Studio Question Types

Question Types

Studio has nine question types and eight presets — a preset is one of the types, pre-configured the way "NPS" is really an 11-point scale. Every type maps onto a class of the siamang engine, so every question you build exports to questionnaire.py, to a labeled .sav, and into the analysis nodes. This page describes each type: what the respondent sees, every option in the Inspector, how the answer is coded, and the rules that apply.

For the Builder screen itself see The Builder; for the codebook side of each question see Codebook and Variables.

---

The types at a glance

+ Question menu label	What the respondent does	Variables written	Scale
Single choice	picks one option	1	nominal
Multiple choice	picks any number of options	1, or 1 per choice in the wide layout	nominal
Likert scale	picks a point on a numbered scale or stars	1	ordinal
Number	types a number or moves a slider	1	ratio
Open text	types text, an email, a phone number, a web address, a date or a time	1	nominal (text)
Matrix	rates several statements on one shared scale	1 per row	ordinal
Ranking	puts items in order	1	ordinal
Best–worst (MaxDiff)	picks the best and the worst of a few items, several times	2 per task + 1	nominal
Conjoint (choice)	picks one of several products, several times	1 per task + 1	nominal

Presets: Yes / No, Rating (stars), NPS (0–10), CES (1–7), Attention check, Date, Email, Phone — see Presets.

---

Fields every question has

Field	Where	Notes
Question text	Question	Shown as plain text: Markdown and HTML are not rendered. Piping works: {answer:variable} inserts an earlier answer, {label:variable} the chosen option's label (for an Other answer, the text the respondent typed).
Hint	Question	Smaller text under the question ("optional guidance shown below the question"); piping works here too.
Required	Question	The respondent cannot continue without answering: Next shows "This question requires an answer." A Matrix asks for an answer in every row (the checkbox's hint says "an answer in every row", or "an answer in every row — N/A counts" when the matrix offers N/A) — see Matrix; MaxDiff and Conjoint ask for every task. Required questions show an asterisk to respondents. The message also appears when the focus leaves a required question that has no answer yet, but not while it moves between the question's own controls (a dropdown's button and its search box, one checkbox and the next, a choice and its Other box). New questions start optional.
Randomize option order	Question	Shuffles the options per respondent. Single choice, Multiple choice and Ranking only. "None of the above", exclusive choices (such as "None of these") and a choice that is the question's Other keep their place; the other options are shuffled around them. The "Other" option Studio adds always comes last.
Attention check	Question	Single choice, Likert scale, Number and Open text — see Attention checks.
Variable	Variable	The codebook entry the question writes — Codebook and Variables. A variable name starting with __ is flagged here ("Variables starting with “__” are the survey runtime’s own: it never submits __x, so these answers never reach the data. Rename the variable."): the survey never submits such a key.
Show if / Hide if / Skip to	Logic	Logic and Branching.
Id	Advanced	The question's own handle ("names the question in scripts and comments; logic and data use the variable"): scripts target it, comments hang on it, checks quote it. Letters, digits and _ (other characters turn into _). Must be non-empty, unique, and not the variable name of another question — the field says "Another question already has this id." or "This is the variable … stores its answer under — the engine refuses an id that is another question’s variable." An Id that differs from the question's variable may not start with __, nor be a name the survey stores something else under (a Matrix row, another question's Other text, an assigned arm, a codebook variable a custom script writes): the field says why, for example "This is where brand stores the text typed into Other — scripts naming it would reach this question’s variable instead." An Id that is the question's own variable is never flagged for this. See Names an Id may not take.
Tags	Advanced	Free labels, comma-separated, for your own organization of the instrument.
Media URL	Advanced	An image, or a video if the link ends in .mp4/.webm, shown with the question ("a public https:// address; a file under Files has no public link"). Use a stable public URL (a download link from Files expires after 5 minutes).

To respondents, questions are numbered Q01, Q02, …

Note — Id and variable name. The answer is stored under the question's variable name: that is its column in your data, and the name conditions, piping and quotas use. The Id may differ from it — a preset starts with Id q5 and variable nps_5 — as long as no question's Id is another question's variable name or, for an Id that differs from its variable, another name the survey stores something under. Details: Question Id and variable name.

The options editor

Single choice, Multiple choice, Ranking, MaxDiff items and conjoint levels all use the same list editor ("code · label"):

• each row has a Code (what is stored), a Label (what is shown), Move up, Move down and Remove option;
• + Option adds "Option N" with the next free code (the highest numeric code + 1);
• Enter in a label adds a blank option under it;
• a code that looks like a number is stored as a number (3), anything else as text (dk).

Codes must be unique and every label must have text — an empty label or a duplicate code makes the questionnaire impossible to save (see What the engine refuses).

Options from the codebook. A Single choice, Multiple choice, Ranking or MaxDiff question does not need choices of its own. Without any, respondents see its variable's value labels (for a MaxDiff, those of its first variable), all of them, including a missing code that has a value label. Imported questionnaires are often built this way. The card sketches those labels under "From the variable’s value labels", and the editor lists them with their codes under "From the value labels of <variable> — the question has no choices of its own, so respondents see these. Any edit here makes them its choices, and the codebook follows." Your first edit, + Option included, copies them into the question's choices with the same codes and then makes the change, so the value labels follow the choices code by code from the first edit on: an option you remove takes its value label with it. Exclusive choices and the assistant's Suggest options work on the same options.

