Survey Measurement of Health: Implications for Social Science Research - Online

The measurement of health in surveys involves collecting data from individuals about their health status, health-related behaviours, and experiences, often as part of multi-purpose surveys. These surveys may include both subjective self-reports (e.g., self-assessed health measures) and objectively measured health data (e.g., physical health assessments, blood-based biomarkers, or DNA data).

Survey measurement of health plays a vital role in advancing social science research on health. It is essential for the analysis of health inequalities, for informing policy decisions, and evaluating the effect of interventions. By capturing individual health behaviours alongside the social, economic, and environmental contexts influencing health outcomes, surveys offer a nuanced understanding of the complex interplay between society and health.

However, measurement error in health data can significantly affect and compromise the quality of social science research in health. Importantly, such errors are not confined to self-reported data. While self-reports are often susceptible to certain types of inaccuracies, other sources of error can arise, including those associated with nurse-administered health assessments or blood-based biomarker data.

Survey mode – e.g. face-to-face interview, telephone interview, online questionnaire - is also well known to affect the distribution of the survey variables so is a special case of measurement error.  Mode effects are hypothesised to be driven by well-known sources of survey bias such as social desirability, positivity and satisficing and the presence (or not) of an interviewer. Mode effects can be defined in the same way as causal or treatment effects and estimated from mixed-mode surveys, and estimated using the same methods.

We provide an overview of methods for understanding measurement error and mode effects. We will also provide practical sessions and illustrative examples demonstrating the impact of measurement error and mode effects in the social and health sciences.

The course covers: 

• Collection of health data in multi-purpose social science surveys

• Measurement errors in health data, including both self-reports and objective measures

• Implications of measurement errors in health data for existing social science research in health

• The nature of mode effects and the connection with classical measurement error

• The definition and identification of different kinds of mode effect from mixed-mode surveys and a case study based on Wave 8 of Understanding Society.

• Practical sessions with illustrative examples of measurement error and mode effects

By the end of the course participants will be able:

• Understand the basis on the process of survey measurement of health, including the collection of both self-reported and nurse-collected health data.
• Explain the role of survey measurement in advancing social science research, particularly in understanding health inequalities, guiding policy, and tracking interventions.
• Recognize the impact of measurement error in health data, including errors in self-reported, nurse-administered, and biomarker data.
• To understand the potential impact of survey mode on survey data, learn how to estimate different kinds of mode effects from mixed-mode surveys, and how to do so robustly using instrumental variable estimation.
• Apply practical knowledge of how measurement error in survey health data can affect the accuracy and interpretation of social science research.

This course is aimed at Post-graduate researchers and analysts, including (but not limited to): Academics, Government Researchers, Third sector organisations, (Health) Consultancy analysts and Survey methodologists. Participants will need a basic knowledge of STATA.

Programme TBC

Day 1

9:00-11:00 Measurement error in self-reported health measures regularly available in large-scale multipurpose datasets

11:00-11:15 (Virtual) coffee break (Q&A session)

11:15-12:45 Beyond self-reported health measures – characterizing and quantifying measurement errors in administrative health data and nurse-collected bio-measures.

12:45-13:45 Lunch

13:45-14:45 Assessing the potential implications for the existing research in economics and social science that rely on health data

14:45-15:45 Practical sessions using Stata and illustrative examples

Day 2

9:30-11:00 Connection between mode effects and measurement error, definition of different mode effects, identification of mode effects from mixed-mode surveys where people non-randomly select survey mode

11:00-11:15 (Virtual) coffee break (Q&A session)

11:15-12:45 Case study: Using regression and instrumental variable-based methods to estimate mode effects from Wave 8 of Understanding Society

12:45-13:45 Lunch

13:45-14:45 Practical session using Stata and illustrative examples

Payment using the Online Store can only be completed via Visa and Mastercard Credit/Debit Card or PayPal.  AMEX is not accepted.
If you have not previously created an account for the Online Store, you will need to create an account to make a booking.