We report on the successful completion of a project to upgrade the positional accuracy of every response to the 1990, 2000, and 2010 U.S. decennial censuses. The resulting data set, called Optimized Spatial Census Information Linked Across Time (OSCILAT), resides within the restricted-access data warehouse of the Federal Statistical Research Data Center (FSRDC) system where it is available for use with approval from the U.S. Census Bureau. OSCILAT greatly improves the accuracy and completeness of spatial information for older censuses conducted prior to major quality improvements undertaken by the Bureau. Our work enables more precise spatial and longitudinal analysis of census data and supports exact tabulations of census responses for arbitrary spatial units, including tabulating responses from 1990, 2000, and 2010 within 2020 block boundaries for precise measures of change over time for small geographic areas.

The U.S. Census Bureau plans to use a new disclosure avoidance technique based on differential privacy to protect respondent confidentiality for the 2020 Decennial Census of Population and Housing. Their new technique injects noise based on a number of parameters into published statistics. While the noise injection does protect respondent confidentiality, it achieves the protection at the cost of less accurate data. To better understand the impact that differential privacy has on accuracy, we compare data from the complete-count 1940 Census with multiple differentially private versions of the same data set. We examine the absolute and relative accuracy of population counts in total and by race for multiple geographic levels, and we compare commonly used measures of residential segregation computed from these data sets. We find that accuracy varies by the global privacy-loss budget and the allocation of the privacy-loss budget to geographic levels (e.g., states, counties, enumeration district) and queries. For measures of segregation, we observe situations where the differentially private data indicate less segregation than the original data and situations where the differentially private data indicate more segregation than the original data. The sensitivity of accuracy to the overall global privacy-loss budget and its allocation highlight the fundamental importance of these policy decisions. Data producers like the U.S. Census Bureau must collaborate with users not only to determine the most useful set of parameters to receive allocations of the privacy-loss budget, but also to provide documentation and tools for users to gauge the reliability and validity of statistics from publicly released data products. If they do not, producers may create statistics that are unusable or misleading for the wide variety of use cases that rely on those statistics.

BACKGROUND Breakfast consumption is associated with better diet quality and healthier weights, yet many adolescents miss breakfast. Nationally, 17.1% of students participate in the School Breakfast Program (SBP). Only 10% of high school students participate. OBJECTIVE Our aim was to evaluate an environmental intervention to increase SBP participation in high schools. DESIGN A group randomized trial was carried out from 2012 to 2015. PARTICIPANTS/SETTING Ninth- and 10th-grade students enrolled in 16 rural schools in Minnesota (median 387 students) were randomized to intervention or control condition. INTERVENTION A school-based intervention that included two key components was implemented over a 12-month period. One component focused on increasing SBP participation by increasing student access to school breakfast through changes in school breakfast service practices (eg, serving breakfast from a grab-n-go cart in the atrium; expanding breakfast service times). The other component focused on promoting school breakfast through student-directed marketing campaigns. MAIN OUTCOME MEASURE Change in school-level participation in the SBP was assessed between baseline (among ninth and tenth graders) and follow-up (among tenth and eleventh graders). School meal and attendance records were used to assess change in school-level participation rates in the SBP. STATISTICAL ANALYSES The Wilcoxon test was used for analysis of difference in change in mean SBP participation rate by experimental group. RESULTS The median change in SBP participation rate between baseline and follow-up was 3% (interquartile range=13.5%) among the eight schools in the intervention group and 0.5% (interquartile range=0.7%) among the eight schools in the control group. This difference in change between groups was statistically significant (Wilcoxon test, P=0.03). The intervention effect increased throughout the intervention period, with change in mean SBP participation rate by the end of the school year reaching 10.3% (95% CI 3.0 to 17.6). However, among the intervention schools, the change in mean SBP participation rates was highly variable (range=–0.8% to 24.8%). CONCLUSIONS Interventions designed to improve access to the SBP by reducing environmental and social barriers have potential to increase participation among high school students.

To measure population changes in areas where census unit boundaries do not align across time, a common approach is to interpolate data from one census's units to another's. This article presents a broad assessment of areal interpolation models for estimating counts of 2000 characteristics in 2010 census units throughout the United States. We interpolate from 2000 census block data using 4 types of ancillary data to guide interpolation: 2010 block densities, imperviousness data, road buffers, and water body polygons. We test 8 binary dasymetric (BD) models and 8 target-density weighting (TDW) models, each using a unique combination of the 4 ancillary data types, and derive 2 hybrid models that blend the best-performing BD and TDW models. The most accurate model is a hybrid that generally gives high weight to TDW (allocating 2000 data in proportion to 2010 densities) but gives increasing weight to a BD model (allocating data uniformly within developed land near roads) in proportion to the estimated 2000–2010 rate of change within each block. Although for most 2010 census units, this hybrid model's estimates differ little from the simplest model's estimates, there are still many areas where the estimates differ considerably. Estimates from the final model, along with lower and upper bounds for each estimate, are publicly available for over 1000 population and housing characteristics at 10 geographic levels via the National Historical Geographic Information System (NHGIS – http://nhgis.org).

