Skip to main content icon/video/no-internet

Mitofsky-Waksberg sampling is a two-stage, clustered approach for selecting a random sample of telephone numbers. Developed by Warren Mitofsky and Joseph Waksberg in the 1970s, this was an innovative approach designed to improve the operational efficiency of telephone samples through reductions in the proportion of unproductive numbers dialed. Prior to the development of Mitofsky-Waksberg sampling, unrestricted random-digit dial (RDD) was used, but this method was operationally inefficient as it led interviewers to call far too many nonworking numbers. Mitofsky-Waksberg sampling (including modified versions of the basic approach) was the predominant approach used for selecting samples for RDD telephone surveys throughout the 1970s and 1980s, but it was largely supplanted by list-assisted RDD by the early 1990s.

An understanding of the various approaches used for RDD sampling requires some knowledge of the structure of a telephone number. In the United States, telephone numbers are 10-digit strings. The first three digits are the area code, and the first six digits are the telephone exchange. A 100-bank is a set of telephone numbers having the same first eight digits. Historically, telephone numbers were geographically clustered. However, under provisions of the Telecommunications Act of 1996, customers are able to retain their telephone numbers when switching from one telephone service provider to another, even when that switch involves a geographic move or a switch between landline service and cellular service. This is called “porting” a telephone number.

Telephone exchanges are designated for particular uses (e.g. cellular only, plain old telephone service [POTS] only, cellular and paging). For RDD surveys, the sampling frame of telephone exchanges has typically been based on those exchanges that are designated for POTS (i.e. landline) use. However, within that subset of exchanges, not every number is assigned, and not every assigned number is residential. Thus, efficiency gains may be achieved by reducing the number of unproductive (nonworking or nonresidential) telephone numbers that are dialed.

Implementation

In the first stage of selection in the Mitofsky-Waksberg approach, the set of telephone exchanges is limited to those exchanges designated for residential use, and a sample of 100-banks is selected for the sampling frame. A random two-digit suffix is appended to each sampled 100-bank to obtain the prime number. Each prime number in the sample is dialed to determine whether it is a residential number. If the prime number is a residential number, the 100-bank is retained in the sample, and in the second stage of selection, additional telephone numbers [secondary numbers) are selected in that 100-bank. If the prime number is not a residential number, then the 100-bank is excluded from the second stage of sampling. Following the second stage of selection, attempts are made to complete interviews until a predetermined fixed number (k) of residential numbers is identified among the secondary numbers in the 100-bank. The total number of residential numbers in the sample is m(k + 1).

A disadvantage of the Mitofsky-Waksberg method is that the selection is sequential; all primary numbers must be resolved before the second stage of sampling can occur, and each secondary unit must be resolved before additional units can be selected. Noncontact cases (ring-no answer and answering machine results) are problematic in that regard. Richard Potthoff, J. Michael Brick, and Joseph Waksberg each developed modified Mitofsky-Waskberg methods to address the sequential nature of the sample.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Sage Recommends

We found other relevant content for you on other Sage platforms.

Loading