A wide Multiple choice shows its choices only when it has one per variable. Otherwise it offers one option per variable, labeled with that variable's label (or its name), and the card and the editor say "One option per variable, labeled with its variable label". Your first edit there makes those options its choices, one per variable, so + Option adds a variable and removing an option drops its variable. With neither choices nor value labels, respondents see no options: the card reads "No answer options yet. Add choices in Options." and Validation → Structure lists "q1: no choices".

---

Single choice

One answer from a list.

The respondent sees radio buttons, a dropdown or a row of buttons. The dropdown is searchable: it shows "— Select —", a search box ("Type to search…") and "No options found" when nothing matches. It works from the keyboard too: Enter, Space or ↓ opens it, ↓ / ↑ move through the options, typing narrows them, Enter chooses, and Esc or Tab closes the list, however long it is. Esc in the open list only closes it; it does not go back a page (see Keyboard and touch). After a click on a radio button or a picture, Enter goes to the next page with the choice kept, while Space only checks the radio.

Inspector → Options

Option	Values	Default
Display	radio · dropdown · buttons	radio
Choices	code · label list	1 Option 1, 2 Option 2
Add “None of the above”	on/off; when on, the hint says what it stores ("stored as -77")	off
Add “Other (please specify)”	on/off; when on, the hint says what it stores and where the text goes ("stored as -66; the text goes to q1_other")	off

Coding. One variable, nominal, with the choices as value labels. The answer is the chosen code. Adding, removing or relabeling a choice updates that code's value label; a label you wrote for another code in the Codebook tab stays.

• None of the above adds an option "None of the above" after the choices, stored as the code -77 and labeled -77 = None of the above in the codebook.
• Other (please specify) adds an option "Other" at the very end; choosing it opens a text box ("Please specify..."). The answer is the code -66, labeled -66 = Other, and the typed text goes to a column of its own, <variable>_other. With Display = dropdown, Other is the last entry of the list and its text box appears under the dropdown.

How these codes are chosen, and what happens when a choice already uses one, is in Codes for Other, None of the above and N/A.

Rules. At least one choice; unique codes; non-empty labels. With Other on, the text column <variable>_other must not be a name a question already stores an answer under. The Builder says so before you save, under Add “Other (please specify)” ("q9 already stores an answer under q1_other, and the engine refuses that — rename this question's variable or q9's.") and in Validation → Structure ("q1: stores its “Other (please specify)” text under "q1_other", which q9 already stores an answer under, and the engine refuses that — rename this question's variable or q9's"). Saved anyway, the Save is marked errors: "Question 'q1' stores its “Other (please specify)” text under 'q1_other', which question 'q9' already stores an answer under." Renaming this question's variable gives its Other text a column of its own under the new name and leaves q1_other with the question that stores it.

---

Multiple choice

Any number of answers.

The respondent sees checkboxes. Picking an exclusive option clears the others, and picking any other option clears an exclusive one — in both data layouts. With Max answers set, further boxes are disabled once the limit is reached and a counter shows "2 of 3 selected" / "Maximum reached".

Inspector → Options

Option	Values	Default
Choices	code · label list	1–3 Option 1–3
Min answers	a whole number, 0 or more	empty
Max answers	a whole number; "empty = no limit"	empty
Exclusive choices	a chip per choice; "picking one clears the rest"	none
Data layout	array · wide ("array: one column of codes · wide: one 0/1 variable per choice")	array
Add “Other (please specify)”	on/off; when on, the hint says what it stores and where the text goes ("stored as -66; the text goes to q7_other")	off

Min answers (2 or more) is checked when the respondent clicks Next: someone who picked some options but too few sees "Select at least 1 more" under the question, and Next does not move on until they pick enough. An exclusive answer ("None of these") is a complete answer on its own — it satisfies the minimum, and the hint disappears once it is picked. Min answers does not make the question required: an optional question can still be left empty, and a required one asks for an answer first ("This question requires an answer."). Under the options, the counter hint "Select at least N more" appears once the question is answered, or at once when it is required; it also shows without Max answers. A Min answers of 1 is not checked on Next — to require an answer, turn on Required.

Coding — array layout (the default). One variable, nominal, with the choices as value labels. The answer is the list of chosen codes. In CSV, Excel, SPSS and Stata exports it is written as codes separated by semicolons (1;3); Parquet keeps a list. Frequency tables of such a question use respondents as the base, so the shares add up to more than 100 %. In conditions, use chose / did not choose — = compares the whole list. A quota on the variable counts a completed response in the cell of every value it chose. With Other (please specify), the list holds Other's code (-66) beside the others and the typed text goes to <variable>_other.

