Network Scale-Up Method
Also known as: NSUM, Scale-Up Method, Aggregate Relational Data Method, Known-Population Network Estimation
The network scale-up method (NSUM) estimates the size of a hidden population — such as undocumented migrants or members of a stigmatized group — by asking ordinary people in a general survey how many members of that population they personally know. Developed by Killworth, McCarty, Bernard, and colleagues and formalized in their 1998 Evaluation Review paper, it rests on a simple bookkeeping idea: if you know roughly how many people each respondent knows in total, and you observe how many of those acquaintances belong to the hidden group, you can scale that fraction up to the whole society. The trick to recovering the total acquaintance count is to ask about several groups whose sizes are already known — people named Michael, nurses, women who gave birth last year — and use the responses to calibrate each respondent's personal-network size. Bernard and colleagues' 2010 review brought the method into mainstream public-health surveillance and emphasized two crucial corrections: transmission bias, because people often do not know which of their acquaintances belong to a hidden group, and barrier effects, because the hidden group may be socially clustered away from typical respondents. For migration research NSUM is attractive precisely because it never requires contacting migrants directly; it infers their numbers from the social fabric of the wider population.
Key highlights
- Estimates hidden-population size without ever sampling or contacting a single member, by leveraging the knowledge of the general public.
- Requires only a standard general-population survey with added 'how many X do you know?' items, making it relatively cheap and scalable.
- Calibrates personal-network size from groups of known size, turning raw acquaintance counts into interpretable population proportions.
- Provides explicit, principled corrections for transmission bias and barrier effects, so its main weaknesses are modeled rather than ignored.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use the network scale-up method when you need to estimate the size of a hidden migrant or stigmatized population and direct enumeration is impossible, but the group is socially visible enough that ordinary people know some of its members. It is well suited to national or regional size estimation where a general-population survey is feasible and where credible groups of known size exist for calibration. It is particularly valuable when contacting the hidden population directly is unsafe, illegal to the respondents, or politically sensitive, since NSUM only interviews the wider public. The method is less appropriate when the hidden population is so concealed that almost no one outside it knows any members, when strong barrier effects isolate the group from typical respondents, or when no reliable known-size calibration groups are available. It also requires that the survey can ask sensitive 'how many do you know?' items honestly, so settings with extreme stigma or fear of disclosure can undermine the transmission assumptions on which it depends.
Strengths & limitations
- Estimates hidden-population size without ever sampling or contacting a single member, by leveraging the knowledge of the general public.
- Requires only a standard general-population survey with added 'how many X do you know?' items, making it relatively cheap and scalable.
- Calibrates personal-network size from groups of known size, turning raw acquaintance counts into interpretable population proportions.
- Provides explicit, principled corrections for transmission bias and barrier effects, so its main weaknesses are modeled rather than ignored.
- Transmission bias is large and hard to measure: respondents often do not know an acquaintance belongs to the hidden group, biasing raw estimates downward.
- Barrier effects and non-random mixing mean the hidden population may be socially isolated from typical respondents, distorting the scale-up.
- Estimates depend heavily on the chosen transmission rate and on the validity of the known-size calibration groups, both of which carry uncertainty.
- Recall and reporting errors in counting acquaintances, including a tendency to underestimate large networks, propagate into the size estimate.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
Why ask about groups of known size instead of just the hidden group?
Because the number of hidden-group members a person knows is meaningless until you know how many people they know in total. The known-size groups solve this calibration problem: since you know, say, that nurses or people with a certain first name make up a fixed fraction of the population, the number a respondent reports knowing in those groups reveals the overall scale of their social network. Once each respondent's personal-network size is estimated this way, the hidden-group count becomes an interpretable proportion that can be scaled up to the whole population. Without the known-size anchors, NSUM has no denominator.
What is transmission bias and why does it matter so much?
Transmission bias is the gap between knowing someone and knowing that they belong to the hidden group. A respondent may have several undocumented acquaintances but be unaware of their legal status, because that information is concealed or simply never discussed. As a result, reported acquaintance counts undercount true ties, and the raw scale-up estimate is biased downward — often severely for stigmatized groups. Bernard and colleagues correct for it by dividing the raw estimate by an estimated transmission rate, the probability that such a tie is actually known. Because this rate strongly affects the answer, measuring it and showing sensitivity to it is one of the most important parts of a credible NSUM study.
How does NSUM compare with directly sampling the hidden population?
Direct methods such as time-location or respondent-driven sampling contact members of the hidden population themselves, which gives rich individual data but is costly, sometimes unsafe, and limited to members who can be reached. NSUM never contacts the hidden population at all; it infers size from what the general public knows, making it cheaper, safer, and able to reach groups that are otherwise inaccessible. The trade-off is that NSUM gives only aggregate size (and a few characteristics), depends on strong assumptions about transmission and social mixing, and cannot collect detailed behavioral data. In practice the two families are complementary and are often used to triangulate the same estimate.
Sources
- 1.Bernard, H. R., Hallett, T., Iovita, A., Johnsen, E. C., Lyerla, R., McCarty, C., Mahy, M., Salganik, M. J., & Stroup, S. (2010). Counting Hard-to-Count Populations: The Network Scale-Up Method for Public Health. Sexually Transmitted Infections, 86(Suppl 2), ii11-ii15.
- 2.Killworth, P. D., McCarty, C., Bernard, H. R., Shelley, G. A., & Johnsen, E. C. (1998). Estimation of Seroprevalence, Rape, and Homelessness in the United States Using a Social Network Approach. Evaluation Review, 22(2), 289-308.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 23). Network Scale-Up Method. ScholarGate. https://scholargate.app/migration-studies/network-scale-up-method