Understanding Statistical Concepts and Terms in Context: The GovStat Ontology and the Statistical Interactive Glossary

	Stephanie W. Haas, Maria Cristina Pattuelli,
	Ron T. Brown, Jesse Wilbur
	School of Information and Library Science
	University of North Carolina at Chapel Hill
	{haas, browr, pattm}@ils.unc.edu, jdwilbur@email.unc.edu
	http://ils.unc.edu/govstat
	Supported by NSF grant EIA 0131824

The GovStat Project
Find what you need,
understand what you find

	Government statistical agencies are a rich source of information, but many barriers  to their use, especially by non-experts, exist.
	Statistical Knowledge Network
		integrate information across agency boundaries
		provide alternate means of finding and viewing information
		provide help when, where, and how it is most useful

Slide 3

Barriers to Using Statistical Information

	Lack of statistical knowledge leads to difficulties:
		searching
		recognizing relevant statistics
		choosing the most appropriate statistics
		understanding what is presented
		using and interpreting the information

Examples from
Metadata User Study

	Why are there so many numbers for employment and unemployment?  Does it matter which I use?
	The table lists mean and median income – which number is right?
	I’ll use the seasonally unadjusted number, because I just want one number, not a set for each season of the year.
	The “E” means it’s an estimate – they don’t know the real number.
	Can I compare a seasonally adjusted number to an unadjusted number?

Basic Principles

	User’s purpose is to find information, not learn more about statistics
	Seeking help shouldn’t mean abandoning the task
	Help should be integrated as seamlessly as possible into the statistical resources

Design Guidelines for SIG
and the GovStat Ontology

	Limited coverage of concept and terms
	Intended for “everyday users”, not experts
	Provide explanations in variety of forms
	Explanations must be attractive
	Explanations should be coordinated with content – context specificity

SIG Framework

Terms

	Frequently encountered terms
	Basic level of statistical literacy
	Strategies for term identification
		examination of frequently-visited pages
		anecdotal evidence from agency and non-agency consultants
		metadata user study
		web crawl of agency sites

Content

	Basic level of explanation
	Variety of delivery “packages”
		definition, example
		brief tutorial, demonstration
		interactive simulation
		combination
	May incorporate related terms and concepts
	May point to more advanced resources

Content
Outline

Context Specificity

	Incorporate explanations in user’s work context
		table- or statistic-specific, e.g., CPI, death rate, gasoline prices, etc.
		agency- or concept-specific, incorporating relevant entities, e.g., employment, benefits, height and weight, etc.
		general, universal context

Slide 13

User Control

	Minimal interruption to user’s task
	Inform user that help is available for a term
	Allow user to select help presentation(s)
	Allow user to control action
		start, stop, repeat, supply values, etc.
	Allow user to explore related terms (ontology)

GovStat Ontology

	Scope of glossary plus connecting concepts and relationships
	Taxonomic relationships support
		provision of context-specific or more general explanations
		concept explanation templates
	Domain relationships support
		combining related concepts in explanation
		integration business rules

Future Directions

	Evaluation of ontology and SIG
	User exploration of statistical concepts and terms in ontology
	Connections to GovStat Metadata Model
	Services for information integration business rules
	Extending context-specific help
		domain terms for statistical agencies
		other arenas

Slide 17