reliability and factorial structure of the chinese version of...

16
Journal of Psychopathology and Behavioral Assessment, Vol. 10, No. 4, 1988 Reliability and Factorial Structure of the Chinese Version of the State-Trait Anxiety Inventory Daniel T. L. Shek ~ Accepted: May 25, 1988 The Chinese version of the State-Trait Anxiety Inventory (STAD was ad- ministered to 2150 Chinese secondary-school students. Reliability data re- vealed that the A-State and A-Trait scales had a high internal consistency, and high item-total correlations were found for most of the items under each scale. Factor analytic data showed that while two factors were abstracted from the A-Trait scale (Anxiety Present and Anxiety Absent), two (Anxiety Present and Anxiety AbsenO or three (Anxiety Present, Calmness, and Hap- piness) factors were abstracted from the A-State scale. By randomly split- ting the total sample into two subsamples, factors extracted from the first two factor-solutions couM be reproduced reliably and high coefficients of congruence were found. These findings generally suggest that the Chinese A-State and A- Trait scales possess acceptable psychometric properties and the factor analytic data tend to support Spielberger's conception of the mul- tidimensional nature of the A-State and A-Trait scales. KEY WORDS: anxiety measurement; state-trait anxiety; State-Trait Anxiety Inventory (STAI); factorial structure; reliability. INTRODUCTION The idea that anxiety is a multidimensional concept has been held by many researchers. Based on extensive factor analytic studies, Cattell and Scheir (1958, 1961) proposed two distinct anxiety factors, namely, state and trait This work was supported by UPGC Grant CPAS/714. tDepartment of Social Work, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong. 303 0882-2689/88/1200-0303506.00/0 1988PlenumPublishing Corporation

Upload: others

Post on 30-Jan-2021

4 views

Category:

Documents


0 download

TRANSCRIPT

  • Journal o f Psychopathology and Behavioral Assessment, Vol. 10, No. 4, 1988

    Reliability and Factorial Structure of the Chinese Version of the State-Trait Anxiety Inventory

    Danie l T. L. Shek ~

    Accepted: May 25, 1988

    The Chinese version o f the State-Trait Anxiety Inventory (STAD was ad- ministered to 2150 Chinese secondary-school students. Reliability data re- vealed that the A-State and A-Trait scales had a high internal consistency, and high item-total correlations were found for most o f the items under each scale. Factor analytic data showed that while two factors were abstracted f rom the A-Trait scale (Anxiety Present and Anxiety Absent), two (Anxiety Present and Anxiety AbsenO or three (Anxiety Present, Calmness, and Hap- piness) factors were abstracted from the A-State scale. By randomly split- ting the total sample into two subsamples, factors extracted from the first two factor-solutions couM be reproduced reliably and high coefficients o f congruence were found. These findings generally suggest that the Chinese A-State and A- Trait scales possess acceptable psychometric properties and the factor analytic data tend to support Spielberger's conception o f the mul- tidimensional nature o f the A-State and A-Trait scales.

    KEY WORDS: anxiety measurement; state-trait anxiety; State-Trait Anxiety Inventory (STAI); factorial structure; reliability.

    I N T R O D U C T I O N

    The idea that anxiety is a multidimensional concept has been held by many researchers. Based on extensive factor analytic studies, Cattell and Scheir (1958, 1961) proposed two distinct anxiety factors, namely, state and trait

    This work was supported by UPGC Grant CPAS/714. tDepartment of Social Work, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong.

    303

    0882-2689/88/1200-0303506.00/0 �9 1988 Plenum Publishing Corporation

  • 304 Shek

    anxiety. While trait or chronic anxiety was conceptualized as a relatively stable and permanent characteristic of people, state or acute anxiety was regarded as a transient condition which varied from day to day or moment to mo- ment. Such a distinction was subsequently shared and amplified by Spiel- berger (1966, 1972), who believed that this distinction applied to both the conceptual and the methodological levels. According to Spielberger, Gor- such, and Lushene (1970), state anxiety (A-State) was regarded as a temporary and fleeting emotional state of the organism and "it is characterized by sub- jective, consciously perceived feelings of tension and apprehension and height- ened autonomic nervous system activity" (p. 3). On the other hand, trait anxiety (A-Trait) was defined as the acquired behavioral disposition which "refers to relatively stable individual differences in anxiety proneness" (Spiel- berger et al., 1970, p. 3). The State-Trait Anxiety Inventory (STAI-Form X) was developed to measure these two distinct concepts. In its later develop- ment, a 20-item revised form (Form Y) was constructed (Spielberger, Vagg, Barker, Donham, & Westberry, 1980). The STAI has been widely employed as an instrument measuring anxiety in epidemiological studies (e.g., Knight, Waal-Manning, & Spears, 1983; Knight, Waal-Manning, & Godfrey, 1983) and in studies investigating the relationship between anxiety and other psy- chological constructs (e.g., Gotlib, 1984; Duckro, Margolis, & Tait, 1985; Ferreira & Murray, 1983). Certainly, the extent to which this instrument can be successfully used in clinical and community settings would depend much on the psychometric properties of the test.

