UN Web Site | UN Web Site Locator
Home Site map Contact 
ESCAP Statistics Division
ESCAP Statistics Division
 
Fourth Meeting    
The Fourth Meeting of the Working Party on the Application of New Technology to Population Data
Manila, 6-9 July 1999

STAT/WPA(4)/INF.1
ENGLISH ONLY

ECONOMIC AND SOCIAL COMMISSION FOR ASIA AND THE PACIFIC

Working Party on the Application of New Technology to Population Data
Fourth meeting
6-9 July 1999
Manila

The development of Computer Assisted Telephone Interviewing (CATI) in Macau
Contents
  1. Introduction
  2. System Overview
  3. The CATI Application Design
  4. The Vacancy and Remuneration Survey (D7TR)
  5. Summary

1.  Introduction

The use of Computer Assisted Telephone Interviewing (CATI) system in statistics is not new, pioneer countries have been successfully deployed CATI in data collection for many year. A CATI system is basically a computer system that helps the interviewers to ask questions over the telephone. With this primary objective, different approach has been adopted to built the system with different level of sophistication in technologies and functionality.

This paper intends to review the experience on the development of the CATI in the Census and Statistics Department of Macau (DSEC) in the year 1999. It explains the overall system configuration, the software employed, the design concept, and the implementation of the first CATI survey. On the time of writing, the CATI project is completed at its first phase, and trial run has been started at July 1999.

2.  System Overview
2.1 System Diagram
Figure 1 - CATI system diagram
Figure 1 - CATI system diagram

There are two servers to support the CATI operation. The first server is the database server, it provides data and storage to the CATI workstation, such as the CATI assignment, the related master files, the log files etc. The second server is the telephone server, this server is installed with special voice/fax boards, its function is to manage the telephone line, such as dial out, hang up, voice record, telephone conference, etc. The CATI workstation is normal LAN workstation running Windows 95, they have no any telephony equipment installed.

From the system diagram we can see that the CATI agents (interviewers) are actually working on two networks, one is the ordinary local network to store and retrieve data, and the second is the telephone network which deals with telephone activities. Two networks are physically separated but are logically connected. The CATI agents run a front-end windows application on the PC and use the headphone for the telephone interview. The following steps show a typical CATI operation and the underlying work flow:

  1. The CATI agent retrieves an assignment by clicking the <next> button on the application screen;
  2. According the priority queue, the application retrieves the next assignment from the database server and displays it on the workstation screen;
  3. The CATI agent reviews the history of the assignment, and click the <dial> button;
  4. The application issues a TCP/IP message to the telephone server to request a telephone call, the message includes the telephone number to be dialled and the extension number of the workstation;
  5. The telephone server receives the message and make the telephone call, upon success of the phone call, the workstation's extension will be connected to the line and the interview starts. If the call failed, the telephone server will send back an error message to the workstation's front-end application.
2.2 The telephone server

This is a Pentium PC running Microsoft NT. It has been installed with two types of boards to support the telephony functions. 1) the voice/fax boards to connect the incoming telephone lines and 2) the extension board to connect the headphone of the CATI agent.

A program called the Ring!impact is running on top of the NT to manage all the telephony activities in the server. This software is a general purpose Computer Telephony (CT) applications development tool, its visual programming interface (Figure 2) allows programmers to construct CT applications by simply connecting pieces of program blocks. A pascal-style programming language called Ring!Talk is employed in the Ring!,providing certain level of flexibility in building CT applciations.

Figure 2 - Ring!impact screen
Figure 2 - Ring!impact screen

One advantage of using the Ring! to develop the CATI system is the save of development time since most of the basic telephony functions required by the CATI system are build-in. Another advantage is that the same CATI computer system can be used to develop other telephony projects easily, like the Interactive Voice Reponse (IVR) system, fax-on-demand, or voice-mails etc.

Drawback on using this software is that it is not a true SDK, programs written on Ring! are imposed certain limitations of driving the voice boards, this limits the functions that the CATI system can has. Another drawback is that Ring! is a single task environment, at any one time it only allow one program running on the server, this create problem if we want to divide the lines/ports to serve two different applications - we have to merge the programs of two applications into a single one in order to have the services on the same time.

2.3 The database server

It is a normal NT server with MS-SQL server installed. All CATI data are saved in this database system for subsequent processing. Front-end applications work with the SQL server in a 2-tier client/server mode. This online approach allows easy management of CATI assignment, monitoring of progress and exchange of data.

2.4 CATI workstation

CATI workstations are normal PCs connected to LAN, no telephone card or modem cards are installed on them. The headphones for the CATI agent are connected to the telephone network, not on the PC.

In order to provide more working space on the desktop, colour monitors are replaced by flat LCD displays for all the CATI workstations. This is good for the CATI agents as they will have to drop down notes or deal with documents during telephone interview.

3.  The CATI Application Design

Since the uses of CATI are expected to be extended to several projects, so the primary design goal is to build a flexible system that can be easily plugged-in different surveys. Under this principal, a three-layers structure of application design is made. These three logical layers are 1) the CATI library, 2) the CATI application, and 3) the survey application.

The CATI Application Design

All programs on the client side are developed in Delphi 4.

3.1 The CATI Library

The CATI Library is a set of functions developed to handle the telephony functions. It acts as the interface between the application program and the telephone server. Upon receiving function calls from the client application, the library generates TCP/IP message to the telephone server to trigger the desired action.

