• Tidak ada hasil yang ditemukan

Research procedure

Dalam dokumen Declaration of originality of research (Halaman 53-91)

CHAPTER 1 INTRODUCTION

4.3 Research procedure

The HR manager or coordinator in the various mining cites in the chroming, platinum and gold mining industry in the North West and Gauteng was contacted. After explaining the research to the HR coordinator permission was asked in order to complete the study. It was important for the coordinator to have a very good relationship with his employees in order to help with the implementation of the study. He had to be able to communicate effectively with the employees in the training centre and be able to group them together according to the various languages used in the study. After this had been done, the reasons for the study were explained. In the centre the GEIS was administered and after completion dropped in an anonymous box. Participation to the study had been voluntary, and the confidentiality and anonymity of participants was emphasised.

The different organisations of the various mines that will participate in the study had received complete feedback, one by one, regarding the results.

Statistical analysis

The IBM SPSS- program version 20.0 (SPSS, 2011), was used to carry out the analysis of the data that is collected. Descriptive statistics was used in this study and it involved the testing of assumptions (Pallant, 2005) and it provided the researcher with a summary of the data he/ she has collected. The purpose of this kind of statistic was to provide the researcher with an overall, logical and simple picture of the data that was collected (Struwig & Stead, 2007). Descriptive statistics that was used included the mean, standard deviation, skewness and kurtosis and the

alpha coefficient (Pallant, 2005). The mean is the sum of the observations that are going to be made which will be divided by the number of observations that will constitute to group (Marascuilo & Serlin, 1988), in other words it can be seen as the average. The standard deviation

“measures the deviation of each score from the mean and the averages the deviations” (Struwig

& Stead, 2007, p. 158). Skewness and kurtosis refers to the distribution of the scores (Struwig &

Stead, 2007). “Skewness refers to the degree of deviation from symmetry, while kurtosis refers to how flat or peaked the distribution is (Struwig & Stead, 2007 p. 159). The degree of reliability was articulated by the Cronbach Alpha Coefficient; it ranges from 0.00 to 1.00 and the closer the alpha coefficient is to 1.00 the closer it will be to the true score (Struwig & Stead, 2007). The Cronbach Alpha Coefficient is acceptable when α > 0.70.

SPSS was utilised to determine item bias by means of ordinal logistic regression. According to Kim (2001) there are various ways to detect Differential item functioning. DIF can be defined as the interference of some demographic characteristic or grouping of the tight relationship between trait level and item responses (Crane, Gibbons, Jolley & Belle, 2006). One of these methods is Zumbo‟s (2009) ordinal logistic regression approach. A variation of ordinal logistic regression to detect DIF between cultural groups was carried out by using several comparisons for different response categories on each item. When applying the regression procedure, outliers beyond the 95% confidence interval was used as DIF items. Crane, Gibbons, Jolley and van Belle (2006) mention that the ordinal logistic regression approaches for testing DIF is not a complicated statistical analysis to accomplish. They further state that because of the ordinal LR framework, many demographic characteristics can be evaluated to determine whether items display DIF.

In order to determine if the items of the GEIS are unbiased a pre-test was conducted by means of various analyses of variance on the four factors of the GEIS. ANOVA showed the expression of the tests of interests in terms of variance estimates (Muller & Fetterman, 2002). In order to determine if there were group differences with regards to the way the participants answered each item, ANOVAS on each item was conducted.

In the AMOS program (Arbuckle, 1997), confirmatory factor analysis and structural equation modelling methods were used to construct and test a four-factor model of emotional intelligence

across language groups. Hypothesised relationships were tested empirically for goodness-of-fit with the sample data. The X and several other goodness-of-fit indices summarise the degree of 2

correspondence between the implied and observed covariance matrices. However, the X test is 2

commonly recognised to be problematic (Joreskog, 1969). It is sensitive to sample size, and could also be invalid when distributional assumptions are violated, leading to the rejection of good models or the retention of bad ones. Due to the drawbacks of X 2

statistics have been developed.

test, many alternative fit

The Goodness-of-Fit Index (GFI) is commonly used to indicate the relative amount of variance and co-variance in the sample predicted by the estimates of the population. It usually varies between 0 and 1, and a result of 0.90 or above indicates a good model.