Coding — wide layout. Clicking wide replaces the question's one variable with one variable per choice, named after the old variable and the choice's code (q7 becomes q7_1, q7_2, q7_3; _2 is added to a name that is taken). Each is nominal, coded 0 No / 1 Yes, and takes the choice's label as its variable label. Studio keeps them in step with the choices: a new choice gets a new variable, a removed choice's variable leaves the codebook, and relabeling a choice relabels its variable, unless you wrote that variable's label yourself. Clicking array collapses them back into one variable (q7) with the choices as value labels. A Multiple choice that takes its options from its variable's value labels (no Choices of its own) gets those labels as its choices when you click wide, one variable per option (aware with the codes 1, 2, 3 and 99 becomes aware_1, aware_2, aware_3 and aware_99), and its exclusive options stay exclusive. A wide question that came from a template or an import without one choice per variable offers one option per variable. Your first edit in Options makes those its choices, one per variable, so + Option adds a variable and removing an option drops its variable. Before any edit, switching it to array replaces them with a single variable and leaves the question without choices until you add some.

In the data, each per-choice variable holds:

Value	Meaning
1	chosen
0	offered and not chosen (the respondent answered the question)
empty	the question was not answered, or this option was hidden from the respondent by its own Show if / Hide if

Whether an option left unchosen is 0 or empty is decided again after every answer, whatever gives it — a click, a rating-scale digit key, a script, a resumed interview. An answer given later, on the same page say, that offers the option makes it 0, and one that hides it clears it. An option whose condition reads another wide question's variable (Show if aware_globex = 0) follows that variable as it is now: when the other question's option is hidden in turn and cleared, this one is cleared too.

Conditions, piping and quotas on a per-choice variable (q7_1 = 1) see the answer, and Exclusive choices work. With Other (please specify), Other has no 0/1 variable of its own: the typed text goes to <Id>_other, named after the question's Id (q7_other), which holds the text while Other is chosen. When a question already stores an answer under that name, or one of this question's own options does, the Builder flags it under Add “Other (please specify)” and in Validation → Structure — ending "…rename q9's variable" (or "…rename that option's variable") — and a Save that keeps it is marked errors. When an analysis of an array question needs one 0/1 column per option (weighting, regression, TURF), you can also add an Explode multiple choice node in the flow — see Node Reference.

Rules. Min answers 0 or more; Max answers at least Min answers ("max_answers must be >= min_answers"); in the wide layout, Max answers no more than the number of per-choice variables ("max_answers cannot exceed number of wide-mode variables"); exclusive codes should be among the choices (otherwise the warning EXCLUSIVE_CODE_UNKNOWN, array layout only).

---

Likert scale

A numbered scale with anchored ends — agreement, satisfaction, likelihood.

The respondent sees a row of numbered buttons (or stars that light up to the one chosen), the end labels under the scale and, if offered, a Not applicable choice. Outside a text field, a digit key 1–9 picks that point on the page's first scale that has it, exactly as a click does (the answer is saved and the question's error message goes). On a scale that starts at 0 the key is the point's number, and a digit the scale does not have (5 on a 0–4 scale) does nothing.

Inspector → Answer

Option	Values	Default
Points	2–11	5
Left label / Right label	text	Strongly disagree / Strongly agree
Offer “N/A”	on/off; when on, the hint says what N/A stores ("stored as -1, a missing code")	off
Display	Numbers · Stars	Numbers
First point	1 · 0 ("0-based: an NPS scale is 0–10 (11 points)")	1

Coding. One variable, ordinal. The answer is the point's number (1–5, or 0–10 with First point 0). Value labels are generated as 1 — Strongly disagree, 2, 3, 4, 5 — Strongly agree. A label you write in the Codebook tab for a point — 3 = Neither agree nor disagree — stays when you edit the question, until an edit changes that point's own label (a new Right label relabels the top point, and only it).

Not applicable is stored as a declared missing code: turning on Offer “N/A” adds -1 = Not applicable to the variable's value labels and declares -1 as its not-applicable missing code, so means and percentages leave it out. If the codebook declares no not-applicable code — a question set up before this was automatic and not edited since, or an import — N/A is stored as the text na instead (hint: "stored as the text “na” — the codebook declares no not-applicable code") and the Save shows the warning NA_STORED_AS_TEXT. See Codes for Other, None of the above and N/A.

Rules. The variable should be ordinal (otherwise the error-level lint INCOMPATIBLE_QUESTION_SCALE); when the number of value labels differs from the number of points you get LIKERT_POINTS_LABEL_MISMATCH.

---

Number

A number: age, hours, income, a count. Menu label Number.

The respondent sees a number box with the unit beside it, or a slider with the current value above it. The slider's range is the variable's valid range (0–10 when none is set); its handle starts in the middle, and nothing is recorded until the respondent moves it.

Inspector → Answer

Option	Values	Default
Display	input · slider	input
Unit	text shown next to the field (placeholder years)	none
Step	a number greater than 0	empty (1)

Inspector → Variable → Min / Max (shown when the scale is ratio or interval) set the valid range, which becomes the input's limits and the slider's range.

Coding. One variable, ratio, stored as a number.

The range is enforced. A typed number below Min or above Max shows "Minimum value is 18" or "Maximum value is 99" under the question as soon as the respondent leaves the field, and again on Next, which does not move on until the number is corrected. The slider cannot leave the range. An empty optional field is not checked. Both messages can be reworded in Theme → Wording ("Number below its minimum", "Number above its maximum").