    Data concerning the psychometric properties of the STAI have been reported and such data are roughly along three lines. The first line of data concerns the reliability status of the STAI. Apart from the data reported in the manual for the STAI, in which it was observed that both the A-State and the A-Trait scales had a high degree of reliability status (Spielberger et aL, 1970), other studies have generally established the reliability status of the STAI. For example, employing test-retest reliability as a measure while Nixon and Steffeck (1977) found that the reliability coefficient of the A-Trait scale ranged between .29 and .54, Metzger (1976) found that the STAI was reliable (A-State = .45 and A-Trait = .97). Further studies employing coeffi- cient alpha as a measure of reliability also tend to suggest that these two scales possess an acceptable reliability status (e.g., Gotlib, 1984; Endler, Mag- nusson, Ekehammar, & Okada, 1976; Martuza & Kallstrom, 1974; Metzger, 1976; Knight, Waal-Manning, & Spears, 1983; Ramanaiah, Franzen, & Schill, 1983). For instance, similar to Knight, Wall-Manning and Spears' (1983) find- ing that the STAI was reliable (alpha = .93 and .87 for A-State and A-Trai t, respectively), Ramanaiah et al. (1983) reported that the alpha values of these two scales were acceptable (A-State = .92 and A-Trait = .92). In addition to the reliability data based on the STAI (Form X), data on the reliability

  • Chinese STAI 305

    of STAI (Form Y)have also been accumulated (e.g., Gotlib, 1984; Ramanalah et al., 1983; Sherwood & Westerback, 1983).

    The second line of data is related to the validity of the STAI (e.g., En- dler et al., 1976; Martuza & Kallstrom, 1974; Knight, Waal-Manning, & God- frey, 1983; Tenenbaum, Furst, & Weingarten, 1985; Donham, Ludenia, Sands, & Holzer, 1984). Concerning the concurrent validity of the test, Spiel- berger et al. (1970) reported that the S-State and A-Trait scores correlated with several measures of personality and psychopathology. Apart from the data in the original validation study, results arising from several studies fur- ther showed that the STAI correlated significantly with other measures of psychological states and psychopathology. For example, Endler et al. (1976) reported that while A-Trait scores correlated with the various measures of the S-R Inventory of General Trait Anxiousness, A-State scores were found to correlate with the Behavioral Reactions Questionnaire. Similarly, while Knight, Manning, and Spears (1983) found that the correlations between the Zung Self-Rating Depression Inventory and the A-State and A-Trait scales were .54 and .70, respectively, Gotlib (1984) found that the A-State and A- Trait scales correlated with other measures of psychopathology. Based on the responses of alcoholics, Donham et al. (1984) further established the con- struct validity of the STAI.

    On the discriminant validity of the test, while Martuza and Kallstrom (1974) found that using the multitrait-multimethod procedure, the STAI was found to be able to measure A-Trait and A-State differences among gradu- ate students in education under different degrees of academic stress; Metz- ger (1976) showed that the STAI was able to differentiate those students under high versus nonstressful situations.

    The final line of data is associated with the factorial structure of the scales within the STAI. Apart from the fact that data in this area would have a direct bearing on the methodological soundness of this instrument, fac- torial data would also provide a test of the validity of the theoretical con- ceptualization underlying the STAI. However, a review of the literature shows that although many studies have been carried out to examine this issue, the existing picture is still unclear and the available data are conflicting.

    Based on the responses of student samples, Endler et al. (1976) found that while three factors could be extracted from the A-Trait scale for the male (Dissatisfaction Distress, Agitated Stress, and Emotive Discontent) and female (Emotive Discontent, Agitated Distress, and Tension Worry) students, three factors could also be extracted from the A-State scale for the male (Emo- tive Discontent, Distress Preoccupation, and Neurotic Depression) and fe- male (Emotive Discontent, Distress Preoccupation, and Avoidance Apathy) subjects. Similarly, while Loo (1979) found that four factors could be ex- tracted from the A-Trait scale, Knight, Waal-Manning, and Godfrey (1983)

  • 306 Shek

    concluded that two factors (positively phrased items and negative phrased items) were found for the A-State scale. When all the items in the A-Trait and A-State scales were subjected to a factor analysis, Barker and Barker (1977) found that three factors (State Anxiety, Generalized Anxiety, and Neu- roticism) could be extracted from the STAI.

    The factorial structure of the STAI under different instructional con- ditions has also been studied in other studies. For example, while Naylor (1978) found that three factors were extracted from the STAI under differ- ent manipulated conditions (A-Trait, A-State nonreversed, and A-State reversed), Naylor, Elsworth, and Astbury (1980) found similar factorial data for the A-State scale. Bartsch (1976) further found that two distinct factors (A-Trait and A-State) emerged from the STAI.

    A further review of the literature shows that data on the factorial struc- ture of the State-Trait Anxiety Inventory for Children (Hedl & Papay, 1982) and Form Y of the STAI have been accumulated (e.g., Spielberger et al., 1980; Vagg, Spielberger, & O'Hearn, 1980; Bernstein & Eveland, 1982). With special focus on Form Y, a generally consistent finding is that while two fac- tors were abstracted from the A-State scale (Anxiety Present and Anxiety Absent), two factors (Anxiety Present and Anxiety Absent) were similarly found for the A-Trait (Spielberger et al., 1980; Vagg et al., 1980; Bernstein & Eveland, 1982).

    The STAI has been translated from English into different languages, and in its different forms, the STAI has been used in different cultures (e.g., Mote, Natalicio, & Rivas, 1971; Sipos & Sipos, 1983), including the Chinese culture (Tsoi, Ho, & Mak, 1986). Utilizing the translated Chinese version of the STAI, Tsoi et al. (1986) found that in comparing different groups of pregnant women, those who did not have a normal child before scored higher in trait anxiety than the controls and those mothers who had already had a normal child.