The Comparative Fit Index (CFI) also compares the hypothesised and independent models, but considers sample size when doing so. The Tucker-Lewis Index (TLI) is a relative measure of co variation explained by the hypothesised model which has been specifically designed for the assessment of factor models (Tucker & Lewis, 1973). In order to acquire good model fit it is recommended for the NFI, CFI and TLI to be acceptable above the 0.90 level (Bentler, 1992), although Hu and Bentler (1999) recommended a cut-off value of 0.95. The Root Mean Square Error of Approximation (RMSEA) estimates the overall amount of error; it is a fu nction of the fitting function value relative to the degrees off freedom (Brown & Cudeck, 1993). Hu and Bentler (1999) suggested a value of 0.06 to indicate acceptable whereas MacCullum, Browne, and Sugawara (1996) suggested that values between 0.08 and 1.00 indicate mediocre fit and values above 1.00 poor fit.

In accordance with Hu and Bentler (1999), a combination approach was used to evaluate the model fit. Following Hu and Bentler (1999), several fit indices (GFI, IFI, TLI, CFI and RMSEA) were used to evaluate the fit of each CFA model. Specifically an absolute close-fit index (RMSEA) and two incremental close-fit indices were chosen (TLI and CFI) because it has been argued that they would provide more stable and accurate estimates than several other indices (Hu

& Bentler, 1999; Maruyama, 1998) and the mentioned three indices have been used in other confirmatory factor analysis studies of emotional intelligence (Gignac et al., 2005). Other fit

indices was included to support the TLI, GFI and RMSEA as they were used in other studies for evaluation of psychological tests (Parker, Taylor, & Bagby, 2003) and provided easy comparisons to the other data sets.

RESULTS

The results will comprise of the descriptive statistics, ANOVAs of each of the items of the GEIS, uniform and non-uniform item bias of the items of the GEIS and finally Goodness-of-fit statistics for the total population, Sotho and West-Germanic groups.

Descriptive statistics

Descriptive statistics describe the core features of the data in a study, by taking into account the mean, standard deviations, skewness, kurtosis, and Cronbach alpha coefficients (William, 2006).

Table 2 presents the mean, standard deviations, skewness, kurtosis, and Cronbach alpha coefficients of the 52 items in the GEIS.

Table 2

Descriptive Statistics of the 52 Items for the Greek Emotional Intelligence Scale (N=357)