Rules. Step greater than 0; Min not greater than Max; the variable must be interval or ratio (otherwise INCOMPATIBLE_QUESTION_SCALE).

---

Open text

Free text, and the typed inputs built on it.

The respondent sees a one-line box, or a four-line box for Multi-line. With a format, the browser shows the matching input (an email keyboard, a phone keypad, a date picker, a time picker). With Max characters, a counter ("12 / 400") appears, and "N characters remaining" once 80 % is used.

Inspector → Answer

Option	Values	Default
Format	Free text · Email · Phone · Web address · Date · Time	Free text
Multi-line	on/off — available only with Free text	off
Max characters	1 or more	none
Placeholder	text shown in the empty box	none

What the respondent must type (checked on Next):

Format	Accepted	Message
Email	name@domain.xx	"Please enter a valid email address."
Phone	an optional +, a digit, then at least five more digits, spaces, dots, dashes or brackets	"Please enter a valid phone number."
Web address	starts with http:// or https://	"Please enter a valid web address (https://…)."
Date	YYYY-MM-DD (what the date picker produces)	"Please enter a valid date."
Time	HH:MM or HH:MM:SS	"Please enter a valid time."

Coding. One variable, nominal, stored as text.

Rules. Multi-line cannot be combined with a format ("a multiline answer cannot have a format"); Max characters at least 1.

---

Matrix

A grid: statements down the side, one shared scale across the top.

The respondent sees a table with one row per statement and one column per scale point (plus an N/A column if offered).

Keyboard. One cell of the grid is in the tab order. ← / → move along the row and answer it with the cell they reach, the N/A column included; ↑ / ↓ move to the same column in the row above or below without answering; Space or Enter chooses the focused cell; Tab leaves the grid. A required matrix can therefore be finished without a mouse. The cell in focus shows the focus ring, a chosen one included.

Inspector → Options

Option	Values	Default
Rows (statements)	a list; "one variable per row"; + Row. Hover a row's number to see the variable it writes	Statement 1, Statement 2
Columns (scale)	a list; + Column	Strongly disagree, Disagree, Neutral, Agree, Strongly agree
Offer “N/A”	on/off; when on, the hint says what N/A stores ("stored as -1, a missing code")	off

Coding. One ordinal variable per row, named <base>_1, <base>_2, … (for a question q4: q4_1, q4_2). Each row is its own column in the data, which is what scale construction and reliability analysis want. Every row variable gets the columns as value labels — coded 1 to n from left to right in a matrix you build — and the answer stored for a row is the code of the chosen column. A matrix inserted or imported with other codes keeps them: the question bank's Trust in institutions, headed 0 … 10, stores 0–10; a Qualtrics matrix stores its recodes.

• Rows. Each row keeps its own variable. + Row adds a row with a new variable (q4_3), which takes the first row's value labels, declared missing codes (a Refused, a Don't know) and valid range — on a matrix without column headers too — so it stores, labels and leaves out as missing the same codes as the other rows; removing a row removes its variable and its codebook entry, and the other rows keep theirs (and their data columns). A row variable's label follows the question text and its statement, as " — ", until you write a label of your own. Rename a row's variable in the Variable section, which has a name field for each row ("Matrix rows write one variable each, in the order of Options → Rows; rename a row's variable here."). A matrix whose rows came from its variables' labels (from an import, say) lists those labels under Rows; editing, adding or removing a row turns them into statements.

• Columns. A column's code follows its header. Renaming a column changes only its label: the code it stores stays, also while you type the new text letter by letter (on the way from "Strongly agree" to "Agree strongly" the header passes through "Agree" without taking that column's code). Moving or removing a column leaves the codes of the others as they are — remove 5 from a 0 … 10 scale and 6 … 10 still store 6 … 10. One rename changes the code: a header renamed to the label of a declared missing code that no column stores (a "Don't know" beside a declared -8 Don't know) takes that code.

• New columns. A column added with + Column takes its code from the text you type into it: the code of the codebook label with exactly that text, if no column stores it yet (a "Refused" column beside a declared missing 9 Refused stores 9, and stays a missing code; "Refusal" and "Don't know" beside ESS-style 77 / 88 store those); otherwise the next scale code of the codebook after the ones the columns store, so headers typed one by one onto a matrix that had none keep the codes its columns showed; otherwise the next code up that nothing uses.

• A missing code without a column. When you remove or retype the column of a declared missing code (77 = Refusal), the code keeps its value label and its missing declaration in the codebook, so answers already stored with it stay labeled and missing.

• Not applicable is an extra column at the end. It stores the row's declared not-applicable code: turning on Offer “N/A” labels -1 = Not applicable and declares it missing on every row variable. A matrix without column headers (possible from an import or the Source tab) keeps storing the text na when declaring a code would change which code its columns store; the Save then shows NA_STORED_AS_TEXT.

