Reliability, validity and structure of the Adolescent Decision Making Questionnaire among adolescents in The Netherlands

Personality and Individual Di€erences 28 (2000) 273±285

www.elsevier.com/locate/paid

Reliability, validity and structure of the Adolescent Decision Making Questionnaire among adolescents in The Netherlands J. Tuinstra*, F.L.P. van Sonderen, J.W. Grootho€, W.J.A. van den Heuvel, D. Post Department of Health Sciences, Northern Centre for Healthcare Research, University of Groningen, 9713 AV, Groningen, The Netherlands Received 2 September 1998

Abstract The reliability, validity and structure of the 30 item Adolescent Decision Making Questionnaire were investigated in a sample consisting of 1642 Dutch adolescents. The original structure of ®ve subscales was not found to be sucient in our data. Therefore a revised ADMQ was proposed, with four subscales and 22 items. A cross validation study proved that this revised ADMQ was improved for its structure and reliability. Also, the revised ADMQ proved to be a convergent and discriminative valid instrument indicated by associations with peer group pressure and di€erences for sex and school level. # 1999 Elsevier Science Ltd. All rights reserved.

1. Introduction Adolescence is a crucial stage in the development of decision making competence. Where children and most young adolescents are incapable of creating options that foresee the consequences of alternatives, evidence suggests that by the age of 15 years many adolescents show a reliable level of competence in decision making (Mann, Harmoni & Power, 1989). Adolescents have, despite this general competence, many di€erent patterns of decision making. These patterns range, comparably with coping styles, from e€ective, functional, or competent * Corresponding author. Tel.: +31-503-632849; fax: +31-503-633082. E-mail address: [email protected] (J. Tuinstra) 0191-8869/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 1 9 1 - 8 8 6 9 ( 9 9 ) 0 0 0 9 6 - 3

274

J. Tuinstra et al. / Personality and Individual Di€erences 28 (2000) 273±285

to ine€ective, dysfunctional, or incompetent (OlaÂh, 1995; Plancherel & Bolognini, 1995; SchoÈnp¯ug & Jansen, 1995; HaÈnninen & Aro, 1996). Mann et al. (1989) developed the Adolescent Decision Making Questionnaire (ADMQ) to measure decision making patterns in adolescence. The authors modi®ed the Flinders Decision Making Questionnaire (DMQ, Mann, 1982) suitable for adolescents. Both the DMQ and the ADMQ are based on the theory of Janis and Mann (1977) concerning decision making styles and self con®dence. Their theory is re¯ected in the ®ve subscales of the ADMQ, namely selfcon®dence, vigilance, panic, evasiveness and complacency. Janis and Mann (1977) classify these ®ve subscales into two categories, `adaptive' and `maladaptive' decision making patterns. An adaptive pattern is carefully deliberated behaviour, such as vigilant and self-con®dent decision making. A maladaptive pattern, on the other hand, fails to meet many of the requirements of high quality information processing. Panic, evasiveness and complacency are maladaptive decision making patterns. The ADMQ has been used in several studies. Ormond et al. (1991) found that middle adolescents (mean 15.6 years old) obtained better scores on each scale of the ADMQ than early adolescents (mean 12.9 years old). Friedman and Mann (1993) compared decision making patterns between Israeli and Australian adolescents. They concluded that the fundamental structure of decision-coping patterns is essentially similar among cultures. Both studies, however, hardly report on psychometric properties of the ADMQ. The only properties mentioned in this respect are the reliability coecients of the subscales. This absence of information is one of the reasons to examine the factorial validity of the ADMQ. Another reason for the examination is provided by a paper of Mann et al. (1997). This paper explored the factorial validity of the original Flinders Decision Making Questionnaire (Mann, 1982) and recommends a revised version of the instrument. The ADMQ has never been used before in the Netherlands and has thus never been validated for Dutch adolescents. In this study we examined three questions concerning the ADMQ. The ®rst question is whether the structure of ®ve subscales of the ADMQ can be found for a population of Dutch adolescents. Second, we evaluated the structure and reliability of a revised version of the ADMQ by means of a cross validation study. Third, we examined the convergent and discriminative validity of the revised ADMQ. For the convergent validity, we used the concept peer group pressure. According to Mann et al. (1989), conformity to peer group pressure can be an important constraint to competent decision-making. Therefore, we hypothesized a positive correlation between peer group pressure on the one hand and the `maladaptive' subscales panic, evasiveness and complacency on the other. No correlation is hypothesized for the `adaptive' decision making styles selfcon®dence and vigilance with peer group pressure. To investigate the discriminative validity of the ADMQ, we examined whether the ADMQ discriminated between groups of individuals by sex and educational level. Earlier studies on the ADMQ concluded that male adolescents outscored females on self-con®dence and on complacency, males scored lower than females on panic (Ormond et al., 1991; Friedman & Mann, 1993). Other studies (Poikolainen, Kanerva & LoÈnnqvist 1995; Schuyt, 1995; Wills, McNamara & Donato, 1995) suggest that adolescents in the higher educational levels are more competent in decision making. Therefore, we hypothesized that the adolescents at the highest

