ADAPTATION PROCEDURE OF LEVEL OF EMOTIONAL INTELLIGENCE (LEI) TEST AND ITS METHODOLOGY PROBLEMS

Research background and hypothesis. There are only few original psychological methodologies developed in Lithuania, therefore most often foreign test procedure adaptation is undertaken. In Lithuania the iniative of psychologists to develop original methodologies is met with hostility though it is universally agreed that tests developed abroad are marked with strong “cultural charge”. Furthermore, some other problematic aspects of foreign test development become obvious, e. g. free treatment of survey construct, etc. Therefore, while adapting new methodology, in addition to correct implementation of all psychometric procedures, detection of problematic areas of the adapted methodology is relevant. Research aim was to present new adapted methodology of Level of Emotional Intelligence (LEI) test and to demonstrate major problems that arise in the adaptation procedure. Research methods. The survey involved 590 students from seven Lithuanian universities. In the five-step survey LEI test validity and reliability were examined using different statistical analysis methods (authors S. Beliayev, S. Yanovitch and A. Mazurov, 2009). Research results . The authors of the methodology maintain that the psychometric parameters of LEI test are sufficient. However, the authors of the present article have uncovered a number of disputable aspects of the said methodology. All required adaptation procedures of LEI methodology were performed: new scale structure was developed and tested using multi-factor analysis, test internal consistence and resolution were checked, test construct validity and re-test reliability were verified, test construction mistakes were spotlighted. Discussion and conclusions. LEI test adaptation procedure has disclosed important methodological problems that have to be addressed while adapting instruments of psychological research: 1) theoretical concept of methodology suggested by the authors and essential conformity of the said methodology to the universally accepted definition of construct in the science of psychology must be evaluated; 2) statement assessment logic proposed by the authors of methodology must be analyzed; 3) validation methods used by the authors must be critically assessed.


INTRODUCTION
T he need for psychometric tests is obvious in various spheres of life: they are most often encountered in professional screening, search for talented, creative people, etc. (Boyatzis, 2008(Boyatzis, a, 2008 b; b;Leonard, 2008).Therefore, problems of both valid methodology development and its adaptation remain relevant.As major mistakes are made in methodology construction and adaptation, psychometry is not always reliable and sometimes it is depreciatingly called "pathology of science" (Barrett, 2003).Scientists themselves with such declarations want to draw attention to poorly developed test methodologies, inadequately performed adaptation procedure, etc.Therefore, essentially the principal question of psychometry today is the same as it was in the last decades: how instruments that would really measure the desired mental attributes could be developed and how it can be ensured that the developed instruments are maximally reliable.It was this scientific problem that inspired the authors, based on specific methodology, to show in more detail problems that are encountered in adaptation procedure.As the authors of the present article currently implement a large scale project "Emotional Intelligence of Young People of Lithuania and its Factors", which is financed by Research Council of Lithuania, in preparation for adaptation of original EI methodology development and standardization procedure various pilot investigations were performed, one of which is presented in this article.
The purpose of the present article was to introduce a new adapted Level of Emotional Intelligence (LEI) test methodology and to unveil the essential problems that arose in the procedure of adaptation.Objectives: 1) presentation of the adapted methodology and its psychometric parameters; 2) highlighting the principle obstacles in test adaptation procedure.

RESEARCH METHODS
Methodology.Level of Emotional Intelligence (LEI) test developed by S. Beliayev, S. Yanovitch and A. Mazurov was selected with the permission of the authors.The said methodology has two scales: personality and communication scale and combined -overall emotional intelligence scale.More detailed parameters of the test will be presented in the results section as in that section adaptation procedure results obtained by the authors will be shown in parallel.
Phases of research.The adaptation procedure of Level of Emotional Intelligence (LEI) test developed by S. Beliayev, S. Yanovitch and A. Mazurov was implemented in several phases.In the first phase, upon the authors' permission, methodology was translated into Lithuanian.The translation was performed independently by two psychologists-experts that hold scientific degree and are experienced in test adaptation.Later translations were compared, discussed and the final version of the methodology was selected.Test was performed with 10 subjects to find out whether methodology statements were understandable and clear to the subjects.In the second phase test questionnaire was prepared.In addition to the methodology prepared for adaptation, social and demographic questions about subject's gender, age, family parameters (is it full, what is its financial status, etc.) and academic achievements were included into the questionnaire.To test construct validity, Social Competence Test (adapted by R. Lekavičienė, 2001) was also included into the questionnaire.In the third phase the survey of 590 subjects was performed.In the fourth phase re-test was performed.82 subjects participated in the re-test procedure.In the fifth phase statistical processing of the obtained data was performed using SPSS 17.0 for Windows.
Methods of mathematical statistics: factor analysis using orthogonal rotation VARIMAX, factor internal consistence reliability was assessed by calculating Cronbach's α coefficient, test validity was verified using dispersion analysis ANOVA and Spearman's correlation coefficient, two measurements were compared using Wilcoxon test.

