Part V: creating the EQ-i 2.0 and EQ 360 2.0

EQ-i 2.0 Stages of Development

Development of the EQ-i 2.0

Stages of Development

The EQ-i 2.0 is a revision of the EQ-i (Bar-On, 1997, 2004). In general, the development of the EQ-i 2.0 adopted many of the same aforementioned stages as the EQ-i, but each stage is explained in greater detail for the revised version. There are five broad stages to the development of the EQ-i 2.0:

Define the goals of the revision (what needs to change and why).
Conceptualize changes to subscales and create new subscales based on the latest EI research and practice using the EQ-i model.
Define subscales.
Build and pilot test the assessment.
Conduct norming and subsequent data analyses to refine and confirm the assessment (e.g., factor analysis, examination of validity and reliability).

Stage 1: Define the Goals of the Revision

As demonstrated throughout this manual, the EQ-i has a long and storied history, attesting to its pivotal role in personal coaching, professional enhancement, selection, and leadership development. Despite the varied application and utility of the EQ-i, many lessons have been learned along the way, providing opportunities to refine the conceptual framework from which the EQ-i was derived. As a result, an extensive revision of the EQ-i was undertaken with the overarching goal of enhancing the psychometric properties, usability, and application of the tool, while staying true to the underlying tenets of emotional intelligence first introduced by Reuven Bar-On (1980, 1997).

First and foremost, the goal of this revision was to preserve the foundation and integrity of the EQ-i. The EQ-i has been extensively used in research and practice since 1997, resulting in more than 200 publications, including research articles and dissertations, books and book chapters, and trade publications, with more than a million applications of the EQ-i, impacting millions of people worldwide. The underlying tenets of the EQ-i are fundamental to its success and warrant retention. Both existing quantitative research and qualitative information collected from over 700 consultants have yielded significant feedback about the strengths and opportunities for improvement within the EQ-i. For example, the strengths have been listed as:

The 1-5-15 Factor Structure
The validity and reliability of the tool
The emphasis placed on well-being and performance
The subset of skills and abilities that are amenable to coaching

Conversely, there exist opportunities to improve and strengthen the EQ-i. As a result, the following goals guided all changes made to the assessment:

Goal	Rationale
1. Aligning items and response options	There is a disconnect between some of the items on the EQ-i and the available response options. For example, some items (e.g., “I don’t do anything bad in my life”), when paired with the existing response options (“Not true of me”), resulted in a double negative, presenting interpretation issues and a higher reading level expected of respondents. Response options need to be restructured to simplify the administration for the respondents and optimize clarity.
2. Cleaning and clarifying item wording Updating language Eliminating clinical language Improving item selection to further enhance cross-cultural applicability	The EQ-i contained clinical or sensitive language within the items making it harder to debrief results in a corporate context. Items such as “I feel comfortable with my body” or “I do very weird things” have less relevance in the workplace and can reduce the face validity of the assessment in that context. Given the worldwide accessibility of the EQ-i, coupled with an ever-growing diversity of industrialized nations, it is critical that the EQ-i 2.0 remain free from cultural bias and jargon. Items such as “I would stop and help a crying child find his or her parents, even if I had to be somewhere else at the same time,” presents a response bias for respondents who come from countries where crying children are commonplace.
3. Addressing non-unidimensional content scales	In the original EQ-i, each of the Emotional Self-Awareness, Impulse Control, and Self-Regard subscales contained multiple constructs. Emotional Self-Awareness addressed emotional expression as well emotional self-awareness. This scale was divided into Self-Expression and Emotional Self-Awareness. Impulse Control assessed impulsivity and anger. The anger component was removed to better address impulsiveness. Self-Regard addressed both body image and a belief in one’s ability. Body image content was removed to more clearly address the notion of Self-Regard as it pertains to inner strength and self-confidence.
4. Reducing interpretation issues associated with scales and their definitions	In the original EQ-i, the Problem Solving subscale addressed a linear, pragmatic approach to solving problems. The EQ-i 2.0 places less value on moving step-by-step; instead it addresses how one applies emotional information when solving problems.
5. Items appear on only one scale	In the original EQ-i, when items appear on more than one scale, they present interpretation problems and decrease the distinctiveness between subscales.
6. Increasing symmetry between EQ-i and EQ 360.	In the original, items on the EQ-i and EQ 360 do not match, making a comparison between self-scores (EQ-i) and other raters’ scores (EQ 360) difficult.
7. Updating and diversifying norm sample	Since the release of the EQ-i in 1997, the population demographics of North America have changed considerably. To remain current and representative, the normative sample (i.e., baseline) must reflect these changes.
8. Refining 1-5-15 factor structure and greater practicality at the composite level	User feedback suggested that the current model is difficult to coach to, particularly at the composite level. Intrapersonal and Interpersonal Composites are often confused with one another and do not resonate well with the client.

