What the Workplace EQ Training Review Leaves Unanswered

What did the latest workplace review leave unanswered?

The 2024 review, “Training emotional competencies at the workplace: a systematic review and metaanalysis,” found average improvement on the emotional-competency measures its studies used. It did not establish whether training made workers more accurate at recognizing emotion or changed how they acted during workplace conflict: the included interventions did not target those outcomes. The distinction matters if you want to know whether a course changes what happens in a difficult conversation, rather than whether a broad score rises.

The review’s last search update was November 2022, so its finding describes the evidence it could identify by that date. Its controlled pooled estimate is also reported two ways within the article: 0.46 standard deviations in the abstract and 0.47 in the results. Both indicate a positive average measured change; the small discrepancy does not resolve the unanswered behavior questions. The review supports cautious optimism about measured competencies, while leaving accuracy and conduct during disagreement open to evidence beyond its search and outcomes. For a reader weighing a course, that means a higher general measure is encouraging but cannot stand in for evidence about the specific response they hope to change at work.

Sources: Training emotional competencies at the workplace: a systematic review and metaanalysis; Can emotional intelligence be trained? A meta-analytical investigation

What did the review include—and what fell outside its answer?

The absence is meaningful within a defined search, not a verdict that no relevant study exists or that emotional-awareness and conflict training cannot work. In “Training emotional competencies at the workplace: a systematic review and metaanalysis,” the authors examined interventions for employed adults that aimed to develop emotional competencies at work. They searched for emotional awareness and conflict management as specific targets, but found no qualifying interventions in either area. Their evidence set therefore cannot answer whether workplace training improved accuracy in reading emotion or behavior in conflict; it can say those questions were not represented among the studies meeting the review’s rules.

Those rules required an intervention to address workplace emotional competencies, a validated measure of the competency targeted, and quantitative results usable in the review. A program could be described as communication training yet still fail to qualify if it did not measure its target competency in that way, or if its results were not reported in analyzable form. The distinction is between a topic mentioned in a course and an outcome tested in an eligible study. A course may include discussion of tension or empathy, for example, without measuring whether participants accurately recognize another person’s emotion or handle a disagreement differently afterward. Requiring a validated measure also means that an intervention’s label alone could not answer the question: a study needed an instrument tied to the competency it set out to change. The review’s absence finding is about this intersection of population, target, measurement, and usable reporting.

The scope also narrows what “workplace” means here: the review sought interventions involving employed adults and work-related competency outcomes. A negotiation workshop might evaluate whether participants reach an agreement, protect a financial interest, or remember a technique. Those could be useful outcomes for negotiation, but by themselves they would not establish change in emotional awareness or conflict management under the review’s criteria. This is an illustration of how an evaluation could fall outside the question, not a claim about a particular excluded study.

The authors searched multiple bibliographic databases, trial registers, and related reviews; two reviewers screened the records independently. That process makes the absence informative for the question and eligibility rules they set. It still has a time boundary: the last search update was November 2022. Studies published or indexed later could not enter, and studies using another population, a different target, or measures and reporting that did not meet the stated requirements could remain outside the synthesis. The sound reading is therefore precise: as of that search, the review found no eligible workplace intervention studies directly addressing emotional awareness or conflict management. It does not establish that no such research exists anywhere, that training has no effect, or that later evidence cannot change the answer. It is a map of what this review could support, rather than a count of every workplace course or every way emotional skills might be assessed.

Sources: Training emotional competencies at the workplace: a systematic review and metaanalysis

How positive is a pooled average when studies disagree?

The positive average matters: across controlled workplace-training studies, participants assigned to training scored higher on measured emotional competencies than comparison participants on average. The review, “Training emotional competencies at the workplace: a systematic review and metaanalysis,” reports 27 controlled studies. Its abstract gives a standardized mean difference (SMD) of 0.46, with a 95% confidence interval from 0.30 to 0.63; its results section gives 0.47, with an interval from 0.31 to 0.62. Those are the review’s two reported point estimates, not a choice between a correct number and an obvious typo. The difference is small and does not change the direction of the result, but readers should be able to see the discrepancy rather than have an editor silently select one figure.

