1 Introduction

This instruction manual discusses the Generalized Electronic Interviewing System (GEIS) with respect to the system requirements, installation, project creation, usage, and maintenance. GEIS is a program for creating and running electronic interviews without the need for a programmed interface.

It is released under the terms of the GNU GENERAL PUBLIC LICENSE described at http://www.gnu.org/copyleft/gpl.html. A copy of the license is included with the program.

The advantages that GEIS may provide are:

• Interviews may be conducted within GEIS using a number of modes, currently including CATI (Computer-Assisted Telephone Interview) and IVR (Interactive Voice Response).

• Questionnaires are scripted using a common script syntax in all supported modes.

• Rapid interview development is achievable without the need for a professional programmer.

• Development of all aspects of a survey is done by the person most likely to understand the survey objectives -- the survey designer.

• The GEIS script contains all the information needed to define the interview, including:
  1. all question texts including variable-texts;
  2. all answer-option texts and codes;
  3. all range check information;
  4. all logical question-branching information;
  5. all interview termination status information;
  6. all DO-loop information; and
  7. all IVR information, including key codes and voice message information.

• The GEIS script also provides a standard, non-proprietorial, open method of project documentation. This means that comparisons between survey projects can be achieved by publishing scripts.

• The script is thoroughly checked by GEIS for logical or syntactical errors.

• Since the GEIS interface is entirely specified by scripting, programming resources can be redistributed to more productive areas, such as reporting, quality control, and system integration.

• The survey data are accumulated within SAS data sets¹.

• The data may be analysed immediately they are collected.

• GEIS can handle very large data sets and any number of interviewers.

• Interview responsiveness is not affected by questionnaire length nor survey size.