    With respect to the psychometric properties of the Chinese version of the STAI, there are still several questions which remain to be answered. First, the reliability status of the A-State and A-Trait scales is in question. Although Tsoi et al. (1986) established the differentiation power of the A-Trait scale, there have been no data published on the reliability of these scales. Second, a review of the literature shows that there have been no data reported on the factorial structure of the Chinese A-Trait and A-State scales.

    The purpose of this paper was to report empirical evidence concerning the psychometric properties of the Chinese version of the STAI with two specific aims the mind. First, this study attempted to understand the relia- bility status of the Chinese STAI using secondary-school students as sub- jects. Second, since the factorial structure of the Chinese A-State and A-Trait scale is still unclear, this study attempted to clarify the factorial structure

  • Chinese STAI 307

    of these scales. The data reported in the paper were based on a study which was designed primarily to investigate student mental health, with the STAI as one of the measuring instruments. Since the data on student mental health and the psychometric properties of other measuring instruments have been discussed elsewhere (e.g., Shek, Lee, & Lam, 1987; Shek, 1987a,b,c; Shek, 1986; Shek, 1988a,b), the present paper focuses primarily on the psychometric properties of the Chinese version of the STAI.

    METHOD

    Instruments

    The Chinese version of the State-Trait Anxiety Inventory (STAI-- Form X) was employed (Tsoi et al., 1986). As indicated, the data reported were based on a study which was designed primarily to investigate student mental health. Thus, apart from the STAI, students were also requested to respond to other clinical scales measuring different areas of psychopathology. These include the General Health Questionnaire (Goldberg, 1972; Shek, 1987a; Shek, 1988b), Beck Depression Inventory (Beck, Ward, Mendelson, Mock, & Er- baugh, 1961; Chan & Tsoi, 1984), Leeds Scale (Snaith, Bridge, & Hamilton, 1976), Purpose in Life Questionnaire (Shek, 1986; Shek, Hong, & Cheung, 1987), Ego Strength Scale (Ko, 1976), and Somatic Scales (Wong & Chan, 1984; Shek & Mak, 1987). These scales have been found to possess adequate reliabilities and validities (Shek, Lee, & Lam, 1987; Shek, 1987c, 1988b) and most of them (including the General Health Questionnaire, Beck Depres- sion Inventory, Purpose in Life Questionnaire, Ego Strength Scale, and So- matic Scales) have been used in adolescent samples.

    Subjects and Procedures

    The study is based upon the responses of 2150 secondary-school stu- dents aged between 13 and 20 years. The students were selected from secon- dary schools in Hong Kong by the multiple-stage stratified random sampling method, with school types and school location as the stratifying factors (Moser & Kalton, 1980). The subjects can be considered heterogeneous, for they came from different areas and socioeconomic classes in Hong Kong. All subjects responded en masse to all the instrument scales in the question- naire. Adequate time was provided for the subjects to complete the ques- tionnaire.

  • 308 Shek

    RESULTS

    Item-total correlation statistics for the items in the A-Trait and A-State scales and the scale reliability data are presented in Table I. While overall reliability analysis using Cronbach's alpha shows that these scales are relia- ble (alpha = .90 and .81 for A-State and A-Trait scales, respectively), alter- native analyses of the data using the split-half reliability technique show that the test is internally consistent (A-State = .89 and A-Trait = .83). A review of the individual items showed that most of the items had high item-total correlations in the A-State (except item 18) and A-Trait (except item 11) scales. When these two items were removed from the reliability analyses, the alpha values were .90 for the A-State and .83 for the A-Trait scales.

    A principal-components analysis was performed on the item responses of the subjects to the A-State scale, yielding three factors with eigenvalues exceeding unity, accounting for 54% of the total variance. To avoid overfactor- ing, further analyses using the scree test (Gorsuch, 1974; Cattell, 1966) showed that two or three factors could be meaningfully extracted. The two-factor solution, which could be considered a relatively adequate representation of the data, was rotated to a varimax criterion for interpretation (Nie, 1983).

    Table I. Item-Total Correlations and Reliabil- ity Measures of the Chinese A-State and A-

    Trait Scales

    Item-total correlation

    1 .4157 .4048 2 .6148 .3575 3 .5525 .3315 4 .4518 .2710 5 .3320 .3124 6 .6453 .4085 7 .4690 .2917 8 .6053 .5299 9 .6346 .4751

    10 .6025 .4586 l l .4252 - .1610 12 .5848 .4664 13 .6267 .3774 14 .4664 .3879 15 .6132 .5269 16 .5065 .3703 17 .6158 .4667 18 .1782 .4554 19 .4923 .2361 20 .5461 .4562

    Cronbach's alpha .8971 .8065 Split-half .8881 .8259

    Item No. A-State A-Trait

  • Chinese STAI

    Table II. Two-Factor Solution of the Chinese A-State Scale for the Vari- ous Samples (Total Sample, Subsamples 1 and 2) ~

    309

    Factor Loadings

    Total Subsample 1 Subsample 2

    Item No. F1 F2 F1 F2 F1 F2 Communality

    1 .09 .57 .12 .56 .06 .59 .3393 2 .26 .68 .28 .66 .24 .70 .5331 3 .66 .20 .66 .20 .65 .20 .4756 4 .56 .14 .51 .15 .60 .13 .3286 5 .02 .48 .03 .48 .00 .49 .2353 6 .72 .27 .73 .24 .72 .30 .5993 7 .58 .15 .58 .14 .59 .16 .3636 8 .22 .71 .25 .71 .19 .72 .5579 9 .75 .21 .76 .21 .74 .21 .6115

