The National RCS and Election Reinterview studies share a common variable catalog, which lists all of the variables measured.

Variable catalog sections: The variable catalog includes two main sections:

• The variable directory provides an overview of variables and the studies and interviewing dates for which they were measured.
• The variable directory links to the detailed variable listing, which provides full information about variables, including question wording, response and other value options, filters and sample splits, and how variables are represented in the data files.

Subject categories: The variable catalog is grouped and ordered by subject categories. The order of variables in the catalog does not necessarily reflect the order in which questions were asked in interviews (see below).

The NAES08-Phone data files are provided in multiple configurations. Both SPSS (.sav) and ASCII tab-delimited formats (.txt) are provided.

The detailed variable listing indicates how each variable is represented in the data files (see below).

Some variables were measured for the full schedule of NAES08-Phone studies and interviewing dates, while others were measured for subsets of studies and/or interviewing dates.

The variable catalog indicates the schedule for each variable, as in the following example:

• The variable schedule is shown graphically to the right.
• In the schedule area, the National RCS schedule is shown first (Nat RCS), followed by the Election Reinterview schedule (R).
• A dark gray bar indicates that the variable was measured for the full study; light gray bars indicate that the variable was measured for a subset of interviewing dates. The absence of a bar indicates that the variable was not measured.
• For the National RCS, the size and position of schedule bars approximate the proportion of the RCS field period for which the variable was measured.

The detailed variable listing provides the wording of questions asked on the survey.

Variation by interviewing date: The wording of some questions changed as the survey was in progress:

• Revised questions: Substantive revisions in question wording are represented as separate variables (example: variable RCb04).
• Adjusted questions: Some questions were adjusted as campaign periods or events passed. These adjustments are indicated in the question text, with the dates of adjustment specified in braces (example: KA01).

The detailed variable listing enumerates the response and other value options recorded for each variable.

Response options read to the respondent: Any response options read to the respondent are included as part of the question text.

In some cases, additional response options were accepted if volunteered; these options are enumerated in the value option list but not included in the question text (example: variable CDb01). In other cases, no response options were read to the respondent, and this is reflected in the question text (example: EB03).

Whenever meaningful, response options were read in random order (see below).

Multiple-answer questions: Some variables represent questions that can have multiple answers. This is indicated in the detailed variable listing by "multiple answers accepted" (example: FAb29).

Verbatim data: In some cases, respondents' verbatim answers were recorded. The detailed variable listing identifies variables with verbatim answers (example: MA01).

Filters: Some variables have filters, meaning they were measured only for respondents who answered questions in particular ways or met other non-random criteria such as residency in a particular state. The detailed variable listing identifies variables with filters and describes the filter criteria (example: variable KB02).

Sample splits: Some variables have sample splits and were limited to random sub-samples of respondents. The detailed variable listing identifies variables with sample splits, and indicates the size and definition of the sub-sample (example: CBa04).

Variation by interviewing date: The detailed variable listing indicates filters and sample splits that changed as the survey was in progress (example: ABo05) and/or were not in effect for the full interviewing schedule (example: SC01).

Data column suffixes: In the data files, data columns representing cross-section variables have a _c suffix, and data columns representing reinterview variables have an _r suffix. Variable RKEY, which uniquely identifies respondents across interviewing components, has no suffix.

Data column layout:

• Single coded data: Each variable that is a single-answer question or records an observation is represented in the data files by a single numeric-code data column. The name of the data column corresponds with the name of the variable (plus a _c or _r suffix; see above), and the detailed variable listing enumerates the numeric codes used.
• Multi-part coded data: Variables that are multiple-answer questions are represented in the data files by a separate numeric-code data column for each possible answer. The detailed variable listing enumerates the names of the data columns corresponding with each answer, and the numeric codes used to indicate if each answer was or was not specified (example: variable FAb29).
• Verbatim data: Verbatim answers are recorded in alphanumeric data columns. The detailed variable listing indicates the numeric codes in the main variable data columns that correspond with verbatim answers (example: MA01). Some multiple-answer questions have verbatim data (example: KG03).

Null values: If a variable was not measured for a respondent because of scheduling or a filter or sample split, the corresponding numeric data column or columns have null (SPSS system-missing) values.

The questionnaires provide the interviewing scripts read to respondents. Questions are numbered sequentially, which indicates the order in which they were asked (not including any randomization; see below).

Tree-format questions: In some cases, questions with similar wording were asked in a tree format. Tree-format questions are indicated in the questionnaire with tree lines. The common wording is provided at the top of the tree, and branching lines identify wording specific to each question (example: questions Q11_c – Q14_c).

Order randomization: Whenever meaningful, the following question randomizations were applied in an effort to reduce response-set bias:

• Adjacent similar questions, including tree-format questions, were read in random order (example: question groups Q134_c – Q136_c and Q137_c – Q141_c were randomized, and questions within the groups were randomized).
• Categorical response options read to the respondent — including names of candidates and political figures, issue position options, and methods of political participation — were presented in random order (example: variable CCa03).
• The direction in which qualitative (but not quantitative) rating scales were presented was randomized (example: variable MA04).