What the Schutte EQ Test Measures and How to Read It

What does the original Schutte scale ask you to rate?

The Schutte Self-Report Emotional Intelligence Test asks a person to describe themselves. Its 33 statements are answered by the respondent, so the resulting number begins as a summary of how that person reports their own emotion-related tendencies. It is not a transcript of a conversation, an observer’s record, or a demonstration made while solving an emotion problem. The distinction matters when someone sees a single total and is tempted to read it as a direct measurement of what they will do under pressure. The score represents responses to a self-description questionnaire; interpreting it as observed conduct requires evidence from somewhere beyond the questionnaire itself.

In “Development and Validation of a Measure of Emotional Intelligence,” the original development began with a pool of 62 items grounded in the authors’ model. Factor analysis using responses from 346 participants informed the shorter 33-item measure. That development process produced a broad total intended to represent emotional intelligence as the study had defined and operationalized it. The relevant object is therefore the scale built in that paper: a particular set of self-ratings organized into an overall score, rather than a universal EQ unit shared by every later instrument that uses the same label.

A respondent’s answer can reflect their memory, self-understanding, and standards for describing their usual tendencies. For instance, two people may encounter a similar disagreement but answer a self-rating differently because they notice different parts of their own response or use different reference points. That possibility does not make self-description meaningless: the person’s view of their typical emotional behavior is itself information. It does mean the original study’s intended broad score should be read as a summary of those answers. The study established the format and the intended construct for its measure; the response format does not turn a reported tendency into direct observation of ability or behavior.

The label “emotional intelligence” can make the total sound more immediate than the process behind it. In this instrument, the process runs from a written statement, through a person’s judgment about themselves, to a combined score. Each step contributes to what the number means. The total can compactly represent the respondent’s pattern of answers, but it cannot show which specific moment or relationship those answers came from. That is why the score’s original intended meaning belongs with its questionnaire format: it summarizes a person’s self-reported tendencies under the study’s definition, rather than certifying a general capacity independently of the respondent’s account.

Sources: Development and Validation of a Measure of Emotional Intelligence

What can a self-report tell you that a performance task cannot?

The Schutte scale’s method answers a question about a person’s self-view: how they describe their usual emotional tendencies. The Measurement of Emotional Intelligence: A Critical Review of the Literature and Recommendations for Researchers and Practitioners distinguishes that kind of self-report from maximal-performance measures, which ask a person to solve emotion-related problems under test conditions. A self-report records what someone says about themselves; a performance task records how they respond to the problems it presents. These methods provide different evidence, so a result from one should not be translated into the other’s terms without support.

That difference becomes practical when a report is used to discuss a difficult exchange at work. A self-rating may prompt a person to examine whether they usually notice tension, pause before replying, or believe they can understand another person’s reaction. It tells the reader how they endorsed descriptions of their own tendencies. A task can show how the respondent handled its particular emotion problems according to its scoring method. Neither result recreates the timing, history, power relationship, workload, or ambiguity of a live conversation. A test response is evidence about the response situation the instrument created; actual conduct in a meeting remains a different observation.

The word “typical” is doing real work here. People may report a pattern they recognize across ordinary situations even if their behavior varies by colleague, topic, or stress level. That self-perception can be useful to development precisely because it reveals the story a person currently tells about their own habits. A discrepancy between that story and a colleague’s specific feedback can become a question to investigate: what happened in that exchange, and what did each person notice? The self-report alone cannot settle whether the account matches behavior, but it can identify the person’s own starting point for reflection.

A performance result has its own boundary. Solving an emotion problem in a quiet assessment setting does not establish that someone will respond the same way during a tense handoff or a disagreement with consequences. Conversely, a person’s self-rating is not invalid simply because it is subjective; subjectivity is part of what the questionnaire captures. The useful distinction is what each method samples: a view of usual tendencies in the SSEIT, and responses to selected problems in a performance task. Neither is a complete record of how someone behaves with other people, so claims about a real interaction need evidence from that interaction rather than a score alone.