    10 .22 .71 .21 .71 .23 .71 .5535 11 .12 .55 .09 .56 .15 .54 .3130 12 .76 .15 .76 .15 .75 .14 .5945 13 .75 .20 .75 .18 .75 .21 .6000 14 .63 .09 .64 .09 .63 .08 .4103 15 .25 .69 .23 .67 .26 .71 .5361 16 .11 .68 .10 .70 .12 .65 .4694 17 .76 .19 .76 .18 .76 .19 .6139 18 .51 - . 2 2 .49 - . 2 0 .52 - . 2 4 .3054 19 .04 .75 .01 .76 .07 .75 .5652 20 .10 .75 .10 .77 .11 .73 .5698

    Eigenvalue Factor 1 = 6.757 (33.8% of variance) Factor 2 = 2.818 (14.1% of variance) Factor 3 = 1.222 (6.1% of variance)

    "Communalities are based on the total sample, and the highest loading obtained by a variable among the factors is italicized.

    The first factor, labeled the Anxiety Present factor, included items 3 (tense), 4 (regretful), 6 (upset), 7 (worrying), 9 (anxious), 12 (nervous), 13 (jittery), 14 (high-strung), 17 (worried), and 18 (overexcited), explaining 33.8070 of the variance.

    Factor 2 explained 14.1 070 of the variance, which included items 1 (calm), 2 (secure), 5 (ease), 8 (rested), 10 (comfortable), 11 (self-confident), 15 (relaxed), 16 (content), 19 (joyful), and 20 (pleasant). Since items in this fac- tor are related to non-anxiety-arousing items, this factor was labeled the Anxi- ety Absent factor.

    To understand further the stability of the factors extracted, the total sample was randomly split into two subsamples (subsamples 1 and 2). Iden- tical factor analytic procedures were then carried out to assess the stability of the factor structures (Gorsuch, 1974; Harman, 1976). The results showed that the factors extracted from the two subsamples resembled those extract- ed from the total sample and high coefficients of congruence were also found

  • 310 Shek

    for the factors across samples. Further analysis on the reliability of the fac- tors showed that these two subscales were internally consistent (Cronbach's alpha = .88 and .87 for Factor 1 and Factor 2, respectively). While Table II shows the varimax-rotated two-factor structure of the A-State scale in the various samples, data on coefficients of congruence across samples are shown in Table III.

    An alternative factor analysis was performed by removing item 18 from the A-State scale. Based on identical factor analytic procedures, the data showed that Factor 1 (including items 3, 4, 6, 7, 9, 12, 13, 14, and 17) and Factor 2 (with items 1, 2, 5, 8, 10, 11, 15, 16, 19, and 20) could be replicated and both scales were found to be internally consistent (.89 and .87 for Fac- tors 1 and 2, respectively).

    Since a scree test showed that three factors could possibly be extract- ed, a three-factor solution for the A-State scale was carried out to under- stand further the factor structure of the scale. The data arising from the three-factor solution showed that while the first factor in the previous ana- lyses was replicated as the first factor (i.e., Anxiety Present), the Anxiety Absent factor was broken down into Factors 2 and 3 in this analysis. While the second factor was named the Calmness factor (including items 1, 2, 5, 8, 10, 11, and 15), the third factor was loaded on by items 16, 19, and 20, which could be named the Happiness factor. The data further revealed that although all the factors extracted were externally consistent (.88, .83, and .81 for Factors 1, 2, and 3, respectively), analysis of the stability of the fac- tors across samples (using procedures identical to those described above) showed that high coefficients of congruence were found for Factors 1 and 2 but not for Factor 3. The rotated varimax three-factor solutions in the var- ious samples are shown in Table IV.

    For the factor structure of the A-Trait scale, a principal-components analysis was performed on the item responses of the subjects, yielding three

    Table I lL Coefficients of Congruence Among the Factors Derived from the A-State and A-Trait Scales Based on the Responses from

    Different Samples

    Factor

    A-State (2)" A-State (3) b A-Trait

    1 2 1 2 3 1 2

    Total sample with subsample 1 .99 .99 .99 .66 .53 .99 .99

    Total sample with subsample 2 .99 .99 .99 .99 .98 .99 .99

    Subsample 1 with subsample 2 .99 .99 .99 .69 .45 .99 .99

    aTwo-factor solution based on A-State scale. bThree-factor solution based on A-State scale.

  • Tab

    le I

    V.

    Th

    ree-

    Fac

    tor

    So

    luti

    on

    of

    the

    Ch

    ines

    e A

    -Sta

    te S

    cale

    fo

    r th

    e V

    ario

    us

    Sam

    ple

    s (T

    otal

    Sam

    ple

    , S

    ub-

    sam

    ple

    s 1

    and

    2) a

    rJo

    Fac

    tor

    load

    ing

    To

    tal

    Su

    bsa

    mp

    le 1

    S

    ub

    sam

    ple

    2

    Item

    No.