Model Factor Mean SD Skewness Kurtosis α

Combined Expression 26.29 4.83 0.57 0.69 0.70

sample and

Recognition

Caring and 52.05 7.96 0.75 2.17 0.72

Empathy

Control of 35.33 6.67 0.43 -0.52 0.82

Emotion Use of Emotion

51.50 7.03 1.23 4.77 0.79

West- Expression 25.18 3.32 0.83 1.87 0.66

Germanic and group Recognition

Caring and 47.65 4.96 0.24 -0.11 0.63

Empathy

Control of 31.65 4.23 0.53 1.00 0.80

Emotion

Use of 47.67 3.48 -0.50 1.72 0.62

Emotion

Model Factor Mean SD Skewness Kurtosis α

Sotho group Expression 27.11 5.62 0.22 -0.60 0.64

and Recognition

Caring and 55.56 8.15 0.43 2.80 0.70

Empathy

Control of 38.24 6.80 -0.12 -0.65 0.79

Emotion

West- Use of 22.65 4.01 -1.13 1.19 0.83

Germanic Emotion Group

Caring and 29.70 5.40 -0.74 0.95 0.83

Empathy

Control of 25.71 4.97 -0.53 -0.15 0.77

Emotion

Table 2 indicates the descriptive statistics for the items of the GEIS. The table shows that Cronbach alpha coefficients varying from 0.62 to 0.82 were obtained for all the GEIS factors of each model. The combined sample as well as the Sotho language group showed acceptable internal consistency. The factors “Use of Emotion to Facilitate Thinking‟ and “Caring and Empathy‟ (0.63) in the confirmatory four factor model of the West-Germanic group (0.62) showed the lowest alpha values. This indicated that the four factor model for the West-Germanic group was not the best fit. It was therefore decided to test a three-factor model for the West- Germanic group. When investigating the three-factor West-Germanic language group, the alphas showed excellent values of 0.83 and 0.77 respectively. It is evident from the table that the scores from the various scales for each model are relatively normally distributed, with low skewness and kurtosis being noted, apart from the following exceptions: “Caring and Empathy‟ and “Use of Emotion to Facilitate Thinking‟ in the combined model, “Use of Emotion to Facilitate Thinking‟ in the West-Germanic group, and “Caring and Empathy‟ in the Sotho group. In conclusion, it can be said that all the factors from the GEIS for the combined sample as well as Sotho language group showed sufficient reliability and validity to be utilised for subsequent analysis. The four-factor model for the West-Germanic group was however under the guideline of 0.70 provided by Nunnaly (1994) and the three-factor model was therefore tested and proved sufficient reliability. .

Next, ANOVAS on each of the items of the GEIS were executed to serve as a pre-analysis for DIF. The following four Tables show the results of the ANOVAs conducted for the items on the

four emotional intelligence factors, namely “Caring and Empathy‟, “Control of Emotion‟,

“Expression and Recognition of Emotions‟, and “Use of Emotions to Facilitate Thinking‟.

Statistical significance, practical statistical significance was presented, as well as the mean differences of the two language groups on each item respectively. According to Cohen (1992), the practical significance of the results were determined by using the cut offs of 0.2, 0.5, and 0. 8 as small, medium, and large.

Table 3

Summary of ANOVAs between cultural groups per item on the Caring and Empathy Factor

Item descriptions

Sum of squares

df Mean

square

f Sig. Eta

squared

Mean WG

Mean Sotho Caring

Empathy EQ33: I cannot stand

Between Groups

18.30 1.00 18.30 15.17 0.00 0.04 2.62 2.17

injustice

Within 428.10 355.00 1.21

Groups

Total 446.39 356.00

EQ37: I am not interested

Between Groups

128.85 1.00 128.85 108.09 0.00 0.23 2.20 3.41

in the problems of others

Within 423.19 355.00 1.19

Groups

Total 552.04 356.00

EQ41: I am open to listen

Between Groups

30.86 1.00 30.86 38.76 0.00 0.10 3.71 4.30

to others

Within 282.59 355.00 0.80

Groups

Total 313.45 356.00

Item descriptions

Sum of squares

df Mean

square

f Sig. Eta

squared

Mean WG

Mean Sotho

EQ24: I am always willing

Between Groups

25.61 1.00 25.61 21.60 0.00 0.06 3.44 3.98

to help

someone who

is confronted

with personal

problems

Within 420.96 355.00 1.19

Groups

Total 446.57 356.00

EQ34: I respond to the

Between Groups

2.49 1.00 2.49 2.60 0.11 0.01 3.55 3.72

emotions of others

Within Groups

339.34 355.00 0.96

Total 341.82 356.00

EQ21: I sympathize

Between Groups

81.30 1.00 81.30 67.91 0.00 0.16 2.87 3.83

with others’

personal problems

Within Groups

425.00 355.00 1.20

Total 506.29 356.00

EQ46: I show my concern to

Between Groups

17.60 1.00 17.60 21.64 0.00 0.06 3.56 4.01

others

Within 288.76 355.00 0.81

Groups

Total 306.36 356.00

Item Sum of df Mean f Sig. Eta Mean Mean

descriptions squares square squared WG Sotho

EQ10: I believe that I am a person who helps and is considerate of others.

Between Groups

6.54 1.00 6.54 5.64 0.02 0.02 3.68 3.95

Within Groups

412.04 355.00 1.16

Total 418.58 356.00 EQ19: I

respect others’

emotions

Between

Groups 48.98 1.00 48.98 49.86 0.00 0.12 3.20 3.95

Within Groups

348.69 355.00 0.98

Total 397.66 356.00 EQ5: I like to

talk with others about their problems.

