Absolute vs. relative
Practical
1. Gender gap
Gender differences in mortality
• Gender differences in all-cause mortality
are not fully understood
• Differences over time and place
• Extensive literature; proposed explanations
include genetic, biological, behavioural,
social and psychological factors
• Gender gap seems particularly large in
eastern Europe
Gender gap in life expectancy at birth by overall (combined)
life expectancy in 1990 (r=-0.64, p<0.001).Male-femaledifferenceinLE(years)
Overall LE(maleandfemalecombined, years)
68 72 76 80
4
6
8
10
12
Austria
Belarus
Belgium
Bulgaria
Czech
Denmark
Estonia
Finland
France
GDR
W.German
Greece
Hungary
Ireland
Italy
Latvia
Lithuani
Moldova
NetherlaNorway
Poland
Portugal
Romania
Russia
Slovakia
Slovenia
Spain
Sweden
Switzerl
Ukraine
UK
Yugoslav
East West
Male-female difference in life
expectancy at birth (years)
8.8
6.5
0
1
2
3
4
5
6
7
8
9
10
East West
Male-female mortality difference
(absolute gap, per 100,000)
1005
530
0
200
400
600
800
1000
1200
East West
Male / female mortality ratio
(relative gap)
2.3
2.1
0
0.5
1
1.5
2
2.5
East West
Peto & Lopez method (Peto et al 1992)
• Absolute lung cancer rates used to indicate the proportions
of deaths from other diseases attributable to tobacco
• Exclude all deaths under 35, all external causes and
cirrhosis
• All the difference between lung cancer rate in each country
and in non-smoker in CPS taken as attributable to tobacco
• Other deaths due to tobacco:
– mixture of smokers and non-smokers with CPS rates to
obtain lung cancer rates equal to those in a country
– this provides a proportion of smokers in a country
– PARF using RRs from CPS, divided by 2
Male / female mortality ratio
(relative gap) after excluding
death attributable to tobacco
1.5
1.6
0
0.5
1
1.5
2
2.5
East West
Questions
• Are the differences in LE between western
and eastern Europe real?
• Absolute or relative?
• Which is more important?
• Did smoking make any contribution to
– LE?
– Differences in LE between east and west?
• The perception that gender gap in mortality
in EE is larger than in WE is driven by the
high mortality rates of both men and
women in EE
• this gave rise to the large LE difference
(analogous to absolute gender gap)
• relative gender gap similar in east and west
• Tobacco contributes to LE and differences
in LE
2. Relative deprivation
Relative deprivation in income and self-rated
health in the United States
Malavika Subramanyam 1, Ichiro Kawachi, Lisa
Berkman, S V Subramanian
Soc Sci Med 2009
Fully adjusted odds ratios (95% CI) of reporting fair-poor health associated
with differences in absolute household income, with and without adjustment
for relative deprivation in individual income.
The first column in each set is from a model that adjusted for age, gender,
education, race, marital status, health insurance status, year of survey and
state of residence.
The second and third columns in each set are from models that additionally
adjust for relative deprivation.
Relative deprivation in income
was derived using the Yitzhaki index
For a person i with income yi who is part of a reference group with N people,
Yitzhaki-index is given as: RDi=1/N∑j(yj−yi), ∀yj>yi, where the amount of relative
deprivation (RD) for individual i is the sum of differences in incomes between
individual i and the j individuals who have incomes higher than individual i.
The summation∑j(yj−yi) is divided by the number of people in the reference
group, N, making the measure invariant to the size of the reference group.
Individual income to the nearest cent was used to calculate RD which was further
divided by 5000 to get estimates in units of $5000.
We make the assumption that an individual compares him/herself to others in
his/her reference group based on their own income. We make this assumption
based on the belief that when it comes to social comparisons, individuals are
more likely to evaluate themselves in terms of their paychecks and other income
Questions
• Interpretation of the graph
• How can the findings be explained?