    F1

    F

    2 F

    3 F1

    F

    2 F

    3 F1

    F

    2 F

    3 C

    om

    mu

    nal

    ity

    1 .0

    7 .7

    9

    .04

    .09

    .12

    .79

    .05

    .79

    .0

    2 .6

    261

    2 .2

    6 .7

    1 .2

    6 .2

    7 .3

    1 .6

    9 .2

    5 .7

    1 .2

    5 .6

    307

    3 .6

    7

    .08

    .17

    .67

    .1

    5 .1

    0 .6

    7

    .07

    .18

    .489

    6 4

    .56

    .0

    7 .1

    0 .5

    1 .0

    8 .1

    2 .6

    1 .0

    4 .l

    l .3

    329

    5 .0

    0 .6

    2

    .08

    .01

    .16

    .59

    -.0

    1

    .64

    .03

    .393

    5 6

    .74

    .1

    2 .2

    2 .7

    5 .2

    2 .0

    7 .7

    4 .1

    6 .2

    2 .6

    156

    7 .5

    9

    .11

    .08

    .58

    .0

    8 .0

    8 .6

    0

    .11

    .08

    .364

    3 8

    .22

    .71

    .30

    .25

    .40

    .65

    .19

    .74

    .25

    .645

    2 9

    .77

    .1

    0 .1

    5 .7

    7 .1

    5 .1

    0 .7

    6

    .10

    .16

    .621

    0 10

    .2

    4 .5

    1 .4

    8 .2

    4 .5

    8 .4

    0 .2

    5 .5

    7

    .42

    .554

    1 11

    .1

    3 .4

    1 .3

    5 .1

    1 .5

    1 .5

    2 .1

    5 .5

    0

    .24

    .315

    1 12

    .7

    6

    .15

    .03

    .76

    .06

    .14

    .75

    .15

    .00

    5950

    13

    .7

    5 .1

    6 .0

    9 .7

    5 .1

    1 .1

    5 .7

    6

    .17

    .09

    .600

    1 14

    .6

    3 .1

    0 -.

    01

    .6

    4

    .03

    .08

    .63

    .10

    -.3

    3

    .410

    8 15

    .2

    7 .4

    8 .4

    8 .2

    4 .4

    9 .4

    5 .2

    9 .5

    0

    .49

    .536

    1 16

    .1

    6 .2

    9 .6

    5

    .14

    .69

    .24

    .16

    .29

    .64

    .529

    0 17

    .7

    7

    .12

    .10

    .77

    .1

    4 .0

    8 .7

    7

    .14

    .08

    .617

    0 18

    .4

    6

    .14

    -.4

    6

    .45

    -.3

    9

    .17

    .48

    .10

    -.4

    9

    .44

    36

    19

    .1

    0 .2

    1 .8

    3 .0

    7 .8

    3 .1

    4 .1

    3 .2

    5 .8

    2 .7

    445

    20

    .17

    .22

    .81

    .15

    .81

    .18

    .23

    .23

    .82

    .7

    334

    aCo

    mm

    un

    alit

    ies

    are

    bas

    ed o

    n t

    he r

    esp

    on

    ses

    of

    the

    tota

    l sa

    mpl

    e, a

    nd

    the

    hig

    hest

    loa

    ding

    ob

    tain

    ed b

    y a

    var

    iabl

    e am

    on

    g t

    he f

    acto

    rs i

    s it

    alic

    ized

    .

  • 312 Shek

    factors with eigenvalues exceeding unity, accounting for 42.30/0 of the vari- ance. Analysis using the scree test showed that two factors could be meaning- fully extracted and the two-factor solution was rotated to a varimax criterion for interpretation. Factor 1, named Anxiety Present, included items 2 (tired quickly), 3 (crying), 4 (as happy as others), 5 (lose out on things), 8 (difficul- ties piled up), 9 (worrying), 12 (lacks self-confidence), 14 (avoids facing cri- sis), 15 (feeling blue), 17 (bothered by thoughts), 18 (disappointments), and 20 (tense), accounting for 22.9~ of the variance. These items are basically related to a person's evaluation of his or her chronic anxiety tendencies.

    Factor 2 explained 13.8% of the variance, which included items 1 (feel- ing pleasant), 6 (rested), 7 (calmness), 10 (happiness), 11 (take things hard), 13 (feeling secure), 16 (content), and 19 (a steady person). Since these items are related to calmness or anxiety-absent states (except item 11, which is a negatively phrased item), this factor was hence labeled the Anxiety Absent factor. Reliability analysis of the factors showed that these two factors are internally consistent (alpha = .81 and .69 for Factors 1 and 2, respectively) and high coefficients of congruence were found for these factors (see Table III). The rotated varimax two-factor solutions for the A-Trait scale in the various samples are shown in Table V.

    Since item 11 on the A-Trait scale loaded negatively on Factor 2 and it had a low item-total correlation, an alternative factor analysis was per- formed by removing this item from the scale. Based on identical factor ana- lytic procedures, the data showed that Factor 1 (including items 2, 3, 4, 5, 8, 9, 12, 14, 15, 17, 18, and 20) and Factor 2 (with items 1, 6, 7, 10, 13, 16, and 19) could be replicated and both scales were found to be internally consistent (.81 and .81 for Factors 1 and 2, respectively).

