Vol. 10, No. 10, October 2014 “Happy Trials to You”

Subscribe free at www.firstclinical.com

© 2014 First Clinical Research and the Author(s)

Making Sense of Biostatistics: Types of Nonprobability Sampling

By Kathleen Mathieson

Probability sampling methods, such as simple random sampling, as discussed in the last

column, are the best way to reduce sampling bias and achieve a representative sample of

patients for a clinical study.1,2,3 These methods, however, require a comprehensive list of

eligible patients, which is rarely available in practice. Therefore, clinical studies must often

use methods of nonprobability sampling, in which subjects are chosen on a basis other than

random selection.4,5 This column describes several common types of nonprobability

sampling methods and factors that can reduce bias and increase generalizability. Regardless

of the sampling method used, assignment to treatment arms should still be random.

Convenience sampling entails selecting patients on the basis of their availability. For

example, patients may be selected from an easily accessible geographic region or specific

clinical site. Patients with certain characteristics are more likely to be chosen via

convenience sampling. For example, in community-based research, healthier patients are

often more accessible than those with more serious illness. This factor and others can lead

to over- or under-representation of certain population attributes that are being studied, and

therefore decrease the generalizability of study results.

The type of convenience sampling used most commonly in clinical studies is consecutive

sampling. In consecutive sampling, each consecutive eligible patient who presents for care

is approached for enrollment. Consecutive sampling provides some structure and thus

additional rigor in that it includes all patients who are accessible within the defined study

time period.4 The resulting sample is thus more likely to represent the target population

than one resulting from simple convenience sampling. When consecutive sampling is used,

researchers must ensure that the study period spans a long enough time period to avoid

potential bias. For example, in studies where seasonal variations may be important (e.g.,

allergy, depression, or arthritis trials), it is important for the study period to be long enough

to achieve a broad, generalizable sample of patients.6

Another way to enhance the rigor of convenience sampling is to use quota sampling to

obtain a certain number of participants with specific characteristics. For example, many

more women than men are diagnosed with lupus. If researchers wish to test the efficacy of

a treatment in both men and women, they may need to set quotas by sex to ensure

appropriate statistical power. Using consecutive sampling, researchers may enroll patients

of both sexes until the quota for women is filled, and then continue recruiting men until

their quota is filled.

Purposive sampling is a precise form of convenience sampling in which patients are hand-

picked based on certain criteria when a researcher is interested in a very specific group of

patients for reasons of feasibility or efficiency.6 Thus, studies with very specific eligibility

criteria are engaging in purposive sampling; it’s a matter of degree.

For studies of rare conditions or for difficult-to-reach populations, snowball sampling may be

an efficient and cost-effective method. This method involves identifying an initial individual

or set of individuals who meet study criteria, then using those individuals to find and contact

others for the study. In the case of a rare illness, a researcher might identify the initial

“seed” of patients via a support group and then use those patients to reach other eligible

patients.

Subscribe free at www.firstclinical.com

© 2014 First Clinical Research and the Author(s)

2

Convenience methods that can be expected to bias the sample should be avoided. For

example, recruiting patients that live near the research site might not introduce substantial

bias, but selecting patients based on their Medicare/Medicaid status might.

Like all study design decisions, choosing a sampling technique for clinical research involves

trade-offs. When choosing a technique, researchers must balance the objective of

generalizability with real-world factors, such as cost, efficiency, feasibility and practicality.

Although probability sampling is the gold standard for ensuring that the study sample

accurately represents the target population, in practice it is usually more of a yardstick

against which more feasible methods can be measured. Nonprobability sampling methods,

such as those described in this column, are frequently used in clinical research, and while

not as rigorous as probability methods, can provide efficient and cost-effective ways to

recruit patients.

References

1. Mathieson, K. Making Sense of Biostatistics: Probability Versus Nonprobability Sampling.

Journal of Clinical Research Best Practices. 2014; 10(8). Available from

https://firstclinical.com/journal

2. Mathieson, K. Making Sense of Biostatistics: Types of Probability Sampling. Journal of

Clinical Research Best Practices. 2014; 10(9). Available from

https://firstclinical.com/journal

3. Kandola D, Banner D, O’Keefe-McCarthy S, Jassal D. Sampling Methods in

Cardiovascular Nursing Research: An Overview. Canadian Journal of Cardiovascular

Nursing. 2014; 24(3): 15-18.

4. Straus, SE, Richardson, WS, Glasziou P, Haynes, RB. Evidence-Based Medicine: How to

Practice and Teach EBM (Third Edition). New York, NY: Elsevier; 2005.

5. Acharya AS, Prakash A, Saxena P, Nigam A. Sampling: Why and How of It? Indian

Journal of Medical Specialties. 2013; 4(2): 330-333.

http://dx.doi.org/10.7713/ijms.2013.0032

6. Portney LG, Watkins MP. Foundations of Clinical Research: Applications to Practice

(Third Edition). Upper Saddle River, NJ: Pearson Prentice Hall; 2009.

Author

Kathleen Mathieson, PhD, CIP, is an Associate Professor in the Doctor of Health Sciences

Program at A.T. Still University (ATSU) in Mesa, Arizona. She is also a Vice-Chair of the

ATSU-Arizona Institutional Review Board and a Still Research Institute Scientist. Contact her

at 1.602.573.6547 or kmathieson@atsu.edu.