Given a transaction database as a global set of transactions and its local database obtained by some conditioning of the global database, we consider pairs of itemsets whose degrees of correlation are higher in the local database than in the global one. A problem of finding paired itemsets with high correlation in one database is already known as discovery of correlation, and has been studied as the highly correlated itemsets are characteristic in the database. However, even noncharacteristic paired itemsets are also meaningful provided the degree of correlation increases significantly in the local database compared with the global one. They can be implicit and hidden evidences showing that something particular to the local database occurs, even though they were not previously realized to be characteristic. From this viewpoint, we have proposed measurement of the significance of paired itemsets by the difference of two correlations before and after the conditioning of the global database, and have defined a notion of DC pairs, whose degrees of difference of correlation are high. In this paper, we develop an algorithm for mining DC pairs and apply it to a transaction database with time stamp data. The problem of finding DC pairs for large databases is computationally hard in general, as the algorithm has to check even noncharacteristic paired itemsets. However, we show that our algorithm equipped with some pruning rules works successfully to find DC pairs that may be significant.

Different perspectives have been offered to explain ethnic labor market concentrations. In most studies, however, residential places are seldom included in the research framework. Using data from 5% Public Used Microdata Samples in 2000, this case study of the San Francisco Bay Area reveals that the robust growth of the new economy is dramatically segmenting the geography of employment and thereby the spatial division of labor in each ethnic group. Living arrangements, such as central-city residence and living in coethnic-concentrated-PUMAs, increase the chances of niche employment for most racial/ethnic groups, even after controlling for human capital and certain local context factors. However, there is a "substitution" effect between personal socioeconomic status and location factors. This study argues that living arrangements can provide a mechanism through which personal characteristics, social networking, and ethnic recourses interact with macroeconomic trends, and thus carve out local labor market experiences across the urban space.

Using 2000 Census data, we describe the relationship between women’s education and marriage outcomes. Women’s education is strongly related to husband's income and marital status. This relationship is highly nonlinear and varies across the distribution of husband's earnings. Roughly half of the correlation between women’s education and consumption operates through the marriage market. Using 1980 Census data and the quarter of birth instruments proposed by Angrist and Krueger, we find that women's education may have a positive causal effect on husband's earnings, though not on probability of marriage.

Recent, post-1980, immigration patterns have had a dramatic effect on U.S. labor markets, leading to considerable debate about the impact of immigration on native-born black workers. This research examines immigrant and black labor markets, across metropolitan areas, using Public Use Microdata and Summary File data from Census 2000 to generate low, mid, and high classifications of immigrant and black occupations based on socio-economic index (SEI). Multivariate findings indicate that the effect of recent immigration on black labor market outcomes differs by occupational level. Competition for low-skilled jobs is identified for native-born blacks in low-level jobs while a bump-up effect is identified for blacks in mid-level jobs. For example, production occupations with low language and skill requirements are shown to be contested among the groups. On the other hand, service and administrative functions emerge as bump-up mechanisms that create opportunity for black workers who amass the human capital required of these occupations. Thus, the ramifications of immigration for native-born blacks are shown to be quite different for low- and mid-SEI jobs

Frequent pattern mining on data streams is of interest recently. However, it is not easy for users to determine a proper frequency threshold. It is more reasonable to ask users to set a bound on the result size. We study the problem of mining top K frequent itemsets in data streams. We introduce a method based on the Chernoff bound with a guarantee of the output quality and also a bound on the memory usage. We also propose an algorithm based on the Lossy Counting Algorithm. In most of the experiments of the two proposed algorithms, we obtain perfect solutions and the memory space occupied by our algorithms is very small. Besides, we also propose the adapted approach of these two algorithms in order to handle the case when we are interested in mining the data in a sliding window. The experiments show that the results are accurate.

With the huge influx of immigrants to the United States, it is a well-observed phenomenon that a large number of ethnic minorities concentrate in a particular set of labor market sectors. Although considerable literature suggests that metropolitan contexts have significant effect on the job earnings of different racial and ethnic groups, there is a missing link between the metropolitan context and the earnings effect of ethnic niche employment. Using data from the 2000 Census data, this study deploys a multilevel research approach to compare job earnings of white, black, Hispanic and Asian workers in their respective niche and non-niche sectors, and to examine how the metropolitan urban labor market contexts influence these earnings. The findings show that engaging in ethnic niches is the main source of earnings inequalities among different ethnic groups and contextual conditions have great impacts on job earnings between ethnic niches and non-niches, and between different groups.

Rising tuition has led many states to offer college prepaid tuition plans. These plans are consistent with the trend in higher education policy toward meeting the needs of wealthier households. The paper argues that the public interest in these plans is hard to find; moreover, median voter theory suggests that prepaid tuition plans may have the unintended consequence of higher tuition inflation, thus making college less affordable for those least able to afford it.

Rapid tuition increases over the last few decades have made public institutions less affordable than they once were. This and other policy changes may be affecting college enrollment across the country. This article examines how student preparation and college supply interact with the usual factors of tuition, financial aid, and family background to explain state by state variation in public college enrollment rates among Black, Hispanic, and White youth over the 1990s. It finds that rapid tuition increases over the 1990s, changes in federal need-based aid, and steady increases in merit-based financial aid cannot explain variation in public college enrollment rates during the 1990s. What can help explain this variation are a states expenditure on state need-based aid and its investment in public higher education capacity. The study also finds that differences in the high school completion rate of Hispanic youth, among states and over time, help explain patterns of Hispanic enrollment in public postsecondary institutions. The article concludes that the current policy emphasis of maintaining low tuition may not be the best most efficient use of public subsidies in terms of promoting equitable access to higher education.