    DISCUSSION

    The results of the reliability analyses show that the Chinese version of the STAI is reliable (alpha -- .90 and .81 for A-State and A-Trait scales, respectively). In addition, the item-total correlation data show that most of the items under the respective scales are significantly correlated with the total. While 95% of the items on the A-State scale had item-total correlations of over .3, 85~ and 95o70 of the items on the A-Trait scale had values of over .3 and .2, respectively. The data generally suggest that while the items on the A-State scale are relatively homogeneous (except item 18), items on the A-Trait scales are measuring the same psychological construct (except item 11). From a cross-cultural perspective, these data compare favorably with the existing findings on the reliability status of the English version of the

  • Chinese STAI

    Table V. Two-Factor Solution of the Chinese A-Trait Scale for the Various Samples (Total Sample, Subsamples 1 and 2) a

    313

    Factor Loadings

    Total Subsample 1 Subsample 2

    Item No. F1 F2 F1 F2 F1 F2 Communality

    1 .16 .65 .12 .65 .18 .65 .4428 2 .45 .09 .47 .10 .44 .07 .2129 3 .45 .06 .42 .07 .48 .06 .2070 4 ,38 .02 .39 - . 0 0 .37 .06 .1460 5 .49 - . 0 4 .43 - . 0 4 .54 - . 0 4 .2423 6 .12 .72 .14 .74 .10 .71 ,5357 7 .06 .59 .07 .60 .05 .59 .3518 8 .63 .15 .63 .15 .63 .14 .4227 9 .63 .07 .62 .08 .63 .06 .3963

    10 .17 .74 .15 .76 .19 .71 .5780 11 .12 - . 5 0 .12 - . 4 8 .11 - . 51 .2606 12 .55 .18 .54 .10 .55 .27 .3287 13 .08 .71 .07 .73 .10 .69 .5105 14 .57 .04 .55 - . 0 0 .58 .10 .3218 15 .64 .13 .60 .18 .68 .10 .4235 16 .12 .64 .09 .66 .16 .62 .4240 17 .64 .04 .62 .02 .66 .06 .4097 18 .63 .02 .61 .03 .64 .01 .3970 19 .02 .57 .03 .58 .00 .56 .3270 20 .64 .02 .65 .02 .63 .01 .4100

    Eigenvalue Factor 1 = 4.590 (22.9% of variance) Factor 2 = 2.754 (13.8% of variance) Factor 3 = 1.119 (5.6% of variance)

    aCommunalities are based on the total sample, and the highest loading obtained by a variable among the factors is italicized.

    STAI (e.g., Gotlib, 1984; Endler et al., 1976; Ramanaiah et al., 1983; Knight, Waal-Manning, & Godfrey, 1983; Martuza & Kallstrom, 1974).

    For the factor structure of the A-State scale, the data showed that either two or three factors could be extracted. Although all the subscales under these analyses were found to be reliable, analyses based on the coefficient of congruence showed that the factors extracted under the two-factor solu- tion were more stable across samples. For the three-factor solution, while the first two factors were found to possess a high degree of stability, the third factor was observed to be relatively unstable across samples. In some respects, Calmness and Happiness (second and third factors under the three-factor solution) might be regarded as subordinate factors which could be incorpo- rated under the Anxiety Absent factor (second factor under the two-factor solution), and there in fact has been evidence suggesting that Chinese people distinguished among happiness, calmness, and contentedness (Tsoi & Tam, 1983). However, since the third factor was found to be relatively unstable,

  • 314 Shek

    it could be parsimoniously concluded that two factors could be abstracted from the Chinese A-State scale. Perhaps future studies can further elucidate the relationship among the dimensions underlying the anxiety-absent, calm- ness, and happiness dimensions of the A-State scale.

    These findings are generally consistent with some of the data reported in previous factor analytic studies in which Anxiety Present and Anxiety Ab- sent were identified as two dimensions underlying the A-State scale (e.g., Bernstein & Eveland, 1982; Knight, Waal-Manning, & Godfrey, 1983; Nay- lor, 1978). The data obtained are also consistent with Spielberger and co- workers' (1980) conception that the A-State scale is composed of State § and Sta te- dimensions.

    For the A-Trait scale, the factor analytic data showed that the two fac- tors extracted had high coefficients of congruence and the two subscales are reliable (except that item 11 loaded negatively on Factor 2 and had a low item-total correlation). One possible interpretation of why item 11 loaded negatively on Factor 2 is related to cultural difference in the comprehension of this item. The content of this item is "I am inclined to take things hard." If an individual gives an affirmative response to this item in the Western culture, this could be interpreted as an indication of "excessive" anxiety. However, based on the fact that Chinese people tend to attach a high value to fulfilling role obligations and they are more "tough-minded" (see, e.g., Yang, 1981), "inclined to take things hard" might be interpreted by Chinese people as an indication of responsible behavior, thus leading to a state of calmness (or at least a non-anxiety-arousing state). By removing item 11, both factors could be replicated and they were loaded on by exactly the same items as in the previous analysis.

    The finding that the A-Trait scale was composed of two factors (i.e., Anxiety Present and Anxiety Absent) is consistent with some of the previ- ous findings (e.g., Bernstein & Eveland, 1982; Spielberger et al., 1980; Sher- wood & Waterback, 1983) and it is generally in line with Spielberger and co-workers' (1980) distinction between Anxiety + and Anxiety- items on the A-Trait scale.

    In summary, the data arising from this study previously give insights into the reliability status and factor structure of the Chinese A-State and A- Trait scales, which have not been elucidated in previous studies. In conjunc- tion with the validity data (e.g., Shek, Lee, & Lain, 1987; Shek, 1987b), the Chinese version of the STAI can be regarded as possessing acceptable psy- chometric properties and it can be used as an objective assessment tool meas- uring anxiety in the Chinese culture, at least for adolescent samples.

