Skip to content
ConductExperience
Menu

Your data, in files you can read

Download a study as one zip of tables: a row for each person, an answers table for each questionnaire, a table for each test, and every event on the session clock. A data dictionary describes every column.

The export comes with files for SPSS, R and Python, and a script can read the same tables through the ConductExperience API with a key.

1,000

people in one export

with up to 600 MB of session files

15

API routes

every one of them a read

60

API calls a minute

4 at once

What the export holds

FileWhat it holdsWhen
participants.csvone row for each personIn every export
parts.csvthe parts of each versionIn every export
stimuli.csveach image and videoIn every export
events.csvevery event on the session clockIn every export
part_windows.csvwhen each part began and endedIn every export
scores.csvquestionnaire and test scoresIn every export
questionnaires/an answers table for each questionnaireIn every export
pages.csva row for each time a person was on a questionnaire page, with how long the page was shown and how many times they gave or changed an answer, but none for the page a person left the study onin a study whose sessions ask a questionnaire
clicks.csva row for each click of a click question, with its place on the picture, its time and the area it hitin a study whose sessions ask a click question
subsets.csva row for each draw a person had: the screens a questionnaire's set drew or the rows its list drew, in the order shown, or the parts a section drewin a study with a set of screens at random, a screen that repeats for rows of a list drawn at random, or a section that draws its parts
tests/a table for each testIn every export
face/face and gaze samplesunless you leave them out
stimuli/the study's own images and videoswhen you ask for them
redcap/files for your own REDCap projectwhen you ask for them
uploads.csveach file people added to a file question, with its type and sizein a study with file questions
uploads/the files people addedin a study with file questions, up to the export's limit
visits.csv, sessions.csveach visit and its sessionin a study with visits
checkins.csva row for every check-in, planned or startedin a study with check-ins
checkin_attempts.csvevery reminder sentin a study with check-ins, unless you leave it out
checkin_answers/the answers to each check-in questionnairein a study with check-ins
checkin_uploads.csveach file people added to a check-in's file question, with its type and sizein a study whose check-ins ask file questions
checkins/the files people added to check-ins, in a folder for each personin a study whose check-ins ask file questions
daily_people.csva check-in summary for each personin a study with check-ins
data_dictionary.csva row for every columnIn every export
analysis/SPSS syntax, and readers for R and PythonIn every export
study/each version of the study, as it was publishedIn every export
export.jsona checksum for every fileIn every export

Attention checks and same answers

Beside pastes, hidden and fast, participants.csv has these two columns, in the data dictionary's words:

ColumnWhat it holds
checks_missedHow many checks the person answered otherwise than a careful person would: questions or grid rows marked Check in the questionnaire. One that rules hid, or that was left blank, is not missed. In a study with later visits, the first visit's. Worked out when the data is downloaded.
same_answersHow many times the person gave one answer throughout: in every row of a grid of 5 rows or more, or in 10 rating questions or more in a row. In a study with later visits, the first visit's. Worked out when the data is downloaded. You can change both numbers on the study's page.

In a study with visits, sessions.csv has both columns for each visit's session:

Column in sessions.csvWhat it holds
checks_missedHow many checks the person answered otherwise than a careful person would in this session: questions or grid rows marked Check in the questionnaire. One that rules hid, or that was left blank, is not missed. Worked out when the data is downloaded.
same_answersHow many times the person gave one answer throughout in this session: in every row of a grid of 5 rows or more, or in 10 rating questions or more in a row. Worked out when the data is downloaded. You can change both numbers on the study's page.

Values from the link, the group, a formula or the round before

participants.csv has a column for each value the study takes from the link, named link_ and the value's name, sorted by name, after its own columns and before the panel's: the value the link page kept when the person started. Each answers table, a check-in's too, has a column for each value after its questions, under the value's own name. For a value from the link named arm, a worked-out value total, a value from the group named cohort and a value from the round before named q1_median, the data dictionary says:

Column of a valueWhat it holdsA blank cell means
link_arm in participants.csvA value from the link the person opened: arm (?arm=), as the link page kept it when they started.The link did not carry it, or the link page could not keep it.
arm in an answers tableFrom the link: ?arm=The link did not carry this value, or the link page could not keep it.
total in an answers tableWorked out: q1 + q2An answer the value reads is missing.
cohort in an answers tableThe person's groupThe person has no group.
q1_median in an answers tableFrom the round before: the median of q1Not rated last round, or this is the first round.

In a study with rounds, an answers table has a row for each round a person took, named in its visit columns, and a value from the round before holds the figure the person was given for that round: round 2's rows hold round 1's figures, frozen when you closed round 1, and round 1's rows are blank. Each closed round's figures, the last round's too, are in the round report on the Participants page.

Repeats, lists, options offered and draws

A repeated question has a column for each option it can repeat for, or each row of a list, named by the question and the option's or the row's number. A cell is blank for an option the person did not choose or a row not drawn for them. A set of screens drawn at random, or a list a screen repeats for that shows each person some of its rows, adds a column that names the screens or rows drawn. A question that offers only some options keeps its own columns, and a cell is also blank where the earlier answer it carries from left that option out. A check-in's answers table has the same columns. For a questionnaire whose question often repeats for each fruit chosen in fruits, whose question best offers only the fruits chosen, and whose set stimuli shows each person 1 of 2 screens, the data dictionary says:

Column in an answers tableWhat it holdsA blank cell means
often__1How often do you eat {{fruits|it}}?: AppleThe option was not chosen, so the screen did not repeat for it; or it was not answered, hidden by a rule, or the questionnaire was ended early.
bestWhich do you like best? (offers the options chosen in fruits)Not offered, since the earlier answer it carries from left it out; or not answered, hidden by a rule, or the questionnaire was ended early.
stimuli1 of these at random: the screens drawn for this person, by the page break that starts each, in the order shown.The answers hold no valid draw of this set.

