Grid Matrix Survey Questions

Grid Matrix questions are designed for repeated measures that share the same response scale. They are common in academic survey software because social science research often uses batteries of agreement items, trust measures, policy attitudes, satisfaction ratings, and Likert-style scales.

Use a grid when the rows are statements or items and the columns are the same answer options for every row. If each row needs a different scale, split the measures into separate questions. For scale-specific guidance, see Likert Scale Survey Questions.

• Radio grids collect one categorical scale point per row.
• Slider grids collect numeric ratings for repeated items. Each row's handle starts unset unless you turn on Show a starting position; see the anchoring caveat below.
• Row and column order controls help reduce order effects.
• Row-subset randomization shows each respondent a random subset of the authored rows, useful for a long battery where every respondent doesn't need every item.
• Row and column locking/anchoring keeps specific rows or columns included under a random subset, or pinned to their authored position under shuffle — independent of each other.
• Checkbox rows (allow multiple selections per row) support an exclusive "None of these" column and a forced-inclusive "Select all" column, the same mechanism used by Multiple Choice.
• Mobile-friendly layout presents rows in a respondent-friendly format on small screens.
• Section breaks split a long grid into pages of a fixed number of rows, reducing straightlining risk on long batteries.
• Row and column shading add zebra striping for readability, and answered rows can highlight once completed.
• An attention-check row instructs respondents to pick a specific column on one statement, flagging anyone who doesn't as a failed attention check.
• Column recodes make Likert exports easier to analyze.

Rows support the same randomization controls as other question types: row order (fixed, flipped, or shuffled) and an optional random subset that shows each respondent fewer than the full set of rows. A row can be locked (always included when a subset is shown) and separately anchored (stays at its authored position under shuffle) — the two are independent, so a row can be locked, anchored, both, or neither. Columns support order and anchoring the same way, but have no subset concept: a shared rating scale should show the same scale points to every respondent.

For a long battery of statements, turn on "Break up block after every N rows" to split the grid into sections of a fixed size. With Page break between sections on (the default), each section is its own survey page; with it off, the sections appear on the same page as separate, smaller grids, each with its own column headers. The editor Preview always shows the sections as separate grids. This is a concrete mitigation for straightlining on long grids, not just detection after the fact — splitting a long battery into smaller sections has been shown to reduce missing data and satisficing compared to one long grid.

When "Allow multiple selections per row" is on, each row renders checkboxes instead of radio buttons. Mark one column as exclusive (e.g. "None of these") to clear a row's other selections when it's chosen, and mark one column as forced-inclusive (e.g. "Select all") to select every other eligible column in that row when it's chosen. This is the same mechanism used by Multiple Choice, so behavior stays consistent across question types.

Selection limits set a minimum and/or maximum number of columns per row, like Multiple Choice selection limits. A limit applies to a row only once the respondent has answered that row. Instrument Review warns when the minimum per row is higher than the maximum or than the number of columns.

Turn on "Add an attention-check row" to designate one statement as an instructed-response item: its text is replaced with an instruction (e.g. "please select Strongly Agree for this item"), and any respondent who doesn't select the required column is flagged as a failed attention/manipulation check. That row is excluded from straightlining detection, since it's an instructed answer rather than a genuine data point.

A Grid Matrix question can also trigger a follow-up question based on a specific row's answer — for example, showing a clarifying question only to respondents who selected a particular column on a particular row. Configure this the same way as any other follow-up trigger, choosing "Grid cell" as the trigger type.

1. Step 1: Open Block Builder. Open the survey and select the block where the battery belongs.

   Step 1: Open Block Builder

2. Step 2: Add a Grid Matrix question. Choose Grid Matrix from the question type selector.

   Step 2: Choose Grid Matrix question type

3. Step 3: Add rows. Create one row for each statement, item, or repeated measure.

   Step 3: Add rows

4. Step 4: Add shared columns. Add the response scale, such as Strongly Disagree through Strongly Agree, and use Likert Scale Survey Questions for scale-specific guidance.

   Step 4: Add shared columns

5. Step 5: Choose the input style. Use radio inputs for categorical scales or slider inputs for numeric repeated ratings.

   Step 5: Choose input style

6. Step 6: Set recodes and order. Configure row order, column order, mobile layout, and recode values before publishing.

   Step 6: Set recodes and order

7. Step 7: Preview mobile layout. Open Preview and check whether the matrix is readable on a small screen.

   Step 7: Preview mobile layout

Grid Matrix exports one response per row, with each row tied to the selected column or slider value. Keep row labels stable after launch so longitudinal or multi-wave social science research data stays comparable.

• Slider Survey Questions
• Semantic Differential Survey Questions for bipolar-adjective rating scales
• Multiple Choice Survey Questions
• Card Sort Survey Questions when grouping items is the construct of interest
• Create Experimental Conditions
• Set Survey Quotas