J. Tuinstra et al. / Personality and Individual Di€erences 28 (2000) 273±285

275

Table 1 Demographic characteristics of the study population

N Sex %Male %Female Age Mean SD Educational levela VBO MAVO HAVO VWO

Total response group

Total ADMQ group

`Experimental' group

`Control'

1984

1642

821

821

51.6 48.4

49.8 50.2

52.1 47.9

47.4 52.6

16.2 0.79

16.2 0.79

16.2 0.78

16.2 0.79

23.8 26.6 27.5 22.1

22.0 27.2 27.0 23.8

22.0 27.3 27.0 23.6

21.9 27.0 27.0 24.0

a

Vocational education, 4 years; MAVO, lower general secondary education, 4 years; HAVO, higher general secondary education, 5 years; VWO, pre-university education, 6 years.

educational levels will outscore on self-con®dence and vigilance and score lower on panic, evasiveness and complacency.

2. Method 2.1. Study population The ADMQ1 was included in a self report questionnaire of a survey among adolescents on social inequality and health. Data were collected in 1994 and 1995 in cooperation with four Public Health Services in four provinces in the north of The Netherlands. All four GGDs drew a random sample out of the secondary education schools in their region. We restricted our study to those adolescents who attend regular education. For this population, a representative sample was realised. Eventually, 1984 of the total sample (N = 2090) of the fourth grade scholars completed a questionnaire. Teachers told us that absence of pupils (non-response of 5%) was mainly due to sick leave, playing truant and other activities, such as school committees. The respondents completed the questionnaire in school, in their classroom, under the guidance of a ®eld worker. In the analyses of this paper we con®ned ourselves to the respondents who entirely completed the ADMQ, N = 1642. This was 83% of the total research group. For the cross validation study, the group of 1642 fourth grade scholars was divided into two groups, an `experimental' group and a `control' group. A random split into two groups however could generate yeargroup e€ects. Therefore, an equal division per yeargroup was 1

See Appendix for the items and scales of the ADMQ.

