Aleksandra Slavkovic

Using a large projection screen as an alternative to head-mounted displays for virtual environments

Head-mounted displays for virtual environments facilitate an immersive experience that seems more real than an experience provided by a desk-top monitor [18]; however, the cost of head-mounted displays can prohibit their use. An empirical study was conducted investigating differences in spatial knowledge learned for a virtual environment presented in three viewing conditions: head-mounted display, large projection screen, and desk-top monitor. Participants in each condition were asked to reproduce their cognitive map of a virtual environment, which had been developed during individual exploration of the environment along a predetermined course. Error scores were calculated, indicating the degree to which each participants map differed from the actual layout of the virtual environment. No statistically significant difference was found between the head-mounted display and large projection screen conditions. An implication of this result is that a large projection screen may be an effective, inexpensive substitute for a head-mounted display.

Bounds for Cell Entries in Two-Way Tables Given Conditional Relative Frequencies

In recent work on statistical methods for confidentiality and disclosure limitation, Dobra and Fienberg (2000, 2003) and Dobra (2002) have generalized Bonferroni-Frechet-Hoeffding bounds for cell entries in k-way contingency tables given marginal totals. In this paper, we consider extensions of their approach focused on upper and lower bounds for cell entries given arbitrary sets of marginals and conditionals. We give a complete characterization of the two-way table problem and discuss some implications to statistical disclosure limitation. In particular, we employ tools from computational algebra to describe the locus of all possible tables under the given constraints and discuss how this additional knowledge affects the disclosure.

Novice heuristic evaluations of a complex interface

In this paper, we describe the ability of evaluators with limited experience to use Heuristic Evaluation (HE) in assessing a complex interface. We analyze our results in terms of the proportion of problems found by different sets of evaluators in different areas of the interface. Our results illustrate that the 5-10 evaluator size advocated by Nielsen and Molich [3] does not generalize to assessing complex interfaces. Evaluators tend to focus on certain sections of the interface and ignore others. Our results suggest that modification to HE is necessary to most efficiently produce a complete set of usability problems in an interface.

Making the Release of Confidential Data from Multi-Way Tables Count

The reader isasked to keep ill min(1tbatt/leconceptof disclosure presented here isa verybroadone. It Wf)LI/~ii(Jft;edesirab/etD requif!.that there beazero.risk of disclosure, asdefined beloW,Jn..yreleaseoftabulations or microda~ files. Such a requirementWCjIJk/endalarge~ofallre/eases howbeingmade. This would be too great a priceto payforcomplete elimination of anyrisk of disclosure. Statistical disclosure limitation (SOL) and confidentiality have often been shrouded with a nonstatistical veil and the methodology for protecting confidential data has produced problematic outcomes for research data users. Here we describe one possible statistical approach to SOL for data in the form of multidimensional contingency tables that illustrates the following points:

O privacy, where art thou?

In last issues column, John Abowd and Lars Vilhuber discussed the interplay between science, confidentiality, and the public good and outlined some of the data access modalities, in particular a synthetic data model implemented by the Cornell Virtual Research Data Center. In this months column, Saki Kinney and Alan Karr from the National Institute of Statistical Sciences give details about how access is operationalized and offer practical guidelines and considerations for researchers seeking access to confidential data from government agencies.

O Privacy, Where Art Thou?: Privacy Protection for the European Census Hub

38 Each member state of the European Union (EU) is responsible for conducting periodic censuses and disseminating the results. There also exists a statistical office of the European Union, named Eurostat, situated in Luxembourg. The task of Eurostat (http://epp.eurostat.ec.europa.eu/ portal/page/portal/about_eurostat/ introduction) is to “provide the European Union with statistics at European level that enable comparisons between countries and regions.” A recent project in this direction is [o privacy, Where Art thou?]

O Privacy, Where Art Thou?: Statistical Disclosure Limitation Research and Practice: Fascinating and Growing Areas of Importance

When releasing data to the public, statistical agencies and other organizations—henceforth all called agencies—are ethically and often legally obligated to protect the confidentiality of data subjects’ identities and sensitive attributes. Stripping unique identifiers such as names, addresses, and government-issued identification codes from the file may not adequately protect confidentiality, because data snoopers may be able to link records in the released data to records in external databases by matching on common values in the two files. For example, the computer scientist Latanya Sweeney famously showed that 97% of the records in publicly available voter registration lists for Cambridge, Massachusetts, could be uniquely identified using birth date, gender, and nine-digit ZIP code. By matching on the information in these lists, she was able to identify then Massachusetts In the previous three columns, established researchers in the area of statistical data privacy presented their views on what is privacy and confidentiality and why society cares for safeguarding and sharing quality confidential data at the same time. They also provided a guide to current data access modalities. In this column, Jerry Reiter, past chair of the American Statistical Association’s Committee on Privacy and Confidentiality, explains how statistical science plays a key role in data dissemination and invites you to begin exploring the many statistical research problems this area presents.

O privacy, where art thou?: Research access to restricted-use data

Official statistics agencies in the United States and other countries have long faced conflicts between two of their many missions. On the one hand, these agencies are charged with collecting vast amounts of high-quality data about individuals and establishments such as businesses, health care providers, and universities in a manner that protects the privacy of data objects and the confidentiality of databases. On the other hand, they must disseminate information for diverse purposes; among the most important are formulation and evaluation of policies and supporting research conducted by academics, other government agencies, and private citizens. More and more, data are also being collected and held by the states. Most notably, state education agencies (SEAs) are building statewide longitudinal data systems (SLDS) containing individual-level data on students and teachers in public schools. Even though SLDS were built under pressure from, and in many cases with funds provided by, the U.S. Department of Education, they are owned and controlled by the states. This column discusses processes by which researchers in the United States can access confi dential data from government Research Access to Restricted-Use Data