Semantic differential · Rating

Semantic differential questions

Put a brand, a product or an idea between pairs of opposite words, such as Cheap and Premium or Boring and Exciting, and let people pick a point on each line. Surveti asks every pair on one screen, lays them out for phones, and gives you a column per pair to analyse.

Available onProfessional & up
The semantic differential as respondents see it: five pairs of opposite words on a laptop, and the same question on a phone with both words above each row

Live demo

Pick a point on each line. Your answer joins the grid.

This is the question your respondents get. Pick a number between each pair of words, and the results grid beside it counts your answer.

How would you describe Northwind Coffee?

  • CheapPremium

    1. 1
    2. 2
    3. 3
    4. 4
    5. 5
    6. 6
    7. 7

  • Old-fashionedModern

    1. 1
    2. 2
    3. 3
    4. 4
    5. 5
    6. 6
    7. 7

  • BoringExciting

    1. 1
    2. 2
    3. 3
    4. 4
    5. 5
    6. 6
    7. 7

  • UnfriendlyFriendly

    1. 1
    2. 2
    3. 3
    4. 4
    5. 5
    6. 6
    7. 7

  • Hard to findEasy to find

    1. 1
    2. 2
    3. 3
    4. 4
    5. 5
    6. 6
    7. 7

In action

The semantic differential in action

Semantic differential in action · A respondent places Northwind Coffee on five pairs of words, by mouse and by keyboard, and the grid updates.
  1. 0:00Five pairs of opposite words, a row of numbers between each.
  2. 0:03A click picks a point.
  3. 0:05The arrow keys move along a row and pick the point.
  4. 0:09With every pair picked, Next moves on.
  5. 0:12Results: the grid shows how many chose each point on each pair.

Every feature

Everything the semantic differential can do

Your own pairs of words, the number of points you want, pairs in random order, and a column per pair in your data.

Build it your waySet-up in the builder.

LiveProfessional+

Your own pairs of words

Type a left word and its opposite for each row, and add as many pairs as the question needs.

LiveProfessional+

3 to 11 points

Seven by default. Pick an odd number for a true middle; every pair uses the same scale.

LiveProfessional+

Randomise the pair order

Show the pairs in a different order to each person. Each pair keeps its words on the same sides.

The order someone saw isn’t saved with the answer.

LiveProfessional+

Required, soft required or optional

Required asks for every pair before Next; soft required warns once, then lets people leave pairs out; optional lets them with no warning.

LiveProfessional+

Translated with your survey

Every pair’s words translate with the rest of the survey.

Screen-reader wording around the words is in English.

On the respondent’s screenWhat people answering see and use.

LiveProfessional+

A row of numbers between two words

Numbered buttons sit between each pair, at least 44px tall by default, in your survey’s theme.

LiveProfessional+

Laid out for phones

On a narrow screen both words sit on one line above the numbers, so a right-hand word never reads as the next pair’s title.

LiveProfessional+

By keyboard

One Tab stop per pair; the arrow keys move along it and pick the point. Each pair is a radio group named by its two words.

LiveProfessional+

Your theme, light or dark

Words and buttons follow the survey’s theme and dark mode.

In your resultsThe Results tab.

LiveProfessional+

A grid for every pair

One row per pair, one column per point, and how many people chose it, shaded so the busiest points stand out.

Counts, not percentages or means.

LiveProfessional+

Filter to a segment

Filter Results to a group and the grid recounts for that group.

In your dataWhat lands in your exports.

LiveProfessional+

A column per pair

One column per pair, such as Q1_1, with both words in its label; 1 is the left word.

LiveProfessional+

Ordinal in SPSS

Each pair is an ordinal variable labelled with its two words.

Not part of the semantic differential yet: a not-applicable option, means and a profile chart, crosstabs and tab books, dashboards and report decks, branching or piping on a pair, and reverse-scored pairs.

Choosing

Semantic differential or a rating grid?

