Next: 6.1 The Local Datasets Up: GEIS Generalized Electronic Interviewing Previous: 5 GEIS libnames Contents

6 GEIS Data Sets

There are a number of data sets and catalogues within the DATACATI library as shown in Table 2. The CONFID data set is the only one created by the user. All others are created by GEIS.

**Table 2:** **GEIS** data sets and catalogues.
Name	Type	Content
CONFID	Data set	Respondent contact details
SCRIPT	Data set	Script questions and logic
ANSWERS	Data set	Answers given by respondents
CONTROL	Data Set	Interview control information
INTRVS	Data Set	Interviewer details
ILOG	Data Set	Interviewer log
PROTECT	Data Set	Interview protection statuses
COMBRESP	Data Set	Data Set used during interviews
CALLS	Data Set	Data Set to store chronological call data
DEFINIT	Catalogue	General project information

The CONFID Data Set

Before running GEIS, the CONFID data set must exist. It is usually created during the survey sample allocation and contains the respondent contact information, including the respondents' names, telephone numbers, and addresses. Deleting all occurrences of the CONFID data set after a survey has finished removes identifying information of respondents from a project.

Each row in the data set corresponds to a respondent. Some variables are mandatory, but it may also contain any number of other variables. The mandatory variables are shown in Figure 4. They must have appropriate values set before compilation. All the mandatory variables are copied from CONFID to the CONTROL data set when the script is compiled. This allows the values in the variable to be preset when the survey starts. Some further explanatory notes on the CONFID variables are given below.

**Figure 4:** *CONFID* structure.
$\begin{figure}\centering \begin{verbatim}-Alphabetic List of Variables an... ...3 STDPHONE Char 20 80 \newcommand{\SILENT} {\SILENT}}\end{verbatim} \end{figure}$

The variables in the CONFID data set are described below.

The ID variable must contain unique numeric identification numbers.
The FULLNAME variable contains the respondent's name.
The STDPHONE variable holds the respondent's telephone number. This is presented to the interviewer in a pull-down list².
The ELIGIBLE variable determines if a case is eligible for interviewing. It may be either zero or non-zero. If set to zero, the case is eligible for interviewing and appears in the pull-down list for interviewers to select. Any other value causes the case not to appear in the pull-down list.
The ELIGIBLE variable is meant to be used to remove individual respondents from the interview list for whatever reason. A local coding system may be used.
The SELECTED may only have the values zero or one. If a case is set to zero, the case does not appear in the pull-down list, and if set to one, it does appear.
The SELECTED variable is often used in programs that select or de-select cases for interviews.
The START variable contains a SAS date-time value representing the earliest date and time that the case may be interviewed. Before this date and time the case does not appear in the pull-down list. However, cases where START is set to missing may be interviewed. The variable is used to preset the date and time for large blocks of numbers to prevent premature interviewing.

An example of a CONFID data set is shown in Table 3.

**Table 3:** An example of a *CONFID* data set.
ID	FULLNAME	STDPHONE	ELIGIBLE	SELECTED	START
123456	George Eliot¹	02 43216789	0	1	1NOV2003
123457	Emily Brontë²	03 1234 5678	0	0	.
123458	Jane Austen³	04 9876 5432	33	1	.
123459	Mary Shelley⁴	06 1234 4321	0	1	.

¹ This case will not be interviewed until after November 1, 2003.

² This case will not be interviewed because SELECTED is set to zero.

³ This case will not be interviewed because ELIGIBLE is non-zero.

⁴ An interview will be attempted with this case.

The CONTROL Data Set

The CONTROL data set is created by GEIS. Important variables are shown in Figure 5. The CONTROL data set is used to control the conduct of interviews. Some variables are copied from the CONFID data set during compilation. Some further explanatory notes on the CONFID variables are given below.

**Figure 5:** Major variables in *CONTROL*.
$\begin{figure}\centering \begin{verbatim}Variable Type Len Format LabelDISTA... ...spondent telephone number SURVTYPE Char 6 Survey type\end{verbatim} \end{figure}$

The ID, STDPHONE, ELIGIBLE, SELECTED, START variables are copied from the CONFID data set, as described in Section 6.

The STATUS variable indicates the interviewing status of a case. It uses a two-character as defined by the $statfmt. format in the file \GEIS\BIN\FORMATS.SAS. Important status codes are shown in Table 4. A full list can be found in the file \GEIS\BIN\FORMATS.SAS. See Section 9.1.

**Table 4:** Selected status codes.
Code	Meaning
AM	Answering machine
CB	Callback arranged
CQ	Completed questionnaire
DO	Dropped part-way
DR	Dropped before starting
DT	Disconnected tone
ER	Error condition
ET	Engaged tone
FM	Fax machine
NA	No attempt made to contact
OS	Out of scope -- Ineligible
PQ	Partly-completed

The ANSWERS Data Set

The ANSWERS data set is created by GEIS. The exact structure depends on the script, but the first variable is always ID. The ANSWERS data set is used to store the responses to interview questions.

The PROTECT Data Set

The PROTECT data set is created by GEIS. The exact structure depends on the script, but the first variable is always ID. The PROTECT data set is used to store the protection statuses of items in the script.

The SCRIPT Data Set

The SCRIPT data set is created by GEIS when a script file is imported. There is no need to pay attention to the structure of this data set unless there is a need to write programs that directly use the script contents. The SCRIPT data set is used to store the script contents.

The CALLS Data Set

The CALLS data set is created by GEIS. The CALLS data set is used to store a chronological listing of all calls made during the project.

The ILOG Data Set

The ILOG data set is created by GEIS. The ILOG data set is used to store a chronological listing of most events that occur while GEIS is running. These include the results of importing and compiling scripts, as well as the outcomes of all interviews conducted by all interviewers.

The INTRVS Data Set

The INTRVS data set is created by GEIS. The INTRVS data set is used to store a monitoring information relating to interviewers, such as the total time logged in, the total time spent interviewing, and the interviewer's password. It is created by the compiler by copying the master copy in the library COMMON.

The COMBRESP Data Set

The COMBRESP data set is created by GEIS. It is created by merging the CONTROL and ANSWERS data sets. It is used as a template for a local data set when an interview starts.

The DEFINIT Catalogue

The DEFINIT catalogue is created by GEIS. It is used to store project settings and the compiled version of the script.

Subsections

6.1 The Local Datasets

Next: 6.1 The Local Datasets Up: GEIS Generalized Electronic Interviewing Previous: 5 GEIS libnames Contents

Ross Corkrey 2006-02-14