Between Groups

26.73 1.00 26.73 18.90 0.00 0.05 3.08 3.63

Within Groups

502.15 355.00 1.42

Total 528.88 356.00 EQ28: When I

am conversing with someone, I am

concentrated to what he/she is telling me.

Between Groups

8.14 1.00 8.14 8.91 0.00 0.02 3.65 3.95

Within Groups

324.35 355.00 0.91 Total 332.49 356.00

Item

descriptions Sum of

squares df Mean

square f Sig. Eta

squared Mean

WG Mean

Sotho

EQ50: I can Between 19.83 1.00 19.83 14.82 0.00 0.04 3.21 3.68

easily Groups

understand what someone

else feels by putting myself

in their position

Within 475.16 355.00 1.34

Groups

Total 495.00 356.00

EQ51: When Between 3.46 1.00 3.46 4.12 0.04 0.01 3.70 3.89

someone is Groups

talking to me

about their problems, I almost feel like

I have experienced these problems

myself.

Within 298.20 355.00 0.84

Groups

Total 301.66 356.00

EQ14: I am Between 23.33 1.00 23.33 3.85 0.05 0.01 2.89 3.41

interested in Groups

others’

psychological

motives.

Within 2153.20 355.00 6.07

Groups

Total 2176.53 356.00

EQ36: I make others feel

Between Groups

168.50 1.00 168.50 126.63 0.00 0.26 2.89 3.41

comfortable with me.

Within Groups

472.38 355.00 1.33 Total 640.88 356.00

Table 3 provides the ANOVA results for the mean item scores between and within groups for the factor „Caring and Empathy‟.

Table 3 indicates that there are significant mean differences (p≤0.01) between the language groups on all the items except for item 34, 10, 51 and 14 on the “Caring and Empathy‟ factor.

The effect sizes (Eta squared) varied between small, medium and a large effect. According to Steyn (2010), 0.20 would be considered to be a small effect size, a medium effect size would be 0.50, while 0.80 would be a large effect size. The majority of the effect sizes had no practical significance, apart from item 37, which had small practical significance of 0.23.

These differences on an item level provide a sound basis for further DIF investigation.

Table 4

Summary of ANOVAs between cultural groups per item on the Control of Emotions Factor

Item descriptions

Sum of squares

df Mean

square

f Sig. Eta

squared

Mean WG

Mean Sotho Control of

emotions

EQ15: I often get angry and afterwards I find my anger inexcusable.

Between Groups

78.19 1.00 78.19 7.99 0.01 0.02 3.79 2.85

Within Groups

3473.39 355.00 9.78

Total 3551.58 356.00

EQ35: I get carried away by emotions of anger.

Between

Groups 91.39 1.00 91.39 77.28 0.00 0.18 3.78 2.76

Within Groups

419.82 355.00 1.18

Total 511.20 356.00

EQ7: When I Between 0.91 1.00 0.91 0.70 0.41 0.00 2.15 2.05 am upset Groups

everything bothers me.

Within 465.83 355.00 1.31

Groups

Total 466.74 356.00

EQ17: I often Between 3.60 1.00 3.60 2.64 0.11 0.01 3.66 3.46

regret things I Groups

say when I get

angry.

Within 484.69 355.00 1.37

Groups

Total 488.29 356.00

EQ6: When I Between 81.94 1.00 81.94 55.01 0.00 0.13 3.63 2.67

am under Groups

pressure I snap.

Within 528.82 355.00 1.49

Groups

Total 610.76 356.00

EQ47: When I Between 11.59 1.00 11.59 11.61 0.00 0.03 3.79 3.43

am Groups

experiencing a

sad event I react intensely.

Within 354.35 355.00 1.00

Groups

Total 365.94 356.00

EQ2: I often Between 12.20 1.00 12.20 8.04 0.01 0.02 2.88 2.51

have Groups

conflicting emotions for

the same persons.

Within 538.45 355.00 1.52

Groups

Total 550.66 356.00

51

Item Sum of df Mean f Sig. Eta Mean Mean

Item descriptions

Sum of squares

df Mean

square

f Sig. Eta

squared

Mean WG

Mean Sotho EQ38: I have

the tendency to

Between Groups

16.13 1.00 16.13 10.13 0.00 0.03 2.01 2.44

show my impatience to

others.