An SMD expresses the gap between group averages in standard-deviation units, allowing a synthesis to combine studies that used different scales. It is not a percentage, the share of employees who improved, or a forecast for any one person. A positive pooled estimate says that the studies’ average comparison favored training. The confidence interval describes uncertainty around that pooled average under the analysis; both versions reported by the review remain above zero. That supports a real average signal in the set of controlled studies, while leaving open how consistently a particular course or outcome will change.

The study-to-study range is less settled than the pooled confidence interval alone might suggest. In the results section, the review gives a prediction interval from −0.34 to 1.29 for the controlled estimate. Unlike the confidence interval around the pooled mean, this interval estimates the range in which an effect from a comparable new study may fall. It crosses zero: some settings may show little or no advantage over comparison, even while the combined average is positive. The review also reports I² of 68%, indicating substantial variation among study estimates beyond sampling error. Differences in training, populations, instruments, and study conditions could all contribute; the pooled number does not identify which difference mattered in a given course.

The separate pre/post synthesis points in the same positive direction but answers a looser question. Across 50 studies and 3,233 participants, “Training emotional competencies at the workplace” reports a mean change of 0.44 SMD, with a 95% confidence interval of 0.29 to 0.59. Pre/post comparisons show change over time without necessarily separating the course from other changes occurring during that period. The controlled average therefore carries more weight for whether training outperformed a comparison condition; the larger set of pre/post studies adds breadth, not an equivalent causal test. These are pooled changes on the emotional-competency measures used in the studies, not direct evidence that people read coworkers’ emotions more accurately or repaired disagreements more effectively.

The review’s appraisal gives reasons to take the positive mean seriously and to avoid treating it as a stable promise. Most studies used self-report measures, while only seven used ability tests. Many studies were judged at serious or critical risk of bias, so weaknesses in design or reporting could affect estimates. The authors’ publication-bias checks were significant, a pattern consistent with published results being more favorable or more visible than missing results. These checks do not calculate exactly how much the pooled estimate is inflated, and a risk-of-bias rating does not establish that every result is false. They do make a precise prediction for an unstudied program harder to defend.

Sensitivity checks preserve a useful part of the conclusion without making the uncertainty disappear. Removing the pre/post outlier reduced that estimate from 0.44 to 0.38 and left high heterogeneity (I²=86%); excluding critically rated studies changed the mean only modestly. The positive direction is therefore not obviously carried by one unusually large pre/post study or solely by the most concerning studies. Still, a stable direction across checks is different from uniform benefit: the controlled prediction interval remains wide, and the measures combine differing competencies. A reader can reasonably conclude that workplace emotional-competency scores improved on average in the reviewed evidence. The review cannot tell a learner that a specific course will work, how large their change will be, or whether a gain on one measure will carry into a difficult conversation.

Sources: Training emotional competencies at the workplace: a systematic review and metaanalysis

What does an awareness measure actually establish?

An awareness score establishes only what its method asks and records. Here, awareness accuracy means how closely a person’s identification or interpretation of emotion matches a defined criterion, such as a scored task response; it is different from feeling confident about one’s emotional judgment. In “Training emotional competencies at the workplace: a systematic review and metaanalysis,” the review includes varied tools, with self-report common and ability tests uncommon. That mix matters because perceived habits, task performance, another person’s report, and observed conduct do not answer the same question.

A self-report is informative when the target is a person’s perception of their own recent habits: whether they usually pause before responding, notice a reaction, or try to understand a colleague’s perspective. A structured performance task instead records responses to specified emotional material under common conditions and scores them against the task’s rules. It can show how someone handled those items; it does not establish that they behaved the same way during a live disagreement. Neither method makes the other redundant. One concerns perceived typical behavior, the other performance on a particular set of problems.

