Matrix question · Matrix

Matrix questions

Ask people to rate several things on the same scale, one row each. In the builder it’s Grid (single) for one answer per row, or Grid (multi) for checkboxes. On a phone the table becomes a stacked list, and every row gets its own percentages, Mean and Top-2-Box.

Available onFree & up· every plan
The Matrix question as respondents see it on a laptop and a phone

Live demo

Answer it. Your answers join the results.

This is the grid your respondents get. Try each layout, rate the rows, and the heat-map beside it recalculates with yours counted. On a phone the table becomes the stacked list unless you choose another layout.

How would you rate each part of your stay?

  • Check-in

    1. Poor
    2. Fair
    3. Good
    4. Very good
    5. Excellent
    6. Didn’t use

  • Room

    1. Poor
    2. Fair
    3. Good
    4. Very good
    5. Excellent
    6. Didn’t use

  • Cleanliness

    1. Poor
    2. Fair
    3. Good
    4. Very good
    5. Excellent
    6. Didn’t use

  • Breakfast

    1. Poor
    2. Fair
    3. Good
    4. Very good
    5. Excellent
    6. Didn’t use

  • Wi-Fi

    1. Poor
    2. Fair
    3. Good
    4. Very good
    5. Excellent
    6. Didn’t use

In action

The matrix question in action

Matrix in action · One guest rates the rows with the keyboard, another answers on a phone and skips a row, and the heat-map updates.
  1. 0:00A guest sees five parts of their stay in one table, each rated from Poor to Excellent, with a Didn’t use column.
  2. 0:03They click Good for Check-in, then press the right arrow to move along the row to Very good.
  3. 0:06Tab takes them to each next row, and the arrow keys rate it.
  4. 0:09On a phone, a second guest works down the same grid as a stacked list and taps Didn’t use for Breakfast.
  5. 0:12They press Next with the Wi-Fi still blank. The message names the row, and the row is marked.
  6. 0:14They rate the Wi-Fi, and the message clears.
  7. 0:16They press Next again and both guests join the results. The heat-map and the Top-2-Box bars redraw, with Didn’t use in Breakfast’s base but not its score.

Every feature

Everything the matrix question can do

A grid that’s quick to build, works on any screen, keeps Didn’t use out of your scores, and reports every row on its own.

BuildRows, columns and which rows each person sees.

LiveFree

Rows and columns, typed or pasted

Add rows and columns one at a time, or use Bulk edit to paste a list, one per line. Write a column as Poor | 1 to set its code.

LiveFree

Statement groups

Put labelled sub-headers over sets of rows, such as Arrival and Your room, inside one grid.

LiveFree

Transpose

Swap the table so the rows run across the top and the points run down. Phones still get the stacked layout.

LiveFree

Shuffle rows and columns

Shuffle the rows, the columns or both for each respondent. The order holds when they go Back or come back later, and the row order each person saw is in your export.

The Didn’t use column always stays last.

LiveFree

Rows from an earlier answer

Build the rows from what someone picked, or didn’t pick, in an earlier choice question. If nothing applies, the grid is skipped.

LiveFree

Show a row only when it applies

Give any row its own show-if logic, and branch later questions on one row’s answer.

LiveFree

Didn’t use, and other answers off the scale

Add a Not applicable column with your own header, such as Didn’t use, or tick Not on scale on a column like Can’t say. Either one sits after a divider, counts in the base, and stays out of the Mean and Top-2-Box.

LiveFree

A code for every column

Each column gets a code no other column uses, even after you delete one and add another, so answers never merge.

For respondentsWhat people answering see and use.

LiveFree

Six phone layouts

On a narrow screen the table becomes a stacked list. Or choose wrapped chips, an accordion, a carousel, a segmented bar or the table itself for one question. For the whole survey, the theme can set the accordion, the carousel or the table.

The switch follows the grid’s own width, not the screen’s, so the builder’s phone preview matches a phone.

LiveFree

Wide grids stack, they don’t hide

A grid with many columns switches to the stacked list below the width it needs, so no column sits off-screen. If you keep the table, it scrolls sideways inside its own box with the rows pinned.

LiveFree

Headers that stay put

Repeat headers on scroll keeps the column labels in view while a long grid scrolls.

LiveFree

Keyboard on every layout

On a single-answer grid each row is one Tab stop, and the arrow keys move along it on every layout, mirrored in right-to-left surveys.

