Competing Explanations

1
Competing Explanations

• Thursday 19 October 2006

2
Todays Class

• Opening Remarks
• Review of Last Week
• Lecture
• Discuss Test

3
Review of Last Week

• Historical change in determinants of health
• Major transitions permanent settlements,
  industrial revolution, last half of 20th century
• Debates about role of medicine, economic growth
• Relative standing vs. absolute deprivation as
  health determinants
• How does inequality in a society influence health
• The Epidemiological Transition
• What is it and why is it important for
  understanding inequalities in health?

4
Today

• Why do I have you read journal articles?
• Two central explanations for understanding
  inequalities in health
• Individual health investment perspective (Smith)
• Social structural disadvantage (Williams)
• Must understand these explanations by examining
  the traditions of thought, methodologies, and
  disciplines they come from.
• Big picture distinguishing between these
  perspectives clarifies the difficulty proving
  health is determined by forces at many levels

5
Tale of Two U.S. States

• Health in Utah Nevada very different
• For Whom? Why?
• Income, schooling, health care, urbanization,
  climate?
• Lifestyle differences in adult years?
• Cigarette/alcohol consumption?
• Mormon discipline?
• Marital instability?
• Geographic instability?
• Selection of unhealthy migrants to Nevada
• Movie Leaving Las Vegas
• Other underlying causes? How do we understand
  individual decisions? Independent? How do we
  change decisions?

6
Individual Health Investment Perspective

• Behaviors and health investments are individually
  determined (budgets constraints of households).
• The path Health ? Income rather than Income ?
  Health explains most of the correlation between
  Income and Health
• Attribute unexplained correlation to selection
  rather than causation
• Health is a Stock

7
Equations
8
Social Structural Reinforcement Perspective

• Behaviors and Health Determinants are patterned
  responses to structural conditions that reflect
  relative standing
• The path Income ? Health rather than Health ?
  Income explains most of the correlation between
  Income and Health
• Attribute unexplained correlation to causation
  rather than selection
• Faults methodological individualism as a
  simplification of the determinants of health

9
(No Transcript)
10
Figure. The relationship between SES and Health
Infectious Diseases Malnutrition Overcrowding
Individuals HHs
Income Wealth
Access/Quality
Medical Care
Health
Space
Utilization Compliance
Psychosocial Risk Factors Behaviors
Work
Environments
Education Family Background Demographics, Genetic
Health Unobserved
Knowledge, Skills, Aspirations Psychosocial
Factors Behaviors
Material Well-being Access to Medical Care
11
Experimental Analogy

• The link between data (what we measure and how),
  how we evaluate what we measure, and what
  explanations we generate from that
• The physics model
• Social Scientists often try to infer experimental
  quantities (treatment effects) using
  non-experimental data.
• Involves thinking about counterfactuals

12
Real Randomized Experiment
Post treatment evaluation
Pre-treatment
Treatment/Placebo
Treatment Group
Complete Population
Randomized Assignment
Control Group
13
Experimental Analogy
HTT -HCC t (HCT -HCC) (tT -tC)
H Health
Subscript denotes membership C control group T
treatment group
Superscript denotes assignment C control
group T treatment group
14
Experimental Analogy

• View1 (Sociological)
• Selection on observed only there is random
  assignment to poor and non-poor after measured
  omitted variables are controlled (or matched).
• No Baseline differences in health predictors
  
• No Differential treatment effects
• Health resilience doesnt vary much over the
  lifecourse.
• Psychosocial factors are identified as
  intermediates between income and health in the
  Causal model

• View 2 (Rational Choice)
• Selection on observed and unobserved nonrandom
  assignment remains after measured omitted
  variables are controlled (or matched).
• Baseline differences in health predictors
• Differential treatment effect
• Health is a stock, and late and inconsistent
  inputs will not affect health capital much.
• Psychosocial factors identify both stable
  confounding characteristics and consequences of
  SES.

15
Incorporating Causation, Selection, and
Reinforcement into an Explanation
LF,0-18
.. LV35 .... LV,45
LV,50
LV,52
I47
I51
Y56
I53

..U18
U45
4 observation points
youth
In labor force
1992
1998
1960
1942
16
(No Transcript)
17
Review Relative deprivation

• How is it manifested after ET?
• Material ? Spatial (segregation of social
  environments)
• Cognitive processes of social comparison are
  involved (70)social meanings attached to
  conditions and how people feel about them.
  Social distinctions, culture, lifestyles, modes
  of living, consumer identities
• Relative income is a social concept. It cannot
  be dealt with at an individual level societies,
  not individuals have income distributions

18
Review Moving Target

• My research addresses historical changes in the
  prices of particular goods, some relevant to
  material well-being, others indicators of social
  deprivation.
• Pricing of space vs. a breadbasket of goods
• Increases in economic residential segregation
• Prices reflect a lot of things, but ultimately
  the value of a good, service, lifestyle, or way
  of living to consumers able to pay
• The sociopolitical economy of pricing of space.

19
(No Transcript)
20
Test Characteristics

• Multiple Choice 30-35
• Short Answer choose 5 of 7

21
Possible Test Questions

• The epidemiological transition saw the so-called
  diseases of affluence transformed into the
  diseases of the poor. Explain why diseases once
  associated with wealth might become more common
  among the poor and give one example.

22
Possible Test Question

• In lecture, Professor Berry described the
  difficulties of attributing causation using
  non-experimental data. This difficulty boils
  down to an essential difference between
  experimental and non-experimental data. What is
  the main thing distinguishing experimental from
  non-experimental data?
• Lack of a double-blind control
• Representative sampling
• Post treatment evaluation
• Random assignment
• The size of the standard errors