Areal interpolation transforms data for a variable of interest from a set of source zones to estimate the same variable's distribution over a set of target zones. One common practice has been to guide interpolation by using ancillary control zones that are related to the variable of interest's spatial distribution. This guidance typically involves using source zone data to estimate the density of the variable of interest within each control zone. This article introduces a novel approach to density estimation, the geographically weighted expectation-maximization (GWEM) algorithm, which combines features of two previously used techniques, the expectation-maximization (EM) algorithm and geographically weighted regression. The EM algorithm provides a framework for incorporating proper constraints on data distributions, and using geographical weighting allows estimated control-zone density ratios to vary spatially. We assess the accuracy of GWEM by applying it with land-use/land-cover ancillary data to population counts from a nationwide sample of 1980 United States census tract pairs. We find that GWEM generally is more accurate in this setting than several previously studied methods. Because target-density weighting (TDW)-using 1970 tract densities to guide interpolation-outperforms GWEM in many cases, we also consider two GWEM-TDW hybrid approaches, and find them to improve estimates substantially.

In this article, the authors describe a new data infrastructure project being developed at the Minnesota Population Center. The Integrated Spatio-Temporal Aggregate Data Series (ISTADS) will make it easier for researchers to use publicly available aggregate data for the United States over a time span that covers virtually the entire life of the nation: 17902012. In addition to facilitating access and ease of use, ISTADS will facilitate the use of thesevarious data sets in mapping and spatial analysis.

The Minnesota Population Center (MPC) provides aggregatedata and microdata that have been integrated and harmonized to maximize crosstemporal and cross-spatial comparability. All MPC data products are distributed free of charge through an interactive Web interface that enables users to limit the data and metadata being analyzed to samples and variables of interest to their research. In this article, the authors describe the integrated databases available from the MPC, report on recent additions and enhancements to these data sets, and summarize new online tools and resources that help users to analyze the data over time. They conclude with adescription of the MPCs newest and largest infrastructure project to date: a global population and environment data network.

The most straightforward approaches to temporal mapping cannot effectively illustrate all potentially significant aspects of spatio-temporal patterns across many regions and times. This paper introduces an alternative approach, bicomponent trend mapping, which employs a combination of principal component analysis and bivariate choropleth mapping to illustrate two distinct dimensions of long-term trend variations. The approach also employs a bicomponent trend matrix, a graphic that illustrates an array of typical trend types corresponding to different combinations of scores on two principal components. This matrix is useful not only as a legend for bicomponent trend maps but also as a general means of visualizing principal components. To demonstrate and assess the new approach, the paper focuses on the task of illustrating population trends from 1950 to 2000 in census tracts throughout major U.S. urban cores. In a single static display, bicomponent trend mapping is not able to depict as wide a variety of trend properties as some other multivariate mapping approaches, but it can make relationships among trend classes easier to interpret, and it offers some unique flexibility in classification that could be particularly useful in an interactive data exploration environment.

Subfamilies--family units residing in someone else's household--are an importantsubject of research, but they have proved difficult to measure. This research documents trends inand dynamics of the Census Bureau's identification of subfamilies by comparing them to highlyrefined and temporally consistent subfamily measures newly available in the Integrated PublicUse Microdata Series (IPUMS). I show that the Census Bureau's measurement of subfamiliesleads to highly unlikely interpretations of family interrelationships and that these apparent errorshave grown worse over time, affecting even the most recent American Community Survey data.Furthermore, errors are particularly high among young adults, nonwhites, and persons without ahigh school diploma--precisely those populations that subfamily researchers are most interestedin. Researchers may wish to consider avoiding the U.S. Census Bureau's subfamily measures infavor of the IPUMS subfamily measures.

With the completion of the U.S. National Historical Geographic Information System (NHGIS), it is now feasible to assemble a large dataset of historical census tract population statistics and boundary data in order to investigate patterns in long-term urban population trends. The present study makes use of this new resource to achieve a broad but concise overview of population trend patterns throughout major U.S. urban areas since 1950. This work thereby makes both methodological and substantive contributions to multiple fields of research, with much of the work dedicated to the development and assessment of new techniques to address two key methodological challenges. The first challenge is to construct a time series of census tract data, which requires linking data through time even where tract boundaries have changed. I present a few relatively simple areal interpolation techniques that can be used to address this problem. Two case studies indicate that a novel technique, cascading density weighting, should be effective both in the present setting and potentially elsewhere. The second methodological challenge is to identify an effective visualization strategy for investigating patterns in long-term trends. I present here a new conceptual framework that identifies a group of mapping techniques--trend summary maps--that should be most useful for visualizing patterns in trends. I provide an overview and assessment of several types of trend summary mapping techniques, and I introduce a novel technique, bicomponent trend mapping, which combines principal component analysis with bivariate choropleth mapping. This technique has several useful advantages not only for visualizing urban population trends but potentially in many other settings of spatio-temporal data visualization as well. Applying the new techniques to historical census tract data enables the central substantive contribution of this research: an overview of population trend variations throughout major U.S. urban cores. This overview supports the standard narrative of recent urban population dynamics--growth on the outskirts, decline in the cores, and some regrowth in centers--but it also reveals many regionally and locally unique patterns, indicating both divergence among cities and increasing heterogeneity within them.