RESEARCH RESULTS
To determine the structure of LEI test variables, factor analysis method was used.Factor analysis was applied to multiplex primary variables of the survey and to generate LEI scales.To process survey data, main component method and orthogonal rotation VARIMAX, e. i. rotation of variable axes to determine maximal dispersion, were used.Sufficiently high Kaiser-Meyer-Olkin (KMO) coefficient, which amounted to 0.71, and the overall descriptive power (dispersion) of the model, which was 54.2%, proved that variables were appropriate for factor analysis.To determine whether variables that fell into the same factor correlated with each other, internal consistence reliability (Cronbach's α coefficient) of each factor was verified.Table 1 shows the following statistical parameters: in the "Scale" column Cronbach's α coefficient, average intercorrelation among items (statements) and minimal as well as maximal correlation are presented; in the "Statements Summarized in the Scale" column averages and standard deviations are shown; the other three columns contain: resolution i/tt, factor loadings L, and model overall explained dispersion in percent.
After test internal consistence assessment and factor validation, nine LEI scales (dimensions) were established.For each cluster of primary attributes that fell into the same factor conditional title was assigned taking into account semantics of the primary statements (see Table 1 • Can you be unfaithful to your partner?(-) (M = 2.97.SD = 1.03).0.30 0.50

V. Ability to Demand
• Will you file a complaint without hesitation if your rights are violated?(M = 2.57.SD = 0.83).0.74 5.6 • In a restaurant a dish is brought that is obviously inadequately prepared.

VII. Tolerance
• Would you calmly accept the news that your friend is homosexual?(M = 2.67.SD = 1.05).0.76 5.2 • Can you be friendly even with those that you are not very fond of?(M = 2.97.SD = 0.84).0.52

VIII. Noncompliance
• Do you defend your attitude to the end?(M = 3.19.SD = 0.76).0.62 11.8 • Do you point to the people surrounding you their shortcomings?

IX. Devotion
• Could you honestly nurse a patient that is tied to the bed? (M = 3.24.SD = 0.85).0.68 6.6 • Do you experience pleasure when buying gifts to your friends?(M = 3.51.SD = 0.72).0.47 to psychometric surveys rule of ignoring factors that explained dispersion lower than 5-10%.
Cronbach's α coefficient, resolution, and factor loading met minimal requirements applied to psychometric tests.As emotional intelligence construct was tightly related to social competence construct (Guralnick, 1999;Weare, Gray, 2003;CASEL, 2008;Mirabile, 2010 et al.), LEI test construct validity was verified by looking for coherence between LEI test and Social Competence Test (developed by R. Lekavičienė, 2001) results.The said test consisted of the following subscales: "Overall Self-confidence" (the scale was characterized by statements like "In general, I rely on my decisions"); "Resistance to Failures and Criticism" (the scale was characterized by statements like "I am constantly afraid that I will say or do something inappropriate" with (-) sign); "Noncompliance" ("I cannot argue with people that are dear to me" with (-) sign); "Not Feeling Guilty" (I am feeling guilty if I do not give pittance to a beggar" with (-) sign); "Ability to Ask for a Favor" ("I feel uneasy when others offer me help" with (-) sign); "Ability to Demand" ("In a restaurant it is very easy for me to call loudly for a waiter"); and "Ability to Express Feelings" ("In the presence of the opposite sex I am always shy" with (-) sign).Test validity was verified using dispersion analysis (ANOVA).The said analysis was applied comparing average differences among several independent samples.The dependent variable was Social Competence test results, i. e. scale averages (Cronbach's α varied within the range of 0.60 to 0.80).At selected level of significance α = 0.01 calculated p = 0.00.Zero hypothesis was rejected, in other words, not all averages were equal.Thus, scores of emotional intelligence test and social competence test were related.After analysis of results by individual emotional intelligence scales statistically significant correlation with social competence scales was determined.Correlation of both test scale data are shown in Table 2.
All seven social competence scales positively correlated with two scales of LEI -"Management of Exterior Related Destructive Emotions" (Spearman's correlation coefficients varied within the range of 0.11 to 0.42, p = 0.00; an example of scale statement could be: "I do not feel discomfort talking to a police officer", etc.) and "Ability to Demand' (Spearman's correlation coefficients varied within the range of 0.16 to 0.37, p = 0.00; an example of scale statement could be: "I will file a complaint without hesitation if my rights are violated").Also, six correlations (out of seven possible) were determined with "Noncompliance" scale (Spearman's correlation coefficients varied within the range of 0.14 to 0.40, p = 0.00; an example of scale statement could be: "I defend my attitude to the end").Some negative correlations were theoretically meaningful, e. g. between "Caution" ("I seldom make spontaneous decisions") in Emotional Intelligence Scale and some social competence scales, such as "Overall Self-confidence", "Resistance to Failures and Criticism", "Noncompliance" and "Ability do Demand".Similarly, "Risk Limitation" scale, which in essence indicated person's fear to take risk, and it negatively correlated with such social competence scales as "Overall Self-confidence", "Ability to Demand" and the like.

