20-question self-construal test. Independent and interdependent self-construal are two separate dimensions — you can be high on both, or low on both.
Self-Construal Test — independent and interdependent
Twenty questions, two scores. One measures independent self-construal, the other interdependent — two separate dimensions rather than two ends of one line, so being high on both, or low on both, is an ordinary result.
Agree with every single statement and both dimensions come out at exactly 30 — that is what balanced keying is for: agreement alone cannot manufacture a high score.
How to take it
Answer for how you generally are
Not how you think you ought to be, and not your best day. The two halves measure two separate things — agreeing with a lot of both, or with little of either, is an ordinary result.
Do not work around the reversed items
Half the statements are written toward the low pole of their dimension. You do not need to handle them carefully; answer them straight. The balance exists so that agreement alone cannot produce a high score.
Read the two numbers first
Each dimension runs 10 to 50 with a midpoint of 30, and the two scores are the result. There is no total and no difference score — either one would erase the two most interesting profiles.
Then treat the quadrant as a reading aid
High means strictly above 30; exactly 30 resolves low. When a score sits close to 30 the label is unstable, so read the neighbouring profile as well.
About this test
Two dimensions, not one spectrum
Self-construal is the psychological term for how a person puts together the answer to who I am. An independent self-construal builds it from internal attributes — preferences, opinions, abilities, goals — which are expected to stay recognisably the same from one situation to the next. An interdependent self-construal builds it from relationships, roles and context: who you are connected to, what is expected of you, and where you stand are not facts about your circumstances but part of the person. Both descriptions come from Markus and Kitayama in 1991, and nearly every popular treatment drops the part that matters most: these are two separate dimensions, not two ends of one line. A person can be high on both, with both frames available, or low on both, with neither doing much work. Drawn as a single spectrum running from Western individualism to Eastern collectivism, neither of those people has a place on the diagram — and neither is rare.
So this test gives you two scores and stops there: no total, no individualism index, no difference score. A difference would look tidier, but it cannot tell a 45/45 respondent from a 20/20 one, and those are the two most interesting people in the table. Each dimension has ten items, five keyed toward it and five toward its low pole and reverse-scored, and each is summed to a score between 10 and 50. High means strictly greater than 30; exactly 30 resolves low. That tie rule is our disclosed convention, matching the one on our attachment test — a product decision, not a research finding. The four quadrant names are reading aids for the two numbers. When either score sits close to 30 the name is unstable, because one answer moved one step can flip it, which is why the durable result is always the pair of scores rather than the label. And because there is no norm sample behind these original items, neither score is ever reported as a percentile — only as a position on the scale we print.
You are not a data point in a culture. You have two scores.
Where the items came from
We wrote the twenty statements ourselves. The field-standard instrument is the Self-Construal Scale (Singelis, 1994), whose copyright is controlled by the publishing journal; we located no commercial-use grant from a primary source, so we did not use it. Rejected in the same sweep was the Schwartz PVQ-21 and ESS Human Values Scale, licensed CC BY-NC-ND 3.0 — non-commercial and no derivatives — which fails twice over, since we need commercial use and a Simplified Chinese version, and a translation is a derivative. Our rule is blunt: a licence that needs an email to confirm counts as unlicensed, not pending. That leaves one route. Constructs are not copyrightable and item text is, so the items are original compositions from the published construct definitions, with the Chinese bank written natively rather than translated. Which instruments were rejected, on what evidence, plus the scoring rule and worked arithmetic, are published in this tool provenance record.
The four profiles
High on both
Two working frames. It buys the ability to hold two accounts of the same situation as legitimate; it costs a second full evaluation of every socially loaded decision.
More independent
The self as its own reference point: legible, portable, clear about what it wants. The blind spot is not coldness but bandwidth on what goes unsaid.
More interdependent
A self constituted in relationships: reads a room before it hardens, carries obligations without evaluating them. The cost is a slow erosion of your own preferences.
Low on both
Neither frame is doing much work. That can mean an anchor elsewhere — a craft, a body of ideas, a long project — or simply a sheet answered down the middle.
Frequently asked questions
-
Because independence and interdependence are two separate dimensions, not two ends of one line. The popular version puts individualism at one end and collectivism at the other, so being high on one means being low on the other by definition. That model cannot represent a person who is high on both, or low on both, and both are ordinary results. Combining the two numbers into an index — or into a difference score — throws away exactly what makes the result worth reading: 45/45 and 20/20 have the same difference and describe very different people. So we print both and never average them.
-
No, and that framing is wrong on its own terms. Cultural averages do differ, measurably, and those differences shape institutions and defaults. But within-culture variation exceeds between-culture variation: the spread of people inside any one national group is much larger than the average gap between groups. A national average therefore tells you very little about an individual, including you. Your result is two scores about you, not a placement of your culture — and a strongly independent person in a strongly interdependent society sits inside the ordinary spread rather than outside it.
