TEDS Data Dictionary

Omitted variables

The table omits variables that were included in the CMS and backup raw data files but have not been included in the dataset. These include the following:

• CMS variables with names ending in "Done" for all items (coded true/false). These are redundant, simply denoting whether or not each item was attempted; and they are absent from the backup and paper data.
• CMS variables with names ending in "Elapsed time" for all items (number of milliseconds). These recorded the response time, in milliseconds, for each item. These have been omitted because they are absent in the backup and paper data, and they would have doubled the number of item variables. Before being dropped, they were used to derive new variables to measure the time taken for each questionnaire section/theme.
• Admin-related items (CMS and backup) containing personal data such as names, address fields, email address, phone numbers and reward preferences.
• Admin-related items (all versions) containing CoTEDS personal information about children of twins and twin pregnancies.
• Derived scores (CMS and backup) used to present feedback in the app. These will be replaced in the dataset with conventional derived scales.
• Questionnaire start and end date-times (CMS and backup). These are dropped from the dataset after using them to derive twin ages and questionnaire durations.
• Various redundant items that do not provide useful information about twins or their responses. These include redundant ID/username fields and variables describing the structure of the questionnaire itself.
• Variables giving detailed information about browser version and language, operating system version and device model name; these contain many unique values that could potentially identify participants, and they are of doubtful value in analysis (also, they specific to the CMS or web data collections).
• Variables describing the status or level of completion of each section/theme in the questionnaire. These have been replaced with derived "status" variables with more conventional coding.
• In the backup file, the date-time of the start and end of each section/theme of the questionnaire. These are omitted because they are absent from the CMS and paper versions. For the backup version, before being dropped, they were used to derive new variables to measure the time taken for each questionnaire section/theme.

Coding changes

Changes in coding from raw data sources to dataset have been designed to ensure consistency between the three versions, and to conform to TEDS coding conventions. Coding conventions used in TEDS datasets include the use of 1/0 coding for yes/no responses, and 0, 1, 2, ... coding for response categories starting with "no", "none" or "never" or similar responses. Where reversed versions of items are needed in the dataset to create scales, the variable is included in its conventional (forward-coded) with the addition of a separate reverse-coded variable; the reverse-coded variables are not generally tabulated below, unless the raw variable itself was reverse-coded. Items with explicitly coded "prefer not to answer" or "not applicable" types of responses have been recoded to missing, generally because they were not used consistently in all versions and so are typically missing in the paper data, for example. In the CMS, optional questions with a "prefer not to answer" label generally produced a missing value in the data.

Table of variables

The table columns are as follows:

• Dataset variable name
• Order number of the variable in the CMS raw data file
• Name of variable in the CMS raw data file
• Order number of the variable in the backup raw data file
• Name of variable in the backup raw data file
• Order number of the variable in the scanned paper raw data file
• Coding used for the dataset variable
• Recoding, if needed, from the three raw data versions

Variables are presented approximately in the order in which they appeared in the questionnaire battery (and in the paper questionnaire).