The two methods also place different demands on the person taking them. A self-report asks for a judgment about oneself, which may draw on remembered examples, self-knowledge, and the meaning the respondent gives to each statement. A performance task asks for a response to a presented problem and is interpreted using that task’s rules. The latter can reduce reliance on a person’s self-description for that particular result, but it still samples a designed set of problems. It does not make the task a neutral window onto every emotion-related choice the person might face. Reading either result well therefore starts by asking what the respondent actually did: endorsed descriptions of themselves, or answered the instrument’s problems. Naming that source of evidence keeps a self-view from being mistaken for demonstrated skill and keeps a task response from being mistaken for a full account of everyday conduct.

Sources: The Measurement of Emotional Intelligence: A Critical Review of the Literature and Recommendations for Researchers and Practitioners

Why can researchers map the same scale into different factors?

The original development paper organized the Schutte Self-Report Emotional Intelligence Test as a broad total. Later researchers asked a more specific question: do the responses in another sample cluster into the same dimensions, or does a different arrangement describe them better? Their answers depend partly on the participants and candidate models included. A factor solution is a proposed map of how item responses relate; a model that fits one dataset gives evidence about that dataset and model comparison. It does not by itself settle the structure of every version for every population. This distinction matters when a reader receives subscale labels: those labels carry an interpretation beyond the broad total, and need evidence tied to the form and people that produced them.

In “Factor structure analysis of the Schutte self-report emotional intelligence scale on international students,” researchers used confirmatory factor analysis with 640 international students studying at U.S. universities. They compared the original structure with five other published models. A modified nested model fit the responses, but the researchers judged it insufficiently parsimonious: its added complexity did not earn its place as the preferred account. They then proposed a different model that also fit and reported convergent and concurrent criterion-related validity, as well as reliability evidence, for that proposal. The repository abstract does not provide the full statistics, so the defensible report is the comparison and its stated outcome. The study shows that alternatives were testable and that the authors supported one proposal in this sample; it does not make that proposal the universal form of the scale.

A separate result came from “Validation of the Schutte Self-Report Emotional Intelligence Scale With American College Students.” This online study analyzed 404 American college students, including 322 women; 88.9 percent of the sample was White. Confirmatory factor analyses compared proposed structures, and the Ng et al. two-level model fit this sample best. “Best” here describes the candidates tested against those responses. It is a comparative result, not a population norm or a demonstration that every respondent’s score can be divided into those same levels. The sample description also sets a boundary around transfer: findings from this group do not automatically establish the same pattern among workers, other age groups, or people completing a changed or translated version.

The two studies therefore do not cancel each other out. Each tested factor arrangements using a defined sample, and each reported a candidate structure with support in that setting. The 640-person comparison rejected parsimony for one modified model before presenting another; the 404-person study identified the Ng et al. two-level model as the best-fitting tested option. Those are useful findings about model performance, not a single agreed map. Factor analysis can organize observed relationships among answers without proving that the resulting labels are distinct emotional capacities or that a reader’s own subscale has the same meaning elsewhere. That further interpretation needs evidence for the particular version, scoring, and population in front of the reader.

For someone looking at a report with several dimensions, the practical question is which version generated those labels and what evidence supports that exact scoring. A report that names a factor structure should make its version and basis clear. If it presents subscales without that connection, the original broad total is the more defensible summary of the instrument’s initial intent; the extra labels are hypotheses to interpret cautiously, rather than established facts about the individual. This does not dismiss the later studies: they reveal promising, competing ways to describe patterns in responses. It keeps the reader from treating a sample-specific fit as a universal key to their emotional behavior. A reader can therefore distinguish a reported score from the map used to divide it: the first is the result of answering the items, while the second is a research-backed proposal about how those answers cluster. If the report does not identify that proposal and its supporting version, the subdivisions add apparent precision without showing why those particular boundaries should guide interpretation.

Sources: Factor structure analysis of the Schutte self-report emotional intelligence scale on international students; Validation of the Schutte Self-Report Emotional Intelligence Scale With American College Students

Why can a consistent score still leave its meaning unsettled?

