Item Text

About these data

IRW item text is reconstructed from published articles, codebooks, and supplementary materials using a largely automated pipeline with partial human review. It is provided for research purposes only. We make no guarantee that the text is accurate, complete, or correctly aligned with the item identifiers in the corresponding IRW response data; the original source document is always authoritative. Users are responsible for verifying item text against that source before drawing substantive conclusions. See Known Issues for documented discrepancies, and please get in touch if you find others.

Note also that the licenses recorded for IRW response data do not extend to the instruments themselves. Inclusion of item text here implies no license to reuse an instrument; copyright and related rights remain with the original rights holders, and users are responsible for obtaining any permissions their intended use requires. The exception is a table whose wording its source released under an open licence (for example CC BY-SA 4.0): that table’s description on Redivis states the licence, the source and IRW’s changes, and that licence applies to the wording.

The standard for item information

We have begun to incorporate the text of the items used to produce stimuli in the IRW. This information is available here. A list of tables with item text can be generated using the irw_list_itemtext_tables() function (via the irw package).

These data are structured as follows:

table This identifier can be used to link to the IRW response data.
section_id This is an identifier for a group of items that share a common context. It is conceptually equivalent to the item_family element in the IRW standard and is used to annotate information about, for example, testlets (i.e., items grouped under a single reading passage).
item This is a persistent identifier for the probe being used to measure, matching the item field in the core IRW dataset.
instrument This is the full, human-readable name or title for the instrument identified by the table column.
language This names the language in which the instrument was administered (e.g. German, Spanish), written plainly rather than as a language code. It is present when the administered language is not English.
instructions This contains the literal text of the instructions provided to the participant for the overall instrument.
section_prompt This contains the literal text of a shared prompt, such as a reading passage, that applies to all items within a given section_id.
instructions and section_prompt are scoped differently: instructions applies to the entire table regardless of section_id; section_prompt applies only to the items sharing one or more specific section_id values. The same span of source text should never be recorded in both fields. If framing or task-level text applies across the whole table, record it once in instructions. If it is specific to a subset of items sharing a section_id (e.g. a passage or context given before a testlet), record it in section_prompt only, even if it superficially resembles instructional language.
item_text This is the literal text of the specific prompt or question associated with an item.
correct_response This is the scoring key for a given item. This field is left blank for items that do not have a correct response. For items with multiple correct answers, values are separated by a semicolon.
option_text This is the literal text for a specific response option available for an item.
resp This is the response value assigned to a specific option_text. It should match the numeric, ordinal resp value found in the response-level IRW dataset.
instructions_translated, section_prompt_translated, item_text_translated and option_text_translated These hold the English translation of the correspondingly named field. They are present only for instruments administered in a language other than English.

Instruments administered in a language other than English

Where a study was administered in a language other than English, the text fields (instructions, section_prompt, item_text, option_text) hold the wording that respondents actually read, verbatim, and language names that language. The English rendering is carried alongside in the parallel _translated fields, so the instrument is usable without reading the administered language. Where the study or its data deposit supplies its own English version, that is the translation we ship, and the source is recorded in the provenance for that table.

item and resp are join keys into the response-level IRW dataset and are never translated. For instruments administered in English, language and the _translated fields are omitted entirely rather than carried empty.

When only an English version survives

Some studies administered an instrument in another language but published only an English rendering of it — sometimes the authors’ own translation in a supplementary file, sometimes English variable labels in the deposited data file, sometimes nothing but the canonical English instrument the study cites. Where the administered wording cannot be recovered from the data deposit or the paper’s own supplements, the English is placed in the base text fields, the _translated fields are left empty, and the table’s provenance records text_source = translated_substitute.

language still names the language the study was administered in. It describes the study, not the contents of item_text, so a table in this situation reads language = Chinese while item_text is English. That combination is deliberate and it is the flag to read:

`language`	`item_text_translated`	what `item_text` holds
empty	empty	the administered wording; the study was in English
populated	populated	the administered wording, verbatim
populated	empty	English — not the wording respondents read

Any table in the third row also gets an entry on the item text issues page saying so in plain words. If you are analysing item wording rather than joining on item, treat those tables with the same care you would give any translation: the response data is unaffected, but the text is a rendering of the instrument rather than a transcript of what was administered.

Item text extracted before this schema was introduced (2026-09-01) predates the language and _translated columns, so a non-English administration from that period may carry English text with no language value. Those tables are being revisited; the issues page is the reliable record in the meantime.

Some additional notes about the item text

In some cases, we do not have ready access to the ‘key’ which would allow us to translate multiple choice options into scored responses. The item response data will have a scored response in the resp column in such cases while the item text will retain the categorical option in the resp_raw column.
When items are prompting some kind of behavior that gets scored, there may not always be text associated with individual resp values.