Noncompliance
All other individual parameters of correlation among various scales of emotional intelligence and social intelligence were also logically valid in the semantic sense, therefore in summary it may be maintained that LEI construct validity is sufficient.
In further calculations (after multi-step factorization) the said nine dimensions (scales) were multiplexed into two larger components (see Table 3).
Proof of factor analysis relevance is the possibility of theoretical interpretation of factors clustered in a given factor.The obtained model satisfied the said condition.It was theoretically valid and statistically meaningful.Two generalized factors were given subtest names of methodology developed by the authors (communicational EI and Personality EI) and the said names in this case were also appropriate.Communicational emotional intelligence encompasses the ability to control impulsive behavior towards the others, take care of surrounding people, i. e. to build and maintain mutually satisfactory relationship and use person's abilities not just for his own good, but also for the good of the surrounding people.Personality emotional intelligence encompasses the ability to understand own positive and negative sides, resist to stress situations, change own attitude when proof of made mistake is provided, to be open and tolerant to various ideas, orientations, methods and customs.
Test reliability was verified using re-test when test procedure was repeated after 5 months.Results obtained in the first and in the second tests were compared using Wilcoxon Signed Rank Test.The said test was handy because of its nonparametric nature and ability to test paired median differences.Antoher advantage of the said test was its logical conformity with t-test, therefore it was a very effective tool for analysis of compatibility of two dependent measurements as well as for fenomenological assessment of possible causes of differences.Hypotheses on two measurement compatibility were developed.Thus, if statistical significance level p ≤ 0.05, hypothesis about compatibility of two measurements was rejected and statement about difference of two measurement results is accepted.Wilcoxon Signed Rank Test showed that neither joint assessment of the first test and re-test nor evaluation of both measurements by individual scales met the condition of p ≤ 0.05, therefore it can be maintained that there were no statistically significant differences between both measurements.

