CONCEPTUALIZATION, OPERATIONALIZATION, AND MEASUREMENT

1
CHAPTER 5

• CONCEPTUALIZATION, OPERATIONALIZATION, AND
  MEASUREMENT

2
Conceptualization - the process of coming to an
agreement about how to use a social scientific
term (e.g., alienation, poverty, social class,
romantic love, democracy).Conceptualization
results in definitions of concepts mental
images of what we mean by a term.
3
Can concepts be defined objectively?

• If not, how can we conduct research using
  concepts?
• Intersubjectivity agreements among subjective
  beings (humans) about the definitions of terms
  used in research.

4
Types of definitions.  Which would be useful to a
researcher?

• Real definition
• Nominal definition
• Operational definition

5

1. Real definition a statement of the essential
   nature or essential attributes of a term
2. Nominal definition a definition that is
   assigned to a term for purposes of communication
   or inquiry a working definition

6

1. Operational definition a specification of how a
   defined term will be measured for purposes of
   research development of specific research
   procedures that will result in empirical
   observations representing the concept in the real
   world

7
Linking conceptualization and data collection -
C-M Diagram

• Conceptualization

Nominal Definition
Operational Definition
Measurements in the Real World
8
Conceptualization practice

• The concept might be defined in an ordinary
  dictionary and might also have been used
  previously in social scientific theory and/or
  research.
• Web of Online Dictionaries
• Sociology Dictionary, Political Science
  Dictionary
• Sociological Abstracts

9

• Consider possible indicators of the concept -
  What would indicate the presence of the concept
  in empirical reality, and what would indicate its
  absence?
• What indicators could you use to determine
  whether people are experiencing the emotion
  called "romantic love."

10

• Consider possible dimensions of the concept -
  different facets or aspects of the concept. 
• Concepts can be unidimensional or
  multidimensional.

11
They make it look easy, but its not!

• Definitions of concepts in the social scientific
  literature
• Romantic love (according to Rubin)
• Workplace alienation (according to Blauner)

12
Operationalization choices

1. Level of measurement
2. Range of variation of attributes (nature of
   extremes)
3. Precision of variations between extremes
4. Use of single or multiple indicators

13
Level of measurement
  Characteristics of Attributes Characteristics of Attributes Characteristics of Attributes Characteristics of Attributes Characteristics of Attributes
Level of measurement Mutually Exclusive Exhaustive Rank-ordered Standard Interval Between Attributes True Zero Point
Nominal X X      
Ordinal X X X    
Interval X X X X  
Ratio X X X X X

14
GSS variables and LOM

• MARITAL
• AGE
• EDUC
• DEGREE
• ZODIAC
• RELIG
• TRAUMA1

• CONLEGIS
• MILQUAL
• PRES96
• WRKSTAT
• INCOME98
• SIBORDER

15
Implications of levels of measurement for data
analysis

• Restrictions on quantitative data analysis
  techniques. The higher the level, the greater
  the number of choices.
• Higher level of measurement may be converted to
  lower level (recoded), but not vice versa

16
Range of variation of attributes (nature of
extremes)

• To what extent are we measuring the extremes of a
  phenomenon?
• GSS examples - POLVIEWS, CONLEGIS, COMMUN

17
Precision of variations between extremes

• How precisely defined are the attributes of the
  variable between the extremes?
• Examples from 1996 General Social Survey 1996
  GSS data

18
Use of Single or Multiple Indicators

• GSS - Attitudes toward abortion
• One item ABLEGAL
• Multiple items - ABNOMORE, ABDEFECT, ABRAPE,
  ABPOOR, ABHLTH, ABANY, ABSINGLE

19
Index construction

• Combining responses to two or more items into one
  ordinal item (index)
• GSS Constructing an index (ABINDX) combining the
  responses to ABNOMORE, ABDEFECT, ABRAPE, ABPOOR,
  ABHLTH, ABANY, and ABSINGLE
• Please tell me whether or not you think it
  should be possible for a pregnant woman to obtain
  a legal abortion if (1Yes, 2No)

20
(No Transcript)
21
Whats a good measurement like? Criteria for
measurement quality

• 1. Precision
• 2. Accuracy
• 3. Reliability
• 4. Validity

22
1. Precision

• Definition - the fineness of distinctions made
  between attributes of a variable
• Which is more precise?
• Political viewsliberal, moderate, conservative
  or POLVIEWS?
• RINCOM98 or Gross income reported to the IRS last
  year?
• Necessity for precision depends on purpose of the
  research

23
2. Accuracy

• Definition - the extent to which the measure is a
  reflection of the real world degree of
  conformity of a measure to a standard or a true
  value.
• Survey example - Which one is likely to be more
  accurate?
• What is your age?
• In what year were you born?