Within 565.12 355.00 1.59

Groups

Total 581.25 356.00

EQ52: When I try to help

Between Groups

80.24 1.00 80.24 58.73 0.00 0.14 3.46 2.51

someone, I unintentionally

give directions

by strongly criticizing them.

Within 485.01 355.00 1.37 Groups

Total 565.25 356.00

EQ49: Prior Between 32.70 1.00 32.70 28.45 0.00 0.07 3.76 3.15

to important Groups

events, I feel

tense.

Within 407.92 355.00 1.15

Groups

Total 440.62 356.00

EQ31: I Between 68.74 1.00 68.74 55.31 0.00 0.14 2.87 3.75

usually control Groups

my anger.

Within 441.24 355.00 1.24

Groups

Total 509.98 356.00

EQ1: I get Between 139.77 1.00 139.77 121.85 0.00 0.26 1.37 2.63

angry easily, Groups

but my anger

does not last

for too long.

Within 407.19 355.00 1.15

Groups

Total 546.96 356.00

Table 4 gives the ANOVA results for the mean item scores between and within groups for the factor „Control of Emotion‟.

Table 4 indicates that there are significant mean differences (p≤0.01) between the language groups on all the items except for items 7 and 17 on the “Control of emotion‟ factor. The effect sizes (Eta squared) varied between small, medium, and a large effect. The majority of the effect sizes (Eta squared) appeared to be of no practical significance, except for item 1, with a 0.26 practical significance.

These differences on an item level provide a sound basis for further DIF investigation.

Table 5

Summary of ANOVAs between cultural groups per item on the Expression and Recognition of Emotions Factor

Item Sum of df Mean f Sig. Eta Mean Mean

descriptions squares square squared WG Sotho

Expression and Recognition of

Emotions

EQ20: I find it difficult to express my emotions to others.

Between Groups

47.55 1.00 47.55 43.77 0.00 0.11 1.88 2.62

Within Groups

385.65 355.00 1.09 Total 433.20 356.00

Item Sum of df Mean f Sig. Eta Mean Mean

descriptions squares square squared WG Sotho

EQ13: I am Between 36.41 1.00 36.41 27.75 0.00 0.17 2.20 2.84

unable to Groups

explain my

emotional

state to others.

Within 465.77 355.00 1.31

Groups

Total 502.19 356.00

EQ11: I am careful not

Between Groups

3.72 1.00 3.72 4.80 0.03 0.01 1.91 2.11

to reveal my

emotions to

others.

Within Groups

275.14 355.00 0.78

Total 278.86 356.00

EQ23: I am Between 8.33 1.00 8.33 7.80 0.01 0.02 2.24 2.55

reserved in Groups

expressing

emotions.

Within Groups

378.94 355.00 1.07 Total 387.27 356.00

EQ18: I believe that few people understand my emotions.

Between Groups

14.31 1.00 14.31 9.78 0.00 0.03 3.13 2.73

Within Groups

519.62 355.00 1.46 Total 533.93 356.00

Item Sum of df Mean f Sig. Eta Mean Mean

descriptions squares square squared WG Sotho

EQ43: I find it difficult to describe exactly what I feel in words.

Between Groups

6.14 1.00 6.14 4.14 0.04 0.01 3.59 3.32

Within Groups

526.99 355.00 1.48

Total 533.13 356.00

EQ4: Most Between 29.33 1.00 29.33 19.24 0.00 0.05 3.60 3.02

people cannot

Groups

understand

exactly what

I feel.

Within 541.16 355.00 1.52

Groups

Total 570.50 356.00

EQ26: I Between 37.91 1.00 37.91 34.16 0.00 0.09 2.12 2.78

rarely Groups

analyse my

emotions.

Within 394.02 355.00 1.11

Groups

Total 431.93 356.00

EQ9: I Between 0.98 1.00 0.98 0.79 0.37 0.00 2.34 2.44

usually keep Groups

to myself when I am

sad.