The original Schutte development paper reported two different kinds of consistency evidence, and the estimates came from separate samples. “Development and Validation of a Measure of Emotional Intelligence” reports internal consistency of .90 for the scale. It also reports a test-retest correlation of .78 over a two-week interval, based on 28 participants. The distinction in sampling matters: the retest estimate is not another summary of the larger development sample, and its small group provides a specific, short-interval result. Both figures are positive findings from the original work. They describe how the responses behaved under those particular analyses; they should not be blended into one general guarantee about the score.

Internal consistency concerns how closely the items’ responses move together in a sample. A high estimate can support the view that the items form a coherent set for that use. Test-retest consistency asks a different question: whether people’s scores are similar when they complete the measure again after an interval. The .78 figure addresses that repeat-score question over two weeks in the 28-person retest sample. The two estimates therefore speak to different forms of steadiness. Neither says that a respondent’s self-description is accurate in every situation, and neither shows that the score directly records emotional behavior. They tell us about patterns of scores, not a direct observation of what someone does in a conversation.

The distinction becomes important when a reader sees a precise-looking total. If repeated answers tend to cohere, the score may be less arbitrary than a collection of unrelated responses; if scores show some short-term stability, the result may persist enough to discuss. That gives a reader a reason to take the result seriously as a description generated by the instrument. It does not answer what the scale’s total represents beyond its intended self-report meaning, whether its proposed subscales are supported in another population, or whether two differently administered versions can be compared. Consistency can make an interpretation more dependable only after the interpretation itself has support.

The critical review “The Measurement of Emotional Intelligence: A Critical Review of the Literature and Recommendations for Researchers and Practitioners” discusses reliability evidence alongside broader questions about how emotional intelligence measures are constructed. Read with the original report, it helps separate a measurement property from the meaning assigned to a score. The .90 and .78 figures are not claims to discard: one concerns relationships among items and the other concerns repeat scores across two weeks. Their scope is simply narrower than statements such as “this measures ability” or “these subscales capture distinct skills.” Those claims require evidence directed at construct meaning, not only evidence that scores show consistency.

For an individual using a result to reflect on work, reliability matters because an unstable score would be a weak basis for deciding what to examine. The original estimates provide some reason to treat the questionnaire’s scores as more than random noise in the samples studied. Yet consistency alone cannot identify the best next behavior to change or establish that a number predicts how a person handles feedback, conflict, or pressure. A score can be coherent and repeatable while its interpretation remains tied to the measure’s design and supporting research. The useful reading is narrow: these studies reported consistency, while the question of what a particular score means still depends on evidence about the construct and version being interpreted. The retest estimate also covers a brief interval, so it cannot tell the reader whether scores remain stable over longer periods or through major changes in circumstances. That is a separate question from whether a total is coherent at one sitting.

Sources: Development and Validation of a Measure of Emotional Intelligence; The Measurement of Emotional Intelligence: A Critical Review of the Literature and Recommendations for Researchers and Practitioners

What can response wording change in a score?

A score can depend partly on how the questionnaire’s answer options work, not only on the broad tendency the items are meant to describe. “A Rasch rating scale modeling of the Schutte self-report emotional intelligence scale in a sample of international students” examined the original five-category response form among international students at U.S. universities. The repository abstract says those categories may not have functioned effectively in that sample. It does not report a sample size in the accessible record, so the result supports a question about this administration and population, not a verdict about every later version of the SSEIT.

Response categories are the ordered choices a respondent uses to indicate how well a statement describes them. A five-category format assumes those options provide useful gradations: that respondents can distinguish them and that moving from one category to the next carries interpretable information. The Rasch study’s concern is that this pattern may not have held for its sample. If neighboring choices are not functioning as intended, a total can still be calculated, but the apparent precision of a one-step difference may exceed what the response process supports. The abstract does not give enough detail to say which categories caused difficulty or how large the effect was; the practical point is to avoid treating the labels as self-validating measurement units.

The same study flagged reverse-keyed items. An item is reverse-keyed when agreement indicates less of the target tendency than agreement with a positively worded item, so its scoring direction must be reversed before combining it with the others. This can help prevent a respondent from receiving the same score simply by agreeing with every statement. It also creates a distinct interpretive demand: a negatively framed statement must be read and processed in the intended direction. The reported problem means responses to those items may have behaved differently in the studied group. It does not show that reverse wording always fails, or that respondents in other populations necessarily misread it.

