How and why market research uses random number generation
You have probably had a call in which a friendly voice at the other end speaks about market research. Have you ever wondered where research agencies get the numbers? Unless it is a specific study for which the client provides the database, the source is randomly generated telephone numbers.
Why do research agencies use randomly generated numbers?
Good research requires a random selection of respondents. SIMAR member agencies are never trying to find out the name of the person they reach, to push a product on them or to obtain any other personal details. Random generation lets an agency reach the owner of any telephone number with the same probability, which is essential when working with aggregated data. Taking part is always voluntary and anonymous. There are certainly people who do not answer unknown numbers, so complete randomness cannot be guaranteed by any form of selection — but random number generation brings us credibly close to it.
The alternative, using thematic databases from brokers, increases sampling error, which is undesirable in market research. Agencies therefore do not use purchased databases, or only to a very limited extent.
How does the generation work?
Random number generation is handled by a computer, which has no trouble producing a batch of several hundred thousand combinations in a few seconds. It takes a known mobile operator prefix and adds a random six-digit combination.
The three largest operators have some 35 prefixes.
Virtual mobile operators have roughly half as many.
The generation itself works by instructing the computer to combine a known operator prefix with further random digits so that the number is exactly nine digits long.
At the same time the computer makes sure it does not generate duplicates and checks the numbers against the list of blocked numbers that SIMAR member agencies maintain.
It will occur to you that this produces many numbers that do not exist. That is true, but today’s digital exchanges deal with it. Dialling in large telephone studios is not done by people but by computers, which discard non-existent numbers within seconds.
What does it look like in practice? Of 100 generated numbers, 45 do not exist, 30 of those reached never answer, and the remaining 25 are available — and only there can an interaction between interviewer and respondent take place and, if it goes well, a research interview happen.
Author: Pavel Čejka, Data Collect s.r.o., e-mail: pavel.cejka@datacollect.cz, tel.: 731 403 556