Each of these goals will be described in more detail. Keep in mind that many of these goals are intertwined such that when a change is made to address one area (e.g., address multi-dimensional content scales) it often impacts other goals (e.g., refining the 1-5-15 factor structure). Therefore, although these goals were established at the onset of the revision process, many of them continued to evolve through to the end of the norming stage.

Stage 2: Conceptualize Changes to Subscales

ITEM LEVEL CHANGES

In response to both qualitative and quantitative feedback, each of the original 133 items was examined by a team of test developers to determine its level of quality, clarity, and relevance. In doing so, a number of the items were revised and many new items developed, resulting in a test pool of more than 220 items. Some items in the development version were revised to address social/cultural bias and item content.

Social/Cultural Bias. Given the worldwide accessibility of the EQ-i, the ever-increasing diversity of industrialized nations, and the multicultural composition of the global workforce, it is critical now, more than ever, that the EQ-i 2.0 remains free from cultural bias and jargon. The following example depicts the type of change made to address social/cultural bias.

Original EQ-i Item	Bias	Decision
It’s hard for me to see people suffer	In countries and cultures where suffering is more commonplace, this item proves difficult to interpret, especially where a tolerance may be developed for human struggles.	Remove item.

Item Content. The original version of the EQ-i consisted of 133 items, including 7 items that made up the Negative Impression scale and Item 133, representing the final validity question. In response to the negative and/or clinical connotations and often ill-perceived implications of the Negative Impression index, the EQ-i: 125, a 125-item version of the standard EQ-i was released minus the Negative Impression items. Although this option was made available to help facilitate administration and improve the EQ-i experience in the corporate context, the exclusion of these items may have sacrificed relevant and often pivotal interpretive information. In response to this dilemma, the EQ-i 2.0 was designed to capture the benefits that a negative impression index offers by creating items for this index that avoided clinical, negative connotations. For example, negative impression items such as “I feel cut off from my body” and “I think I have lost my mind” were removed and replaced with “I have bad days” and “Things bother me.”
To complement the changes to the Negative Impression index, each remaining item was scrutinized and amended if too lengthy, objectionable, or containing religious, sexual, suggestive, or political content. Results of the post-data collection item analyses resulted in moderate modifications to the content of each scale, unifying and clarifying scale content. In direct response to the empirical, theoretical, and practical implications of the item changes, the data collection item pool was eventually reduced to 133 items. In light of these alterations to the items and amendments to the response format as highlighted above, the resulting EQ-i 2.0 is still statistically comparable to the EQ-i (see Standardization, Reliability, and Validity and Appendix A).

SCALE LEVEL CHANGES

In addition to addressing several concerns at the item level, there remained room to refine the EQ-i at the composite and subscale levels using knowledge that has accumulated over time regarding emotional intelligence in general and the EQ-i specifically. Of primary concern was the frequency of double content scales and subscales with interpretation difficulties.

MULTI-DIMENSIONAL CONTENT SCALES

The goal of any one scale should be to measure a unified construct that captures the essence of a sound theoretically defined target, in this case a distinct component of emotional intelligence. Internal consistency (Cronbachs’s Alpha, Cronbach, 1951 or K-R 20 Kuder & Richardson, 1937) refers to the extent to which all items on a given scale are related. However, the larger issue for the practical application of an assessment is the homogeneity (i.e., sameness) and unidimensionality (i.e., oneness) that signify the extent to which all items on a scale measure a single underlying factor (Clark & Watson, 1995). Although a scale can consist of a number of related items without being homogeneous, a scale cannot be homogeneous without sufficiently similar items (Clark & Watson, 1995). Because the EQ-i is designed to measure several related facets of emotional intelligence, it is essential that each subscale remain unidimensional and each composite scale intercorrelated. By adhering to this simple principle, the overarching construct of emotional intelligence can be reliably measured by its well defined and unified component parts.
Notwithstanding the rigorous psychometric, theoretical, and applied steps taken during the development of the EQ-i, the Emotional Self-Awareness, Impulse Control, and Self-Regard subscales maintained excellent internal consistency but had distinct sub-components. For example, the Emotional Self-Awareness subscale was comprised of eight items, four of which addressed one’s self-awareness while the remainder addressed one’s ability to express oneself. In essence, Emotional Self-Awareness captured two highly related but separate skills. Impulse Control contained items that measured impulsivity and anger control, and Self-Regard captured one’s level of personal confidence and comfort with one’s physical self. Content scales that contain multiple facets present interpretation concerns and, more importantly, provide coaching and development challenges. The EQ-i 2.0 was revised with these limitations in mind and, as a result, each of the 15 subscales presents with one unified construct underpinning the general construct of emotional intelligence.