On a checkbox grid every box is a Tab stop and Space ticks it.

Try it in the demo: Tab to a row, then use the arrow keys.
LiveFree

The page stays where you are

On iPhone, answering a row updates it in place, so the page doesn’t jump back to the top as you work down the grid.

LiveFree

Messages that name the rows

Ask for every row, or at least a number of rows. Press Next too soon and the message says how many rows are done and names the ones left, and each of those rows is marked.

Data qualitySpot careless answers and bad data.

Live

Catch straightliners

A single-answer grid fails the check when it has 4 or more rated rows that all got the same answer. Didn’t use and Not on scale answers don’t count as rated. A failed grid shows as a Straightliner signal on the Fraud page, and by default it takes 2 failed checks to flag a respondent. The check is on by default.

Checkbox grids aren’t checked.

LiveFree

Checked on the server

Every answer is checked against the grid’s own rows and columns. An unknown row or column, Didn’t use when the column is off, or None of us with another tick is refused.

AnalysisEvery row reported, cut and piped on its own.

LiveFree

A heat-map in Results

Results shows each row as percentages of its base, with a Mean and Top-2-Box. Didn’t use gets its own column, counted in the base, rows nobody answered show at 0, and with weighting (Research Team) the heat-map is weighted too.

LiveCrosstabs · Solo+Letters · Professional+

Each statement in a crosstab

Each row of the grid is its own crosstab row choice: one row per point and a Didn’t use row, with Top-2-Box and the Mean underneath and letters for significant differences.

LiveTab books · Research Team

A grid summary in tab books

One table with every statement against your banner, on Top-2-Box, Bottom-2-Box or the Mean, then a table per statement.

LiveFree

A bar per statement in your deck

The report slide draws a Top-2-Box bar for each statement, from the people who rated it, in PDF and PPTX too.

LiveFree

Pipe one row

Insert one row’s answer into a later question. The pipe menu lists each row by its label, such as Wi-Fi.

LiveFree

Filter by a statement

Filter Results to the people who gave one row a particular answer, such as everyone who rated the Wi-Fi Poor or Fair.

LiveFree

Names, not a scale

When the columns are names or categories, such as brands, each row reports percentages only, with no Mean or Top-2-Box, and exports as nominal in SPSS.

Not part of the matrix question: ranking rows against each other (see Ranking), text boxes or dropdowns in every cell (Textbox Grid and Dropdown Grid are their own question types), and a slider per row (use Slider List).

Checkbox grids

When more than one column applies

Choose Grid (multi) and each row takes as many ticks as apply. Use it for “who used what” rather than a rating.

  • None of us, per row. Mark a column Exclusive and ticking it clears the rest of that row, while any other tick clears it. The server refuses a row that has both.
  • Shares of each row. Results shows, for each row, the share of people who answered it that ticked each column. In crosstabs (Solo+) and tab books (Research Team) each row is a multi-response table, with no Top-2-Box or Mean, because ticks aren’t a scale.
  • One 0/1 column per tick box. Each row and column pair exports as its own column, the question’s column name plus _1_1, _1_2 and on: 1 when ticked, 0 when not.
  • Keyboard. Every box is a Tab stop and Space ticks it.

Try it

Grid (multi) · None of us is exclusive

Who in your party used each of these?

  • Pool

    1. Me
    2. My partner
    3. Children
    4. None of us

  • Gym

    1. Me
    2. My partner
    3. Children
    4. None of us

  • Spa

    1. Me
    2. My partner
    3. Children
    4. None of us

  • Restaurant

    1. Me
    2. My partner
    3. Children
    4. None of us
Q1_1_1Q1_1_2Q1_1_3Q1_1_4Q1_2_1Q1_2_2Q1_2_3Q1_2_4Q1_3_1Q1_3_2Q1_3_3Q1_3_4Q1_4_1Q1_4_2Q1_4_3Q1_4_4
1000110000011100
0001011000011010

Two guests’ answers in the CSV: one column per row and column, named by the row’s position and the column’s code (1 Me, 2 My partner, 3 Children, 4 None of us).

Build it

Add a matrix question

  1. 1

    Add the question

    In the builder, add a question and pick Grid (single) from the Matrix group, or Grid (multi) for checkboxes. Write the question, such as “How would you rate each part of your stay?”

