Research design

1
Research design operationalization of variables

• Chapter 6 in Babbie Mouton (2001)
• The construction of actual, concrete measurement
  techniques the creation of operations that
  will result in the desired measurements.
• The development or choice of specific research
  procedures (operations) that will result in
  representing the concepts of interest.

2
Operationalization

• An operational definition is a procedure for
  classifying, ordering, or quantifying something
• Classifying - crowded or not crowded
• Ordering - uncrowded, mildly crowded, severely
  crowded
• Quantifying - measure crowdedness in terms of the
  number of residents per square kilometre
• Focus on questionnaires other
  operationalization techniques in section on types
  of research design

3
Choices to be made about operationalization

• The range of variation how large should your
  categories be?
• Depends on the purpose of your study pragmatic
  considerations (e.g. income)
• Variation between the extremes how fine are the
  disctinctions you want to make in your study?
• e.g. age
• Again, depends on the purpose of your study
• (Why research is such a challenging task very
  few recipes)
• Single or multiple indicators of variables
• Some straightforward, such as gender
• But others benefit from multiple indicators

4
Levels of measurement

• Of great concern in psychometrics
• We look for three elements
• A zero point
• Distances between categories equal
• Categories can be rank-ordered

5
Nominal measures

• Reflects only categories
• The variable gender has two attributes, male and
  female
• They are distinct from one another, but they have
  no additional structures
• Also political party affiliation, birthplace, etc.

6
Example

• The next question is on the subject of work.
  People look for different things in a job. Which
  of the following five things would you most
  prefer in a job?
• 1. work that pays well
• 2. work that gives a feeling of accomplishment
• 3. work where there is not too much supervision
  and you make most decisions yourself
• 4. work that is pleasant and where the other
  people are nice to work with
• 5. work that is steady with little chance of
  being laid off

7
Ordinal measures

• Include the above
• Plus we can logically rank-order the attributes
• e.g. social class, prejudice
• We can rank order people according to the amount
  of education they had

8
Example

• 1. very satisfied
• 2. fairly satisfied
• 3. neither satisfied nor dissatisfied
• 4. fairly dissatisfied
• 5. very dissatisfied

9
Interval measures

• Include the above
• Plus the intervals between the attributes have
  meaning
• e.g scores on an intelligence test
• BUT
• We cannot say that a person with an IQ of 120 is
  TWICE as intelligent as one with an IQ of 60
• And the 5 points difference between 110 and 115,
  and 95 and 100?

10
Ratio measures

• Include all of the above
• Plus they have a true zero point
• e.g.
• how long you have lived in Cape Town?
• age
• the number of times you were married
• income

11
Implications

• Mainly for the analysis of data (the statistics
  part of the course)

12
Indexes and scales

• Social scientific concepts have complex and
  varied meanings
• Often difficult to capture all the dimensions of
  a concept
• As a result, we normally make multiple
  observations, and use more than one indicator of
  a variable
• Indexes and scales are techniques to combine
  indicators of a variable into a single measure of
  that variable
• They are composite measures of variables
  measurements based on more than one data item
• Typically they are ordinal measures of variables
• Typical in quantitative research
• Indexes and scales differ through the manner in
  which scores are assigned to individual attributes

13
Indexes

• Constructed through the simple accumulation of
  scores assigned to individual attributes
• e.g. we add the number of items a person agreed
  with to obtain a total score
• e.g. How politically active are you? We give you
  six statements of political activities you could
  have been involved in, and if you indicate that
  you have been involved in three of them, you will
  obtain a score of 3 on our index of political
  activism.

14
Scales

• A scale recognizes that not all responses reflect
  the same degree of the presence of a variable
• e.g. women are different from men is a much
  weaker statement of the variable sexism than
  women should not be allowed to vote
• Also, it assigns scores to patterns of responses

15
Scaling procedures

• Bogardus social distance scale
• Thurstone scales
• Likert scaling
• Semantic differential
• Guttman scaling
• The format of the Likert-type scaling is commonly
  used in questionnaire design today

16
Shopaholicism

• 1 I enjoy shopping
• SAgr Agr Neither Dis Sdis
• 2 I look forward to going shopping
• SAgr Agr Neither Dis Sdis
• 3 I shop whenever I have the opportunity
• SAgr Agr Neither Dis Sdis
• 4 I avoid shopping if I can
• SAgr Agr Neither Dis Sdis
• 5 When I visit a town or city I don't know well,
  I always want to see the shops
• SAgr Agr Neither Dis Sdis
• 6 Shopping is a chore that I have to put up with.
• SAgr Agr Neither Dis Sdis

17
Shopaholicism

• Scoring - note 4 and 6
• Level of measurement - ordinal or interval/ratio