Both put several items on one screen. A semantic differential gives each row its own pair of opposite words. A rating grid asks every statement on the same scale, such as agree to disagree.

  • Semantic differential · Professional+

    Pick a semantic differential when

    • You want to know how people see something: its image, personality or feel.
    • Each quality has a natural opposite, such as Cheap and Premium.
    • You don’t want to put words in their mouths with a statement to agree with.
  • Rating grid · Free

    Pick a rating grid when

    • Every item fits one scale, such as satisfaction or agreement.
    • You want a mean and top-2 box per statement in Results, and statement rows in crosstabs (Solo+).
    • You need a not-applicable column.

Rating just one thing on one scale? Use a rating scale (Free).

Build it

Add a semantic differential

  1. 1

    Add the question

    In the builder, choose Semantic Differential from the Rating group, and name what people are rating in the question, such as “How would you describe Northwind Coffee?”

  2. 2

    Type your pairs

    Under Attribute Pairs, replace the starter pairs with your own: the left word, and its opposite on the right. Choose Add pair for each one after that.

  3. 3

    Choose the number of points

    Set Scale Points. Seven is the default; an odd number gives a true middle.

  4. 4

    Shuffle the pairs, if you like

    Tick Randomize pair order to show the pairs in a different order to each person. Each pair keeps its words on the same sides.

  5. 5

    Decide whether every pair is needed

    Required asks for every pair before Next. Soft required warns once, then lets people leave pairs out. Optional lets them with no warning.

The Semantic Differential editor with the question text, 7 in Scale Points, the five pairs under Attribute Pairs, the Randomize pair order box and a preview of the first two pairs
EditorPairs, points and order in one place.
The question’s Rules tab with Response requirement set to Required
RulesRequired means every pair.
The question on a phone: both words above each row of seven points, and a point picked on every pair
On a phoneBoth words above each row.

The data you get

A column per pair, 1 at the left word

Each pair lands in its own column, named after the question, with both words in the column label. 1 is always the left word and the highest number the right.

respondent_idQ1_1Q1_2Q1_3Q1_4Q1_5
47762a85-0001-4000-8000-00000000000155362
47762a85-0002-4000-8000-00000000000275561
47762a85-0003-4000-8000-00000000000365513
47762a85-0004-4000-8000-00000000000467731
47762a85-0005-4000-8000-00000000000576747
47762a85-0006-4000-8000-00000000000655356
47762a85-0007-4000-8000-00000000000765476
47762a85-0008-4000-8000-00000000000865574

Trimmed to the question’s columns. Every export also carries the respondent columns (status, language, timing and more).

One column per pair

Named after the question and the pair’s place, such as Q1_1 to Q1_5, labelled “How would you describe Northwind Coffee? Cheap – Premium”.

1 is the left word

Points run from 1 at the left word to 7 (or your top point) at the right. If you mix sides, recode those pairs in your stats tool.

Skips stay empty

On an optional or soft required question, a pair someone left out is an empty cell, never a guess.

Shuffled order, same columns

Randomising the pairs never moves the columns: each answer stays with its pair.

Profiles in your stats tool

Average each column of the CSV or SPSS export to draw the classic profile line of means, brand against brand.

Charts & analysis

Every pair, every point, at a glance

Results shows a grid for each semantic differential: one row per pair, one column per point, and how many people chose it. Filter to a segment and the grid recounts.

Northwind Coffee: how people describe itResults · semantic differential grid · 200 answers

The Results grid over 200 completed responses: a row per pair, from Cheap ↔ Premium to Hard to find ↔ Easy to find, the points 1 to 7 across the top, and how many people chose each one, shaded by count
Stronger shading means more people chose that point. Read across a row to see where opinion sits between the two words.

Where it charts

  • ResultsA grid of counts per pair
  • ExportsA column per pair
  • DashboardNot yet
  • Report deckNot yet
  • CrosstabsNot yet
  • Tab booksNot yet

Analysis that applies

  • Filter to a segment

    Professional+

    Filter Results to a group, such as people who buy coffee weekly, and the grid recounts for that group.