Within 439.81 355.00 1.24

Groups

Total 440.79 356.00

EQ30: I overlook my

Between Groups

24.59 1.00 24.59 22.18 0.00 0.06 2.19 2.72

emotions.

Within Groups

393.57 355.00 1.11 Total 418.16 356.00

Table 5 gives the ANOVA results for the mean item scores between and within groups for the factor “Expression and Recognition of Emotions‟.

Table 5 indicates that there are significant mean differences (p≤0.01) between the language groups on all the items, except for items 11, 23, 43, and 9 on the “Expression and Recognition of Emotions‟ factor. The effect sizes (Eta squared) varied between small and medium effect. The majority of the effect sizes (Eta squared) appeared to be of no significance.

These differences on an item level provide a sound basis for further DIF investigation.

Table 6

Summary of ANOVAs between cultural groups per item on the Use of Emotion to Facilitate Thinking

Item descriptions Sum of df Mean f Sig. Eta Mean Mean

squares square squared WG Sotho

Use of Emotion to Facilitate Thinking EQ42: I am usually

pessimistic about future

accomplishments.

Between Groups

1.91 1.00 1.91 1.85 0.18 0.01 2.10 2.24

Within Groups

365.94 355.00 1.03

Total 367.85 356.00 EQ3: I have the

tendency to focus on the negative side of things.

Between Groups

28.87 1.00 28.87 22.78 0.00 0.06 3.26 3.83

Within Groups

449.93 355.00 1.27 Total 478.80 356.00

EQ44: It’s Between 8.78 1.00 8.78 6.63 0.01 0.02 3.10 3.41 difficult for me to Groups

be optimistic.

Within Groups

469.76 355.00 1.32

Total 478.54 356.00

EQ22: I think of Between 24.16 1.00 24.16 4.62 0.03 0.01 3.85 4.37

the positive side Groups

of things.

Within 1854.84 355.00 5.23

Groups

Total 1879.00 356.00

EQ39: I always Between 65.85 1.00 65.85 48.07 0.00 0.12 2.84 3.71

try to see the Groups

good side of things.

Within 486.35 355.00 1.37

Groups

Total 552.20 356.00

EQ8: I deal with Between 75.12 1.00 75.12 48.21 0.00 0.12 2.81 3.73

my problems in a Groups

positive way by

trusting myself.

Within 553.19 355.00 1.56 Groups

Total 628.31 356.00

EQ40: I function Between 9.25 1.00 9.25 11.62 0.00 0.03 3.84 4.17

more based on the hope for

Groups

success and less

with the fear of

failure.

Within 282.57 355.00 0.80

Groups

Total 291.82 356.00

57

Item descriptions Sum of df Mean f Sig. Eta Mean Mean

Item descriptions Sum of df Mean f Sig. Eta Mean Mean

squares square squared WG Sotho

EQ12: Most of Between 13.98 1.00 13.98 11.61 0.00 0.03 3.65 4.05

the time I believe Groups

things will go well for me.

Within Groups

427.35 355.00 1.20 Total 441.33 356.00

EQ32: I feel Between 10.14 1.00 10.14 10.37 0.00 0.03 3.65 3.98

confident before Groups

important events

in my life.

Within 347.11 355.00 0.98

Groups

Total 357.25 356.00

EQ48: I trust my Between 52.28 1.00 52.28 49.81 0.00 0.12 3.00 3.77

abilities and I Groups

undertake the resolution of difficult situations.

Within 372.59 355.00 1.05

Groups

Total 424.87 356.00

EQ45: When faced with

Between Groups

89.10 1.00 89.10 58.26 0.00 0. 141 2.31 3.32

failure, I tend to

behave

energetically by

designing a new

plan of action.

Within 542.96 355.00 1.53

Groups

Total 632.06 356.00

EQ27: I adapt Between 39.78 1.00 39.78 31.09 0.00 0.08 2.83 3.50

easily by reacting Groups

creatively to any obstacles.

Within Groups

454.25 355.00 1.28 Total 494.03 356.00

Item descriptions Sum of df Mean f Sig. Eta Mean Mean

squares square squared WG Sotho

EQ29: I find Between 1.52 1.00 1.52 2.25 0.14 0.01 3.92 3.46

various Groups

alternative solutions to a problem.

