Chapter 4 Flashcards
operationalization
The process of systematically observing some feature or characteristic of the world and then recording it in the form of a number or category
Must be able to measure theoretical concepts of interest in order to test for suspected cause and effect
Without good measurement, inference is suspect (i.e. theory testing suffers)
performance measurement
Use of measurement for administrative purposes or leadership strategy
Focuses on measuring activities outputs, and outcomes of programs, initiatives or even entire organizations
construct (trait)
Concept or thing that we seek to measure
conceptual clarity
Define characteristics and boundaries of a concept or construct of interest
Know your unti of interest (individuals? Firms? agencies?)
Know your variation of interest (over time? Between units?)
Be precise!
conceptualization
Defining carefully and precisely what it is you seek to measure
where do conceptualizations come from
Legislations, regulations, policy debates
Insurance coverage/underinsured
Management initiatives
Customer satisfaction
Academic theory
PSM
manifest constructs
Things that are factual
More directly observable than others
EX. height and weight of child
latent constructs
Things that are not easily measurable
Factors that cannot be observed directly
EX. child’s knowledge of mathematics or language arts, political ideology, self esteem etc.
proxies
A proxy measure
EX. eligibility for free school lunch is a proxy measure of the family income of students
indicators
Observable measure of an abstract construct
The tradeoff - we get to measure something abstract, usually the cost of increase error
scales and indexes
measures composed of multiple indicators
EX. Grade point average
Index formed from grades in all classes
Measures overall academic performance
validity
Extent to which your instrument measures the construct of interest
Does what you are measuring map onto the theoretical construct you intended to measure?
Assessment: is your measure of a construct related to other measures (of variables of interest) as predicted by theory?
face validity
Based on looking at measure, how well does it get at what we want to measure
(is it valid or face?)
content validity
Includes all important dimensions of the construct
Depends on whether it captures the full range of variations of construct
(does an IQ test have items covering all areas of intelligence?)
construct validity
Seeing how well our measures correspond with variables that are logically or theoretically related to the underlying construct we purport to measure
(to what extent does this questionnaire actually measure intelligence versus other related variables?)
concurrent validity
How well the measure agrees with a current measure of the same concept
predictive validity
How well the measure predicts the logical consequences
EX. job satisfaction → quitting (one year)
limitations of validity
Validity is not all or nothing
Evidence can be less than completely convincing
There is often no magic test for validity
A measure can be valid for one purpose and not another
Often validity is a matter of subjective judgment
random measurement / error noise
Random and average out to 0
Unpredictable and uncontrollable errors
EX. bathroom scale arrow bent
Systematic measurement error / bias
Errors that are systematic and on average bend the measure in a particular direction
reliability
Extent to which re-application of a measurement method produces identical values for a variable
If you cannot generate the same values for independent or dependent variables successively, confidence in results is diminished
bias
Extent to which measure is consistently off of the mark (low or high)
Can still uncover associates between IV and DV
But must be skeptical of size of descriptive and estimated relationship
test-retest reliability
Redo and get same results
interrater reliability
How similar results are from same researcher when measuring same person or object
discrete and continuous numbers
Discrete is full numbers, continuous has decimals
The mathematical qualities of values assigned
nominal
Measures or variables in which the numbers refer to categories that have no inherent order to them
Can be arranged in any order
discrete
Cannot be ranked or operated on by any mathematical function
EX. type of car, blood type
ordinal
Measures or variables in which the numbers refer to categories that do have an inherent order to them
Numbers convey that order
Discrete
Categories can be ranked
The distance between those ranks is undefined
EX. military rank
interval ratio
Numbers the distance between values is the same across all values
Can be discrete or continuous
Constant distance between values
EX. temperature
Interval: arbitrary zero point (temp in c or f)
Ration: zero is meaningful (temp in k)
dummy variables
Only have two values, 0 and 1
Represent single unit of something
split-half reliability
Divide items randomly into two halves and look at correlation between two halves
cronbachs alpha
Averages out variation due to luck of the draw
qualitative variables
Numbers refer to actual quantities of something
categorical variables
Numbers stand for categories
unit of measurement
Prices meaning of numbers that appear in data set
simple random sampling
Selecting people or elements from a population in such a way that each individual has an equal chance or probability of selection
Assumed in most basic statistics formulas and statistical software
More complex forms exist - not in this class
sampling error
A statistic from a simple random sample is an unbiased estimate for parameter of the population
Proportion unemployed in sample is our guess at proportion unemployed in population
This is why we see things like +/- 3% in surveys
But a random sample results in the estimates that vary somewhat from the true mark, just by luck of the draw
parameters
Traits that can be quantified like averages, differences between groups, and relationships among variables
central limit theorem
All sampling distributions follow a normal distribution in the limit 9i.e. The larger they get the moral normal and narrow )
Big reason why quantitative research wants lots of observations