  • Translated words in Results

    Professional+

    Pick a survey language in Results and each pair you translated shows in that language.

  • Means and profiles from the export

    Professional+

    Each pair is its own column, ordinal in SPSS, so a mean per pair or a profile line is one step from the CSV or SPSS export, in SPSS, R or Excel.

How Surveti compares

Semantic differentials across the common survey tools

Semantic differential capabilities in Surveti, SurveyMonkey, Qualtrics, Typeform and Alchemer, as of 7 November 2026
CapabilitySurvetiSurveyMonkeyQualtricsTypeformAlchemer
A semantic differential question typeyesIts own type · Professional+no1Matrix / rating scale insteadyes3Matrix, Bipolar variationno7Opinion scale per pairpaid10, 11Full Access only
Its own pair of opposite words on each rowyesno2Rows are items; shared columnsyes3partial7One pair per opinion scaleyes10left|right rows
Number of points set by the authoryes3–11, default 7partial2Matrix columnsunclear3partial75–11 steps, one pairyes10Editable columns
Randomise the pair orderyesOrder isn’t savedpartial2Matrix rowspartial3Matrix tables in generalunclearunclear10
Not-applicable optionnopartial2With weights onlyunclear3unclearunclear10
Require every rowyesEvery pair before Nextyes2unclear3unclearunclear10
Per-row distribution in resultsyesGrid of countsyes2Counts or %unclear3unclearyes10Values and %
Mean per row in resultsnoyes2With weightsunclearunclearunclear10
One export column per rowyesWords in the labelpartial2Coded by positionyes3unclearyes10
SPSS exportyesOrdinal · Professional+unclearpartial5On the free accountno9CSV or XLSX onlypaid11Full Access
On the entry planpaidProfessional+unclearpartial5Matrix listed; variations notno7No such typeno11Not on Collaborator
As of 7 Nov 2026, from each vendor’s public help, developer and pricing pages.yespartialpaidunclearno

Partial means partly built in or possible only with a documented workaround; paid means built in on paid plans or licences; unclear means the vendor’s public docs don’t settle it. QuestionPro has a Bipolar Matrix with a slider answer type; Alchemer codes its points −2 to +2 by default and hides the numbers from respondents. Google Forms has no semantic differential type.

Sources (8)
  1. [1] SurveyMonkey: Question types. https://help.surveymonkey.com/en/surveymonkey/create/question-types/ · accessed 7 Nov 2026
  2. [2] SurveyMonkey: Matrix / rating scale. https://help.surveymonkey.com/en/surveymonkey/create/matrix-rating-scale/ · accessed 7 Nov 2026
  3. [3] Qualtrics: Matrix table (Bipolar). https://www.qualtrics.com/support/survey-platform/survey-module/editing-questions/question-types-guide/standard-content/matrix-table/ · accessed 7 Nov 2026
  4. [5] Qualtrics: Free accounts. https://www.qualtrics.com/support/survey-platform/managing-your-account/trial-accounts/ · accessed 7 Nov 2026
  5. [7] Typeform: Create API reference. https://www.typeform.com/developers/create/reference/create-form/ · accessed 7 Nov 2026
  6. [9] Typeform: Download your data. https://help.typeform.com/hc/en-us/articles/360029253572 · accessed 30 Oct 2026
  7. [10] Alchemer: Semantic Differential. https://help.alchemer.com/help/semantic-differential · accessed 7 Nov 2026
  8. [11] Alchemer: Compare licences. https://www.alchemer.com/plans-pricing/compare-licenses/ · accessed 7 Nov 2026

What’s different here

A real question type

A semantic differential is its own question, with its own pairs, not a grid or a stack of single questions dressed up as one.

Built for phones

On a narrow screen both words sit on one line above each row of numbers, so a right-hand word never reads as the next pair’s title.