276

J. Tuinstra et al. / Personality and Individual Di€erences 28 (2000) 273±285

made for the experimental and the control group. Table 1 shows some demographic characteristics of our study population. As can be seen in Table 1, all groups are similar for the distribution of sex, age and educational level. 2.2. Instruments The Adolescent Decision Making Questionnaire is a self-report questionnaire consisting of 30 items. Respondents could mark one of the four responses to each item: `always' (scored 1), `often' (scored 2) `sometimes' (scored 3), `never' (scored 4). All scores were reversed, so that a high score represents a high level of the respective subscale. There were three exceptional items for the subscale self-con®dence which were not reversed, because of their negative formulation. The ADMQ2 was translated into Dutch using the prescribed procedure for this purpose, with native speakers and both forward and backward translations. Social pressure from the peer group was measured by ®ve items of van der Linden and Dijkman's scales (1989). The items contain reasons for joining the peer group. For each of the ®ve reasons the respondent is asked whether he `never', `sometimes', `regularly' or `often' joins the peer group. An example of an item is: ``I want to be just like the others''. A mean sumscore was computed for these items (range 1±4). A high score represents a high level of pressure from the peer group. Educational level was measured by one single item about the type of school. The Dutch secondary educational system is very clearly organised. Pupils know which of the four regular types of school they attend (VBO, MAVO, HAVO or VWO). 2.3. Analytical procedure The entire analytical procedure can be divided into three steps. The ®rst step was the examination of the psychometric properties of the original ADMQ in the experimental group by the internal consistency of the subscales (Cronbach's a and interitem correlations), a Simultaneous Component Analysis (SCA), a Principal Component Analysis (PCA) and a scree plot (Cattell, 1966). The SCA tested whether the ®ve subscale structure of the ADMQ explains enough variance of possible maximal explained variance. In addition, the SCA tested whether the items load highest on their intended component. The PCA of the ADMQ was examined in a three-stage iterative process by judging the content of the components, the content of the items, the item loadings on the components (>0.40) (Stevens, 1992) and the internal consistency of the components. Finally, a revised structure of the subscales of the ADMQ was proposed. The second step was the investigation of the revised structure of the ADMQ in the control group on their psychometric properties. Both the structure of the components and the psychometric properties such as Cronbach's a and inter-item correlations should be similar in the experimental and the control group. In addition an SCA was conducted in the control 2

Taken from the paper of Friedman and Mann (1993).

J. Tuinstra et al. / Personality and Individual Di€erences 28 (2000) 273±285

277

group to test the structure of the revised ADMQ. This second step concerns a cross-validation (Stevens, 1992) and is a well-used method in studies of factorial validity (e.g. Mann et al., 1997). The third and ®nal step in the analytical procedure was the examination of convergent validity and discriminative validity of the revised ADMQ in the total group. As already mentioned in the introduction, the convergent validity was investigated by examining the associations between the four revised ADMQ subscales and social pressure from the peer group. The discriminative validity was investigated by examining di€erences in scores of the revised ADMQ between groups of individuals by sex and educational level.

3. Results 3.1. Subscale structure in the experimental group The ®rst step of the analysis in the experimental group was the examination of the internal consistency of the ®ve subscales of the ADMQ, self-con®dence, vigilance, panic, evasiveness and complacency (Friedman & Mann, 1993). Table 2 shows for each subscale both the Cronbach's a and the mean inter-item correlation. The Cronbach's a of the ®ve subscales is rather low (Nunally, 1978). Causes for these low as may be either the few number of items of the subscales and/or the inter-item correlations. Further inspection of the mean inter-item correlations shows that some of them are also relatively low. Two of the ®ve subscales have a mean inter-item correlation below 0.20. The ®rst column of Table 3 presents the unrotated loadings on the ®rst factor. As can be seen, the instrument represents one common underlying component. Although there are some low loadings, these are not from a particular subscale. This means that no particular subscale needs to be excluded a priori. The results of the PCA (also Table 3) show that the ®ve subscales are not suciently recognised in our data. Twelve of the 30 items do not load on their intended component. Especially items of the subscales vigilance and complacency often mis®t on their intended component. This notion of insucient structure is supported by an SCA. The original ®ve subscale structure of the ADMQ explains 36.1% of the variance, while 39.2% explained Table 2 Internal consistency (Cronbach's a ) and mean inter-item correlation per ADMQ subscale (experimental group N = 821)

Self-con®dence Vigilance Panic Evasiveness, cop out Complacency

Cronbach's a

Mean inter-item correlation

No. of items

0.63 0.55 0.64 0.65 0.59

0.22 0.17 0.31 0.24 0.15

6 6 4 6 8

278

J. Tuinstra et al. / Personality and Individual Di€erences 28 (2000) 273±285

Table 3 Principal component analysis of the ADMQ in the experimental group; Varimax rotated on ®ve factors (N = 821)a

SELFC1 SELFC2 SELFC3 SELFC4 SELFC5 SELFC6 VIGIL1 VIGIL2 VIGIL3 VIGIL4 VIGIL5 VIGIL6 PANIC1 PANIC2 PANIC3 PANIC4 EVASIV1 EVASIV2 EVASIV3 EVASIV4 EVASIV5 EVASIV6 COMPLA1 COMPLA2 COMPLA3 COMPLA4 COMPLA5 COMPLA6 COMPLA7 COMPLA8 Eigen value %Var. Cum% Var. a b