INTERPRETATION RELATED CONCERNS

A distinct benefit of the EQ-i is the applied nature of the conceptual model; it lends itself to coaching, and through it personal and professional development can be measured. Given that applicability, it was a goal of this revision to simplify and clarify interpretation ambiguity associated with Problem Solving and Happiness.

Problem Solving. The Problem Solving subscale placed value and importance on one’s ability to remain pragmatic and didactic throughout the problem solving process. For example, a linear, step-by-step process for gathering information and weighing the pros and cons of a given situation, while considering alternative courses of action, was considered ideal. However, not all effective problem solvers engage in this style of problem solving. In fact, many top decision makers are required to make rapid decisions and must use emotional information to facilitate their problem solving. As a result, the EQ-i 2.0 was modified to address the process by which one engages emotions during problem solving, instead of focusing on one’s respective style for solving problems.
Happiness. Happiness has proven to be fundamentally related to emotional intelligence. Although happiness has long been considered a component of emotional intelligence, research and application suggest that happiness is more likely a by-product of enhanced emotional intelligence. Furthermore, happiness itself is a difficult construct to operationalize and directly coach to. However, if you consider that factors such as Self-Regard, Interpersonal Relationships, Optimism, and Self-Actualization are intimately linked to happiness, one can better develop a plan for increasing feelings of happiness and contentment by overtly addressing these other behavioral aspects of emotional intelligence. As a result, happiness is viewed both as an indication of one’s well-being and a product of emotional intelligence, simplifying the interpretation and applied implications of one’s results.

Stage 3: Define Composite Scales and Subscales

The composite and subscales changes that occurred from a re-conceptualization of the EQ-i framework and item level adjustments resulted in the EQ-i 2.0 composite scales and subscales being operationally refined. The test developers aimed to be as comprehensive as possible by incorporating the most significant aspects of contributions made by theorists and researchers over the last several decades. Another important aim was to define scales as clearly, operationally, and distinctly as possible. This step was undertaken to strengthen the foundations of the assessment from the onset so that its validity would stand up to examination by subsequent statistical procedures (e.g., reliability, validity; see Jackson, 1971; Loevinger, 1957).
Professional and theoretical jargon was avoided to create clearly worded and comprehensible definitions that would resonate with the test-taker and the person providing feedback. Definitions that were too abstract, esoteric, and cryptic were rejected. Scales undergoing significant changes to their definitions are shown below:

Self-Perception (new composite scale)
Self-Expression (new composite scale)
Decision Making (new composite scale)
Emotional Expression (new subscale)
Problem Solving
Impulse Control

Stages 1 to 3 laid the theoretical foundations upon which the EQ-i 2.0 was developed. Building on this framsework, the next two stages describe how the EQ-i 2.0 was actually constructed and designed to measure the psychological characteristics of EI, while staying true to the underlying tenets of the original EQ-i.

Stage 4: Build and Pilot Test the Assessment

The fourth stage of development consisted of the actual construction of the first (i.e., pilot) version of the EQ-i 2.0. The construction was carried out by selecting items based on the above-mentioned operationally defined factors, deciding upon the exact placement of these items within the inventory, developing a response format, and creating validity scales. This process is similar to that used for the original EQ-i, described in detail by Bar-On (2004). Each step is briefly described below.
For the EQ-i 2.0, the pilot item pool consisted of 221 items. This number was selected to provide adequate coverage of all content areas, while ensuring the length of the inventory was practical for respondents in the pilot testing phase to complete in a reasonable amount of time. A balance was maintained between items phrased in a positive manner and items stated in a negative fashion, so that individuals who respond positively or negatively to all items indiscriminately can be easily detected.

Pilot testing (see the EQ-i 2.0 Pilot Study and Standardization section in Standardization, Reliability, and Validity) was performed to ensure the practical functionality of the assessment (e.g., administration time, instructions, etc.) and to examine the psychometric properties of the items and scales. Statistical analyses were conducted to eliminate items that may have been too confusing, showed little variability in responses, lacked face and construct validity, or were redundant with better performing items.