• Required on a matrix asks for an answer in every row; a row answered N/A counts. The Inspector says so beside the checkbox: "an answer in every row", or "an answer in every row — N/A counts" when Offer “N/A” is on. A matrix has no conditions on its rows, so every row is one the respondent sees. On Next:

  • a required matrix with no row answered shows "This question requires an answer.", as any required question does;
  • one answered in some rows but not all shows "Please answer every row." (Theme → Wording → Required matrix: rows left, where {n} is the number of rows left);
  • in both cases Next waits, and the rows still without an answer are marked: the statement turns the error color with a bar on its left, and the row's cells get error-colored outlines. Each mark goes when its row is answered; the message goes with the next answer, as any question's does.

  Leaving the matrix with no row answered shows "This question requires an answer." at once; rows left empty wait for Next, since the respondent is still working down the grid. A row a script cleared (set to empty) is unanswered. In the Walkthrough, a matrix counts toward "A/V answered" only once every row is answered.

• Skip to on a matrix fires as soon as any row is answered — on an optional matrix, one row is enough; a required one is held by Next until every row is answered, and then skips.

• Conditions and piping later in the survey can use a row's variable (Show if q4_1 ≥ 4, {label:q4_1}).

Which code a column stores. The survey reads each column's code off the first row variable's value labels and declared missing codes. The Builder follows the same rules, and so does Studio when it reads an older response that stored a column's position (for a matrix with headers). They matter most for a matrix from an import or the Source tab, whose codebook was not built with it. The rules are tried in turn:

1. Every header is the text of exactly one value label: each column stores that label's code.
2. The codebook declares missing codes (N/A, a refusal, a don't know): a header whose text is one of their labels stores that code, wherever the codebook lists it, and the other headers take the codes of the remaining labels in order, when there are as many of each. Codebooks that come from SPSS often list -8 "Don't know" before 0 … 10; a matrix headed "0" … "10", "Don't know" over such a codebook stores 0 … 10 and -8.
3. There are as many value labels as headers: the columns take the labels' codes in order (a 0–10 scale headed "0" … "10" over labels "No trust at all" … "Complete trust").
4. Otherwise the codebook gives nothing to go by, and the columns store 1, 2, 3 … from left to right.

A matrix without column headers shows the first row's value labels as its columns, in code order; with Offer “N/A” on, the N/A code is left out of them and has its column at the end. "In order" is the codebook's order; for labels written as an object in the Source tab, see Value labels in the Source tab.

Rules. The number of rows must equal the number of variables.

---

Ranking

Put items in order.

The respondent sees the items under "Tap or drag to rank". Tapping an item moves it into the ranked list with its position number; ranked items can be dragged to reorder or removed with ✕; the rest stay under "Remaining options".

Inspector → Options

Option	Values	Default
Items to rank	code · label list	1–3 Option 1–3
Rank at most	1 or more; "empty = all"	empty

Coding. One variable, ordinal, with the items as value labels. The answer is the ordered list of codes, first = rank 1; text exports write it as codes separated by semicolons.

Rules. At least one item; Rank at most 1 or more.

---

Best–worst (MaxDiff)

Shows a few items at a time and asks which is best and which is worst.

The respondent sees all tasks on one page. Each task is a small table: the Best column, the item, the Worst column. One item cannot be both best and worst in the same task. The question counts as answered only when every task has both picks. A pick is a button: clicked again, or pressed with Space or Enter once the keyboard has moved to it, it is released. Pressing Enter right after clicking a pick goes to the next page and keeps it (see Keyboard and touch). A chosen pick in keyboard focus shows the focus ring.

Inspector → Options: Items (default Item 1–6), Items per task (default 4), Tasks (default 8), Versions (default 20), Best is called, Worst is called.

Coding. Two variables per task plus one for the design version: <name>_t1_best, <name>_t1_worst, …, <name>_version — 17 variables for 8 tasks. They cannot be renamed; lowering Tasks removes the variables (and codebook entries) of the tasks that went. Later pages can test and pipe them ({label:q7_t1_best}).

The design, the lints and the analysis are described in MaxDiff and Conjoint.

---

Conjoint (choice)

Shows whole products side by side and asks which one the respondent would choose.

The respondent sees all tasks on one page. Each task shows "1 / 10", a grid with the attributes as rows and one column per product, a button under each product and, if configured, a "none of these" choice. Every task must be answered.

Inspector → Options: Attributes (default Brand: Brand A, Brand B; Price: Low, High), Products per task (default 3), Tasks (default 10), Versions (default 20), “None of these” (empty = force a choice).

Coding. One variable per task — which product was chosen, 1 to n from left to right, n+1 for "none" — plus one for the design version: <name>_t1 … <name>_t10, <name>_version. As for MaxDiff, lowering Tasks removes the variables of the tasks that went, and later pages can use the task variables in conditions and piping.

Details in MaxDiff and Conjoint.

---

Codes for Other, None of the above and N/A

Three answers are added with a switch rather than typed as choices: Add “Other (please specify)” (Single choice and Multiple choice), Add “None of the above” (Single choice) and Offer “N/A” (Likert scale and Matrix). Each is stored as a code of the question's variable, and Studio writes that code into the codebook for you, so exports label it and the Save has nothing to warn about. The hint beside the switch always says what is stored.

