Steps 45: Analyzing Data and Drawing Conclusions

1
Steps 4-5 Analyzing Data and Drawing
Conclusions Special Focus on making causal
arguments
2

• Step 5 Analyzing the Data and Drawing
  Conclusions
• Ch. 2 says little about data analysisbut youll
  learn a lot about it through the MicroCase
  exercises
• Univariate Analysis Three Forms of Data
  Presentation
• GIS (Geographic Information Systems) data
  analysis and presentation. How many variables in
  maps on p. 63?

3
Drawing Conclusions

• Description. Gives us the conceptual tools to
  see social reality more fully. E.g. present a
  typology of responses to job loss, or of ways
  organizations respond to change, etc.
• Explanation. Most sociological theory attempts
  to identify causal interrelationships

IndependentVariable
DependentVariable
4
Criteria of Causation
1) Time Order independent variable must be an
antecedent variable 2) Correlation a necessary
but not sufficient condition 3)
Non-spuriousness ruling out alternative
explanations
5
Time Order is Not Always Obvious
Tolerance
Web Use
A finding from the 2000 General Social Survey But
which comes first?
6

• Measuring Correlations
• Measures of correlation measure the strength of
  the association of two or more variables
• Correlation coefficients vary from 1.0 to 1.0
• They must be examined for statistical
  significance
• Positive and negative correlations can both
  indicate causal relations
• But correlation does not always indicate
  causation

7
Nonspuriousness Is the correlation spurious?
Variable A
is not causally related to
is correlatedwith
A Third Variable
Variable B
8
What is spurious about a spurious correlation
is not the correlation. The correlation is real.
It is the causal interpretation which is
spurious.
9
Question Is this a causal relationship or a
spurious correlation?
Amount of Fire Damage
Fire Trucks
10
Some Spurious Correlations

• hand size and shoplifting

Large Hands
X
Gender
Shoplifting
11
Some Spurious Correlations

• marijuana use and health

Marijuana Use
X
Youth
Good health
12
Do music lessons make better students?
13
Is this correlation spurious?
Music Lessons
What variables might explain the relationship?
?
High SAT Scores
14
(No Transcript)
15
Is this correlation spurious?
Web Use
What variables might explain this relationship?
?
Tolerance
16
From Ferrante Text (p. 60)
Investing in employees human capital
Higher stock price (shareholder return)
Could this involve either a time order or
spurious correlation problem?
17

• Generalizability
• what is this issue about?
• Discuss the importance of sample and response
  rate