Observer report and direct behavior observation add a different vantage point. A colleague or supervisor can describe conduct they have seen, such as whether someone invited clarification or returned to a tense discussion. That report depends on what the observer had a chance to witness and how they interpreted it. Direct observation can record a defined action in a specific interaction, but a single exchange does not establish a person’s general emotional skill across colleagues and conditions. Evidence about conduct is most useful when the behavior and situation are stated clearly, rather than inferred from a broad label such as “good at reading people.”

Consider a hypothetical exchange: a coworker replies to a proposal with a curt message. The recipient feels certain the coworker is angry and reports strong confidence in reading the tone. That confidence is evidence of the recipient’s belief, not confirmation of the coworker’s emotion; the message could reflect irritation, haste, or something else. A task score on recognizing emotional cues would provide evidence about the tested task, but would not establish that the recipient checked their interpretation, asked what was wrong, or repaired the exchange. For that workplace question, the relevant evidence concerns what happened in the conversation and how the other person responded.

This is why a self-reported awareness gain can be meaningful without proving improved recognition accuracy or changed conflict behavior. If the aim is to notice one’s own reactions, perceived recent habits may be the right target. If the aim is to interpret others accurately, the assessment needs a criterion tied to recognition; if the aim is repair at work, the outcome must include the relevant interaction behavior. The measures can complement one another, but the conclusion should travel only as far as the outcome measured.

Sources: Training emotional competencies at the workplace: a systematic review and metaanalysis

What changed in the online trial—and what did not?

The randomized trial “Training Emotional Intelligence Online: An Evaluation of WEIT 2.0” gives a useful test of whether reported change and performance change appear together after training. Researchers assigned 214 participants to the online program or a waiting-list control: 91 entered the training group and 123 the control group. Outcomes were measured after the intervention and again eight weeks later. That design gives the comparison a clearer time order than a before-and-after account of trainees alone: the study could ask whether the assigned groups differed at follow-up, while retaining a later check on whether any measured differences remained.

At the eight-week follow-up, significant effects appeared on selected self-reported outcomes concerning emotion perception and emotion regulation. The performance-based outcomes in those areas did not show significant treatment effects. The split matters because the trial did not report one undifferentiated “EQ improved” result. It reported a pattern across outcome types: participants’ accounts of some emotional tendencies shifted relative to the comparison condition, while scores on the study’s performance measures did not produce a statistically significant treatment effect. The precise conclusion is therefore about which outcomes registered a detectable difference in this sample and study, not a blanket judgment about whether the intervention worked.

The control condition also defines what the result can compare. Because participants were assigned to training or a waiting list, the contrast concerns access to this program during the study period, not a head-to-head test against another course or a particular workplace practice. The eight-week measurement adds a later observation after the intervention; it does not amount to an extended record of daily behavior, and it cannot show whether effects persisted beyond that point. Those design details make the result informative about this program and these measures at the reported checkpoints, without turning it into a general estimate for every form of emotional-skills training.

Nor does the nonsignificant performance result establish that the program had exactly zero effect on the abilities sampled. Statistical nonsignificance means the analysis did not establish a difference at its chosen threshold; it is not itself a demonstration of equivalence. The study’s design supports a stronger comparison than self-report alone, but the outcome pattern leaves room for uncertainty about the size and precision of performance change. It also gives no basis for attributing the divergence to a particular mechanism. The data described here show different statistical results across selected self-report and task measures; they do not explain why those measures differed.

WEIT 2.0 adds a focused answer to the workplace review’s measurement question: reported emotional change and demonstrated performance need not move in lockstep in a single intervention study. Yet the trial tested online training with general-population participants, not a workplace course evaluated during disagreement. Its follow-up did not establish whether a participant checked an interpretation with a colleague, set a boundary, or returned to repair after tension. Random assignment and the eight-week interval strengthen this study’s comparison; neither turns it into evidence about observed workplace conflict. For that question, the trial identifies a useful gap between kinds of outcome while leaving the behavior at work to be measured directly.

Sources: Training Emotional Intelligence Online: An Evaluation of WEIT 2.0

Does a workplace transfer measure capture conflict repair?