  2. 2

    Add rows and columns

    Type the rows under Row Labels and the scale under Column Headers, or use Bulk edit to paste one per line. For columns, write Poor | 1 to set each code yourself.

  3. 3

    Add Didn’t use

    Tick Add a “Not applicable” column and type the header people see, such as Didn’t use. For a column like Can’t say, tick Not on scale. Both stay out of the Mean and Top-2-Box.

  4. 4

    Shuffle, group and filter rows

    Tick Randomize rows or Randomize columns. Use Build rows from a previous question to ask only about what someone used, and Show this row if… for a single row.

  5. 5

    Choose the layout

    Keep Display on Table (grid) and Mobile display on Auto for the stacked list on phones, or pick chips, accordion, carousel or segmented bar. Tick Repeat headers on scroll for a long grid.

  6. 6

    Set the rules and preview

    In Rules, require every row, set Minimum rows required, or make it soft required or optional. Preview at phone width to see the layout your respondents get.

The Grid (single) editor for “How would you rate each part of your stay?”: Row Labels Check-in, Room, Cleanliness, Breakfast and Wi-Fi, Column Headers Poor, Fair, Good, Very good and Excellent, each list with its Bulk edit link, then Display set to Table (grid), Mobile display on Auto, Randomize rows ticked, and Add a “Not applicable” column ticked with the header Didn’t use
EditorRows, columns and a Didn’t use column.
The grid’s display settings: Display set to Table (grid), Mobile display on Auto (stacked list, or theme default), Randomize rows ticked, Randomize columns, Transpose and Repeat headers on scroll left unticked, Add a “Not applicable” column ticked with Didn’t use, and Minimum rows required left at All
LayoutOne setting for desktop, one for phones.
The builder’s phone preview of the stay question as a stacked list: a card for Check-in and one for Room, each with Poor, Fair, Good, Very good and Excellent as full-width buttons, then Didn’t use after a dashed divider
Phone previewThe stacked list a phone shows.

The data you get

One column per row, with every point labelled

Each statement exports as its own column holding the code of the point chosen. Your column labels come along as value labels, and Didn’t use gets a code of its own that never clashes with a point.

respondent_idQ1_1Q1_2Q1_3Q1_4Q1_5Q1_order
47762a85-0001-4000-8000-00000000000154443rows:Breakfast|Cleanliness|Room|Check-in|Wi-Fi
47762a85-0002-4000-8000-00000000000254452rows:Breakfast|Check-in|Cleanliness|Wi-Fi|Room
47762a85-0003-4000-8000-00000000000344532rows:Check-in|Cleanliness|Room|Wi-Fi|Breakfast
47762a85-0004-4000-8000-00000000000444333rows:Check-in|Room|Cleanliness|Wi-Fi|Breakfast
47762a85-0005-4000-8000-00000000000544343rows:Room|Check-in|Cleanliness|Wi-Fi|Breakfast
47762a85-0006-4000-8000-00000000000625551rows:Breakfast|Cleanliness|Wi-Fi|Check-in|Room
47762a85-0007-4000-8000-00000000000734552rows:Wi-Fi|Room|Cleanliness|Check-in|Breakfast
47762a85-0008-4000-8000-000000000008345NA5rows:Wi-Fi|Cleanliness|Room|Breakfast|Check-in

Trimmed to the question’s columns. The columns ending _1 to _5 are Check-in, Room, Cleanliness, Breakfast and Wi-Fi, in the order you wrote them. Points export as 1 to 5 and Didn’t use as NA. Rows were shuffled for each guest, so the _order column records the order each one was shown. Every export also carries the standard respondent columns (status, timing, weight and more).

One column per statement

Columns are named by position, the question’s column name plus _1, _2 and on, so rewording a row never renames its column or breaks your syntax.

Labels as value labels

In SPSS and the Excel Value Labels sheet every point carries its label, so 4 reads as Very good. CSV and JSON carry the codes only.

Ordinal in SPSS

A rating grid comes measured as ordinal. A grid of names or categories comes as nominal and gets percentages only, with no Mean.

Didn’t use

CSV writes NA. SPSS codes it 97, outside the scale, labelled with your header and declared missing when you ask for it.

The order each person was shown

With rows shuffled, the _order column records the order each respondent was shown.

Weighting

Weighted exports carry a weight column, and the SPSS syntax sets WEIGHT BY weight. Weighting is on Research Team.