F1b

Component 1

Component 2

Component 3

Component 4

Component 5

ÿ0.54 ÿ0.45 ÿ0.43 ÿ0.47 ÿ0.38 ÿ0.49 ÿ0.27 ÿ0.16 ÿ0.22 ÿ0.40 ÿ0.55 ÿ0.21 0.50 0.39 0.49 0.44 0.41 0.41 0.56 0.60 0.32 0.50 0.53 0.42 0.00 0.29 0.13 0.30 0.24 0.50

0.60 0.48 0.23 0.16 0.61 0.73 0.23 0.26 0.18 0.56 0.63 0.30 ÿ0.16 ÿ0.02 ÿ0.20 ÿ0.18 0.04 0.01 ÿ0.12 ÿ0.18 0.07 ÿ0.09 ÿ0.19 ÿ0.12 0.26 ÿ0.07 0.10 ÿ0.00 ÿ0.06 ÿ0.18 5.10 17.0 17.0

ÿ0.22 ÿ0.25 ÿ0.28 ÿ0.17 0.04 ÿ0.10 ÿ0.06 0.06 0.01 0.02 ÿ0.35 ÿ0.07 0.23 0.09 0.15 0.12 0.51 0.41 0.67 0.72 0.16 0.64 0.51 ÿ0.01 0.09 0.05 0.05 0.24 0.03 0.40 2.22 7.4 24.4

ÿ0.19 ÿ0.07 ÿ0.20 ÿ0.46 ÿ0.12 ÿ0.07 0.01 ÿ0.13 0.02 ÿ0.12 0.00 ÿ0.10 0.59 0.68 0.53 0.52 0.21 0.33 0.13 0.12 0.44 0.11 0.14 0.57 ÿ0.07 0.17 0.18 0.02 0.12 0.15 2.12 7.1 31.4

0.05 0.08 ÿ0.03 ÿ0.10 ÿ0.13 ÿ0.04 ÿ0.70 ÿ0.17 ÿ0.74 ÿ0.15 ÿ0.01 ÿ0.09 ÿ0.19 ÿ0.15 0.02 ÿ0.14 0.16 0.08 0.11 ÿ0.03 0.11 0.04 0.02 0.21 0.51 0.39 ÿ0.00 0.60 0.21 0.09 1.37 4.6 36.0

0.07 ÿ0.14 ÿ0.15 ÿ0.12 0.17 0.15 0.15 0.32 0.08 ÿ0.07 0.02 0.45 ÿ0.04 ÿ0.08 0.09 0.14 ÿ0.12 ÿ0.20 0.06 0.10 0.07 0.03 0.33 0.13 0.15 0.39 0.44 0.22 0.62 0.42 1.15 3.8 39.9

Component, 1, self con®dence; 2, evasiveness; 3, panic; 4, vigilance; 5, complacency (Friedman & Mann, 1993). Unrotated factor loadings on ®rst factor.

variance is possible. This discrepancy of 3.1% is too high for a sucient structure (personal communication, Dr Kiers). Looking at Table 3, we see that the ®rst component represents a self-con®dent way of decision making. Two of the six items of self-con®dence (SELFC3, SELFC4), however, do not load on this self-con®dent component, whereas two other items of vigilance do. Looking at the content of these two items VIGIL4 and VIGIL5, it is not surprising that they attach to the component self-con®dence. Also a third item of vigilance, VIGIL6, belongs to the ®rst component both for content and loading. The second component represents an evasive way of