DISCUSSION
As problematic issues of the researched methodology validity will be discussed in this section, the reader may find it interesting to learn about the procedure of "birth" of methodology developed by S. Beliayev, S. Yanovitch, and A. Mazurov, therefore we will concisely present the most important phases of the development of the said methodology and its standardization as they were described by the test authors.Initially, while developing the said methodology so called rough version containing 177 questions was prepared.Content validity was verified using the method  The following procedures were performed when verifying compatibility of scales and reliability of methodology: 1) correlation links between both scales were determined (Pearson's correlation coefficient r = 0.39, p ≤ 1.001); 2) correlation between overall emotional intelligence level and individual scales was determined (Pearson correlation coefficients were r = 0.88 and r = 0.78 for personality and communicational EI correspondingly, p ≤ 1.001).
Re-test with 65 subjects was performed 2.5 months after the first survey; the results confirmed sufficiently high reliability of the methodology and stability of attributes (the following Spearman's correlation coefficients were obtained: personality EI r = 0.87, communicational EI r = 0.89, overall EI r = 0.91, p ≤ 0.05).Using factor analysis relations among scales were examined.The authors specified the following weights of factors: personality EI L = 0.86; communicational EI L = 0.80, overall EI L = 0.99 and claimed that factor structure of the methodology was stable.Construct validity of the methodology was repeatedly verified, but this time verification was based on Eysenck EPQ (The Eysenck Personality Questionnaire) methodology.Such choice of the authors was determined by their treatment of temper as a natural structure and support of the opinion that emotional intelligence was of physiological nature.Developers of the test determined the following scale correlations between methodology developed by them and Eysenck EPQ: personality EI, communicational EI and overall EI with extroversion-introversion r = 0.82, r = 0.15 ir r = 0.72 (p ≤ 0.05) correspondingly; personality EI, communicational EI and overall EI with neurotism r = 0.38, r = 0.69, r = 0.66 (p ≤ 0.05).
Thus the authors of methodology maintained that construct validity of methodology was sufficient.
In the adaptation of Level of Emotional Intelligence (LEI) test in Lithuania, all required procedures were performed and problematic areas were identified that will be discussed shortly.After detailed qualitative analysis of statements became obvious that major part of statements could hardly be attributed to identification of EI level.We would like to remind that classical concept of emotional intelligence contains four major aspects that express effective understanding of own and others' feelings and emotions (understand my own strengths, weaknesses, needs and drives, recognize how feelings affect me, listen to others, understand others' perspectives, sense how others are feeling) and management (manage feelings and impulses, choose words carefully, avoid hasty judgments, motivate self to achieve, recognize others' specific strengths, inspire and motivate others, overcome obstacles that prevent improvements, nurture relationships, promote a friendly, co-operative climate) (Jordan et al., 2002;Schutte et al., 2002;Boyatzis, 2007;Palmer, 2007;Wong et al., 2007;Koman, Wolff, 2008) (see Figure 1).
After statement analysis it became obvious that some of them were related to the assessment of the level of emotionality, but not emotional intelligence.For example, "Are you venturesome?"High score on this statement could hardly be an indicator of high personality EI.On the contrary, a person capable to control his/her emotions is more intelligent emotionally.Some of the statements are also ambiguous and disputable, e. g. "Do you tell the truth into collocutors face?" (unambiguous answer is impossible: a person who is emotionally intelligent will consider the situation and the personality of collocutor), "Do you give pittance?"(unambiguous answer also is impossible).Some statements are altogether astonishing as it is not clear on what basis they were included into EI test: "Do you observe fashion changes?", "Will you pick up without hesitation trash that you missed throwing into wastebasket?"etc.When it became obvious that LEI methodology measured the level of emotionality, but not emotional intelligence statements were submitted for assessment of psychologists-experts.The purpose of the said assessment was to reject very ambiguous questions and assess which questions might be analyzed by calculating the scores assigned to respondent's questions (no, rather no, rather yes, yes) and answers to which questions should be re-coded, i. e. the opposite score calculated if the answer was to be rated as emotionally intelligent instead of emotional.For example, if answers to such statements as "Do you feel embarrassed when you have to submit your urine for test in clinic?", "Do you feel discomfort talking to a police officer?" were positive, the lowest score should be assigned meaning that such answer, as was mentioned above, expressed high tension of the subject rather than emotionally intelligent behavior.The methodology was fundamentally rearranged using statistical procedures.The problematic areas highlighted in the article should help those psychologists who intend to undertake test adaptation procedures.

CONCLUSIONS AND PERSPECTIVES
1.All necessary LEI methodology adaptation procedures were performed: communication, EI and emotional EI Cronbach's α values, resolution and factor loading indices that were calculated using multi-factor analysis met the minimal requirements for psychometric tests; LEI construct validity and reliability verified using re-test were sufficient.
2. The adaptation procedure unveiled important methodology problems that need to be addressed in the adaptation of psychological investigation instruments: 1) theoretical concept of construct and conformity of the said concept to the principal construct definition in the science of psychology must be evaluated; 2) statement assessment logic proposed by the authors has to be analyzed; 3) validation methods used by the methodology authors must be critically assessed.
As some psychometric parameters of LEI methodology meet only minimal requirements, it would be expedient to continue validation of the present methodology, e. g. verify its discriminant and criterion validity. Acknowledgement.

Figure
Figure.Classical concept of the structure of emotional intelligence

Table 2 . Correlation between LEI and Social Competence Test Subscales Note
: Spearman's correlation coefficient r and statistical level of significance p are specified in the table.

Table 3 .
The The research has been supported by Research Council of Lithuania (Contract No. MIP-106/2012).Classical concept of the structure of emotional intelligence