Charts & analysis

Every row reported on its own

Results shows the whole grid as a heat-map with a Mean and Top-2-Box for each row. Crosstab any statement against another question, compare every statement at once in a tab book, and show a Top-2-Box bar per statement in your deck.

The Surveti Results card for the stay question as a heat-map: a row for each part of the stay (Check-in, Room, Cleanliness, Breakfast and Wi-Fi), each with its base n of 150, a column for each point from Poor to Excellent and a Didn’t use column, each cell the share of that row’s answers, darker for more, then each row’s Mean and Top-2 (Check-in 63%, Room 63%, Cleanliness 77%, Breakfast 58%, Wi-Fi 45%)
ResultsEach cell is a row % of everyone who answered the row, darker for more. Didn’t use has its own column, counted in the base, and stays out of each row’s Mean and Top-2.

Where it charts

  • ResultsHeat-map with a Mean and Top-2-Box per row
  • Report deckA Top-2-Box bar per statement, in PDF and PPTX
  • CrosstabsEach statement as its own row
  • Tab booksA summary of every statement
  • FiltersResults for people who gave a row an answer
  • Client portalThe report deck
A report slide from the PDF export: “How would you rate each part of your stay?”, Top-2-Box (% favourable) by statement, a bar for each row: Check-in 63%, Room 63%, Cleanliness 77%, Breakfast 58%, Wi-Fi 45%
Report deckEach bar is the share of that row’s rated answers that were Very good or Excellent. Didn’t use is left out of the bar.
A Surveti tab-book grid summary of Top-2-Box % by statement, by trip type: a row for each statement (Check-in, Room, Cleanliness, Breakfast and Wi-Fi) against Total, Business, Leisure and Family lettered A to C, each cell its Top-2 count above its percentage, significance letters under the cells that differ, and the note that the top and bottom box leave Not applicable and non-answers out
Tab bookEvery statement against the banner in one table. Each statement shows its Top-2 count, then that count as a share of the guests who rated the row, so Didn’t use is left out. Family guests rated breakfast higher than Business guests (72% against 44%).

Wi-Fi, by trip typeColumn %, letters at 95% and 99%

Crosstabs · Solo+Letters · Professional+
The Surveti Crosstabs view of the Wi-Fi statement by trip type: Total, Business, Leisure and Family lettered A to C, a row per point from Poor to Excellent, a Didn’t use row, then Top-2-Box and Mean rows, in column percentages with the base for each column, significance letters beside the cells that differ and none on the Didn’t use row, and the note that the top and bottom box leave Not applicable and non-answers out
The point rows and Didn’t use add to 100% of everyone who answered the Wi-Fi row. Didn’t use counts in the base and stays out of Top-2-Box and the Mean, which use the rated answers only. A letter marks a figure significantly higher than the column with that letter: Leisure and Family guests rated the Wi-Fi higher than Business guests did (53.2% and 53.3% Very good or Excellent, against 25%). Letters need Professional+. The table itself needs Solo+, and without letters it still runs.

Analysis that applies

  • Crosstabs

    Solo+

    Pick any statement for the rows: one row per point and a Didn’t use row, with Top-2-Box and the Mean underneath. Didn’t use counts in the base and stays out of both.

    Crosstabs

  • A means test and a proportions test

    Professional+

    The Mean row is tested with Welch’s t-test, and the point rows and Top-2-Box with a test of proportions, in the same table.

    Significance testing

  • Tab books

    Research Team

    A grid summary with every statement against your banner, on Top-2-Box, Bottom-2-Box or the Mean, then a table per statement with its own Didn’t use row.

  • Weighting

    Research Team

    Weights carry through to the heat-map, crosstabs, tab books, the report slide and exports.

