Longitudinal Surveys of Children (1998)

procedures, to name a few, are both highly salient and very sensitive with regard to the protection of confidentiality.

The Statistics Canada survey—Canadian National Longitudinal Survey of Children and Youth—addresses the issue of confidentiality by offering to run programs for its customers. In the United States, it is common practice for a federal agency such as the Census Bureau to swear in a data user as a special employee if microlevel data files rather than aggregated data files are to be analyzed. The Census Bureau has set up two research data centers—one at Carnegie Mellon University and one at the bureau's Boston regional office, and a third center is being created—where researchers work directly with confidential data.

The NCES has a licensing arrangement with organizations that have access to restricted-use data files. These files contain individually identifiable information. NCES will lend restricted-use data only to qualified organizations in the United States using a strict licensing process. Under the terms of the license, NCES has the right to conduct unannounced, unscheduled inspections of the data user's site to assess compliance with the provisions of the license.

Data disclosure becomes even more controversial when the data collection process entails videotaping interviews or behavior. For example, several studies collect data by observing mother-child interactions. When videotaped, respondents are inherently identifiable. In such situations it is often necessary to obtain several types of consent. One type might be to collect the data; another might be to use the data (e.g., use of the data might be allowed solely for educational purposes). Furthermore, consent alone may not always be enough. Respondents may not be aware of the full implications of consent when the data are public, which may lead to a level of disclosure of the data that was not intended. As stated above, when collecting data on minors, the consent issues of adoption, custody, and juvenile crime may be of concern.

Workshop participants proposed a number of ideas for meeting the challenges facing longitudinal survey research on children, including the following:

1. Developing a guide for interviewing children and youth. Workshop participants emphasized the usefulness of developing a guide for conducting research on children and youth in the context of survey research. Such a guide would facilitate interviews and ensure that data collection is based on tested principles and best practices in the field. Though the issue has been addressed in various settings and has appeared in different research papers, participants were

• not aware of any systematic attempt to synthesize existing knowledge on how best to collect survey data on children and youth. Participants were unaware of guides or handbooks on this issue, only clinical assessment guides.
• Developing short-item assessments. The current practice for collecting data on child development is to apply existing standardized assessment scales. Such scales are often too long and too difficult to administer in a survey where only a short period of time is available for data collection. Workshop participants called for research to develop short-item assessments and learn which constructs are amenable to short-term assessments and which are not. The tools must then be validated and recognized as standard measurement tools. Furthermore, short forms and adaptive testing strategies need to be developed to be utilized with current standard instruments.
• Embedding studies and experimental components into longitudinal surveys. Workshop participants discussed the value of embedding smaller methodological studies or data collections into longitudinal surveys on children. For example, it may be necessary to embed smaller studies into ongoing national initiatives in order to ensure that the data are relevant and that they address current policy questions. Such efforts may involve embedding observational modules or small-scale experiments. Participants called for research into the best strategies for incorporating smaller methodological research projects into larger studies, citing the Wisconsin Longitudinal Survey as a good example of this strategy. They also stressed the need for funding for such methodological research, suggesting that agencies that fund larger ongoing studies be urged to set aside money for embedding studies and experimental components in them.
• Ensuring timely dissemination of data while preserving confidentiality. While workshop participants agreed that data should be made available to the public on a timely basis, they noted that subjects' confidentiality must always be of utmost concern. Preliminary files could be released as research files with appropriate caveats, they suggested. They further agreed that, because some users may have difficulty understanding how to use data from longitudinal datasets, different strategies for ensuring that data users analyze and interpret data properly should be investigated. Strategies to educate data users on how to properly interpret data include workshops at professional conferences, short courses offered by sponsoring agencies, and intern and visiting scholar (fellowship) programs.
• Adding topics to existing and new datasets. Balancing coverage and content and the need to stay current in data collection is a major challenge to longitudinal studies. At times, researchers might find it necessary to augment the scope of a study already in progress. A central core of topics may be addressed initially and consistently and then new ones might be addressed in supplementary modules over the course of the survey. Issues discussed in this context at the workshop were (1) the role of fathers; (2) the well-being of immigrant children