Providing a unified and comprehensive treatment of the theory and techniques of sub-national population estimation, this much-needed publication does more than collate disparate source material. It examines hitherto unexplored methodological links between differing types of estimation from both the demographic and sample-survey traditions and is a self-contained primer that combines academic rigor with a wealth of real-world examples that are useful models for demographers.
Between censuses, which are expensive, administratively complex, and thus infrequent, demographers and government officials must estimate population using either demographic modeling techniques or statistical surveys that sample a fraction of residents. These estimates play a central role in vital decisions that range from funding allocations and rate-setting to education, health and housing provision. They also provide important data to companies undertaking market research. However, mastering small-area and sub-national population estimation is complicated by scattered, incomplete and outdated academic sources—an issue this volume tackles head-on.
Rapidly increasing population mobility is making inter-census estimation ever more important to strategic planners. This book will make the theory and techniques involved more accessible to anyone with an interest in developing or using population estimates. Skip to main content Skip to table of contents.
Advertisement Hide. Subnational Population Estimates. Dispatch time is working days from our warehouse. Book will be sent in robust, secure packaging to ensure it reaches you securely. Book Description Sage Publications, Inc , Norfleet W. Rives; William J. Rives ; William J. This specific ISBN edition is currently not available. View all copies of this ISBN edition:.
Synopsis About this title Applied demography is a technique which can handle small geographic areas -- an approach which allows market segments and target populations to be studied in detail. About the Author : William J. Buy New Learn more about this copy. Customers who bought this item also bought. Stock Image. Rives, William J. New Paperback Quantity Available: 1. Seller Rating:. New Paperback Quantity Available: The same quantities for migration were set to 0.
- Analysis of Taste and Aroma (Molecular Methods of Plant Analysis).
- Population Review.
- The 60 Second Procrastinator: Sixty Solid Techniques to Jump-Start Any Project and Get Your Life in Gear.
For fertility, MAE ; similarly for population count. Level 4 hyperparameters and selected implied quantiles of the Mean Absolute Error MAE of the respective demographic parameters used in the simulation study and the application to Burkina Faso. The remaining initial estimates were drawn from distributions derived from Equations 9 — 12 in an analogous manner. The coverage of central marginal Bayesian confidence intervals or credible intervals under the model was estimated by the following experiment.
Check that Equation 14 is satisfied by the initial estimates; if not return to Step 1. Draw a large MCMC sample from the joint posterior and find the 0. The estimated coverage is then the proportion of the J Bayesian confidence intervals containing the known, true value for each parameter. Initial values for the population counts, vital rates, and migration proportions were set to the initial estimates. Start values for the variances were arbitrarily set to 5 as they appeared to have a negligible effect on the final results.
Rives, Norfleet W.
Point estimates of the coverage of the marginal 0. These are all close to 0. In practical applications with real datasets, where the true parameter values are unknown, interest will be in interval estimates of the demographic parameters. These should be based on the joint posterior distribution. For illustration, we have plotted central marginal Bayesian confidence intervals of a selection of age-specific and age-summarized parameters that might be of interest based on the MCMC sample from a single replicate of the simulation study Figure 2. For comparison, we have also plotted the true parameter values used throughout the simulation and the noisy initial estimates generated under the model.
Quality of demographic data
Ninety-five percent Bayesian confidence intervals for selected parameters from a single replication of the simulation study. The online version of this figure is in color.
- Shop by category.
- ESTIMATION METHODS, DEMOGRAPHIC.
- Applied Demography Toolbox!
- Race Matters: Indigenous Australians and Our Society?
Bayesian confidence intervals can be plotted for age-specific parameters as has been done for age-specific fertility rates in Figure 2 a. Confidence intervals for any function of the age-specific parameters can be obtained immediately by transforming each vector of age-specific values in the MCMC sample and computing the sample quantiles.
Quantities of particular interest are the summary measures defined in Section 3. We show TFR, e 0 , and the average annual total net number of migrants in Figures 2 b — d. We now illustrate the method by reconstructing the female population of Burkina Faso from to Uncertainty in this case is nonnegligible due to the fragmentary nature of the available data. This application shows how our method is able to quantify this appropriately by producing probabilistic interval estimates.
Brief descriptions of our initial estimates are given below. Further details are in the online supplementary materials, including the initial estimates themselves. The United Nations UN figures are preferred over the raw census counts because important adjustments were made for underenumeration. This form of bias is more common in certain age-groups and efforts to reduce it are based on postcensal surveys. Burkina Faso experienced a high level of population growth between and ; the total female population increased from 2.
The population has a young age structure, as illustrated by the age-specific population counts in Figure 3. Population counts by five-year age group for the female population of Burkina Faso. Alkema et al. Therefore, we took Alkema et al. Age patterns sum to one and indicate the share of fertility attributable to each age-group. We obtained a separate pattern for each projection interval in the reconstruction period by smoothing the available point estimates over age, within interval, using loess Cleveland ; Cleveland, Grosse, and Shyu and normalizing. The loess method performs a series of locally weighted regressions.
Smoothing within five-year sub-interval Figure 4 yielded trends that were also sensible, a priori, when viewed by five-year age group see supplementary materials. No point estimates were available for the period — To generate initial estimates for this period, we multiplied the — age-pattern by Alkema et al.
Data points for the initial estimates of age-specific fertility patterns of Burkina Faso women, —, grouped by five-year sub-interval. The lines are the within-time loess smooths. Standard methods were used to derive survival proportions from the life tables. Estimates of migration of comparable detail to those for vital rates are seldom available for many countries.
In addition, whole-population estimates for — are available from the literature by the United States Census Bureau and United Nations a that indicate sustained net out-migration over the period — The latter source suggests a reversal to net in-migration over — We designed our initial estimates to reflect the direction and approximate magnitude suggested by these sources. They are coherent because uncertainty about all other parameters is accounted for. Our simulation study suggests that the intervals are also well calibrated in that they achieve their nominal coverage over repeated random sampling of the initial estimates and census data, under the model.
The posterior medians for population counts at baseline Figure 5 are very close to their initial estimates; across age groups, the maximum absolute difference is 2. All of the intervals have half-widths less than 7.
The Springer Series on Demographic Methods and Population Analysis
Ninety-five percent Bayesian confidence intervals for the female population of Burkina Faso by age in Also shown are the initial estimates. A similar plot for age-specific fertility rates is shown in Figure 6. This is because most of the information about fertility in the nonfertility parameters comes from the population counts in the age range 0—5, and this depends mainly on the level of fertility, not how it is distributed across age group of mother.
Information about the age pattern of fertility in our posterior distribution comes mostly from the data-derived initial estimates. The interval widths narrow as age-specific fertility approaches zero, which occurs at the extremes of the age range of nonzero fertility. Due mainly to biology, human fertility is known to be low at the extremes of this range with a high degree of certainty. The shape of our posterior intervals reflects this. Ninety-five percent Bayesian confidence intervals for age-specific fertility rates for the female population of Burkina Faso, — Posterior median estimates of TFR Figure 7 a increased from 7.