For a questionnaire whose list juices shows each person 2 of its 3 rows at random, and whose question taste repeats for each juice drawn, the data dictionary says, naming each row by its first field:

Column for a listWhat it holdsA blank cell means
juices2 of these rows at random: the rows of the list juices drawn for this person, in the order shown.The answers hold no valid draw of this list.
taste__3How do you rate {{juice|it}} by {{maker|its maker}}?: PlumThe row was not drawn for this person, so the screen did not repeat for it; or it was not answered, hidden by a rule, or the questionnaire was ended early.

subsets.csv has a row for each draw in a person's sessions: each section that showed them some of its parts, and each set of screens or list a questionnaire drew. After participant_code and participant_status, its columns are:

Column in subsets.csvWhat it holdsA blank cell means
step_idThe part that asked the questionnaire. It is a row of parts.csv.The row is a section's draw of its parts, or the questionnaire was not asked by a study part.
questionnaire_idThe questionnaire that holds the set or the list, by its ID in the version. A part's follow-up questionnaires share its step ID; this tells their sets and lists apart.The row is a section's draw of its parts.
setA questionnaire's random set of screens, by the set's key; a list that shows each person some of its rows, by the list's key; or a section that shows some of its parts at random, by the section's ID.Never blank.
drawnWhat was drawn for the person, joined by ;. For a set of screens, the page break that starts each group of screens drawn, in the order the person saw them. For a list, the name of each row drawn, in the order the person saw them. For a section, the step ID of each part drawn, in the section's order; each is a row of parts.csv.Never blank.

The data dictionary

data_dictionary.csv is the study's codebook: it has a row for every column of every table, with these columns:

ColumnWhat it says
fileThe table the column is in.
columnThe column's name.
typeWhat it holds, such as text, a whole number, a day, a time, or true or false.
unitIts unit, such as ms, seconds or degrees of visual angle.
allowed_valuesThe values it can take.
meaningWhat it means, in a sentence.
blank_meansWhat an empty cell means, such as "The camera was off."

Files for analysis

Every export has an analysis/ folder for SPSS, R and Python, written from its own data dictionary:

  • An SPSS syntax file for each table, with variable labels, value labels and missing values.
  • read_study.R, in base R, and read_study.py, with Python's standard library, or DataFrames when pandas is installed.
  • They pair columns by name, and stop with a plain message when a table and its dictionary disagree.
  • SPSS reads a line as a row, so it reads its own copies of the tables whose written answers hold line breaks, with each break made a space. The main tables keep their line breaks.

What you choose to include

Choice on the Data screenStartsWhat it adds
Include face and gaze dataOnThe face/ folder, one file for each person.
Include panel IDsOnEach panel's participant ID, in participants.csv.
Include images and videosOffThe study's own images and videos, in stimuli/.
Include pilot peopleOffThe people from pilot links and pilot lab sessions.
Include send attemptsOnEvery reminder sent, in a study with check-ins.
Add files for REDCapOffThe redcap/ folder, in the layout you choose.
What to include, on a study's Data screen: face and gaze data and panel IDs ticked, pilot people and images and videos not, and version 1 with 4 people
What to include, on a study's Data screen. Example data.

Checks on the files

  • export.json holds a checksum for every other file, and the versions of the code that made them.
  • Text that a spreadsheet would run as a formula stays text.
  • A person whose files cannot be read is left out, with a line that names them.
  • study/ holds each version of the study as it was published, so the data names the exact design each person ran.

The ConductExperience API

  • Make a key on the Integrations page, for all your studies or up to 100 of them.
  • A key lasts 1, 3 or 6 months, 1 year or 2 years, and you can have 10 in use. You can revoke a key at any time.
  • Up to 60 calls a minute and 4 at once. A table comes in pages of up to 500 people.
  • Every table is byte for byte the export's table. The API reads your data and never changes it.
  • The Integrations guide has an example script in Python, with the standard library only.
Route, under /api/v1What it gives
GET /openapi.jsonDescribe this API
GET /meDescribe this key
GET /studiesList studies
GET /studies/{id}Get a study
GET /studies/{id}/versions/{n}Get a published version
GET /studies/{id}/participantsList the people in a study
GET /studies/{id}/participants/{code}/planGet a person's check-in plan
GET /studies/{id}/sessionsList session files
GET /studies/{id}/sessions/{session}Summarize a session file
GET /studies/{id}/sessions/{session}/fileGet a link to a whole session file
GET /studies/{id}/checkinsList check-ins
GET /studies/{id}/checkins/{code}/{occurrence}/answersGet the answers to one check-in
GET /studies/{id}/tablesList the data tables of a study
GET /studies/{id}/tables/{name}Get a data table
GET /studies/{id}/exportsList earlier exports

Where your data is kept

  • Study data is kept in Microsoft Azure in the United States, in the account of the researcher who made the study.
  • Only the people on the study's team can open it, each with the role you give them.
  • Video never leaves the computer that runs a session. Only the face and gaze measures are saved.
  • A study or session you delete is kept for 7 days in case of a mistake, then removed. Deleting a study removes its people's check-in data too. Backup copies of deleted data are removed within 30 days.

Limits

  • One export holds up to 1,000 people, 600 MB of session files and 1,024 MB of images and videos.
  • When the files people added come to more than 1,024 MB, the export keeps every table, leaves the files out and says so. They stay in your account's storage.
  • The API has no events, face or subsets table. Download the export for those.

Related

Planning a study?

Start one now, or tell us what you want to measure.