Below are example of the library functions :

Function Description Parameter
DialOut  Make phone call Ext, TelephoneNo
Monitor  Silent monitoring of an ext. Ext, ExtToBeMonitor
HangUp  Hang up the phone Ext
GetStatus  Return to workstation status Ext
JoinConf  Join the conference group Ext,ConferenceGroupNo
LineStatus Return the status of the line LineNo
HoldLine Hold the line in wait status Ext,LineNo
SupHelp Alter supervisor for assistence Ext
3.2 The CATI Application

This is the basic module of a CATI application. Inside this module all CATI functions has been coded, only left an windows opened for plugging in new questionnaire.

Figure 3 - CATI Application Module
Figure 3 - CATI Application Module

The buttons on the right are the standard functions that a CATI application will have for any establishment related survey, these buttons are:

Next Case [F2] Get the next assignment from the queue
Dial [F4] Make the phone call
Appointment[F5] Make the appointment with the respondent 
Estab.Info.[F6] Information of the establishment for verification purpose
Questionnaire[F7] Bring up the questionnaire for interview
Result [F9] Fill the interview result
Playback [F10] Playback the pervious voice records of the interviews
Search [F11] Find a particular establishment
Save [F12] Save whatever have been changed
Exit Leave the program

This CATI application will only works with the supporting tables like the Assignment table, the Establishment master table, and the log table etc.. The handling of survey questionnaire will be the responsibility of the survey application.

3.3 The Survey Application

With the skeleton of the above mentioned CATI application, survey applications can be easily built. One great feature found in Delphi is called Form Inheritance, this feature allows the programmer to build a new form(screen) based on a previously developed one, it is not copy of the previous one, but inherit from the previous one. Copy and inherit make a significant difference in terms of programming: inherited form will carry the characteristic of the master form, if sometime has been changed or added to the master form, the inherited forms will automatically have these changes or new functions.

Form Inheritance is the way that the CATI application is built, all the CATI telephony functions are built into the master forms, and the survey applications inherit these functionality and implement their own business logic.

The following screen show a form inherited from the CATI master form.

igure 4 - Establishment verification form
Figure 4 - Establishment verification form
4.  The Vacancy and Remuneration Survey (D7TR)

D7TR is the first trial project running on CATI in DSEC. It is a quarterly survey to collect data of labour market with a sample size of about 800 establishments. In order to update the central establishment database, this survey also has the function of establishment verification. According to the new plan, all establishments will be interviewed for every 1.5 year, it means a total number of around 4000 establishment will be interviewed each quarter. As the size of the sample has been increased significantly, new method of data collection was studied. Finally, the CATI approach was chosen for this survey.

Figure 5 - Vacancy and Remuneration Survey
Figure 5 - Vacancy and Remuneration Survey
Figure 6 - CATI screen for the questionnaire of the Vacancy and Remuneration Survey
Figure 6 - CATI screen for the questionnaire of the Vacancy and Remuneration Survey

Paper questionnaires are first sent to the respondents, and a CATI follow-up is started three days after mailing to confirm the receipt of questionnaire. According to the size and characteristic of the establishment, different groups are defined in the CATI assignment list. Some large companies are not included in telephone interview as personal interview conducted by field agent is preferred.

Figure-6 shows the data entry screen of the project. The screen is basically identical to the paper questionnaire. Each row and column is marked with a clear identification number to help communication during telephone interview.

Since the CATI workstation has online access to the database, some enhanced validation function become feasible:

Online coding Retrieve classification code by filling in a part of the description, such as street code, ocupation code etc.
Range/Cross check Multiple range and cross checks are programmed into the data entry to minimise possible human errors.
Correspondence check Inputted data will be immediately validated with the data collected in last quarter, exaggerated difference will generate warning to the CATI agent for immediately clarification.
5.  Summary

Due to the very tight manpower for program development, the development of the CATI system for the phase I can only included those basic functions mentioned in the previous session.

In phase II, the following CATI functions will be addressed:

  • Incoming call queue
  • Incoming call transfer
  • Group conference
  • Supervisor monitoring
  • Performance statistics

It is foreseeable that CATI will become an important means for data collection in DSEC, and we are optimistic that this new technology will eventually improve the quality of our statistic production.

Census and Statistic Department of Macau
26 June 1999


 
Pop-IT project (1997-2001)
Project Objectives
Working Party Members
Working Party Meetings
First meeting, Bangkok, 24-26 September 1997
Second meeting, Singapore, 1-3 April 1998
Third meeting, Bali, 7-9 January 1999
Fourth meeting, Manila, 6-9 July 1999
Ffth meeting, Bangkok, 21 October 1999
Sixth meeting, Bangkok, 26 March 2001
Workshops
Application of New Information Technology to Population data, Bangkok, 12-20 October 1999
Population Data Analysis, Storage and Dissemination Technologies, Bangkok, 27-30 March 2001
Guidelines
Population data collection and capture (BBS - Statistics Indonesia)
GPS in modern mapping and GIS technologies to population data (Bangladesh Bureau of Statistics)
Population data dissemination (Statistics New Zealand)
Project Newsletter
Contact us
   
Copyright (c) 2013 ESCAP  |  Legal Notice