How Surveti compares

Matrix questions across the common survey tools

Matrix capabilities in Surveti, SurveyMonkey, Qualtrics, Typeform and Alchemer, as of 30 October 2026
CapabilitySurvetiSurveyMonkeyQualtricsTypeformAlchemer
Matrix question on the free (or entry) planyesGrid (single) and Grid (multi) on every planunclear12Basic names four other typesyes29Matrix table on the free accountyes38Matrix on Freeno61No free plan
Checkbox grid (several answers per row)yesWith an exclusive None column per rowyes1Multiple responses per rowyes18Allow multiple answersyes35Multiple selectionyes46Checkbox Grid
Phone layout that stacks the rowsyesStacked list by defaultunclear1, 2Large matrices may scroll sidewaysyes18Accordion or carouselunclear43Community reports sideways scrollingyes45, 46, 48One row at a time
Author chooses the phone layoutyesSix layouts per question, four in the themeunclear1No setting documentedyes18Choose among formatsunclear35Mobile not coveredyes45One at a time can be turned off
N/A column kept out of averagesyesIn the base, out of the Mean and Top-2yes1Only with weights onyes18, 30Exclude From Analysisunclear35N/A not coveredpartial45Counted when its value is numeric
Row and column randomisationyesBoth, every plan; row order exportedpaid1, 16Rows; columns not mentionedyes18Statements and scale pointsno35Can’t randomise a Matrixyes53Rows and columns, with limits
Top-2-box for grid rowsyesResults, crosstabs, tab books, decksunclear14No top-box feature foundyes27, 34Statistics Tableunclear36Not documentedyes58Per-element setting
Crosstab using grid rowspaidEach statement a row · Solo+no10Matrix not a crosstab typepaid28, 29Not on the free accountunclear36Not documentedyes59Grid rows as rows
Significance testing on grid rowspaidProportions and a means test · Professional+paid11Per-row use not explainedpartial28Matrix handling not describedunclear36Not documentedpartial59No test of means
Tab book / banner-table exportpaidGrid summary of every statement · Research TeamunclearNot on the pages readunclearNot on the pages readunclearNot on the pages readunclearNot on the pages read
Export: one column per rowyesNamed by position (_1, _2 …), every planpaid8Rows span columns; export is paidyes18One column per statementunclear39Matrix layout not describedyes45One column per row
Straightlining flagyesOn by default; single-answer grids, 4+ rated rowspaid6English, US data centre onlyyes24, 253+ matrices and 80% of them; off by defaultunclear36Nothing documentedyes57Grids with more than 4 rows
As of 30 Oct 2026, from each vendor’s public help pages and pricing.yespartialpaidunclearno

Partial means partly built in or possible only with a documented workaround; paid means built in on paid tiers or licences only; unclear means the vendor’s docs don’t settle it. Google Forms, QuestionPro and SurveySparrow were also reviewed. Google Forms has a multiple choice grid and a checkbox grid, with an option to limit one choice per column. QuestionPro offers an N/A option across a matrix, validation for rows, columns or both, and a check for straight-line answers. SurveySparrow’s matrix grid offers radio button, checkbox, text, dropdown and rating scale cells.