Option	Stored as	Added to the codebook
Other (please specify)	-66 in the question's column; the typed text in <variable>_other (wide layout: <Id>_other)	the value label -66 = Other (not in the wide layout, which has no code column for it), and a variable <variable>_other — nominal, text, labeled " — other (please specify)" — whose Used by is the question
None of the above	-77	the value label -77 = None of the above
Not applicable	-1	the value label -1 = Not applicable and the missing code -1 Not applicable of kind not applicable — on the variable, or on every row variable of a matrix

• When a code is taken. If one of the question's choices already uses -66 or -77, Studio uses the next free code down (-67, -78, …) and the hint says so ("stored as -67; the text goes to q1_other"). Adding a choice coded -66 to a question that offers Other moves Other to -67; the new choice stays an ordinary option. N/A starts at -1 and moves down the same way when the scale uses -1.
• A choice as the Other option. In the Source tab, other_code in the question's metadata can name the code of one of its choices: that choice becomes the Other option — it opens the text box, and no second "Other" is added, on the card either. A Qualtrics import does this for a text-entry choice. A question without choices does the same with a value label under the code Other stores (-66, or the code other_code names): with Add “Other (please specify)” on, that label is its Other option. The card shows such an option once, in its place, with no separate "Other (please specify)". The options editor leaves a value label with the Other or None of the above code out of its list, because the switch stands for it. After your first edit there, the survey adds its own Other at the end, stored under the same code.
• Options that come from the codebook. A Single choice, or a Multiple choice in the array layout, can have no Choices of its own: the survey then offers its variable's value labels, and the card and the options editor show them (see The options editor). Switching Other or None of the above on for such a question first copies those labels into its Choices list (without the Other or None code itself), as editing them in the options editor does, so the new option is added beside them and the Save stays valid.
• Scales with a valid range (an NPS 0–10, a CES 1–7): the Other and None codes are also declared as missing codes, so means and the range check leave them out.
• Switching an option off removes its value label (and, for Other, the <variable>_other entry) unless the code is declared missing. Switching Offer “N/A” off keeps the "Not applicable" label and the missing code, so answers already stored with it stay missing; delete them in the Codebook tab if you do not want them.
• Wording. Respondents see "Other", "Please specify...", "None of the above" and "Not applicable". Replace them in Theme → Wording (“Other” option, “Other” text box, “None of the above” option, “N/A” option) — see Theme and Branding.
• Questions set up before these codes were written for you. A question that offered Other, None or N/A and has not been edited since has no labels for these codes. The Save then warns ADDED_CODE_WITHOUT_LABEL ("Question 'q1' stores “Other (please specify)” as -66, which variable 'q1' has no value label for.") or NA_STORED_AS_TEXT, and its N/A is stored as the text na. Edit the question once in the Builder — switching the option off and on again is enough — and Studio adds the labels, the <variable>_other entry and the N/A missing code.

---

Older surveys and responses

The survey runtime now stores several answers differently from earlier versions of Studio. Two things follow.

A survey published before the change keeps its old build until you republish it: Save (a Save with no changes will do), then Republish #N on the environment's card. Until then its respondents get the old behavior: no Other entry in a dropdown, wide Multiple choice variables and exclusive choices that do not work, Min answers and a Number's range not checked on Next, conditions on matrix rows and MaxDiff or conjoint tasks that never fire, a required Matrix that lets the respondent on after one row, and no page Body above the questions.

Responses already collected are read in today's layout — in the Data tab, every export, flows and the research bundle — so old and new responses sit in the same columns:

Stored by an earlier build	Read as
Other, which showed up as columns named code (__other__) and text for a Single choice, or selected and otherText for a Multiple choice	the question's Other code (-66 unless the question names another) in the question's own column, and the text in <variable>_other (<Id>_other for a wide question)
None of the above: the text __none__	the None code (-77 unless the question names another)
N/A: the text na	the variable's declared not-applicable code, once the codebook declares one; without one it stays na
Wide Multiple choice: one column named after the question's Id, listing the chosen options' variables	1 in each chosen option's variable and 0 in the question's other variables
Matrix: the chosen column's position (1 … n)	the column's code — a 0–10 scale that was stored as 1–11 reads 0–10

An export you make now can therefore differ from one you downloaded earlier for the same responses (0–10 where it said 1–11, -66 where it said __other__). Two things cannot be recovered: an older wide answer cannot tell an option hidden from the respondent from one they did not choose, so both read 0; and quota cells on a per-choice variable or a matrix row count only responses collected by a survey built after the change.

---

Presets

The Presets column of the + Question menu. After insertion a preset is an ordinary question of its type — change anything you like.