24
3. Reliability

• Definition - whether a measurement technique,
  applied repeatedly to the same phenomenon would
  yield the same result
• Types of reliability
• a. Test-retest
• b. Split half
• c. Use of established measures
• d. Reliability of research workers

25
3. Reliability - a. Test-retest

• Definition - testing for reliability using two
  tests, separated by a time interval, in a
  situation in which the researcher does not expect
  the variable of interest to change in value.
• Examples knowledge of research methods,
  attitudes toward abortion (ABDEFECT), level of
  formal education (EDUC)

26
3. Reliability - b. Split half

• Definition - testing for reliability of a
  multiple-item measure by splitting it randomly
  into two halves and seeing if the two halves
  correspond in value.
• Examples 150-item test of knowledge of research
  methods, 50-item romance attitudes scale

27
3. Reliability - c. Use of established measures

• Many measures of social scientific concepts have
  already been tested and shown to have adequate
  reliability by previous researchers.
• Examples
• Rubin's definition and operationalization of
  romantic love
• Sroles definition and operationalization of
  anomia

28
3. Reliability - d. Reliability of research
workers

• Research workers may make mistakes which produce
  unreliability (e.g., interviewers reading
  questions incorrectly coders entering
  information incorrectly into data files).
• Careful training and supervision of research
  workers is essential to maximizing reliability.
• If the research workers are the data collectors,
  this is also called Inter-rater or
  Inter-observer reliability

29
4. Validity

• Definition - whether a measurement technique
  adequately reflects the nominal definition of the
  concept

30
Validity on the C-M Diagram

• Conceptualization

Nominal Definition
Does the measurement technique reflect the
nominal definition?
Operational Definition
Measurements in the Real World
31
Different Approaches to Validity

• 1. Face Validity
• 2. Content Validity
• 3. Criterion-related Validity
• 4. Construct Validity

32
1. Face Validity - Definition

• Definition - does the measurement technique fit
  common agreements and mental images concerning a
  particular concept does it seem reasonable?

33
1. Face Validity - Example

• Nominal Definition Religiosity - extent to which
  a respondent adheres to a set of cultural ideas,
  symbols, and practices that focus on the meaning
  of life and the nature of the unknown.
• Examples
• What is your shoe size?
• How well is the President performing?

34
2. Content Validity

• Definition - the degree to which a measure
  covers the range of meanings included within a
  concept
• Examples
• Does an exam assess all aspects of course
  content?
• Do questions asking respondents opinion of
  abortion in case of rape and in case the mothers
  health is endangered assess all aspects of
  abortion issues?

35
3. Criterion-related Validity

• Definition - the degree to which a measure
  relates with some external criterion measuring
  the same concept
• Examples
• Does SAT score predict performance in college?
  Some research data
• Does an index of marital adjustment predict which
  couples later divorce?

36
4. Construct Validity

• Definition - the degree to which a measure is
  related to other variables within a system of
  theoretical relationships as the researcher
  hypothesizes it should be

37
4. Construct Validity - Example

• Hypotheses
• Women should support abortion more frequently
  than men do.
• Catholics should oppose abortion more frequently
  than those of other or no religious affiliation.
• Research results
• Hypothesis is supported -gt Construct validity of
  measurement technique is supported.
• Hypothesis is not supported -gt Construct validity
  of measurement technique is not supported or
  research is poorly designed and executed.

38
Whats wrong with this picture? (Figure 5-2, p.
149)
39
An interesting comparison from two methods
textbooks
Earl Babbie, The Basics of Social Research, 3rd
edition, Wadsworth, Belmont, CA, 2005, p. 149.
Royce A. Singleton, Jr., and Bruce Straits,
Approaches to Social Research, Oxford U. Press,
N.Y., 1999, p. 117.
40
A Caution!

• Validity has to do with assessing whether the
  measures of our concepts reflect the nominal
  definition we have assigned, NOT the "real"
  definition!

41
Go forth and measure and operationalize in peace!