What Counts as a Good EQ Score, and Why the Number Is Soft

Why there is no equivalent of an IQ of 100

IQ has a standardised metric because the field converged on one: a mean of 100, a fixed spread, and large reference samples that are periodically renormed. Emotional intelligence never converged. Competing research groups defined the construct differently — as a set of cognitive abilities, as a cluster of personality-like traits, or as a mix of competencies — and each built instruments on their own definition with their own scoring.

The practical consequence is that a score of 75 means whatever the test that produced it decided it means. Some report percentiles against their own sample, some report a 0–100 index with no sample behind it at all, and many commercial tests do not publish their normative data. If a test reports a percentile without saying which population it was drawn from, the precision is decorative.

This is why interpreting bands rather than numbers is the honest approach, and why any test that tells you your EQ to a specific point should be read with suspicion about the point, not the direction.

Ability tests and self-report tests are measuring different things

An ability measure — the MSCEIT is the best-known — works like a cognitive test. It asks you to identify the emotion in a face, or to judge which response would best regulate a described feeling, and answers are scored against consensus or expert judgement. There are right answers, which means the test can catch you being worse at something than you believed.

A self-report questionnaire asks you to rate your own emotional skills. That is a different measurement: it captures your confidence in those skills, which is related to but not the same as having them. Self-report EQ correlates strongly with personality traits — particularly low neuroticism, high extraversion and high conscientiousness — to the point where some researchers argue it is largely a rebranding of existing trait measures rather than a distinct ability.

Neither is worthless and the distinction matters for how you read your result. Almost every free online EQ test, including the one on this site, is self-report. What it gives you is a structured account of how you see your own emotional functioning across separable skills, which is genuinely useful for spotting an imbalance — and it cannot tell you that you are good at reading people, only that you believe you are.

Read the four skills, not the total

The total is the least useful number a well-built EQ test produces, because the component skills are only loosely related and averaging them hides the thing you would act on. Four separate abilities are usually distinguished, and being strong at one says little about the others.

Noticing your own emotion as it happens is the first and it is more variable than people expect — plenty of people can describe their feelings well in retrospect and cannot identify them in the moment. Regulating what you notice is a distinct skill: knowing you are angry and choosing what to do next are different capacities, and the gap between them is where most interpersonal trouble lives.

Reading other people is a third, and it dissociates from the first two more than any of them — being highly accurate about others while poor at your own internal signals is a common and coherent profile. Acting on what you read is the fourth, and it is where accuracy becomes useful or does not: knowing a colleague is upset and doing something workable about it are separate abilities.

A profile that is high on noticing and low on regulating calls for entirely different work than one that is high on regulating and low on reading other people. That is the actionable output, and it is why a single EQ number is the wrong place to look.

Three different things are called an EQ test

Part of why an EQ score is hard to interpret is that the field never settled on one definition, and the three that survived are measured in incompatible ways. Knowing which kind you took is the difference between reading your number correctly and reading it as something it is not.

Ability models treat emotional intelligence as a genuine mental aptitude, on the same footing as verbal or spatial reasoning, and measure it with tasks that have better and worse answers scored against expert consensus (Mayer et al., 2016). Trait models treat it as a constellation of emotional self-perceptions sitting inside normal personality space, and measure it by asking you — which is why trait EQ correlates substantially with established personality dimensions, particularly low neuroticism and high extraversion, and in factor analyses lands largely inside that space rather than outside it (Petrides et al., 2007). Mixed models bundle emotional skills together with motivation, optimism and social competencies, which makes them popular in corporate training and the hardest of the three to pin down.

The practical consequence shows up when the three are compared on the same outcome. In a meta-analysis of 158 studies covering 42,529 students, ability tests predicted academic performance best at a correlation of about .24, mixed measures came next at .19, and self-report was weakest at .12 (MacCann et al., 2020). Free online tests, including the one on this site, are essentially all self-report, because ability testing needs scored items and a validated consensus key.

So the meaning of a free EQ test result is narrower than the label suggests, and it is not nothing. It is a structured account of how you currently see your own emotional functioning — which is the strongest suit of self-report measures, since they correlate more closely with well-being and life satisfaction than with performance outcomes.

So what is the highest score, what is average, and what does out of 100 mean?

These are the three questions that follow immediately from being handed a number, and all three have the same root answer: they are properties of the test you took, not of emotional intelligence. There is no ceiling, no population mean and no standard denominator, because there is no standard scale for any of them to be defined against.

The highest possible score is whatever the instrument's own scale tops out at. On a questionnaire that sums agreement ratings, the maximum is every item answered at the top of the range — which is reachable, and reaching it tells you that you endorsed every statement rather than that you have hit some limit of the construct. That is a different kind of maximum from an ability test's, where the ceiling is a count of correct answers. Neither maximum transfers to the other test, and neither is an EQ of 100 in the sense that an IQ of 145 is a position in a known distribution.

Out of 100 is the most misleading of the three, because the denominator looks like IQ's and is usually doing something much simpler. A 0–100 display is often a rescaled sum of Likert responses: the test added up your answers and stretched them onto a familiar-looking range. That is a within-test position, not a percentile, and 72 out of 100 on one test can sit above or below 72 on another. The distinction worth checking is whether the test names the reference sample its number is drawn from. If it does not, treat the two digits as a location on that test's own range and nothing more.

Average is the one with the most interesting answer. If a test reports percentiles against its own respondents, then average is the fiftieth percentile by construction — it cannot be anything else, and it moves as the sample changes. For a free online test that means the average describes the people who chose to take a free online emotional intelligence test, which is not a general population. Self-selection runs in a particular direction here: the people who seek out this measurement are disproportionately those who already think about their emotional functioning. So a score at the average of that group is not a score at the average of people in general, and there is no way to convert between them without normative data the test would have to publish.

What emotional intelligence does and does not predict

The honest summary is that it predicts real things by modest amounts, and that the popular claims run well ahead of the evidence. The widely repeated figure that EQ accounts for the large majority of career success does not come from peer-reviewed research, and no study supports a number of that size.

What meta-analyses do find is that emotional intelligence predicts job performance with small to moderate effects, larger in roles with high emotional demands — nursing, teaching, sales, customer-facing work — and close to negligible in roles with few. It also shows incremental prediction beyond personality and cognitive ability in some analyses, though how much is contested and depends heavily on which instrument was used.

It is also more trainable than most traits, which is the genuinely encouraging part. Emotion regulation strategies in particular respond to practice, and the skills a good report separates out are specific enough to work on individually. That is a better use of a result than the number: treat it as identifying which of four skills to practise, not as a rank.