A longitudinal experiment among government officials gets closer to work transfer because it followed participants after emotional-intelligence training and assessed behavior in their jobs. “Does Social and Organizational Support Moderate Emotional Intelligence Training Effectiveness?” reports randomized allocation of 176 officials, with 88 assigned to training and 88 to control. The study examined outcomes over multiple periods, including follow-up three months after training. This makes it more relevant to whether learning carries beyond a course than a result measured only on the training material itself.

The transfer outcomes were organizational citizenship behavior and counterproductive work behavior, assessed through participants’ self-evaluations and supervisors’ ratings. The self-evaluation sample included all 176 officials; the supervisor-rating sample at three months was 78. The study reported positive training outcomes in both evaluation samples. Its moderation findings diverged: perceived support moderated effects in self-evaluations, while the corresponding moderation was not found in supervisor evaluations. That difference belongs to the pattern of results across the two samples. It does not establish that one evaluator was right and the other wrong; they supplied distinct reports, with different access to behavior and a smaller supervisor sample.

The study’s longitudinal structure is valuable for this reason: it treats transfer as a question about outcomes after learning, rather than assuming that an immediate course response is equivalent to changed conduct at work. Still, the reported categories aggregate behavior across situations. An overall citizenship or counterproductive-behavior rating does not tell a reader which episode changed, how often it occurred, or whether the change involved the emotional judgment and repair steps at the center of this article. The timeline makes work outcomes relevant; the breadth of the outcomes limits their specificity.

These outcomes indicate transfer in a broad sense: the researchers looked for changes in discretionary conduct at work, rather than relying solely on an emotional-intelligence score. Citizenship and counterproductive behavior ratings can register consequential patterns in how people contribute or disrupt work. But they do not specify whether someone noticed their own rising frustration, tested an interpretation before acting, stated a limit clearly, or returned to a damaged exchange to repair it. Those actions could matter greatly in conflict, yet they are more particular than the study’s selected outcome categories. A favorable broad rating therefore cannot answer every question about which emotional skill changed or what happened in a difficult conversation.

Observer access is part of the measurement. A supervisor may have repeated opportunities to see some work behaviors and little chance to witness a private disagreement; an employee may notice internal reactions that never become visible. Differences between reports can therefore reflect what each measure can capture, as well as the people who completed it. The study’s supervisor sample also fell below the full self-report sample, so the two sets of estimates do not rest on equivalent numbers of raters. These conditions counsel against treating either view as a complete record of transfer.

For evaluating conflict training, the practical implication is to name the behavior and the evaluator before drawing a conclusion from transfer. If the intended change is repair, an outcome could record whether a person reopened a strained exchange and whether the other participant experienced the attempt as responsive. That is an editorial proposal for a more targeted measure, not an outcome tested in this government-official study. The experiment shows that training-related outcomes can be followed into work through broad ratings and that moderation can differ by evaluator. It remains indirect evidence for awareness and repair because neither rating category isolates those particular actions.

Sources: Does Social and Organizational Support Moderate Emotional Intelligence Training Effectiveness?

Does the older training synthesis change the workplace conclusion?

It makes the case for trainability of measured emotional-intelligence scores harder to dismiss, while leaving the workplace behavior question where it was. The earlier meta-analysis, “Can emotional intelligence be trained? A meta-analytical investigation,” brought together 58 published and unpublished adult studies using pre/post or treatment-control designs. Its abstract reports a moderate positive effect on EI scores. That is a meaningful counterweight to reading the later workplace review’s unresolved target outcomes as evidence that emotional skills cannot change: across a wider adult evidence base, training was associated with gains on the measures used.

The same abstract reports larger effects in published studies than in dissertations. That difference matters because a positive overall synthesis does not erase the possibility that the visible research record gives a more favorable impression than less-publicized work. The finding does not tell us how much the difference reflects publication selection, study quality, participant or program variation, or other factors; the accessible summary does not provide detail to settle those explanations. It does, however, make the broad conclusion more careful: EI measures appear trainable in the studies gathered, with the published subset showing stronger results.