Sources (30)
  1. [1] SurveyMonkey: Matrix/Rating Scale. https://help.surveymonkey.com/en/surveymonkey/create/matrix-rating-scale/ · accessed 30 Oct 2026
  2. [2] SurveyMonkey: Matrix of Dropdown Menus. https://help.surveymonkey.com/en/surveymonkey/create/matrix-of-dropdowns/ · accessed 30 Oct 2026
  3. [6] SurveyMonkey: Response Quality. https://help.surveymonkey.com/en/analyze/response-quality/ · accessed 30 Oct 2026
  4. [8] SurveyMonkey: XLS exports. https://help.surveymonkey.com/en/surveymonkey/analyze/xls-exports/ · accessed 30 Oct 2026
  5. [10] SurveyMonkey: Crosstabs. https://help.surveymonkey.com/en/surveymonkey/analyze/crosstabs/ · accessed 30 Oct 2026
  6. [11] SurveyMonkey: Significant differences. https://help.surveymonkey.com/en/analyze/significant-differences/ · accessed 30 Oct 2026
  7. [12] SurveyMonkey: Free vs paid plans. https://www.surveymonkey.com/product/free-vs-paid-plans/ · accessed 30 Oct 2026
  8. [14] SurveyMonkey: Combining answer choices. https://help.surveymonkey.com/en/surveymonkey/analyze/combining-hiding-answer-choices/ · accessed 28 Oct 2026
  9. [16] SurveyMonkey: Randomizing answers. https://help.surveymonkey.com/en/create/randomizing-answers · accessed 28 Oct 2026
  10. [18] Qualtrics: Matrix table. https://www.qualtrics.com/support/survey-platform/survey-module/editing-questions/question-types-guide/standard-content/matrix-table/ · accessed 30 Oct 2026
  11. [24] Qualtrics: Response quality. https://www.qualtrics.com/support/survey-platform/survey-module/survey-checker/response-quality/ · accessed 30 Oct 2026
  12. [25] Qualtrics: Response Quality functionality. https://www.qualtrics.com/support/survey-platform/data-and-analysis-module/data/response-quality-functionality/ · accessed 30 Oct 2026
  13. [27] Qualtrics: Statistics table. https://www.qualtrics.com/support/survey-platform/reports-module/results-section/visualizations/tables/statistics-table/ · accessed 30 Oct 2026
  14. [28] Qualtrics: Cross tabulation overview. https://www.qualtrics.com/support/survey-platform/data-and-analysis-module/cross-tabulation/cross-tabulation-overview/ · accessed 30 Oct 2026
  15. [29] Qualtrics: Free accounts. https://www.qualtrics.com/support/survey-platform/managing-your-account/trial-accounts/ · accessed 30 Oct 2026
  16. [30] Qualtrics: Formatting answer choices. https://www.qualtrics.com/support/survey-platform/survey-module/editing-questions/formatting-answer-choices/ · accessed 30 Oct 2026
  17. [34] Qualtrics Community: top/bottom box for a Likert matrix (search summary). https://community.qualtrics.com/survey-platform-54/grouping-answer-choices-from-a-likert-matrix-in-a-qualtrics-report-33002 · accessed 28 Oct 2026
  18. [35] Typeform: Matrix question. https://help.typeform.com/hc/en-us/articles/360054731432-Matrix-question · accessed 30 Oct 2026
  19. [36] Typeform: Results summary. https://help.typeform.com/hc/en-us/articles/360029577071-Your-Results-summary · accessed 30 Oct 2026
  20. [38] Typeform: Free plan. https://help.typeform.com/hc/en-us/articles/360032972852-Free-plan · accessed 30 Oct 2026
  21. [39] Typeform: Export responses. https://help.typeform.com/hc/en-us/articles/360029253572 · accessed 30 Oct 2026
  22. [43] Typeform Community: matrix questions on mobile. https://community.typeform.com/build-your-typeform-7/matrix-questions-don-t-appear-properly-on-mobile-3348 · accessed 30 Oct 2026
  23. [45] Alchemer: Radio Button Grid. https://help.alchemer.com/help/radio-buttons-grid · accessed 30 Oct 2026
  24. [46] Alchemer: Checkbox Grid. https://help.alchemer.com/help/checkboxes-grid · accessed 30 Oct 2026
  25. [48] Alchemer: Star Rating Grid. https://help.alchemer.com/help/star-rating · accessed 30 Oct 2026
  26. [53] Alchemer: Randomization. https://help.alchemer.com/help/randomization · accessed 30 Oct 2026
  27. [57] Alchemer: Quarantine Bad Responses (PDF export). https://help.alchemer.com/help/pdfexport/id/5f2ed3aded121cdc14845355 · accessed 30 Oct 2026
  28. [58] Alchemer: Answer grouping. https://help.alchemer.com/help/answer-grouping-recoding · accessed 30 Oct 2026
  29. [59] Alchemer: Cross tab report. https://help.alchemer.com/help/cross-tab-report · accessed 30 Oct 2026
  30. [61] Alchemer: Plans and pricing. https://www.alchemer.com/plans/ · accessed 30 Oct 2026

What’s different here

Six phone layouts, on every plan

A grid becomes a stacked list on a narrow screen without any setting, judged by the grid’s own width. Pick chips, an accordion, a carousel or a segmented bar instead for one question. For the whole survey, the theme can set the accordion, the carousel or the table.

Didn’t use stays out of every score

It counts in the base and stays out of the Mean and Top-2-Box in Results, crosstabs, tab books and the report slide, whatever code it has. In SPSS it gets its own labelled code, declared missing when you ask for it.

Every statement in a crosstab, with a means test

Each statement is its own crosstab row, with Top-2-Box, the Mean, and letters from a test of proportions and Welch’s t-test. Crosstabs are on Solo+, letters on Professional+.

A grid summary in your tab book

Every statement against your banner in one table, on Top-2-Box, Bottom-2-Box or the Mean, with letters. Tab books are on Research Team.

Ordinal in SPSS, labels included

Each statement exports as an ordinal variable with every point labelled, named by position so your syntax keeps working. SPSS export is on Professional+.

A straightlining check on each grid

On by default, every single-answer grid with 4 or more rated rows is checked on its own. Same answer on every rated row, and the respondent gets a Straightliner signal on the Fraud page.