Item targeting raised another concern. In a Rasch analysis, targeting asks whether the difficulty or location of the items matches the range of the people being measured. If the items cluster away from respondents’ levels, the questionnaire may distinguish some parts of that range better than others. A total can then conceal uneven information: two people may receive comparable sums even though the items were more informative for one than the other. The repository abstract identifies targeting as a concern but does not provide enough accessible statistics to quantify where the mismatch occurred. That makes it a reason to inspect the evidence behind a particular interpretation, rather than a basis for reconstructing an individual’s measurement precision.

The finding was not that the scale had no coherent signal. “A Rasch rating scale modeling of the Schutte self-report emotional intelligence scale in a sample of international students” also reports that most items contributed to one underlying construct. That positive result belongs beside the response-category, reverse-keying, and targeting concerns: item contribution to a common construct and the functioning of the response process are different questions. A set of items can broadly align while some answer options or item properties still complicate how finely a total should be read.

For a reader comparing a report with published findings, the practical check is version-specific. Confirm whether the report uses the original 33-item form or a changed, translated, or differently scored version, and whether its response categories match the form that was studied. Then ask whether the evidence being cited actually tested that form in a comparable population. A published result about the original five-category format among international students cannot automatically validate another response scale; nor does its sample-specific difficulty invalidate every version. The relevant evidence must travel with the exact form and scoring claim being made.

Sources: A Rasch rating scale modeling of the Schutte self-report emotional intelligence scale in a sample of international students

How far can a group association carry an individual interpretation?

An observed difference between groups can raise a useful question about what a questionnaire’s score is associated with; it cannot explain why one person received a particular result. “The construct validity of the Schutte Emotional Intelligence Scale in light of psychological type theory: a study among Anglican clergy” examined 364 Anglican clergy serving in the Church in Wales. The researchers paired the 33-item Schutte scale with the Francis Psychological Type Scales and reported higher Schutte scores among clergy classified as extraverts than introverts, intuitive than sensing, and feeling than thinking. The judging-perceiving comparison showed no significant difference. These are findings about groups in that specific study, not a diagnostic key for an individual respondent.

The paired measure matters because the result is an association between two sets of questionnaire classifications and scores. It can inform a construct question: do SSEIT scores vary alongside preferences measured by another instrument in this population? The study’s reported differences give a qualified yes for three of its comparisons, while the null judging-perceiving result shows that the pattern did not extend uniformly across every comparison. The result is therefore informative precisely at its stated level: certain group averages differed in this sample, and one tested comparison did not reach significance. It does not establish that the preference categories caused the score differences or that those categories reveal what any clergy member can do in a real interaction.

A group average and a person’s result answer different questions. A group comparison describes how scores were distributed across categories in the sample. It does not tell a reader which experiences, response habits, or circumstances produced a particular person’s score. Someone whose score sits above or below another respondent’s cannot infer from this paper that a preference label explains the difference. The paper does not report an individual-level mechanism that would support that leap. Even when a group association is genuine, people within each group can vary; the relationship between categories and scores does not turn the category into a personal account.

The design also limits what can be concluded. The study reports survey associations in one religious and occupational population at one period. It does not establish the direction of influence: the measured preferences could not be treated as a cause of higher or lower emotional-intelligence self-ratings on these findings alone. Nor does a difference in self-report scores demonstrate a difference in observed emotional behavior, because the Schutte measure records respondent ratings. The study did not test whether one group communicates better at work, handles disagreement more effectively, or produces stronger outcomes. Using it to rank colleagues or predict their conduct would substitute a new claim for the one the researchers examined.

The absence of a significant judging-perceiving difference deserves equally careful reading. It means this study did not find a statistically significant difference on that comparison; it does not prove that the categories are identical in every population or that no relationship could appear under another design. Conversely, the three reported differences should not be converted into a complete typology of emotional skill. The paper’s title and analysis use psychological-type theory to examine construct validity, but the pattern is not an endorsement of a general personality taxonomy or a set of work-style descriptions. Its scope remains the pair of measures, the specified comparisons, and the clergy sample.