-
No. Neither dimension is the healthy one and neither is the problem. Both describe complete ways of being a person, each with genuine strengths and genuine costs, and the result pages spend equal effort on both. There is no research finding that a strong lean in either direction is better for wellbeing, and none making a low score on both a deficit to be corrected. If a page ever reads like a verdict on you, that is a failure of our writing rather than a finding.
-
No. This is not a clinical instrument and it cannot identify any condition. It measures neither ability nor wellbeing, and nothing here is assessed by a professional. Our items have no validation studies and no norm sample, which is why we never print a percentile — only a position on the 10 to 50 scale we show you. Treat the result as a vocabulary for thinking about yourself. If something in your life feels persistently hard, a person who can ask follow-up questions will see far more than twenty statements can.
-
To stop agreement itself from producing a result. Within each dimension, five statements are worded toward it and five toward its low pole and reverse-scored, so someone who agrees with everything gets exactly 30 on both dimensions rather than a high score. The balance also lets us compute a response-style check: the average of your raw answers should sit near the middle, and when it does not we say so on the result. Reverse items are written as real statements of the low pole, not as forward items with a negation added.
-
We wrote them. The 20 statements are original RECATOOLS work, composed from the published definitions of the two constructs. The field-standard instrument, the Self-Construal Scale (Singelis, 1994), is journal-copyrighted and we located no commercial-use grant from a primary source, so we did not use it — our rule is that a licence needing an email to confirm counts as unlicensed, not pending. Constructs are not copyrightable; item text is. The full sourcing record, including which instruments we rejected and why, is published in this tool provenance file.
-
No. Everything is scored in your browser. Your answers are written to your device local storage only so you can close the tab and resume where you left off, and that copy is cleared when you finish. Nothing is sent to us, nothing is attached to an account, and there is no record on our side that you took this test at all. That also means we cannot recover a result for you later — if you want to keep it, save or share the page yourself.
-
Probably a little, and possibly more than a little. Self-construal is more context-dependent than a personality trait: it moves with the situation you have in mind while answering, and in the priming research it moves with the language you answer in. Answering with your family in mind gives a different sheet from answering about work. That is a finding about how the self works rather than a fault in the test — and it is why retaking with one specific situation in mind is more informative than retaking in general.
-
A dimension counts as high only when it is strictly greater than 30, so exactly 30 resolves low. That is our disclosed convention, chosen to match the tie rule on our attachment test rather than be argued case by case. More usefully: any score within about three points of 30 makes the quadrant name unstable, because one answer moved one step can change which page you land on. The two numbers are the durable result, so read the neighbouring profile too when a score sits near the line.
-
Neither. It does not sort you into a type — there are no types here, just two numbers on two scales. It also does not measure what you value or what you think matters in life, which is a different research tradition with its own instruments. What this describes is narrower: how your sense of yourself is put together, out of internal attributes, out of relationships and roles, out of both, or out of something these twenty questions did not ask about.
Method & sources
What this tool implements
- Independence and interdependence are scored as TWO INDEPENDENT DIMENSIONS. The quadrant is a reading aid over the two scores, never a fifth number, and the page says so
- The tie rule is disclosed, not incidental: a dimension counts as high only when STRICTLY above the 30 midpoint, so a score exactly on 30 resolves LOW. This matches the attachment test's convention and is pinned by test on both sides plus exhaustively across all 1681 cells of the decision surface
- The label is named as the least stable part of the result. A dimension within 3 of the midpoint flips the quadrant on retake, and the borderline note says WHICH dimension rather than hedging about both
- The East-versus-West corrective is stated accurately as a claim about VARIANCE — within-culture variation exceeds between-culture variation — and not as a claim that cultural differences do not exist
- Neither dimension is positioned as modern, advanced, traditional or limiting. The audience is overwhelmingly Chinese and the page must be equally interested in both, with zero condescension in either direction
- The undifferentiated quadrant is presented honestly as BOTH a real profile and a likely artefact of consistently mid-scale answering, and is never framed as a deficiency
Sources
- Markus HR, Kitayama S. Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review 1991;98(2):224-253 — the framework and the two-dimensional structure (framework only; no item text used)
- Singelis TM. The measurement of independent and interdependent self-construals. Personality and Social Psychology Bulletin 1994;20(5):580-591 — REJECTED as an item source: SAGE, journal-controlled copyright, no commercial-use grant located from a primary source. Verified 2026-08-11
- ERQ (Gross & John 2003) — REJECTED: the Stanford lab grants free use but requires permission to translate, adapt or modify, and we ship zh-CN
Abridged — the full review record for this tool runs longer than the list above.
Pick up where you left off
Stored only in this browser — never sent to our servers.