Skip to main content

Producing Statistics about Hard-to-Reach Populations through Adaptive and Network Sampling

Posted by Matt Gregg, Mathematical Statistician, National Agricultural Statistics Service in Research and Science
Feb 21, 2017
Dr. Steve Thompson at the 25th annual Morris Hansen Lecture
Dr. Steve Thompson headlined the 25th annual Morris Hansen Lecture, November 17, 2015.

The number of people who had heart disease related surgeries, the percentage of Americans who take anti-depressants; the number of women who opt for natural childbirth, these are health statistics you likely hear about in the news frequently. But how do public health researchers obtain data about hard-to-reach, marginalized populations such as the homeless at-risk of contracting specific things like HIV/AIDS?

Producing statistics about hidden, underserved populations was one of the topics explored by Dr. Steve Thompson, professor of statistics at Simon Fraser University during the 25th annual Morris Hansen Lecture hosted by USDA’s National Agricultural Statistics Service (NASS). The lecture series was established by the Washington Statistical Society to honor Morris Hansen and his pioneering contributions to survey sampling and related statistical methods during his long and distinguished service at the U.S. Census Bureau. More than 200 people attended this year’s lecture at USDA’s Jefferson Auditorium in Washington.

Statistics are powerful tools in all disciplines, including public health, wherein researchers seek patterns of illnesses in groups. As a mathematical statistician, I’m proud to participate in education and outreach events, such as the Morris Hansen Lecture series, that invite the general public to learn more about the importance and usefulness of official government statistics.

Family members attending the 25th Annual Morris Hansen Lecture
(Left to right): Heidi Harrison (Morris Hansen’s granddaughter), Ms. Evelyn Allin (Morris Hansen’s daughter), and Dana Allin (Morris Hansen’s grandson, Ms. Evelyn’s son) attended the 25th Annual Morris Hansen Lecture, established by the Washington Statistical Society to honor Morris Hansen and his ground-breaking contributions to survey sampling and related statistical methods during his long and distinguished service at the U.S. Census Bureau.

Health statistics can help determine who is at risk for various diseases, find ways to prevent illnesses and decide which diseases should be further studied. As Thompson pointed out during his lecture, however, with hard-to-reach populations, traditional and standard survey methods don’t work too well.

These populations may include unauthorized aliens, informal sex workers, illicit drug users and people who live ‘off the grid’ – such as those who live off the land in the wilderness or in informal settlements and slums. According to Dr. Thompson there are instances where a researcher may want to collect data from hidden populations, but a sampling frame – a comprehensive list of all the members of the population the researcher wants to study – may not exist. In such cases using adaptive and network sampling techniques makes surveying hard-to-reach populations more feasible. Combining both statistical techniques often better captures information about hidden populations.

Thompson explained that adaptive and network sampling have interesting and useful applications in the fields of human health, environmental science, natural resources management, agriculture and much more. As government agencies, non-profits and private sector enterprises in the public health arena seek to improve the health of all their stakeholders, including those otherwise hidden or marginalized, adaptive and network sampling may help identify, explore and assist the most challenging groups.

A representation of network sampling
This is a representation of network sampling, a statistical method that makes studying rare, hidden populations more feasible. Typically, sampling frames do not exist for these populations because usually there is little information on the size and magnitude of the population. (USGS image).
Category/Topic: Research and Science

Write a Response

CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.