rank 1 could be 10s faster or 1 hour faster, but it will still be shown as rank 1 and 2 without any meaningful difference
What kind of evidence supports criterion validity?
Known-groups paradigm - where researchers see whether scores on the measure can discriminate among two or more groups whose behaviour is already confirmed
When do you try to determine discriminant validity?
When you are looking at a concept that is similar but still different to another concept, and need to be able to differentiate the two measures
