Week 3: From Ideas to Measures

1
Week 3 From Ideas to Measures
2
Conceptualization

• process whereby fuzzy or imprecise ideas are
  made specific and precise
• (Babbie p. 124)
• Includes creating a set of indicators
• Which can be grouped into dimensions
• eg

3
49-UPThe richer the child at birth, the higher
his/her quality of life will be

• By rich we mean
• Income of parents
• spiritual strength
• parents own their home/ car
• Good relationships with family
• Number of holidays/year

4
49-UPThe richer the child at birth, the higher
his/her quality of life will be

• By rich we mean
• Income of parents
• spiritual strength
• parents own their home/ car
• Good relationships with family
• Number of holidays/year

These are all Indicators
5
49-UPThe richer the child at birth, the higher
his/her quality of life will be

• By rich we mean
• Income of parents
• spiritual strength
• parents own their home/ car
• Good relationships with family
• Number of holidays/year

These are economic dimensions of the indicators
These are all Indicators
6
49-UPThe richer the child at birth, the higher
his/her quality of life will be

• By rich we mean
• Income of parents
• spiritual strength
• parents own their home/ car
• Good relationships with family
• Number of holidays/year

This is a social dimension of the indicators
These are all Indicators
7
49-UPThe richer the child at birth, the higher
his/her quality of life will be

• By rich we mean
• Income of parents
• spiritual strength
• parents own their home/ car
• Good relationships with family
• Number of holidays/year

This is a spiritual dimension of the indicators
8
With three keys dimensions economic, social and
mental quality of life

• By higher quality we mean
• Low level of stress
• Rewarding relationships with friends
• Strong mental stability
• Successful marriage
• Access to basic needs
• Early retirement.

9
Operationalization

• development of specific measuring techniques and
  decisions on how the data will be collected
  (what method to use)
• We operationalize our variables and
  (specifically) their attributes

10
The richer the child at birth, the higher his/her
quality of life will be

• By rich we mean
• Income of parents
• spiritual strength
• parents own their home/ car
• Good family relations
• Number of holidays/year

Ratio
(0-10,000, 10,001- 20, 000, 20,001-30,000)
(attends religious service 1/month, 2/month,
3/month)
Ordinal
(yes/no)
Nominal
(argue never, sometimes, often, always)
Ordinal
(0,1,2,3.)
Ratio
11
Measurement QualityBabbie p143

• A study is reliable if
• A study is valid if

12
Measurement QualityBabbie p143

• A study is reliable if
• When we repeat it many times we get the same
  results/ observations
• eg the bathroom scale

13

• A study is valid if
• The measure used accurately reflects the concept
• eg IQ as an accepted measure of intelligence

14
Validity and Reliability
Which bullseye represents which statement?
Discuss with your neighbor

• Neither Valid nor reliable
• Reliable, not valid
• Both Valid and reliable
• Valid, not reliable

15
Validity and Reliability
To be continued..!
16
Professor Sarah Elwood

• I am interested in understanding the social and
  political impacts of spatial technologies such as
  GIS, and the changing role and power of
  community-based planning and local activism in
  shaping urban geographies
• Social, urban, GIS Geographer
• Speaking in our class week 3!

17
Validity and Reliability
To be continued..!
18
How do I know my measure is reliable?

• (Babbie p.145-146)
• 1. Test-Retest method eg Health surveys, weighing
  yourself
• 2. Split-half method
• for complex concepts eg fear, prejudice, social
  anxiety
• Measure in different ways/ using different
  questions
• 3. Use established measures
• eg the IQ test for intelligence
• 4. Check on your researcher!
• Re-ask a sample of your questions to a sample of
  the respondents to see of they answer in the same
  way.
• Have the same set of results coded by different
  people (eg transcripts, newspaper articles etc)
• Discuss at length with researchers goals,
  methods, train them etc.
• ..eg Myers-Briggs Personality Test

