Identifying good measurements Flashcards
concpetual definition
is the researchers definition of the construct at the theoratical level
operational defintion
linked to the decisions on measuring and manipulating these concpetual varaibles
what are the three common ways to measure
self reporting; operationalising a variable by recording peoples answers to question about themselves in a questionnaire or interview. For children this can be replaced by parent or teacher support
observational measurements: the researcher records observable behaviour or physical traces of behaviour. For example, happiness can be measured by obsering and counting the number of times someone smiles
physiological measurement: recording of biological data, such as brain activity, hormone level or heart rate, usually requiers the use of specific equipment.
the two variables
categorical variables: variables whose levels vary into categories, such as male/female. Also called nominal variables.
quantative variables: variables that can be measured in numbers with a meaningful distinction, such as height or IQ
three quantative variables
ordinal scale: the numbers of variables represent an order
interval: the numbers represent equal intervals between levels and there is no real zero
ratio scale: the numbers have equal intervals and the value of zero means ‘nothing’
reliabilty
how consistent the results of measurement are (whether the researcher can trust the results). An operationalisation that is reliable will always produce a consistent pattern of scores.
three types of reliabilities
test-retest: to what extent a measurement can be reproduced consistent scores with each use. most relevant for constructs that are relativelt stable.
interrater reliability: whether independent observers show consistent or similar scores in a measurement. Most relevant for observational measures.
Internal reliability: whether a participant will give a consistent pattern of answers, regardless of how the question is formulated. Also called internal consistancy, only applies to self-report scaled with multiple items.
how to visualise reliability
scatterplot
most common and efficient way to show reliability
correlation coefficient or r
correlation coefficiant or r
a single number that indicates how close the points on a scatterplot are to a line drawn through it
Direction of the slope (correlation coefficiant)
slowing upwards - positive correlation
not sloping - zero correlation
sloping downwards - negative correlation
strength of the relationship (correlation coefficiant)
+1.0 = strong positive correlation
0 = no correlation
-1.0= strong negative correlation
validity
indicates whether the operationalisation of variable measures what it is supposed to measure. to ensure that researchers are using the intended conceptual variables, it is important that researchers check for construct validity before operationalisition.
how is validity achieved in two ways?
reliability of measurements: consistency of results
measurement validity: whether the measurement is related to the construct
measuring validty for abstract constructs
construct validity is highly important in psychological research, especially when its not observable, like happiness is hard to measure.