This research investigates the effect of economic inequality on political participation. Differential rates of political participation are an important way in which citizens act politically. A large literature on democratization argues that inequality erodes the sense of commonality and mutual trust that underpins democratic self-governance, and there is some evidence that inequality lowers rates for political participation even in modern democratic societies. But these claims have recently been contradicted by other researches who claim that inequality fosters the sort of conflict and debate on which democracy thrives. I propose that these two opposing literatures can be reconciled by attending to the scale at which inequality exists. Inequality among close neighbors might lower participation by reducing interpersonal trust and the opportunities for political discussion, inequality among people in different neighborhoods might stead increase participation by heightening political conflict, and inequality among people in different counties might reduce participation rates by undermining the feeling of consensus and shared fate on which democratic governance rests. I attempt to unravel this complex dynamic with measures of economic segregation, created by decomposing total inequality into three portions: inequality across counties and within states; inequality across neighborhoods and within counties; and inequality across individuals within neighborhoods. I then link these measures to individual-level survey data from the American National Election Studies and the roper Social and Political Trends dataset. Contrary to my suppositions, inequality among close neighbors increase voter registration while decreasing participation in endeavors other than political campaigns. Furthermore, these effects are particularly strong among-low-income people, thus altering income differences in participation rates. Inequality across neighborhoods, in contrast, tends to exacerbate the tendency of rich citizens to vote more than poor citizens. And inequality across counties undermines participates in electoral campaigns, regardless of one's income. Although the mechanisms underlying these relationships are unclear from these analyses, it is evident that "inequality" has no univalent effect on political participation; rather, it depends on the scale at which inequality exists.

We typically conceptualize patterns in geographic time series in one of two ways: as trends in spatial patterns or as spatial patterns in trends. Chronological map serieseither animations or small multiples of chronologically-sequenced mapshave become a standard approach to mapping geographic time series. I propose that this approach is generally effective for visualizing trends in patterns but not for visualizing patterns in trends. On the other hand, trend summary maps (e.g., change maps, time-series glyph maps, trend cluster maps, and trend component maps) are generally effective for visualizing patterns in trends but not trends in patterns. Trend summary map techniques deserve more attention because they can help us identify and illustrate spatio-temporal patterns in ways that chronological map series cannot.

Conducting temporal analysis of census data often requires applying areal interpolation to integrate data that have been spatially aggregated using incompatible zoning systems. This article introduces a method of areal interpolation, target-density weighting (TDW), that is useful for long-term temporal analysis because it requires only readily available historical data and basic geographic information system operations. Then, through regression analysis of a large sample of U.S. census tract data, a model is produced that relates the error in TDW estimates of tract population to four basic properties of tracts. An analysis of model residuals combined with theorized absolute limits on interpolation error yields formulas with which we can compute upper and lower prediction bounds on the population in a tract of one census at the time of a different census. These prediction intervals enable the interpretation of different interpolated estimates with appropriately varying degrees of uncertainty.

This paper investigates as part of the National Historical Geographic Information System project (http://www.nhgis.org/) the creation of a multi-scale database of historical US census units. This database will include, at a minimum, three different scales (1:150,000, 1:400,000 and 1:1,000,000) and boundary data for all documented census since 1790. Besides the commitment to the production need, the main challenge in the generalization of these spatio-temporal data is the maintenance of geometric and topological consistency both within a dataset and between datasets for one target scale. We propose to address this challenge through: (1) a generalization framework based on the constraint based generalization paradigm and the active object approach; and (2) a topological data model linking all datasets, which represent different census years, for one target scale. The framework is implemented in ESRIs ArcGIS environment using ArcGIS 9.0, Oracle, C# and ArcObjects. The implementation of the model generalization process was completed and successfully tested for the three target scales of the final database. Model generalization accomplishes the removal of redundant points and the removal of boundary-change sliver polygons. The implementation of the cartographic generalization process is still on-going and has focused, until recently, on different approaches for the enlargement and elimination of too small census units or detached parts of a census unit as well as on the reduction of the outline granularity of census units boundaries. Results that were automatically generalized with the current version of the prototype exhibit satisfying quality based on a preliminary visual evaluation.