Inadequate parity standardization in Hispanic birth projections
The primary aim of this analysis is to examine the role of inadequate birth parity standardization as a potential source of error in US Hispanic birth projections. It is proposed that inadequate standardization for birth parity results when the effects of birth parity distributions on subsequent births are not included as part of the methodological assumptions of a projection.
Specifically, because individual fertility decisions are framed within a sequential decision- making model, they alter the perceived costs and benefits of marginal births, and by extension, the probability of these births occurring. At the aggregate level, these individual decisions are manifest in birth parity distributions, which introduce non-uniform probabilities for additional births among women within an analytical group. To the extent that the methods used to project subsequent births are unable to account for these variations, inadequate parity standardization will be a source of error.
Because methods that do not account for the effects of birth parity distributions, by default, assume no corresponding effect, this analysis first examines the birth parity distributions of US Hispanics and Non-Hispanics to determine if their birth parity distributions differ from the null assumption of a uniform distribution. Moreover, to the extent that the birth parity distributions of US Hispanics differ from non-Hispanics, the potential for inadequate birth parity standardization as a disproportionate source of error in US Hispanic birth projections is introduced. A method that accounts for variations in birth parity distributions is tested to assess improvements in projection accuracy, when compared to traditional methods.
Differences in birth parity distributions between the groups confirm inadequate birth parity standardization as a potential source of error in US Hispanic birth projections. Projection results comparable to official US projections were obtained, with the use of fewer resources and less data.