This changes the weight of one claim, not the target of the evidence. The earlier synthesis combines adult training research around EI scores; it does not isolate workers’ accuracy in identifying a colleague’s emotion or their actions during a disagreement. Those are different endpoints from an overall test or questionnaire score. The studies can support a general expectation that some measured emotional competencies respond to training, but that expectation cannot be converted into a finding about whether a workplace course changes checking an interpretation, regulating a response, or repairing a strained exchange.

Read beside the workplace review, the two syntheses point in compatible directions at different levels. Broad adult evidence supports score change; the more narrowly relevant workplace review still leaves the specified awareness and conflict outcomes unresolved. This comparison is an inference from their scopes and reported outcomes, not a new pooled analysis. A reader choosing a course can reasonably expect that measured EI scores may move, but should ask what the program assesses and practices before treating a score gain as evidence of better conduct under pressure. The broader review strengthens the trainability premise without supplying the missing workplace test.

Sources: Can emotional intelligence be trained? A meta-analytical investigation

Does healthcare training evidence answer a cross-workplace question?

Healthcare training is a relevant test of whether the broader pattern appears in a demanding work setting. “Emotional intelligence training among the healthcare workforce: a systematic review and meta-analysis” reviews 17 longitudinal studies of healthcare workers and identifies emotional-intelligence measures as the primary outcomes. Its accessible summary reports broad post-intervention increases. That evidence gives the trainability argument a concrete occupational context: care work includes consequential interactions where noticing and responding to emotion can matter. A reader should therefore take the review seriously when considering whether EI training can change measured outcomes in applied settings.

But healthcare workers do not stand in for every occupation, and change on an EI measure is still distinct from a record of what happened in a conflict. The review’s reported outcome focus does not establish that participants became more accurate at reading another person’s emotional state, checked uncertain interpretations more often, or repaired disagreements more effectively. Those behaviors require measures aimed at them. A broad scale can register a change relevant to emotional skills without identifying which action changed or whether a coworker experienced a different interaction.

The occupational setting makes transfer plausible as a question, not automatic as a conclusion. Emotional demands in care work may resemble some pressures in management, customer-facing roles, or team coordination, yet the situations and opportunities to observe behavior differ. Applying the healthcare result elsewhere would require evidence that connects the populations, training, and outcomes: for example, comparable measures and follow-up of specific conduct in the other job setting. The healthcare synthesis, as summarized in the accessible record, does not provide that cross-occupation comparison.

So this adjacent evidence narrows the uncertainty only in one respect: positive change in broad EI outcomes has also been reported across a longitudinal healthcare-training literature. It does not fill the gap around awareness accuracy or observed conflict behavior across workplaces. That distinction matters for a learner evaluating a course advertised for communication: healthcare findings may support the possibility of measurable development, while a claim about handling disagreement needs evidence that follows disagreement-related behavior. The practical judgment depends on the outcome the course actually measured, not simply on the emotional intensity of the occupation in which training occurred.

Sources: Emotional intelligence training among the healthcare workforce: a systematic review and meta-analysis

A useful next step for an individual learner

Choose a response you can recognize in a real interaction, then make that the learning target. If disagreement is where you want to improve, the target might be pausing long enough to check what you understood before replying. Look for practice that rehearses that response in a relevant situation, rather than relying on a broad promise to improve communication. Later, ask for feedback close to the behavior: did you check your interpretation, and did that change how the conversation continued? That gives you something specific to adjust on the next attempt, and helps you notice what made the response easier or harder. The feedback is evidence about that interaction, not a final rating of your emotional skill.

A private Emotional Skills Profile can help you reflect on recent behavior and judgments about authored scenarios, then select a focus for practice. Its 32 items offer a structured starting point for that reflection; they do not establish ability or show that a course changed workplace behavior. If you would like that individual reflection aid, explore the profile at [/assessment](/assessment). Keep the chosen response small enough to notice, and let later practice and feedback tell you whether it is becoming more available when the conversation gets difficult.

Sources: Emotional Skills Profile

Questions readers ask