Page Comparison

Four validity indicators are measured:

General frame of mind

Three items are rated at the start of the survey using the same Likert-type response scale as the other survey items:

I feel great
I think positively
I am in a good mood

We expect students to provide a positive score on all three items; scores ranging from 1 to 3 should be explored further during the required debriefing of the SEI-YV report. It is also a good idea to consider responses to these three items together with the next validity indicator, and/or the EQ component EOP (Exercise Optimism).

Positive impression

Average performance and the distribution of scores for positive impression is closely aligned with that of other EQ scales. This is a standardized score with a mean of 100 and a standard deviation of 15, similar to that of the EQ components and Life Barometers. This means that a markedly high, standardized positive impression score, as identified by standard deviation points, remains a valid gauge for further follow-up as indicated in SEI-YV reports across the different survey versions. Scores larger than 1SD are labeled as “possibly invalid”; scores larger than 2SD are labeled as “probably invalid”. The reasons for a high positive impression (PIN) score should be explored further with the youth, parent, counselor, and/or teacher.

Number of missing items

We require a minimum of 94% completion of the survey items in order to validate the EQ profile of the youth. This means that a maximum of six missing items across the survey is allowed. When the maximum is reached, no zero point or midpoint on the response scale is allocated, as we feel that such a method may put the youth at a disadvantage and can lead to misleading scale results. Instead, scale scores are calculated with the remainder of the items that contribute to each scale. While we lose a tiny bit of the psychometric strength in the scale(s) where the missing response occurs, the calculation of the scale score is the most accurate reflection of that youth that we have at that time.

Response inconsistency

Dispersed throughout the survey, five pairs of items have similar wording. These item pairs are intended to assess the consistency with which the youth responded to the survey. Youth who fail to read the survey items carefully, or who respond at random or in some other systematic fashion may miss these pairs, potentially causing them to have largely different ratings from one another. Tracking the differences in these paired ratings, researchers expected that the ratings would be generally very close under normal assessment circumstances. The mean difference between these five item pairs together is 0.76 (less than one rating point) with standard deviation 0.48. Average scores on all five individual inconsistency pairs lie below a one-point difference.

When IC is larger than five, we question the validity of the youth’s EQ profile of. This number remained stable across different survey and report versions. There is a slightly noticeable sensitivity in the IC score related to age: as can be expected, younger children (who have less life experience), are less consistent in their item ratings.

Overall, girls tend to score slightly higher than boys. The highest difference between the two genders lies in ICE (increase empathy), again, which can be expected. The score difference is about seven raw points (Cohen’s d < 0.40). When this is translated into a standardized score, the difference becomes less than one standard point and hence is negligible.

Construct Validity

The objective of construct validity is to find empirical confirmation of EQ conceptualized as consisting of three pursuits and further segmented into eight EQ components. As with most self-judged (self-report) measures of Emotional Intelligence, we found that the factorial structure of the SEI-YV is uni-factorial. This is also a point of critique against EQ measures in general, which is well debated in academic literature. However, through statistical factor analysis with varimax rotation, we also found mild support for a two-dimensional scope, as shown by the scree plot of eigenvalues below.

A total of 13 factors had an eigenvalue greater than 1 and accounted for 47.91% of the common variance. However, in this case Kaiser’s criterion may not be the best guide for extracting this many meaningful factors with the number of items available. One factor shows a large positive eigenvalue and accounts for 20.33% of the common variance alone. The next three factors have an eigenvalue greater than 2; the common variance that these four factors account for is 31.59%. In addition, the scree plot shows a reasonable bend at five factors (with common variance 34.03%). Eight factors together account for 39.92% of the common variance in the factor structure of the SEI-YV.

We need to recognize that due to the target population of the SEI-YV where the attention span is typically shorter than in the case of adult surveys, we are attempting to confirm the theoretical structure within the limitations of only 68 EQ scale items. (Other items are used specifically for calculating the validity indicators.) The factor structure revealed below also helps explain the internal consistency statistics (Cronbach’s coefficient alpha) that are reported later.

In an attempt to confirm the K-C-G pursuits, interpretation of three extracted factors reveal a grouping of the eight EQ scales as follows:

...

Factor number

...

Dominant EQ component

...

EQ Pursuit

...

Factor 1:

...

EOP
EEL (basic knowledge) EIM
NVE (open sharing)

...

Choose Yourself Know Yourself

...

Factor 2:

...

ICE
EEL (interactive knowledge)
PNG
NVE (conflict resolution) RCP

...

Give Yourself Know Yourself

...

Factor 3:

...

ACT

...

Choose Yourself

Empirically, the allocation of specific EQ components to one of the three pursuits appears to be somewhat forced. There is growing argument that elements of all three pursuits can be found in each of the eight scales. Using the norm data, we plan to explore this empirically in further detail. For example, we want to investigate the possibility of treating and measuring the three pursuits as a horizontal dimension and complementary to the existing eight EQ components, rather than in a hierarchical fashion as is the current practice. At the very least, this will help us understand the three pursuits with renewed insight.

An eight-factor structure largely confirms the following EQ scales in this order of strength:

...

Factor number

...

Dominant EQ component

...

Common Variance Explained

...

Factor 1:

...

EIM

...

5.3775

...

Factor 2:

...

ICE

...

5.1616

...

Factor 3:

...

EOP

...

3.1336

...

Factor 4:

...

RCP

...

2.8836

...

Factor 5:

...

NVE (open sharing)

...

2.8474

...

Factor 6:

...

EEL (complex)

...

2.7905

...