Many believe that equality of opportunity will be achieved when the prospects of children no longer depend upon the wealth and education of their parents. Many also believe that democracy is the political institution that will bring about justice. This study asks whether democracy, modeled as competition between political parties that represent different interests in the polity, will result in educational funding policies that will, at least eventually, produce citizens with equal capacities (human capital), thus breaking the link between family background and child prospects.

Most discretization algorithms are univariate and consider only one attribute at a time. Stephen D. Bay presented a multivariate discretization(MVD) method that considers the affects of all the attributes in the procedure of data mining. But as the author mentioned, any test of differences has a limited amount of power. We present OMVD by improving MVD on the power of testing differences with a genetic algorithm. OMVD is more powerful than MVD because the former does not suffer from setting the difference threshold and from seriously depending on the basic intervals. In addition, the former simultaneously searches partitions for multiple attributes. Our experiments with some synthetic and real datasets suggest that OMVD could obtain more interesting discretizations than could MVD.

Differences in career incentives make elected public officials more likely to engage in vote-buying compared to appointed bureaucrats. Sustaining excessive public employment is an example of an inefficient policy that is used by public officials to increase their chances of staying in office. I present a model in which patronage is used by public officials to overcome the commitment problem in vote buying. The model predicts that elected executives choose higher level of public employment than their appointed counterparts and that the difference in employment levels is larger for those public employees that are harder to layoff. I use panel data on local governments in the U.S. to test the predictions of the model.

The Integrated Health Interview Series (IHIS) is a five-year NICHD-funded collaboration between the School of Public Health and the Minnesota Population Center at the University of Minnesota to harmonize, document, and disseminate data from the National Health Interview Survey for the period from 1963 to the present. The IHIS builds on the model of the Integrated Public Use Microdata Series (see www.ipums.org) which spurred substantial new research based on harmonized microdata of the U.S. census from 1850-2000. The IHIS will enhance the value of NHIS data and promote new research by allowing researchers to make consistent comparisons across four decades of dramatic change in public health, and thus to study the health status of Americans as a dynamic process. The public-use NHIS files are presently restricted to the period from 1969 forward. Thus, in addition to harmonizing, documenting, and disseminating the currently available public use files of the NHIS, we propose to extend the time series backward by working in conjunction with NHIS staff to create new public-use files for NHIS data prior to 1969.

In this study, we estimate that in 2005, the average monthly employment rate of noninstitutionalized, black men, age 25-34, with a high school degree or less in New York State was 64.4% percent. In other words, on average, more than one man in three in this population was without employment in any given month. In contrast, the average monthly employment rate of similar white and Hispanic men was 84.6% and 84.3% respectively. In addition, approximately 40% of the less educated black men in this age group were idle for the entire previous calendar year.1 And when the institutionalized part of the population is included, the employment rate of black men, age 25-34, with a high school education or less declines to 57.3%, a stunningly low employment rate for young men, many of them fathers, in the prime of their working lives.

Medicare and Social Security often are assumed to provide universal coverage for the population age 65 and older. Evidence from New York City raises doubts. Data from the Statewide Planning and Research Cooperative System, the Centers for Medicare and Medicaid Services, the Social Security Administration, and the U.S. Bureau of the Census provide evidence that 16% to 20% of New York City residents age 65 and older lack such coverage. Noncoverage is not unique to this city, but it may be particularly common there. Noncoverage is pronounced in, but not limited to, certain immigrant groups. Because the population share covered by Medicare increases with age and most hospitalizations not covered by Medicare are paid by Medicaid, Medicaid gradually may be replacing Medicare as the payer for hospitalizations for a substantial share of the 65+ population in New York City.

Reduced-rank restrictions can add useful parsimony to coefficient matrices of untiltivariate models, but their use is limited by the daunting complexity of the methods and their theory. The present work takes the easy road, focusing on unifying themes and simplified methods. For Gaussian and non-Gaussian (GLM, GAM, mixed normal, etc.) multivariate models, the present work gives I unified, explicit theory for the general asymptotic (normal) distribution of maximum likelihood estimators (MLE). MLE can be complex and computationally hard, but we show a strong asymptotic equivalence between MLE and a relatively simple minimum (Mahalanobis) distance estimator. The latter method yields particularly simple tests of rank, and we describe its asymptotic behavior in detail. We also examine the method's performance in simulation and via analytical and empirical examples.

This paper presents a quantitative theory of development that highlights three mechanisms that relate schooling, fertility, and growth. First, we point out that in the early stages of development, fertility and schooling may rise together as the schooling of younger children increases their relative contribution to family income when they turn working age. Second, the model contains a supply-side theory of schooling that generates a rise in schooling independent of technological change. Third, we introduce a direct negative effect of industrialization on fertility that does not operate through human capital and the quantity-quality tradeoff. An initial quantitative assessment of the theoretical mechanisms is conducted by calibrating and applying the model to United States history from 1800 to 2000. We find that the demise in family production is an important factor reducing fertility in the 19th century and schooling of older children is dominant factor reducing fertility in the 20th century. The same model is applied to England from 1740 to 1940, where we offer two complimentary explanations for the rise in fertility from 1740 to 1820. The first is based on the rapid expansion in the cottage industry and the second on the increased relative productivity of children. We also find that the subsequent fall in fertility from 1820 to 1940 cannot be explained without introducing child labor/compulsory schooling laws.