Measurement Chapter 6
Measurement
Chapter 6
Introduction
Measurement Classifying units of analysis by categories to
represent variable concepts Example
You may classify individuals into the categories, such as--satisfied with life or not satisfied with life--to represent the variable concepts of life of satisfaction
Introduction
Quantitative research Research focused on variables, including their
description and relationships
Introduction
Qualitative research Research focused on the interpretation of the
action of, or representation of meaning created by, individual cases
Introduction
Measure A specific way of sorting units of analysis into
categories
Introduction
The concepts in social science research can often be difficult to measure Concept examples
Feelings regarding fatherhood, satisfaction with life, suicide, and power
Measurement is a way to define a concept
Conceptualization
The process of clarifying what we mean by a concept
Conceptualization
Conceptual Definition A definition of a concept through other
concepts
Conceptualization
Concepts Words or signs that refer to phenomena that
share common characteristics
Conceptualization
Dimensions If a concept is so large that it may need to be
divided into different aspects or dimensions to further clarify it
Dimensions are aspects or parts of a larger concept
Conceptualization
Multidimensionality A concept that refers to a phenomenon that is
expected to have different ways of showing up or manifesting itself is said to be multidimensional
Multidimensionality is the degree to which a concept has more than one discernible aspect
Operationalization
The process of specifying what particular indicator(s) one will use for a variable
Operationalization
Indicators Observations that we think reflect the
presence or absence of the phenomenon to which a concept refers
Operationalization
Example – Age Conceptualize age – as chronological age Still need to determine an operational
definition
1. Are you old?
2. How old are you? Circle the letter of your age categorya. 25 or younger b. 26 to 64 c. 65 or older
3. How old are you?
Operationalization
Operationalization involves providing operational definitions Declarations of how a specific phenomena
described by concepts are determined for a specific instance.
Operational Definitions
Example – age Several ways to conceptualize age
Chronological age (time since birth) Functional age (the way people look or the things
they can do) Life stages (adolescence, young adulthood, and
so on)
Operationalization
An Example of Conceptualization and Operationalization Quality of Life Measure Study interested in quality of life and how
stress and coping resources affect life satisfaction, with a particular interest in comparing younger, middle aged, and older adults
Operationalization
Index A composite measure that is constructed by
adding scores from several indicators
Operationalization
Operationalization
Composite Measures A measure with more than one indicator Designed to solve problems of ambiguity that
are associated with single indicators by including indicators of a variable in one measure
Operationalization
Composite measure Scale
Indexes in which some items are given more weight than others in the determination of the final measure of a concept
Quiz – Question 1
Professor Johnson is interested in examining the connection between age at first sexual experience and high school completion. Which of these concepts must be operationalized?
a. First sexual experience
b. High school completion
c. College enrollment
d. A and B
e. None of the above
Focal Research
“Impostor Tendencies and Academic Dishonesty” How is the impostor phenomenon
conceptualized by Ferrari and by other social psychologists?
Are you surprised by this way of thinking about “impostors”?
Focal Research
Thinking about ethics Students were given informed consent forms
to sign before they were asked to fill out Ferrari’s questionnaire.
By not putting their names on the questionnaires, the anonymity of respondents is guaranteed.
Focal Research
Measurement error The kind of error that occurs when the
measurement we obtain is not an accurate portrayal of what we tried to measure
Measurement in Visual Analysis and Qualitative Research Visual analysis
A set of techniques used to analyze images] refers to a set of techniques used to analyze images
Measurement in Visual Analysis and Qualitative Research Coding
Assigning observations to categories
Exhaustive and Mutually Exclusive Categories Exhaustive
The capacity of a variable’s categories to permit the classification of every unit of analysis
Exhaustive and Mutually Exclusive Categories Mutually exclusive
The capacity of a variable’s categories to permit the classification of each unit of analysis into one and only one category
Quiz – Question 2
What is the problem with the following survey response set for the question “how old are you”?
25 or less 25 to 65 65 and older
a.The concept is not operationalizedb.The response set is not mutually exclusivec.The constructs are multi-dimensionald.The conceptual definitione.None of the above
Quality of Measurement
Reliability & Validity Reliability - the degree to which a measure
yields consistent results A measure is reliable if it yields consistent results
time after time Validity – the degree to which a measure taps
what we think it’s measuring A measure is valid if you measure what you think
your are measuring
Quality of Measurement
Checking Reliability A yard stick will consistently measure 30
inches time after time, so it is reliable—but it is possible to have the first inch sawed off, so we need to check reliability.
