Other types of data used in health research

Other major types of data used in population-based health research include survey and census data. We’ll highlight some of the key distinguishing characteristics of these data sources.

Survey Data

Survey data are collected from a target population for the purpose of answering a particular research question. As data collection can be designed around the research question, surveys can be used to capture specific variables of interest. This includes information on attitudes and health-related behaviours, which are not captured in administrative data. The Canadian Community Health Survey and the Canadian Health Measures Survey are examples of national health surveys. Weighting is often applied in order to make the results more representative, which adjusts for differential representation of groups in a given survey sample. However, because survey data are self-reported, they introduce the potential for non-response and response bias (e.g., social desirability bias), are time-consuming, expensive, and require a large number of respondents to be able to adequately answer the proposed research question (i.e., sufficient statistical power). Moreover, survey data are cross-sectional in that the information collected provides a snapshot of one point in time. Although it is possible to construct longitudinal data sets when surveys are conducted in the same sample over several points in time, it is difficult to precisely ascertain the timing of exposures and outcomes (causality).

Census Data

Census data comprise basic information about every person in Canada. The Canada census is a mandatory questionnaire that runs every 5 years, with 20% of households receiving the long-form census and 80% receiving the short-form, and provides a snapshot of the Canadian population. Advantages of the census are that the data are accurate, complete, and allows for the opportunity to collect specific information that could be used in health research. However, conducting a census is time-consuming and expensive.

Overall, the features that distinguish administrative data from survey or census data are that administrative data:

1. provide a wide range of information on large populations
2. facilitate the creation of data sets for longitudinal studies more easily than with primary data
3. are relatively efficient and cost-effective to use and
4. are collected primarily for the purposes of health care administration, rather than answering a particular research question

With the use of appropriate health research methods administrative data can help us derive meaningful insights on the healthcare system and the comparative effectiveness of interventions. We’ll discuss in more detail both the strengths and limitations of administrative data in Section 4: Strengths and Limitations of Administrative Health Data.