A reader can use the finding to ask whether a proposed interpretation has evidence beyond the instrument that generated the score. The clergy study supplies one external association check for one population, and the mixed pattern—three differences, one nonsignificant comparison—helps show what such a check can and cannot contribute. It can sharpen questions about construct boundaries and motivate a replication in another group. It cannot explain an individual’s response history, identify a causal process, or establish a workplace consequence. Those claims would require different evidence, such as direct observation or studies designed around the outcome in question. The association is real within the report’s stated design; transfer to a particular worker remains untested by it.

Sources: The construct validity of the Schutte Emotional Intelligence Scale in light of psychological type theory: a study among Anglican clergy

How can you turn the result into a useful reflection?

Use a self-description as a prompt for examining one interaction, then separate what you remember from what you inferred. The point is not to decide whether the score was right. It is to find a small action you can notice in a similar moment. Keep four notes: the cue you remember, the meaning you gave it, what you did that another person could have observed, and what followed. These details make the reflection specific enough to learn from without turning a single recollection into a verdict about your character.

For illustration only, imagine a colleague asks for a late change to a handoff. You remember your shoulders tightening and the thought, “They think I missed something.” Your observable response is that you answer quickly, list the work already completed, and do not ask what prompted the change. The immediate consequence is that the request remains unclear and the exchange becomes defensive. This invented scene is not evidence about how people with a particular SSEIT result behave; it simply shows how to keep an internal reaction distinct from an action and its consequence.

Now turn the notes into a question you could check next time: was the tension a cue that you felt frustrated, or might it have come from surprise, time pressure, or uncertainty about the request? You do not need to settle that explanation to choose a useful behavior. For example, you could pause before answering and ask, “What changed that makes this update necessary?” That question is observable. It gives the other person room to explain and gives you better information before you decide how to respond. The experiment is the behavior, not a promise that the conversation will go smoothly.

Specific feedback can add another account of the same exchange. You might ask a colleague who was present what they noticed about your reply, naming the moment and the question you are trying to answer. “I felt defensive” describes your experience; “I interrupted before you finished” is a behavior another person may be able to confirm or correct. Their view can reveal something you missed, while still reflecting their position and memory. If accounts differ, keep the difference visible instead of averaging them into a new score. Ask what each person observed and what detail would help distinguish the explanations.

Afterward, record whether you tried the pause and clarifying question, what happened next, and what you would adjust. One interaction cannot establish a stable pattern, and one colleague’s account cannot settle what you intended. The exercise remains useful because it asks for a modest, testable change: in the next comparable handoff, notice the tension and ask one question before defending the previous work. If the situation differs, adapt the behavior to the cue rather than forcing the same script. You can then decide whether the action helped you understand the request, regardless of whether it changed the other person’s response. Keep the record brief: a sentence about the cue, a sentence about your response, and a note on what the question clarified are enough. A longer account can tempt you to argue your case after the fact. If you cannot remember the exchange clearly, mark that uncertainty and choose a future moment to observe instead of filling gaps with a confident story. That keeps attention on what you can practice, rather than on proving a preferred explanation of the earlier event.

What should you do after an SSEIT result?

Start with the report in front of you: check which SSEIT version it names and how it says the result was scored. Then choose one recent interaction and write down one behavior you want to try in a similar moment. Make the behavior small enough that you can tell whether you did it: pause before replying, ask what the other person needs, or name the feeling before explaining your position. Attach it to a cue you are likely to notice, such as a change request or a sharp disagreement, so it can guide an actual response rather than remain an intention. You can take that step without another questionnaire. The result is a starting point for reflection; the action comes from the interaction you want to handle differently. Give the experiment one comparable opportunity, then note what you did and what became clearer. The aim is a more deliberate response, not a target score or a promise about how anyone else will react.

If a separate guided reflection would help, explore EQ Test’s private 32-item [Emotional Skills Profile](/assessment). It is an educational self-reflection resource with its own questions and report, separate from the Schutte test; it is not a validated ability instrument.

Questions readers ask