Census-based measures of fertility, mortality, and migration

1
Census-based measures of fertility, mortality,
and migration

• Hist 5011

2
Fertility Yasuba 1962

• Measure ratio of women aged 15-49 to children
  aged 0-4 for each state (or county)
• Correlate with characteristics of state (e.g.
  land availability, sex ratio, ethnicity)

3
Fertility Coale and Zelnik 1963

• Begin with single-year age distribution
• Adjust for mortality
• Adjust for census underenumeration
• Yield number of births in each year, back to
  1800
• Suggested very early fertility decline

4
Limitations

• No information on marital fertility
• No age-specific rates (cannot look at stopping
  vs. spacing)
• Cannot study differentials between population
  subgroups (e.g. different occupations)

5
Microdata allows own-child fertility analysis

• Retherford and Cho, 1978
• Calculate mean number of children of each age
  living with mothers of each age
• Adjust mean upwards to reflect mortality of
  children, underenumeration, and children residing
  without mothers
• Yields estimates of age-specific marital fertility

6
Simple own-child approach

• Even if we can make rough estimates of
  adjustments for whole populations, we cannot do
  so for population subgroups
• Therefore, adjustments make no sense when
  studying fertility differentials
• Simple own-child approach uses no adjustments
  just measure mean children under 5 with mothers
  of each age.

7
Children ever born

• Limitation we dont know when they were born
• Best for study of completed fertility
• Can calculate cohort parity distributions for
  older women
• Allows cohort-parity analysis (David and
  Sanderson 1987)

8
Mortality Two-Census methods

• Get two adjacent censuses
• Adjust population counts at each age for
  immigration, emigration, and changes in net
  underenumeration
• Subtract to estimate number of deaths
• Divide by midpoint of population to estimate
  age-specific death rates
• Rough estimates only, since effects of
  adjustments are large

9
Mortality Children-ever-born and children
surviving

• Calculate percent of children born surviving by
  age of mother
• Standardize or focus on a particular age group
• Can be used to study differentials
• With fancier techniques, can be used to estimate
  age-specific death rates for young people
  (Preston and Haines 1991)

10
Migration Net migration estimates from
aggregate data

• Eldridge and Thomas 1960
• Similar to two-census mortality estimation
• Get age distributions by state
• Adjust for mortality and differential
  underenumeration
• The remaining difference between time periods is
  net migration
• Result slow upward trend in migration since 19th
  century

11
Migration Using birthplace information

• Calculate percent of native-born persons residing
  out of their state of birth
• Works especially well for lifetime migration
  (Kelly Hall and Ruggles forthcoming)

12
(No Transcript)
13
Migration Record linkage

• Thernstrom 1963 and many others
• Ferrie 2004
• New 1880 linkage project

14
(No Transcript)
15
Migration Using children present

• Can be tricky
• Easiest measure, if you have enough cases just
  look at persons who have children of a particular
  age
• Large potential for selection bias

16
Migration residence 5 years ago

• Good since 1940 U.S., 1960s or 1970s in other
  countries
• Focus on recent migration
• Allows many methods