Work with interactive mapping tools from across the Census Bureau.
Read briefs and reports from Census Bureau experts.
Watch Census Bureau vignettes, testimonials, and video files.
Read research analyses from Census Bureau experts.
Developer portal to access services and documentation for the Census Bureau's APIs.
Explore Census Bureau data on your mobile device with interactive tools.
Find a multitude of DVDs, CDs and publications in print by topic.
These external sites provide more data.
Download extraction tools to help you get the in-depth data you need.
Explore Census data with interactive visualizations covering a broad range of topics.
Information about the U.S. Census Bureau.
Information about what we do at the U.S. Census Bureau.
Learn about other opportunities to collaborate with us.
Explore the rich historical background of an organization with roots almost as old as the nation.
Explore prospective positions available at the U.S. Census Bureau.
Information about the current field vacancies available at the U.S. Census Bureau Regional Offices.
Discover the latest in Census Bureau data releases, reports, and events.
The Census Bureau's Director writes on how we measure America's people, places and economy.
Find interesting and quirky statistics regarding national celebrations and major events.
Find media toolkits, advisories, and all the latest Census news.
See what's coming up in releases and reports.
Suppose that the log-likelihood-ratio sequence of two models with different numbers of estimated parameters is bounded in probability, without necessarily having a chi-square limiting distribution. Then BIC and all other related consistent model selection criteria, meaning those which penalize the number of estimated parameters with a weight which becomes infinite with the sample size, will, with asymptotic probability 1, select the model having fewer parameters. This note presents examples of nested and non-nested regression model pairs for which the likelihood-ratio sequence is bounded in probability and which have the property that the model in each pair with more estimated parameters has better predictive properties, for an independent replicate of the observed data, than the model with fewer parameters. Our second example also shows how a one-dimensional regressor can overfit the data used for estimation in comparison to the fit of a two-dimensional regressor.
Model selection, linear regression, misspecified models, AIC, BIC, MDL, Hannan-Quinn criterion, overfitting
This symbol indicates a link to a non-government web site. Our linking to these sites does not constitute an endorsement of any products, services or the information found on them. Once you link to another site you are subject to the policies of the new site.