Construct and discriminant validity of Emotional intelligence measure

1
Construct and discriminant validity of Emotional
intelligence measure
By Laura Guillén, Willem E. Saris and Richard E.
Boyatzis

• SMABS Budapest, 2006

2
OBJECTIVE
Increase knowledge of emotional intelligence and
contribute to offer an appropriate model to be
used and developed within Spanish organizational
settings.
3
Why?

• The concept of emotional intelligence (EI)
  resonates in the business world (Domagalski,
  1999) and many authors have called for more
  research that clearly conceptualized it (e.g.
  Sala, 2002 Becker, 2003, Day and Carroll, 2004).
  
• Construct validity is still a limitation.
• Discriminant validity is challenged by the
  statements that claim that EI does not explain
  much more than other well-known psychological
  constructs (Davies, Stankov and Roberts, 1998
  Schulte et al., 2004)
• EI has proven resistant to adequate measurement
  (Schutte, Malouff, Hall, haggerty, Cooper, Golden
  and Dornheim, 1998 Dulewicz, Higgs and Slaski,
  2003).

4
Specific Objectives of the paper

• EI Theory of Performance Model of Goleman and
  Boyatzis (Goleman 1995, 1998 Boyatzis, 1982 and
  Goleman, Boyatzis and McKee, 2000).
• Big Five Personality Model (e.g. Costa and
  McCrae, 1992).
• Construct validity
• Internal structure of a self-report EI
  instrument.
• The notion of consensus scores
• Discriminant Validity
• EI and Personality discriminant validity.
• Low to moderate correlations are expected 1)
  Component of emotions 2) Self- vs. Consensus-
  scores 3) Contextualization of the models 4)
  Descriptive vs. Predictive approximations and,
  5) Other studies (e.g. Murensky, 2000).

5
Methods

• Participants
• 120 employees of two medium-sized public
  companies within the Spanish context.
• 120 reported gender (77 women and 43 men) and 110
  reported age (mean 38.44 SD 9.45)
• Measurement Instruments
• ECI-2 72 items with response categories based on
  frequency of demonstration or observation (each
  item on a scale of 1 to 5) of specific
  behaviours.
• Eighteen competencies result the test that are
  grouped into four theoretical clusters.
• Spanish translation provided by HayGroup.
• NEO-FFI 60 items of the Revised NEO Personality
  Inventory (NEO-PI-R)
• Five domains of adult personality Neuroticism,
  Extroversion, Openness to Experience,
  Agreeableness, Conscientiousness.

6
Methods

• Procedure
• 120 completed the ECI-2 (self-, others- and
  consensus scores).
• 82 completed the NEO-Five-Factor Inventory.
• Design and Strategy of Analysis.
• CFA General EI model was divided into simplier
  ones (CFA with 72 items in small samples presents
  serius problems, Boomsma, 1993).
• Groups were formed following the empirical
  clustering reported by Batista-Foguet et al.
  (2006) and respecting the theoretical clusters
  (Goleman et al., 2002).

7
Methods

• EI Model (Goleman, Boyatzis and McKee, 2002)

8
Methods

• Groups

Group 1
Group 2
Group 3
Group 7
Group 4
Group 6
Group 5
9
Methods

• Design and Strategy of Analysis.
• CFA without restricting the relationships between
  the items and the factors, except for
  identification restrictions that were chosen in
  line with the factor structure expected.
• Looping process adding restrictions
  progressively.
• To adjust the models and modify them deleting
  invalid variables (Batista-Foguet and Coenders,
  2000) (1) items with loadings lower than .40 in
  the correspondent factor were deleted due to poor
  quality and, (2) items with loadings higher than
  .40 in other factor that is not theoretically
  appropriate were deleted due to theoretical
  invalidity.
• Unweighted summated scales were calculated.
• Then, the structure of subsequent models with the
  appropriate items was assessed.
• Unweighted summated scales were calculated.

10
Results
Three competences of the model (Accurate
Self-Assessment, Influence and Conflict
Management) were omitted.
11
Results Internal structure of the ECI-2
Three competences of the model (Accurate
Self-Assessment, Influence and Conflict
Management) were omitted.
12
Results Internal structure of the ECI-2

• Reliabilities and Correlation matrix

13
Results

• EFA The Big Five personality model was
  assesed.

14
Results

• Correlations between personality traits and
  emotional competencies were calculated

15
Results

• Regressions predicting Emotional Competencies

16
Results

• Coefficients

17
Discussion

• Construct Validity of EI
• The findings are in agreement with the
  behavioural competencies conceptual framework of
  EI.
• A major limitation is sample size.
• Consensus scores
• Competence correlations
• Invalid or poor items/competencies

• Construct Validity of Personality

18
Discussion

• Discriminant Validity
• Correlations between personality traits and
  emotional competencies are low to moderate
• Extroversion and Conscientiousness are purported
  to be the most important personality factors to
  predict emotional competencies
• The percentage of total variation of the
  competencies explained by the personality traits
  ranged from 0,074 to 0,185.

19

• Questions?