Quality of Measurement
Checking Reliability Test-retest method Split-half method Interobserver Reliability Interrater Reliability method
Quality of Measurement
Checking Reliability Test-retest
A method of checking reliability of a test that involves comparing its results at one time with results, using the same subject at another time
Quality of Measurement
Checking Reliability Test-retest
Measuring the width of a desk two or more times in close succession with the same yard stick
Problems: Phenomenon under investigation might actually
change between the test and the retest – change in life satisfaction
Earlier test results may influence you the second time you test
Quality of Measurement
Checking Reliability Split-half method
A method of checking the reliability of several measures by dividing the measure into two sets of measures and determining whether the two sets are associated with each other.
Deals with the problems of the test-retest method Making more than one measurement of a
phenomenon at the same time
Quality of Measurement
Checking Reliability Interobserver reliability method – compares
the results obtained by one observer with results obtained by another using exactly the same method.
Assessing the traits of individual characters in children's’ books using a pre-defined measurement scheme, they agreed 88 percent of the time on average
Quiz – Question 3
If we administer a survey to measure school-related stress and re-administer it one month from today to the same group and obtain different results, what does this indicate to us about our measurement?
a. It may not be a reliable measureb. Stress may have changed in the groupc. We may need to re-administer the survey again to determine
more conclusively if the measure is reliabled. All of the abovee. None of the above
Quality of Measurement
Checking Validity Does this measurement strategy feel as if it’s
getting what it is supposed to?
Face validity Content validity Predictive validity Construct validity
Quality of Measurement
Checking Validity Face Validity
The degree to which a measure seems to be measuring what it’s supposed to be measuring
Quality of Measurement
Checking Validity Content validity
How well a measure covers the range of meanings associate with a concept
If most people know something about a concept, feel that a given set of questions “gets at” that concept and the set of questions asked about a question cover the usual range of “content” implied by the concept
Quality of Measurement
Checking Validity Predictive validity
How well a measure is associated with future behaviors you would expect it to be associated with
Involves comparing the results of your measurement scheme with a criteria variable that it should predict
Example Compare the SAT scores of high school seniors with their
performance in college and found that students with higher test scores did better in college, then you could say the SAT test has predictive validity
Quality of Measurement
Checking Validity Difficult to validate social science measures
using behavioral criteria, b/c limited in finding behaviors that they obviously predict to
Many of the social science indicators are meant to measure abstract concepts
Social scientists tend to use construct validity
Quality of Measurement
Construct Validity How well a measure of a concept is
associated with a measure of another concept that some theory says the first concept should be associated with
Level of Measurement
Nominal level variables Describes a variable whose categories have
names Nominal level variables permit us to sort our
data into categories that are mutually exclusive
Level of Measurement
Nominal Level Variables Said to be the lowest level of measurement,
b/c it is virtually impossible to create a variable that is not nominally scaled
Examples Sex (Male, Female) Religion (Christian, Muslin, Jewish, Other) Cause of Death (Suicide, Homicide, Accident)
Level of Measurement
Ordinal Level Variables Describes a variable whose categories have names
and can be rank-ordered in some way Example
Social Class (upper, middle, lower) Ordinal level variables guarantee that categories are
mutually exclusive and that the categories can be ranked
All ordinal variables can be treated as nominal, but not all nominal variables may be treated as ordinal
Sex vs. Age
Level of Measurement
Interval Level Variables Describes a variable whose categories have names,
can be ranked-ordered, and whose adjacent categories are a standard distance from another
All interval variables can be treated as nominal or ordinal variables, but ordinal and nominal variables cannot be treated as interval
Level of Measurement
Interval Level Variables Categories of interval variables can be
meaningfully added and subtracted Fahrenheit temperature scale SAT score
The difference between 550 and 600 on the math aptitude test can be seen as the same as the difference between 500 and 550
Level of Measurement
Ratio level variables Describes a variable whose categories have names,
can be rank-ordered, has a standard distance from another AND HAS AN ABSOLUTE ZERO
Absolute zero is the point at which there is a complete absence of the phenomenon in question
Examples Weight Income Length
Ratio level variables can also be interval, ordinal, and nominal level variables
Quiz – Question 4
Letter grades given at the end of the term are an example of _____________ measurements; calculated GPAs are a type of _________________ measurements.
a. ordinal, interval
b. ordinal, ratio
c. interval, ratio
d. Nominal, ratio
e. None of the above
Quiz – Question 5
A researcher designs a survey question asking respondents to indicate their class standing as either: First-Year, Sophomore, Junior, Senior, or Graduate School.
a. nominal
b. ordinal
c. interval
d. ratio
Level of Measurement
The practical significance of level of measurement Researchers find themselves wanting to summarize
the information that has been collected Certain statistics require different levels of
measurement Three ways to describe central tendency
Mean – interval, ratio Median – ordinal Mode - nominal
Level of Measurement
Given a choice, measure a variable in a higher level of measurement than a lower one
The higher the level of measurement, the more statistics that can be performed on the data
Summary
The importance of measurement