Class 6 Notes - Identifying Good Measurement Flashcards
in psych…
often interested in latent concepts, that cannot be directly observed [happiness, stress]
ways to measure variables
- self report measure
- observational measure
- physiological measure
self-report measure
- Self-efficacy: general belief that I can confident in a particular domain - subjective/personal interpretation
- Recording peoples answers about themselves
- Questionnaire/interview: subjective thoughts/feelings/desires
- Limitations
People might not want to be honest about a certain question (racial bias/prejudice — uncomfortable with saying true thoughts on that topic)… social desirability might influence our answer [measuring the way in which one wishes to be perceived]
How much fast-food do you eat a month?
Often are not consciously aware of many of our true feelings/beliefs
Unconscious
We often don’t know, and underestimate, the things that motivate us, why we like things, or why we ascribe to certain beliefs
Self-report on a process that they are unaware — cannot possibly be described
observational/behavioural measures
~ Pain: societal masculinity means pain tolerance will be inflated when a man is among a group of women
Self-report might not be the most accurate
Micro-facial expression: eyebrows down-inside
~ Recording observable behaviours; pressing keys on a keyboard; rate what emotion do u think this person is feeling
~ Can train observers to just look at participants and judge
~Important when ppl have desire to self-represent — gets around the limitations of self-report method
~Limitations
If you have to train observers, to recognized a certain behaviour: requires money, resources, time
Best practice to have multiple observers observing at the same time
Trained observers still disagree — further training required to establish a communal conception
physiological measures
~ Craving: measure of brain waves to quantitatively express how much craving is this person experiencing
Self-report difficult to determine: do they have perfect conscious interpretation of their cravings? — people have different scale
More accurately quantify — linked to standardized processes that we assume to be the same across different people
~Recording biological data
~How much of a molecule is in the blood stream; how much sweat; how much are their pupils dilating — that we know are associated with certain traits
~Limitations
Do not give any insight into subjective experiences
Expensive: fMRI (300-400$); magnetic tools
quantitative variables are …
coded with MEANINGFUL NUMBERS
categorical (nominal) variables have
discrete categories as levels
quantitative scales
- ordinal : ordered
- interval : ordered, equidistant, no meaningful zero
- ratio: ordered, equidistant, meaningful zero
examples of interval and ratio scales
Interval: temperature, GPA, IQ — no meaningful zero
Ratio: meaningful zero — reaction time, Kelvin scale
reliability of measurement
test retest
interrater
internal
Test-retest
~ Construct to be stable over time
* Measure today, and measure tomorrow: good measure should be the same * Ex. satisfaction with life
Interrater
~ Trained 2 observers to quantify some behaviour that they see in the outside world
* How much do each of these observers agree: did they say the same thing * If so, measures are reliable/consistent * If not, measure are not very reliable
Internal
~ Specifically when asking participants questions; each of those questions being a test of the concept
* Each of those questions seems to be assessing something similar to the other questions * Correlation between all of the different items; do they point to the same measure of the concept?
Evidence for reliability is a special example of an association claim
• These two things are linked
o Test with retest
o Quantity of 1st rater with quantity of 2nd rater
o Internal reliability: is item 1 associated with item 2
Scatterplots + correlations can be used to evaluate reliability
• Any deviants from the line indicative of disagreement between results