|
| Summary |
Extrapolation of mean annual migration rates leads to an underprediction of
population stability. This is because in a heterogeneous population, the individuals
who are most likely to move, and who contribute to the mean annual migration rate will
have moved away, leaving behind those who are less likely to move.
Cross-sectional analysis of this data set does not indicate any
systematic variation of the mean migration rate with time. Even for
data sets which showed evidence of temporal variation, there would be
no indication of whether this was due to age, cohort or inertial
effects.
Even though the standard Poisson model seems
misspecified, because all the explanatory variables are
exogenous
the parameter estimates are
consistent, ie. they tend to the true values when sample size is
increased. However, standard errors are underestimated and may lead us to
conclude that an explanatory variable is significant, when in fact it
is not. For instance, results for the standard Poisson model suggest
that educational qualifications do affect the likelihood of migration;
the Poisson mixture model
does not indicate significant educational qualification effects.
There is evidence that the likelihood of migration varies markedly
between individuals and that the sample contains a number of "stayers",
individuals likely never to move.
With a single count of outcomes for each individual, it is impossible to
distinguish between a heterogeneous population, with some individuals having
a consistently high and others a consistently low propensity to migrate, and
a truly contagious process, in which an individual's experience of migration per se
increases the probability of subsequent migration.
Next:Introduction to longitudinal data analysis |
Home page | Contents | Previous |