Preset	Type	Configured as	Variable
Yes / No	Single choice	buttons; 1 Yes, 0 No	yes_no_<n>
Rating (stars)	Likert scale	5 points, Stars, no end labels	rating_<n>
NPS (0–10)	Likert scale	11 points, First point 0, "Not at all likely" / "Extremely likely", text "How likely are you to recommend us to a friend or colleague?"	nps_<n>
CES (1–7)	Likert scale	7 points, First point 1, "Strongly disagree" / "Strongly agree", text "The company made it easy for me to handle my issue."	ces_<n>
Attention check	Single choice	text "To show you are reading carefully, please choose “Rarely”.", Required, 1 Never … 5 Always, marked as an attention check with 2 as its acceptable answer	attention_<n>
Date	Open text	Format Date	date_<n>
Email	Open text	Format Email, placeholder name@example.com	email_<n>
Phone	Open text	Format Phone, placeholder +1 555 000 0000	phone_<n>

<n> is the question's number, and the question's Id is q<n>.

Note. A preset's Id (q5) and variable (nps_5) differ on purpose. The answer is stored under the variable, nps_5 — the column in your data and the name to use in conditions — so there is nothing to fix. Rename either one if you like; see Question Id and variable name.

The question bank's NPS, CSAT and CES blocks are Single choice questions with an open follow-up, not Likert scales — see Question Bank, Templates and Library.

---

Attention checks

An instructed-response item ("please choose Rarely") tells you who is reading, and so does a statement nobody reading it agrees with ("I swim to work every day"). It is available on Single choice, Likert scale, Number and Open text questions, in the Question section:

1. Turn on Attention check ("scored in the flow, not in the survey").

2. Set Acceptable answers, the answers a reading respondent gives. On a Single choice or Likert scale question, check them among the question's codes, shown as "code · label" ("2 of 5 answers" when several are checked), the codes the survey stores for Other (-66), None of the above (-77) and N/A among them where the question offers them; a Likert scale also has a button for each side of the scale, named by its end labels — "The “Strongly disagree” side (1, 2)", "The “Strongly agree” side (4, 5)" — with an odd scale's middle point on neither. On a Number question, type them separated by commas (7, 8), kept as numbers; on an Open text question, type the one answer, kept as text. Until one is set the Inspector warns "Until an answer is set, this question checks nothing."; with every code checked, "Every answer is acceptable, so nobody fails this check." One answer is stored as expected: 2 in the question's metadata, several as a list, expected: [1, 2].

3. Optionally end the survey for respondents who fail: End the survey after N failed attention checks counts the survey's checks together (see After several failed checks); for this check alone, turn on Also end the survey for respondents who fail (hint: "adds a screen-out branch", then "branches to page"). Studio adds an ordinary Branch (next if) rule, first among the page's branches, to the question's page that sends anyone who answered with none of the acceptable answers to a Screen-out page, and writes it again when the acceptable answers change. While End the survey after N failed attention checks is on, this switch is off and disabled. The rule is evaluated when the page is left, a respondent who skipped the question is not screened out, and a failing respondent's answers are still collected and counted as screened out. You can edit or delete the rule (or change its target page) in the page's Logic section like any other; turning the option off removes the rule and leaves the page. Only a branch of that shape to a Screen-out page is the switch's: one you wrote to another page stays.

   The Screen-out page the rule points at sits after the last Final or Redirect page, so that only the rule leads there: respondents who pass finish on the Final page and are recorded as completed.

   • If such a Screen-out page already exists (one without a Show if or Hide if of its own, after the Final page), the rule points at it.
   • Otherwise Studio adds one, named disqualification ("Thank you" / "You do not qualify for this study."), right after the last Final or Redirect page. A conditional Screen-out page elsewhere — such as the consent screen-out of the built-in templates — keeps its own job and is not used.
   • If no Final or Redirect page without a condition ends the survey, Studio first adds a Final page ("Thank you" / "Thank you for taking part.") after the last content page and any end pages right behind it.

   The checkbox's (i) reads: "Adds a branch on this page to a Screen-out page, taken by a respondent who gives another answer. It is an ordinary page branch, editable under Logic, and comes first among the page’s branches, so another one does not send a respondent who fails past it. It is checked when the respondent leaves this page, so they finish the page first, and their answers are still collected and counted as screened out. Leave it off to keep everyone and decide in the analysis instead."

Questionnaires saved with an older placement. Earlier versions of Studio put the Screen-out page in front of the Final page, or pointed the rule at an existing conditional Screen-out page. The Inspector says what goes wrong, under the checkbox — as it does, in any questionnaire, when a Skip to on the same page keeps the rule from firing (the last row below) — and Validation → Structure lists it as "q7: the screen-out branch for failing this attention check does not work. …":

Message	What happens	Fix
"Respondents who pass reach “page” too: it comes after this page with no Final page in between, and pages are shown in order."	everyone who passes is screened out as well	Fix the branch moves that Screen-out page, with its wording, right after the last Final or Redirect page (adding a Final page first if none ends the survey) and points the rule at it
"“page” has a show if or hide if of its own. Where it is hidden, a respondent who fails is sent on to the next page shown after it instead of being screened out."	failing respondents are not screened out	Fix the branch points the rule at a Screen-out page after the Final page, adding one if needed; the conditional page keeps its job
"Skip to on question is checked before branch rules, so for anyone who answers question this branch never fires."	the rule never fires for anyone who answers that question	no automatic fix: clear that question's Skip to, or move the check to another page

Walk through a passing and a failing answer in Preview before you publish; details in Attention checks.