19
How do I know my measure is valid?

• Babbie p. 146-147
• Face-validity Common sense agreement that the
  measure is a good one
• eg IQ is a good measure of intelligence, better
  than say number of times you rented books from
  the library last year
• eg Census definitions for family, race etc
  have a workable validity
• Criterion-related validity/ predictive validity
  Based on some external criterion
• eg the validity of a written drivers test is
  determined by the relationship between the scores
  received and the subsequent driving record
• Construct validity Variable in question is
  related to other similar variables
• eg Measuring marital satisfaction? See if your
  measure correlates with of infidelities. They
  should be correlated
• Note Both use comparison variables
• 4. Content validity How much does your measure
  cover the range of meanings included within a
  concept.
• eg quality of life are we measuring all
  variants of quality or just economic quality of
  life? What about spiritual, social, cultural etc
• eg 2. testing health inequities access to
  health care? Morbidity? Infant mortality?
  Psychological inequalities etc? How valid is our
  measurement in terms of covering all these
  aspects? It is fine if it only focuses on one or
  two but we need to acknowledge this as a
  limitation in our study.

20
3 key terms (Babbie p.190)

• Element
• An element is a unit about which information is
  collected, provides the basis for the study
• eg a young gamer or,
• a 315 student we interview about working in the
  fast food industry
• Population
• The aggregation of study elements, the group we
  are interested in generalizing about.
• eg young gamers,
• all the 315 students who work in the fast food
  industry

21
Sampling

• Today Non-probability sampling
• When to use it?
• 1. Available subjects
• 2. Purposive/ Judgmental sampling
• 3. Snowball sampling
• 4. Quota sampling
• .

22

• Available subjects
• Convenient and easy
• Can offer useful insights
• Not representative, we cant generalize
• Risky
• 2. Purposive/ Judgmental sampling
• Picking the most useful subjects/ objects
• (usually involves a pre-test or pre-survey to
  find them)
• Focused insight into your question
• Not representative, we cant generalize

23

• 3. Snowball sampling
• Do you know anyone else who would be a good
  interviewee?
• Useful when subjects are hard to find, suspicious
  of researchers or difficult to access
• Useful for exploratory work
• Possibility for bias
• Not representative, we cant generalize
• 4. Quota sampling
• Sample population has same proportion of
  characteristics as real population
• eg if 40 315 class are female, 40 of our sample
  of the class should be female
• Can provide ore varied and nuanced results
• Can offer better measure of representation
• Hard to get accurate initial data on your
  population
• Bias is still possible since researcher is not
  selecting randomly

24
Probability Sampling

• Humans are biased! (p. 188)
• Non-probability samples cannot guarantee the
  sample is representative!
• So we cant generalize using non-probability
  methods
• Random selection is basis for representation and
  ? generalizability

25
Random Selection

• Key to probability sampling
• Flip a coin
• Use of random number lists
• (Babbie appendix)
• Takes human decision-making out of the process
  and thus human bias

26
Sampling Frames

• Random sampling requires some Sampling Frame
• List of elements composing a population from
  which your sample is selected (p. 199)
• eg telephone book, class list, census block,
  school register, students in UW directory, list
  of members of an organization
• All elements must have an equal chance of being
  selected
• ? Note there may be omissions!

27
4 kinds of Probability Sampling

• Simple random sampling (p. 202)
• A number is assigned to each element in the
  sampling frame
• A table of random numbers is used to select
  elements
• eg with a study of 315 students
• Simple
• Generalizable
• Can be slow

28

• 2. Systematic sampling (p. 202)
• Every nth element is chosen
• Sampling ratio proportion of elements chosen
  eg if every 10th student is selected then the
  sampling ratio is 1/10
• eg with a study of 315 students
• Quick
• Generalizable
• Watch out for periodicy!

29

• 3. Stratified Sampling
• Elements in a population are grouped into
  homogenous units before sampling
• eg Babbie p. 207
• Modification of 1 and 2, not an alternative
• More representative than 1 and 2
• Useful if your population varies a lot
• Can be generalizable

30

• 4. Cluster sampling
• (Babbie p. 209)
• Multi-stage sampling
• Natural groups sampled first
• Elements within these sub-populations are then
  sampled
• eg. Almost impossible to sample inhabitants of a
  city based on a list/ sampling frame
• BUT we can
• 1. sample certain residential blocks in the city
• 2. list the houses on those blocks
• 3. take a sample of those houses
• 4. list the inhabitants of those houses
• 5. Take a sample of those inhabitants

31
Professor Kam Wing Chan

• migration labor market, urban finance China,
  Chinese cities
• Review his article
• Focus on his chose sample
• Extra info on Hukou migration available!
• Come with your questions!
• (Speaking in our class on Friday!)