|
<< Click to Display Table of Contents >> Navigation: User Guide > CATI > CATI introduction |
A Computer Assisted Telephone Interview is when an interviewer calls a respondent to do an interview and the questions asked are controlled by a computer program and the answers typed in directly.
As with CAPI (Computer Assisted Personal Interview) the questions can be presented in a random order and filtered (not everyone answers all the questions).
Data is entered using QPSMR Telin which opens a CATI QIF file.
There are some extra files used for a CATI project:
•Sample file. This holds all the information about a list of respondents that will be called including their telephone numbers.
•Quota file. This holds information about quotas and how many records have been saved for each quota.
•Appointments file. This "Partial" file contains records waiting for an appointment.
See CATI data files for details of these CATI data files.
See also Sample point entries and Quota point entries.
The sample is an ASC data file which contains all the contacts for the survey.
The telephone number(s) must be set in TELNUM entries and any other information known about that record, for example company name. This information is called sample data.
Some sample data may also be marked as sample points which can be used by interviewers to select sample to interview.
Each record in the sample data file is given a unique serial number which will be used for any data collected for that record.
For more information see Quota control
A "quota" is a preset number of respondents required to be interviewed in a particular category. This list of quotas is stored in a Quota file and can be created, viewed and edited with Quota file view.
For example, a survey may be commissioned where the customer requires 200 interviews of people who have used brand A. The client further requires that 150 of these must be women and 50 must be men. For this survey we have a "quota target" of 150 female users and a "quota target" of 50 male users. We are also saying that interviewees that do not use brand A are "ineligible". The purpose of quota control is to ensure that no more than 150 female users are interviewed and that no more than 50 male users are interviewed. Quota control is also responsible for ensuring that interviewees not using brand A are not interviewed. Ineligibility can be set with the quota control file or it can be done by simply marking a response as ineligible.
Quota control prevents an interview proceeding once it becomes clear that the interview is not required, either because the quota is full or because the respondent is ineligible. In the example above if 50 male user interviews have already been completed then any male respondent will now be treated as "out of quota".
IMPORTANT: Quota control is only used to prevent an interview proceeding - it cannot be used to make choices during an interview.
If questions are to be used for quota control they should appear at the earliest point that is practical in the questionnaire. In our above example, Gender would probably be the first question (usually not actually asked) and the usage of brand A within the first few questions. The actual question may be a list of brands ever used and a variable will need to be made with users of brand A.
Single-coded questions and variables from sample data can be used for both quota control and as sample points. This is very useful because it prevents "out of quota" sample points being retrieved from the sample file.
Up to 100 single-coded entries may be used to set the quotas you require. These entries are independent from each other, and an interview will be halted if it is out of quota (over quota) or ineligible in any of the quota entries.
For interlocking (nested or matrix) targets it is necessary to create single-coded variables to combine responses from a number of different entries. For example, if you needed to interview specific numbers of male and female respondents, within certain age groups, you would make a new single-coded variable with responses which identified all the quotas required. The target number of respondents needed for each "gender and age" category, determined by the responses in the variable, can then be entered into the quota file.
IMPORTANT: If a quota entry is left blank (empty) then it will automatically be treated as ineligible. For this reason it is important that quota entries are never filtered and have a valid response for every respondent. If quotas are being set for a filtered question then a variable will need to be created with an extra response for anyone not passing the filter. This extra response can be set to "Accept all" if no specific quota is needed for those who do not pass the filter.
Quota control is only used if it is requested when generating the Questionnaire Input file (QIF).
Quota files can be edited from the main menu at any time.