Without step 3 nobody is screened out during the interview: the check is scored afterward by the Response quality node, whose Attention checks parameter is filled from these settings, lists included. See Data Quality.

With Prolific as the panel provider (Distribute → Panel), the Inspector also shows Prolific's rules: on a check that screens out after one failure in a study of 5 minutes or more, that Prolific needs at least two failed checks; on a check that screens out, that the checks must come at the start of the study; on a Likert check with an odd number of points, "If this is a nonsensical statement: Prolific requires a scale without a neutral option, and this scale’s middle point (3) is one. Give the scale an even number of points."; and under Format of an Open text question — "Prolific does not allow asking participants for a personal email address without its prior written approval. Contact them through Prolific’s messages instead." (Email), "Prolific does not allow asking participants for a phone number." (Phone) and "If this asks for a date of birth: Prolific does not allow that without its prior written approval. Ask for age in years instead." (Date). The Validation tab lists them under Prolific. They never stop a Save; see Prolific.

With Respondent as the panel provider, the Inspector warns on a check that screens out ("Respondent has one completion URL: a participant screened out on failed attention checks stays on Studio’s page and never reaches it; …"), and under Format of an Open text question — "Respondent recognizes participants by the IDs it adds to the survey link, not by an email address. Ask for an email only if the project uses Respondent’s email visibility option, in which participants who opt in enter the email of their Respondent account; for participants sourced through Respondent’s API, requesting personal data is a violation of its terms. Otherwise contact participants through Respondent’s messages." (Email), "Respondent prohibits collecting a participant’s phone number, and removes researchers found collecting personal data from the platform. Contact participants through Respondent’s messages." (Phone), and "If this asks for a date of birth: Respondent prohibits collecting it. Ask for age in years instead." (Date). The Validation tab lists them under Respondent; see Respondent.

---

Converting a type

The type dropdown at the top of the Inspector converts the question in place.

Kept	Dropped
question text, hint, Id, Required, Show if, Hide if, Skip to, Randomize option order, the variable name, and the choices when both types have them (Single choice, Multiple choice, Ranking, MaxDiff)	tags, Media URL, the attention-check setting, Other and None of the above, display and all other type-specific settings

Converting to or from MaxDiff or Conjoint replaces the question's variables with a fresh set. Check the Variable section afterward: the codebook entry keeps its old scale (a Single choice turned into a Number is still nominal, which the engine flags as INCOMPATIBLE_QUESTION_SCALE).

---

What the engine refuses

These make the questionnaire malformed: Save is refused with "Save failed." and the engine's reason, and the Preview cannot be built. Fix them before you Save.

• an empty survey title, question text or question Id;
• an option with an empty label, or two options with the same code;
• Multi-line together with a text Format;
• Step of 0 or less; Rank at most of 0; Max characters of 0;
• Max answers smaller than Min answers, or — in the wide layout — larger than the number of per-choice variables;
• a valid range whose Min is greater than its Max;
• a redirect Delay (s) that is not a whole number;
• a missing code without a label (possible only in the Source tab — the Codebook tab always writes one; see Missing codes);
• a MaxDiff with fewer than three items; a conjoint with fewer than two attributes, an attribute with fewer than two levels, an attribute name that is not a plain identifier, or two attributes with the same name.

Tip. The message names the field that is wrong and where it is, by page name and question Id — for example "questionnaire: pages/0/items/0/text: must not be empty (page 'page1', question 'q1')", or "…/choices: needs at least 3 entries (page 'page1', question 'q7')" for a MaxDiff with two items. pages/0/items/2 counts from zero: page 1, item 3. Inside a block the place reads "(page 'page1', block 1, question 'q1')".

Other problems — a duplicate question Id, a question Id that is another question's variable name, an Id that differs from its question's variable and is a name the survey stores something else under (a Matrix row, another question's Other text, an assigned arm; see Names an Id may not take), Skip to a page that no longer exists, a page nobody can reach, an Other whose default code -66 is already a choice's code (possible only from the Source tab or an import), a "None of the above" code that is one of the answers, an Other text column that another question writes — do not stop the Save; the Save is marked errors and cannot be published. The Validation tab lists them; see Testing Your Survey.

---

What is deliberately missing

• Loop & merge — not in the engine, so not in Studio.
• Multi-language questionnaires — one questionnaire is one language. To run a survey in a single other language, write the questions in that language and replace the runtime's phrases (buttons, the required-answer message, the resume prompt, "Other", "Please specify...", "None of the above", "Not applicable", the minimum and maximum messages, the text-format messages, …) under Theme → Wording, and the default thank-you text under Theme → Respondent experience → Completion screen (Theme and Branding). Labels that only screen readers announce — such as "Task 1" on a MaxDiff task, "Choice 1" on a conjoint task or "Loading survey" — stay in English.
• Exotic widgets — constant sum, side-by-side grids, drill-downs, file upload, signatures, heat maps. When you import a questionnaire that has them, they are listed as not imported.

If your design needs one of these, say so through Profile → Support.

See also

• The Builder
• Codebook and Variables
• MaxDiff and Conjoint
• Logic and Branching
• Data Quality

---

← The Builder · Studio contents · Codebook and Variables →