ITEM ORDER

After items were selected for the inventory, they were arranged in a random order. This order was then revised slightly based on practical considerations. Specifically, the least threatening, negative, or emotionally-laden items were presented at the beginning of the inventory. This was done to increase rapport with the respondents and eliminate any potential for intimidation. Within this random order, a rotation process was used to ensure items from identical subscales did not appear in succession.

RESPONSE OPTIONS & DOUBLE NEGATIVES

Throughout the history of the EQ-i the response options have been subject to examination and have been amended based on feedback from respondents, research findings, as well as practical and applied considerations. The original response format consisted of a 7-point Likert self-rating scale:

1 = Always True of Me
2 = Very Often True of Me
3 = Often True of Me
4 = Sometimes True of Me
5 = Seldom True of Me
6 = Very Seldom True of Me
7 = Never True of Me

The first modification to the response options included reversing the direction of the response format to better align with a more intuitively perceived progression of behavior from less to more. As a result, the response option “Never” was adjusted to align with the smallest response option value (i.e., 1) on the rating scale, while the ”Always” response option was moved to align with the highest value on the rating scale (i.e., 7). A second revision to the rating scale resulted in a shortening of the response options from a 7-point scale to a 5-point scale, with moderate modifications made to the extreme poles of the rating scale. The idea of joining “True of Me” with “Very True of Me” and “Not True of Me” with “Very Seldom True of Me” was a way of reducing the burden to the respondent while maintaining a discriminating response set. The changes resulted in:

1 = Very Seldom or Not True of Me
2 = Seldom True of Me
3 = Sometimes True of Me
4 = Often True of Me
5 = Very Often True of Me or True of Me

Despite the positive changes to the response set, there remained a disconnect between some of the items on the EQ-i and the available response options. For example, several items, when paired with the original EQ-i response options, resulted in a double negative which was problematic for some respondents.

EQ-i Item	Response	Meaning
“I don’t do anything bad in my life”	“Very Seldom or Not True of Me”	The respondent does not do this behavior, which in this case means that he would in fact do bad things in his life.

To resolve these issues, the response set was simplified and items were carefully worded or re-worded to avoid double negative situations and make the tool as easy to use as possible. The new response options are:

1 = Never/Rarely
2 = Occasionally
3 = Sometimes
4 = Often
5 = Always/Almost Always

Stage 5: Conduct Norming and Subsequent Data Analyses to Refine and Confirm the Assessment

Various statistical analyses were performed during the course of EQ-i 2.0 construction to ensure that the assessment is of the highest statistical and measurement standards. A majority of these analyses were run following the collection of the normative sample.

NORMING

Since the release of the EQ-i in 1997, the population demographics of North America have changed considerably. To remain current and representative, the normative sample (i.e., baseline) must reflect these changes. In response, the EQ-i 2.0 normative sample consists of 4,000 respondents who are representative of the current North American population, including by age, gender, race, ethnicity, and educational level. For a more detailed breakdown of the EQ-i 2.0 demographics and normative process please refer to Standardization, Reliability, and Validity.

DATA ANALYSES

Once the normative sample was complete, multiple analyses were conducted to establish the psychometric properties of the EQ-i 2.0 and further refine the assessment to reach its present form. Standardization, Reliability, and Validity provides a thorough explanation of the analyses conducted, including:

Standardization
Reliability
Content Validity
Factor Structure (1-5-15)
Relationships between EQ-i 2.0 and Other Measures
Group Difference in EQ-i 2.0 Scores
Validity Scale Research and Analysis

The final form of the EQ-i 2.0 has 133 items, composed of 15 subscales, 1 Well-Being Indicator, and 3 validity scales, with 6 to 9 items per subscale. The 15 subscales are further grouped into 5 composite scales, with 3 subscales contained in each. The composite scales each contain 23 or 24 items. The subscales were grouped into these composite scales based on their statistical and theoretical relatedness and are combined to create the overall EQ-i 2.0 Total EI Score. A sixteenth subscale, Happiness, is not scored as part of this overall score but is considered a highly relevant and related Well-Being Indicator. The Positive Impression and Negative Impression validity scales contain the same 6 items; these items do not overlap with any of the items in the subscales/composite scales. The Inconsistency Index contains 20 items drawn from items in the subscales/composite scales.

The English language version of the EQ-i 2.0 is simple and straightforward and has been determined to sit between a third and fourth grade reading level at 3.6 (Flesch, 1948; Kincaid, Fishburne, Rogers, & Chissom, 1975). As described in the Introduction, statistical analyses were performed that describe the EQ-i 2.0’s standardization and examine its reliability and validity.