    A C K N O W L E D G M E N T S

    The author wishes to thank S. T. Kwan, Lorine Lam, and Elbert Lee for thier assistance in the data collection process.

  • Chinese STAI 315

    R E F E R E N C E S

    Barker, B. M., & Barker, H. R. (1977). Factor analysis of the item in the State-Trait Anxiety Inventory. Journal of Clinical Psychology, 33(3), 450-455.

    Bartsch, T. W. (1976). A manipulative study of the effect of instructional set on the convergent and discriminant validity of questionnaire measures of trait and state anxiety: A com- parative factor analytic approach. Educational and Psychological Measurement, 36, 885-898.

    Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 12, 57-62.

    Bernstein, I. H., & Eveland, D. C. (1982). State vs trait anxiety: A case study in confirmatory factor analysis. Personality and Individual Differences, 3, 361-372.

    Cattelt, R. B. (1966). The scree test for the number of factors. Multivariate Behavioral Research, t(2), 245.

    Cattell, R. B., & Scheier, I. H. (1958). The nature of anxiety: A review of 13 multivariate ana- lyses comparing 814 variables. Psychological Reports: Monograph Supplement, 5, 351-388.

    Cattell, R. B., & Scheier, I. H. (1961). The meaning and measurement ofneuroticism and anxi- ety. New York: Ronaid Press.

    Chan, C. M., & Tsoi, M. M. (1984). The BDI and stimulus determinants of cognitive-related depression among Chinese college students. Cognitive Therapy and Research, 8(5), 501-508.

    Donham, G. W., Ludenia, K., Sands, M. M., & Holzer, P. D. (1984). Cross-validation of the State-Trait Anxiety Inventory with an alcoholic population. Journal of Clinical Psychol- ogy, 40(2), 629-631.

    Duckro, P. N., Margolis, R. B., & Tait, R. C. (1985). Psychological assessment in chronic pain. Journal of Clinical Psychology, 41(4), 499-504.

    Endler, N. S., Magnusson, D., Ekehammar, B., & Okada, M. (1976). The multidimensionality of state and trait anxiety. Scandinavian Journal of Psychology, 17, 81-96.

    Ferreira, R., & Murray, J. (1983). Spielberger's State-Trait Anxiety Inventory: Measuring anxiety with and withou~ an audience during performance on a stabilometer. Perceptual and Motor Skills, 57(1), 15-18.

    Goldberg, D. P. (1972). The detection of psychiatric illness by questionnaire. Oxford: Oxford University Press.

    Gorsuch, R. L. (1974). Factor analysis. Philadelphia: W. B. Saunders. Gotlib, I. H. (1984). Depression and general psychopathology in university students. Journal

    of Abnormal Psychology, 93(1), 19-30. Harman, H. H. (1976). Modern factor analysis. Chicago: University of Chicago Press. Hedl, J. J., & Papay, J. P. (1982). The factor structure of State-Anxiety Inventory for chil-

    dren: Kindergarten through the fourth grades. Personality and Individual Differences, 3, 439-446.

    Knight, R. G., Waal-Manning, H. J., & Godfrey, H. P. D. (1983). The relationship between state anxiety and depressed mood: A validity study. Journal of Behavioral Assessment, 5(3), 191-201.

    Knight, R. G., Waal-Manning, H. J., & Spears, G. F. (1983). Some norms and reliability data for the State-Trait Anxiety Inventory and the Zung Self-Rating Depression Scale. Brit- ish Journal of Clinical Psychology, 22, 245-249.

    I(o, Y. H. (1976). The mental health of junior and senior high school boys. Acta Psychologica Talwanica, 18, 37-52.

    Logan, P., & Loo, R. (1979). Source of covariation between the State and Trait scales of ~the State-Trait Anxiety Inventory. Journal of Psychology, 103, 3-5.

    Loo, R. (1979). The State-Trait Anxiety A-Trait scale: Dimensions and generalization. Journal of Personality Assessment, 43(1), 50-53.

    Martuza, V. R., & Kallstrom, P. W. (1974). Validity of the State-Trait Anxiety Inventory in an academic setting. Psychological Reports, 35, 363-366.

    Metzger, R. L. (1976). A reliability and validity study of the State-Trait Anxiety Inventory. Journal of Clinical Psychology, 32(2), 276-278.

    Maser, C. A., & Kalton, G. (1980). Survey methods in social investigation. London: Heinemann Educational Books.

  • 316 Shek

    Mote, T. A., Natalicio, L. F. S., & Rivas, F. (1971). Comparability of the Spanish and English editions of the Spielberger State-Trait Anxiety Inventory. Journal of Cross-CulturalPsy- chology, 2(2), 205-206.

    Naylor, F. D. (1978). Success and failure experiences and the factor structure of the State-Trait Anxiety Inventory. Australian Journal of Psychology, 30(3), 217-222.

    Naylor, F. D., Elsworth, G. R., & Astbury, J. A. (1980). Interactions of factorial components and testing occasions in the determination of scores on the A-State scale of the State- Trait Anxiety Inventory. Australian Journal of Psychology, 32(3), 217-223.