Within 239.48 355.00 0.68

Groups

Total 241.00 356.00

EQ16: My worry Between 67.43 1.00 67.43 45.01 0.00 0.11 2.58 3.46

and stress do not Groups

decrease my ability to complete any task.

Within 531.82 355.00 1.50

Groups

Total 599.25 356.00

EQ25: Usually Between 87.97 1.00 87.97 53.57 0.00 0.13 3.92 2.92

my personal Groups

problems do not

affect my performance at work.

Within 582.96 355.00 1.64

Groups

Total 670.93 356.00

Table 6 shows the ANOVA results for the mean item scores between and within groups for the factor „Use of Emotions to Facilitate Thinking‟.

Table 6 indicates that there are significant mean differences (p≤0.01) between the language groups on all the items except for item 42, 22 and 29 on the “Use of Emotions to Facilitate Thinking‟ factor. The effect sizes (Eta squared) varied between small, medium and a large effect.

The majority of the effect sizes (Eta squared) appeared to be of no significance.

These differences on an item level provide a sound basis for further DIF investigation.

Table 7 indicates the uniform and non-uniform biasness of the 52 items of the GEIS. Zumbo (1999) suggested that for an item to be classified as DIF (i.e., a collective of uniform and non- uniform DIF), the 2-degree of freedom Chi-squares test conducted between steps 1 and 3 have to have p-values of less than or equal to 0.01. When referring to R² effect sizes, Jodoin and Gierl (2001) classify R² < 0.035 as negligible, 0.035 <R²<0.070 as moderate, and R² >0.07 as large effects sizes.

Table 7

Uniform and Non-uniform Item Bias of the items of the GEIS

Item Chi-

Square

DF Sig. Nagelkerke R² Δ

Chi- Square

DF Sig Nagelkerke R² Δ Caring and

Empathy

51 24.81 1 0.00 0.06 24.82 2 0.00 0.00

50 115.17 1 0.00 0.24 116.27 2 0.00 0.00

46 36.93 1 0.00 0.04 42.56 2 0.00 0.02

41 55.80 1 0.00 0.04 67.37 2 0.00 0.03

37 179.75 1 0.00 0.20 200.27 2 0.00 0.04

36 196.99 1 0.00 0.18 224.19 2 0.00 0.05

34 36.36 1 0.00 0.10 39.14 2 0.00 0.01

33 3.35 1 0.07 0.00 15.05 2 0.00 0.03

28 30.11 1 0.00 0.06 30.51 2 0.00 0.00

24 117.84 1 0.00 0.24 118.03 2 0.00 0.00

21 230.09 1 0.00 0.34 230.34 2 0.00 0.00

19 183.13 1 0.00 0.29 184.52 2 0.00 0.00

14 122.89 1 0.00 0.29 136.81 2 0.00 0.03

10 102.66 1 0.00 0.25 110.57 2 0.00 0.02

5 131.22 1 0.00 0.26 133.00 2 0.00 0.00

Control of Emotions

52 296.38 1 0.00 0.45 296.38 2 0.00 0.00

49 114.96 1 0.00 0.23 114.96 2 0.00 0.00

47 105.66 1 0.00 0.25 115.05 2 0.00 0.02

38 1.71 1 0.19 0.00 3.40 2 0.18 0.01

35 271.23 1 0.00 0.39 273.02 2 0.00 0.00

31 168.38 1 0.00 0.25 171.21 2 0.00 0.01

17 40.60 1 0.00 0.11 44.67 2 0.00 0.01

15 184.08 1 0.00 0.37 191.21 2 0.00 0.01

7 7.11 1 0.01 0.02 9.44 2 0.01 0.01

6 219.62 1 0.00 0.36 219.90 2 0.00 0.00

2 74.32 1 0.00 0.18 77.02 2 0.00 0.01

1 48.48 1 0.00 0.00 123.87 2 0.00 0.18

Dalam dokumen Declaration of originality of research (Halaman 53-91)

Dokumen terkait