Keyboard friendly

One Tab stop per pair, arrow keys to move along it, and a visible focus ring on every point.

Your words, in their language

Every pair translates with the rest of the survey, and Results shows each pair you translated in the language you pick.

Shuffle without losing track

Randomise the pair order; each answer still lands in its own pair’s column.

Clean columns out

One column per pair, both words in the label, ordinal in SPSS.

When to use it

Use it to see how people picture something

When to use it

  • Brand image and personality: modern or old-fashioned, friendly or cold.
  • First impressions of a pack, an ad or a product concept.
  • Comparing how two brands are seen, one question per brand.
  • Attitudes in academic work, where Osgood-style scales are the standard.

When to reach for something else

  • Every item fits one agree-to-disagree scale. Use a rating grid.
  • Some people can’t judge some pairs. There’s no not-applicable option, so make the question optional rather than forcing a middle answer.
  • You need a mean per pair on a dashboard or in a client deck. Those charts aren’t available for this type yet.
  • You want to rank qualities against each other. Use MaxDiff or ranking.

Sample wording

“How would you describe Northwind Coffee?” Cheap … Premium · Old-fashioned … Modern · Boring … Exciting · Unfriendly … Friendly · Hard to find … Easy to find

Best practices

  • Use true opposites, not two related words: Cheap and Premium, not Cheap and Low quality.
  • Pick an odd number of points, 5 or 7, so there is a real middle.
  • Keep the positive word on the same side, or note the pairs you flip and recode them when you analyse.
  • Keep it to around five to ten pairs, and shuffle them if order could sway the answers.
  • Finish your pairs before the survey goes live, and preview it on a phone.

Accessibility

Plain radio groups, one per pair

Named pairs

Each pair is a radio group named by its two words, inside a group named by the question.

One Tab stop per pair

Tab moves between pairs; the arrow keys move along a pair and pick the point.

Visible focus, tall targets

Every point shows a focus ring and is at least 44px tall by default.

Readable on phones

On narrow screens both words sit above the numbers, so the row reads in order.

Use it in

Where the semantic differential fits

Solution

Market research

Brand image and concept tests, side by side with MaxDiff, conjoint and grids.

Solution

Academic

Osgood-style attitude scales, with clean columns for SPSS and R.

Solution

Agencies

How a client’s brand is seen against its rivals, one question per brand.

Example

Northwind Coffee

“How would you describe Northwind Coffee?” on five pairs, from Cheap … Premium to Hard to find … Easy to find.

Frequently asked questions

What is a semantic differential question?

A question that puts the thing you’re asking about between pairs of opposite words, such as Cheap and Premium, with a row of points between each pair. People pick the point that fits best. It was developed by Charles Osgood to measure what things mean to people, and it’s widely used for brand image.

How many points can each pair have?

From 3 to 11, with 7 as the default. An odd number gives a true middle point. Every pair in the question uses the same number of points.

How do the answers come out in my data?

One column per pair, numbered 1 at the left word up to your top point at the right, with both words in the column label. In SPSS each column is an ordinal variable.

Does Surveti draw the profile chart of means?

Not yet. Results shows a grid of how many people chose each point on each pair. For a mean per pair or a profile line, average the columns of the CSV or SPSS export in SPSS, R or Excel.

Can I show the pairs in a random order?

Yes. Tick Randomize pair order and each person sees the pairs shuffled, with each pair’s words on the same sides. Answers always land in their pair’s column. The order someone saw isn’t saved with the answer.

Can people skip a pair or answer “not applicable”?

There’s no not-applicable option. A required question asks for every pair; make it optional or soft required to let people leave pairs out, and those cells stay empty in your data.

Which plan do I need?

The semantic differential is on Professional+, with every setting, the Results grid and CSV, Excel and SPSS exports.

Related question types

MatrixRating scale
  • Slider List A slider per row on a numeric range. Professional+
  • Star Rating Grid Stars for each row. Free

Add a semantic differential question in minutes.

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