    Nie, N. H. (1983). SPSS-X user's guide. New York: McGraw-Hill. Nixon, G. F., & Steffeck, J. C. (1977). Reliability of the State-Trait Anxiety Inventory. Psy-

    chological Reports, 40, 357-358. Ramanaiah, N. V., Franzen, M., & Schill, T. (1983). A psychometric study of the State-Trait

    Anxiety Inventory. Journal of Personality Assessment, 47, 531-535. Shek, D. T. L. (1986). The purpose in life questionnaire in a Chinese context: Some psycho-

    metric and normative data. Chinese Journal of Psychology, 28(1), 51-60. Shek, D. T. L. (1987a). Mental health of secondary school students in Hong Kong: An epidemi-

    ological study using the General Health Questionnaire. Paper presented at the Work- shop on Current Research on Youth in Hong Kong, Social Science Research Centre, The Chinese University of Hong Kong.

    Shek, D. T. L. (1987b). A study on secondary school students" mental health using the General Health Questionnaire and the Purpose in Life Questionnaire. Paper presented at the Cairo World Congress on Mental Health, World Federation on Mental Health, Cairo.

    Shek, D. T. L. (1987c). Reliability and factorial structure of the Chinese version of the General Health Questionnaire. Journal of Clinical Psychology, 43(6), 683-691.

    Shek, D. T. L. (1988a). Reliability and factorial structure of the Chinese version of the Pur- pose in Life Questionnaire. Journal of Clinical Psychology, 44(3), 384-392.

    Shek, D. T. L. (1988b). Validity of the Chinese version of the General Health Questionnaire. Journal of Clinical Psychology (in press).

    Shek, D. T. L., & Mak, J. W. K. (1987). Psychological well-being of working parents in Hong Kong. Hong Kong: Hong Kong Christian Service.

    Shek, D. T. L., Hong, E. W., & Cheung, M. Y. P. (1987). The Purpose in Life Questionnaire in the Chinese context. Journal of Psychology, 12(1), 77-83.

    Shek, D. T. L., Lee, E. S. P., Lam, L. K. H. (1987). Adolescent mental health: An epidemiological study in Hong Kong. Paper presented at the 4th International Symposium on Adolescent Health, Australian Association for Adolescent Health, Sydney.

    Sherwood, R. D., & Westerback, M. E. (1983). A factor analytic study of the State-Trait Anxi- ety Inventory utilized with pre-service elementary teachers. Journal of Research in Science Teaching, 20(2), 225-229.

    Sipos, K., & Sipos, M. (1983). The development and validation of the Hungarian Form of the State-Trait Anxiety Inventory. Series in Clinical and Community Psychology (Vol. 2, pp. 27-39).

    Snaith, R. P., Bridge, G. W. K., & Hamilton, M. (1976). The Leeds Scale for self-assessment of anxiety and depression. British Journal of Psychiatry, 128, 156-165.

    Spielberger, C. D. (1966). Theory and research on anxiety. In C. D. Spielberger (Ed.), Anxiety and behavior. New York: Academic Press.

    Spielberger, C. D. (1972). Anxiety as an emotional state. In C. D. Spielberger (Ed.), Anxiety: Current trends in theory and research. New York: Academic Press.

    Spielberger, C. D., Gorsuch, R. C., & Lushene, R. F. (1970). Manual for the State-Trait Anxi- ety Inventory. Palo Alto, CA: Consulting Psychologists Press.

    Spielberger, C. D., Vagg, P. R., Barker, L. R., Donham, G. W., & Westberry, L. G. (1980). The factor structure of the State-Trait Anxiety Inventory. In I. G. Sarason & C. D. Spiel- berger (Eds.), Stress and anxiety (Vol. 7). Washington, DC: Hemisphere.

    Tenenbaum, G., Furst, D., & Weingarten, G. (1985). A statistical reevaluation of the STAI Anxiety Questionnaire. Journal of Clinical Psychology, 41(2), 239-244.

    Tsoi, M. M., & Tam, W. Y. K. (1983). A Chinese visual analogue mood scale for subjective feelings. Acta Psychologica Taiwanica, 25(2), 67-74.

  • Chinese STAI 317

    Tsoi, M. M., Ho, E., & Mak, K. C. (1986). Becoming pregnant again after stillbirth or the birth of a handicapped child. In L. Dennerstein & I. Fraser (Eds.), Hormones and be- havior. Holland: Elsevier.

    Vagg, P. R., Spielberger, C. D., & O'Hearn, T. P. (1980). Is the State-Trait Anxiety Inventory multidimensional? Personality and Individual Differences, 1, 207-214.

    Wong, C. K., & Chan, T. S. C. (1984). Somatic symptoms among Chinese psychiatric patients. Hong Kong Journal of Mental Health, 13, 5-21.

    Yang, K. S. (1981). The formation and change of Chinese personality: A cultural-ecological perspective. ,4cta Psychologica Taiwanica, 23, 39~55.

  • 本文献由“学霸图书馆-文献云下载”收集自网络,仅供学习交流使用。

    学霸图书馆(www.xuebalib.com)是一个“整合众多图书馆数据库资源,

    提供一站式文献检索和下载服务”的24 小时在线不限IP

    图书馆。

    图书馆致力于便利、促进学习与科研,提供最强文献下载服务。

    图书馆导航:

    图书馆首页 文献云下载 图书馆入口 外文数据库大全 疑难文献辅助工具

    http://www.xuebalib.com/cloud/http://www.xuebalib.com/http://www.xuebalib.com/cloud/http://www.xuebalib.com/http://www.xuebalib.com/vip.htmlhttp://www.xuebalib.com/db.phphttp://www.xuebalib.com/zixun/2014-08-15/44.htmlhttp://www.xuebalib.com/