Factor 7:

...

PNG

...

2.7420

...

Factor 8:

...

ACT

...

2.2084

All item factor loadings are above 0.30, indicating a good placement within the factor to which it contributes. Once the norm data has grown to be reasonably free from demographic skews, we will further investigate the multi-factorial structure for improved theoretical refinement and fit. The common variance that the factors currently account for also needs to be monitored. There are common psychometric issues in self-assessment questionnaires:  Personal bias, answer style, and inconsistency.  Steps have been taken to insulate the SEI scores from these obscuring influences:

• Personal bias: the SEI has been tested to consider the effects of these biases by utilizing a “positive impression” scale. To a very large extent, the SEI functions effectively without correction. However, the Positive Impression factor is reported on the data sheet to provide useful insight to a SEI Assessor interpreting SEI results.

• Answer style: Another common psychometric issue is that different people assign a Likert scale (e.g., 1-5) with different meanings.  Some rarely use extremes, others “always leave room for improvement.”  To compensate for these differences, the SEI includes an Answer Style index.

• Inconsistency: some test takers are inconsistent in their answers which can reveal a lack of understanding or a lack of focus.  These can reduce the value of the results.  The SEI includes a test of consistency that also evaluates completion time.

POSITIVE IMPRESSION

Average performance and the distribution of scores for positive impression is closely aligned with that of other EQ scales. This is a standardized score with a mean of 100 and a standard deviation of 15, similar to that of the EQ components and Life Barometers. This means that a markedly high, standardized positive impression score, as identified by standard deviation points, remains a valid gauge for further follow-up as indicated in SEI-YV reports across the different survey versions. Positive impression range can be: Very Low, Low, Average, High or Very High. 

EXECUTION TIME

As in the other SEI indices, the individual test-taker’s behavior is compared to a large international sample; typically individuals take around eight minutes to complete the SEI. If completion time is unusually fast or slow, it’s a signal of a potential issue. The completion time index is calculated based on main questionnaire, starting after the personal data page, and ending before the final, optional, mood question.

If the completion time is:

Extremely short: Red Light

Short: Yellow Light

Average: Green Light

Long: Yellow Light

CONSISTENCY ANSWERS (CA)

The consistency index evaluates the frequency of answer choices that the test-taker uses in the 5-point Likert Scale (e.g, “I Agree,” “I Disagree”). This evaluation is based on the elaboration of two indicators:

1. DENSITY INDICATOR - how often does the person use option 1, 2, 3, 4, or 5? Their frequency is compared to the international standard. If one or more options is significantly over-used, the system identifies a potential problem.

2. RANDOM INDICATOR - this test compares way the test taker answers every item with  the international standard. If they follow a random pattern in answering, the system detects a potential problem. To increase accuracy, this indicator is also linked to completion time to make the final random feedback more accurate.

These indicators are summarized in a single CA index, reported in the Data sheet through 3 lights:

Red Light - Low Consistency (problems in one or both indicators)

Yellow Light - Moderate Consistency (potential problem in the random index)

Green Light - High Consistency (both indicators in a normal range)

Concurrent Validity

In the SEI-YV survey, the assessment measures Life Barometers under a separate section, which asks the youth about current performance in specific arenas that impact their lives daily. The overall Life Barometer score breaks down into five different Barometers. Youth tend to rate themselves slightly lower on the Life Barometer items than on the EQ component items.

Importantly, the Life Barometers afford us a built-in opportunity to measure concurrent validity via multiple regression analysis. The validity is said to be concurrent rather than predictive due to the fact that youth complete the second section of the survey immediately after the first section. The insights gained from this analysis are also reported in a powerful EQ Yardstick that is extremely useful for purposes of further development. Many practitioners view the EQ Yardstick as the most powerful part of the SEI-YV report.

The variance in scores within each of the Life Barometers is explained as follows by the EQ components:

Overall, the regression results, and hence the concurrent validity of the SEI-YV, look favorable. The results for the two outer Life Barometers are only moderate due to a slight bimodality in the responses: youth tend to have either good health habits or not so much, and are either moderately self-efficacious or slightly more. This latter aspect is addressed via refinements in the item wording of the current edition of the SEI-YV, V2.11. While we include good health as a Life Barometer in the SEI-YV in recognition of this being a growing issue among youth in many parts of the world, we are still working to fully understand the relationship between physical health (eating and exercise) and emotional intelligence. The most important EQ scales contributing to each Life Barometer are determined by a statistical technique called stepwise regression analysis (forward procedure). The top three with the highest contributing value to each of the five Life Barometers are shown below. Note that some EQ components repeat across the Life Barometers, though their order of importance may differ. It is also helpful to keep in mind that the EQ components contribute value as they work together to explain the variance in Life Barometer scores from the norm base. Therefore, all indicated EQ components should be considered together when compared against the corresponding Life Barometer.

Ultimately, all eight EQ components are important for further development of the Life Barometers. The top EQ components that statistically contributed most to the regression model for each Life Barometer were selected, hence the order in which they are listed is meaningful. Note that the same EQ component can contribute to different Life Barometers, but in a different order of importance, and in combination with different EQ components each time. From the perspective of further developing the Life Barometers then, Assessors should pay particular attention to these combinations of EQ components.

Since the beta values in the respective regression equations of all the identified EQ components are positive, one should interpret the relationship that high scores in these EQ components are generally associated with high scores in the Life Barometers. The statistical selections above help users target their development with an enhanced chance of success. Ideally, the Life Barometer and three EQ component scores grouped together should be in balance with one another.