When to use it

Use a matrix to rate several things on one scale

When to use it

  • You’re rating several items, like parts of a service, on the same scale.
  • You want to compare the items side by side, with a score for each.
  • You’re asking a battery of agree statements. See the Likert scale page.
  • You need to know who used what, several answers per row. Use a checkbox grid.

When to reach for something else

  • You’re rating just one thing. Use a rating scale.
  • You want the items put in order of preference. Use ranking.
  • Each row needs different answer choices. Ask separate questions.
  • The list runs past about ten rows. Split it across two grids or two pages.

Sample wording

“How would you rate each part of your stay?” Check-in · Room · Cleanliness · Breakfast · Wi-Fi, each on Poor · Fair · Good · Very good · Excellent · Didn’t use

Best practices

  • Keep it short: five to ten rows is plenty.
  • Label every column, and keep labels short so they fit on a phone.
  • Add a Didn’t use or Not applicable column when a row may not apply, so nobody is pushed onto the middle point.
  • Shuffle the rows so no item always gets the first, most careful answer.
  • Put the best answer at the high end, so Top-2-Box reads as positive.
  • Check the Fraud page for straightliners before you report.

Accessibility

Usable without the mouse or the screen

Every cell named

Each answer is announced with its row and column, such as “Wi-Fi: Good”, and whether it’s chosen. On phone layouts each row is a group named by its statement.

One Tab stop per row

On a single-answer grid Tab lands on each row once, and the arrow keys move along it, on every layout and mirrored in right-to-left surveys. On a checkbox grid every box is a Tab stop.

Room to tap

Every answer has a 44px target. On a narrow screen the table becomes full-width rows, so nothing shrinks to fit.

Messages that say which row

If a row is missed, the message names it and the row itself is marked, so nobody has to hunt for the gap.

Use it in

Templates and solutions that use matrix questions

Template

Customer Satisfaction (CSAT)

“How would you rate the following aspects?” Product quality, customer support, value for money and ease of use, from Very poor to Excellent.

Template

Brand Awareness & Perception

“How would you rate each brand on the following attributes?” Quality, value, innovation, customer service and trustworthiness.

Solution

Customer feedback

Rate every part of the experience in one question.

Solution

Market research

Attribute ratings with tab-book summaries and SPSS-ready data.

Frequently asked questions

What is a matrix question?

A grid of rows and columns: each row is something to rate, and the columns are the answer scale they share. In the Surveti builder it’s called Grid (single) when each row takes one answer, and Grid (multi) when each row takes checkboxes.

How does a matrix question look on a phone?

When the screen is narrower than the table needs, each row becomes its own card with the points stacked as full-width buttons. You can choose wrapped chips, an accordion, a carousel, a segmented bar or the table instead for one question. For the whole survey, the theme can set the accordion, the carousel or the table.

How do I keep “Didn’t use” out of the average?

Tick Add a “Not applicable” column and type your own header, such as Didn’t use. It counts in the base but not in the Mean or Top-2-Box. In the example, 24 of 150 guests didn’t use breakfast, so breakfast’s Top-2-Box of 58% is out of the 126 who rated it. A column you mark Not on scale, like Can’t say, works the same way.

What does matrix data look like in an export?

One column per row, named by position, the question’s column name plus _1 to _5, holding the code of the point chosen and NA for Didn’t use. In SPSS (Professional+) each one is an ordinal variable with every point labelled, and Didn’t use is code 97, declared missing when you ask for it.

How do I compare rows across groups?

In Crosstabs (Solo+) pick a row, such as Wi-Fi, and cut it by any banner. On Professional+, letters mark differences in each point, Top-2-Box and the Mean. To see every row at once, a tab book (Research Team) adds a grid summary.

How many rows should a matrix have?

As few as you need, and rarely more than ten. Long grids invite the same answer all the way down. Surveti checks every single-answer grid with 4 or more rated rows for that, on by default, and adds a Straightliner signal on the Fraud page.

Can people pick more than one column in a row?

Yes, with Grid (multi). Mark a column such as None of us as Exclusive and it clears the rest of the row. Crosstabs show the share of each row ticking each column, and the export has one 0/1 column per row and column.

Related question types

Likert scaleRating scaleRanking

Add a matrix question in minutes.

Start free — no credit card. 50 question types, logic, and analysis in one place.