CATI Overview

The exact procedure to be used by an interviewer while conducting an interview will of course be planned and administrated by the individual call center. However the procedure will probably follow the general format as below:

1. The interviewer logs on to the system.
2. A survey is assigned to the interviewer.
3. There will normally be an introductory screen with the potential respondent's details, including of course the respondent's phone number, and an introduction the interviewer must read out to introduce themselves, the company they are working for and the reason they are calling.
4. The interview will call the number displayed, or a dialer will call the number and pass the line to the interviewer.
5. If the interviewer cannot get through to the potential respondent for any reason (no reply, busy, unobtainable, fax, modem etc), they will code the call outcome accordingly on a sample outcome screen. If they are connected to the respondent they read out the introduction, and if the respondent agrees to take part then the survey can start.
6. The survey script will be coded to follow a particular path (routing) depending on the answers the respondent gives. This means that certain questions or blocks of questions may be answered while others might be skipped. All of this is done automatically, and the interviewers only have to code the respondent's answers to the questions.
7. When the interviewer reaches the end of the survey, the system will flag the current interview and sample record as completed. This means that the sample record (the respondent) will not be used again for this survey.
8. If the survey terminates prematurely for any reason, an exit code will be assigned to the survey depending on the reason for the termination and the sample record will be flagged accordingly. This process can be automated or manual.
9. Surveys can be completed in more than one stage. For example, if the respondent is busy he/she may ask the interviewer to call back at a more appropriate time. The interviewer then terminates the survey and sets up an appointment.
10. As the planned appointment time approaches, CATI will try to send the "started" interview to the same interviewer who originally booked the appointment, if he/she is logged in and available; otherwise it will pass the appointment on to the next available interviewer.
11. When an interviewer has completed a survey and is ready to start a new one, the interviewing system will query the database for the next available sample record to be fed to the interviewing program, and the interviewing working flow will start over again.

Sample records are flagged with different priority status. The system usually queries the database for sample records tagged with an appointment flag first, then no-reply, busy numbers and so on, before releasing a fresh sample. In this way, all sample records with a higher priority status are used before the fresh (still unused) sample. The sample priority can be customized by the CATI system administrator according to global or survey-based requirements.

If there are screening questions linked to quotas (for example age, sex, etc.) and the relative quota cell is already full (go to About Quotas for more information), the system will automatically route the interviewer to a closing statement. The survey will then terminate and the call and sample outcomes (quota fail) will be coded automatically.

Initially each sample record has a status of “New”. The status will be updated every time the sample record is passed on to an interviewer, until a survey is completed for that respondent or the sample record is eventually disposed of. CATI administrators can customize the rules which affect the way sample records are used and disposed of.

Sample can be flagged and moved according to requirements. For example, if a quota cell is closed and there are still records in the sample file for that quota cell, such records can be flagged or moved to an unused queue so that they will no longer be sifted through. If the sample includes tens of thousands of records, this can save a considerable amount of computer resources. Note that this is only done if a quota option is checked for a CATI channel survey (by default it is not) (go to Survey Channels for more information).

Time Zones and Shifts

Surveys can be assigned to a time zone depending on where the respondents are located. This means you can avoid calling them at inappropriate times - for example in the middle of the night. The desired time zone is set in the sample and then fixed in to the script, and the system will then allow or stop calls depending on the time zone the respondent is in.

It will normally be sensible (necessary) to specify which hours of each day of the week are to be available for interviewing. The days can be divided into "shifts" which can be classified as for example morning, afternoon, evening, or weekend. This classification may also be useful for recall purposes.

Note: When interviewing is being conducted across time zones, the shift times refer to the current time in the respondent’s time zone.

For further information, refer to the CATI Supervisor documentation.

Call Disposition

The calls will be disposed of depending on the survey and the outcome of the call. This means that if a respondent's telephone is busy or there is no answer when it is called, the interviewer will code the survey outcome as “busy” or “no reply”. The survey will then terminate and the sample record will be flagged in the database as “busy” or “no reply”. If a dialer is being used, most of the call dispositions are made automatically (connected call, engaged, no reply, unobtainable, fax) and this will dramatically increase the productivity of the call center. As the information is stored in the database, supervisors can extract reports to see for example productivity, interviewers' strike rate, calls per hour, etc.

Respondents are often able to see the phone number or a name for an incoming call, and can decide not to answer a call from an unknown number or that originates from an usual area code. When the respondent sample is selected for a survey, the sample can include caller identification information (CLID) to tell the respondent who you are. This can be set independently for each respondent. A special sample field named 'ExtensionNumber' must added to the respondent database into which the CLID values can be imported, and this information is then sent to the dialer (go to Preparing the Respondent List for more information).

Note: On a supporting dialer the system field for ‘ExtensionNumber’ will be sent to the dialer as a call property to specify the desired Caller ID (CLID). This means that individual CLIDs can be set by supplying the desired values in a sample file (using a column headed ‘ExtensionNumber’). Contact support to check if your system supports this functionality.

Dialler

A dialer is a computer server which dials telephone numbers on behalf of the interviewers. This removes any dead time, and therefore dramatically improves the call center's productivity.

A dialer is typically a hardware add-on to the CATI system. It is a computer server full of telephony hardware which links to the telephone network, dials numbers automatically and feeds the connected calls to the CATI system. It allows the CATI system to offer features such as real time monitoring, full call recording and verbatim recording, which significantly increase call center quality and productivity.

A large number of different dialer systems are available on the market - refer to the documentation for your specific system for further details.

Scheduling

The Scheduling functionality allows the CATI interviewer to set up rules to process finished or interrupted calls and reschedule them as necessary. A Scheduling Script is made up of a number of rules which are applied to the finished call. This script is typically used to reschedule the finished call according to this set of rules. A scheduling script can consist of:

• Shift types - defines the types of shifts specified for use in the scheduling scripts.
• Shifts – defines the hours worked by the interviewers during a week.
• Rules – describe sub-rules that trigger recalls.
• Sub-rules – describe the actions that should be performed for a specific sub-rule.
• Actions - describe the actions that should be performed for a call.
• Time Zones – defines a set of time zones that are then used in the scheduling script.
• Custom script (optional) – custom script can be used to perform specific actions not available via the GUI interface.

Scheduling scripts can be saved to the database and stored as XML descriptions. Saved scripts can be edited or deleted at any time. The user can also export and import Scheduling script descriptions in XML format.

Monitoring

CATI provides supervisors with full control of the interviewing process, and allows them a direct view of surveys as they are conducted. For example, a supervisor can view reports that show:

• The number of completed interviews
• Mid-interview results for all questions or selected questions
• Mid-interview results for all respondents to date or a sub-set of respondents

While working in a live environment, the supervisor can check interview qualifications. For example, when looking at the sample information, they can allocate French-speaking respondents to the interviewers who speak French, or send certain samples to their "persuader interviewers", interviewers who have low refusal rates.

The supervisor can also review samples to ensure enough numbers are available, and then instantly make changes when needed. For example, if there is a shortage of telephone numbers, the supervisor can adjust the sample management rules. They might decide to re-try the "no answer" numbers or re-try the "busy" numbers five times instead of four before discarding them.

The supervisor can monitor the interviewers while the surveys are being conducted - both video and audio monitoring are available. Note however that audio monitoring can be only achieved if the CATI center is either using a dialer or has an ad-hoc audio monitoring system in place.

After choosing a user or user group, the supervisor can then view the user name, ID and their status in the panel.

The supervisor can choose which interviewer to monitor from a list of the interviewers currently logged on. Once they have selected one user, they can see on their screens exactly what the interviewer is seeing on his/her screen. If audio monitoring is enabled, the supervisor can hear the conversation between an interviewer and a respondent but he/she cannot be heard by the interviewer.

The supervisor can oversee the interviewer's activities such as keystrokes, entered answers and mouse clicks in real time, and can see all the questions and the answers the interviewer is typing, in sequence, in real time.

Productivity Reports

There are two different reports available to the supervisors to monitor the productivity rates. These reports can be printed or exported to suitable formats such as ASCII, CSV, Crystal, Excel, etc. The reports available are:

• Sample Status Summary report - a summary of the status of all records within a survey.
• Productivity report - a report based on productivity of all interviewers working on particular surveys.