J. Tuinstra et al. / Personality and Individual Di€erences 28 (2000) 273±285

279

decision making. Other people are important in decision making. The respondent avoids deciding for him-, or herself. Five of the six items of evasiveness load on this component. Two items of complacency also load on the evasive component, COMPLA1 and COMPLA8. These two items stress the importance of other people in decision making. The third component is panic. All four items of the subscale panic load high on this component, added by SELFC4, EVASIV5 and COMPLA2. The fourth and the ®fth component are less clear to identify in Table 4 Principal component analysis of the ADMQ in the experimental and control group; Varimax rotated on four factors, on 22 items (N = 821) Experimental group C1 C2 C1 avoidance EVASIV1 EVASIV3 EVASIV4 EVASIV6 COMPLA1 COMPLA8

C3

0.46 0.67 0.72 0.63 0.64 0.56

C2 self-con®dence SELFC1 SELFC5 SELFC6 VIGIL4 VIGIL5 VIGIL6

Control group C1 C2

C3

0.58 0.67 0.76 0.51 0.59 0.49

0.63 0.71 0.72 0.59 0.46 0.42 0.64 0.68 0.60 0.60 0.56

C4 impulsive, thoughtless VIGIL1 VIGIL3 COMPLA3 COMPLA4 COMPLA6

0.64 0.65 0.65 0.46 0.61 ÿ0.66 ÿ0.73 0.55 0.48 0.63

4.29 19.5 19.5 0.72 0.30

C4

0.35 0.64 0.75 0.67 0.59 0.56

C3 panic PANIC1 PANIC2 PANIC3 PANIC4 COMPLA2

Own value %Var. Cum% Var. Alpha -Inter-item corr

C4

2.08 9.4 29.0 0.70 0.28

1.89 8.6 37.5 0.65 0.27

1.28 5.8 43.4 0.62 0.25

ÿ0.67 ÿ0.71 0.58 0.41 0.59 4.25 19.3 19.3 0.71 0.29

2.02 9.2 28.5 0.70 0.28

1.84 8.4 36.9 0.64 0.26

1.38 6.3 43.1 0.59 0.23

280

J. Tuinstra et al. / Personality and Individual Di€erences 28 (2000) 273±285

terms of the original ®ve subscales. The items with the highest loading on the fourth component, VIGIL3 and VIGIL1, concern a deliberate way of decision making. The opposite way of deliberate decision making, namely, an impulsive and thoughtless way of decision making, is the content of the item with the highest loading on the ®fth component, COMPLA7. Both the fourth and the ®fth component refer to the same concept, namely, precision in taking decisions, deliberate or impulsive and thoughtless. Besides this theoretical motive, there is a methodological motive to propose a four component structure as a better solution than a ®ve component solution. Also, a scree test (Cattell, 1966) indicates that ®ve components in the ADMQ are too many. Therefore the same PCA was conducted, now for four components. 3.2. The revised subscale structure Three criteria were used to omit an item: because of their discrepancy of content, their low loadings, and the increase of the internal consistency of the component by omitting this particular item. Eventually, we omitted eight items. Five items, EVASIV2, SELFC2, SELFC3, COMPLA5, and COMPLA7 were deleted because of their low loadings (<0.40). VIGIL2 was deleted because of the increasing internal consistency of the component by omitting this particular item. SELFC4 and EVASIV5 were deleted because of their discrepancy of content in comparison with the other items of the component. Table 4 shows the revised structure of the ADMQ with 22 items. The ®rst component contains four items of the original subscale evasiveness and two of complacency. Considering the content of the items, the main idea of these items is that the respondent avoids deciding and follows decisions of other people. So avoidance is a better description than `evasiveness' of this ®rst component. The second component contains three items of the original subscale selfcon®dence and three of the subscale vigilance. The description self-con®dence can remain, because it represents the content of all six items: a self-con®dent way of decision making. The third component can maintain the name of the original subscale panic. All four items of the original subscale panic are included, and an additional item of complacency, COMPLA2. COMPLA2 ®ts perfectly in the description panic. The fourth and ®nal component contains two items of the subscale vigilance and three items of the subscale complacency. The items of these two original subscales represent opposite ways of decision making, illustrated by the two minus signs for the loadings of both VIGIL1 and VIGIL3. These two items represent a deliberate way of decision making, while the three other items represent an impulsive, thoughtless way of decision making. In constructing a score for this subscale we need to reverse the scores of both VIGIL1 and VIGIL3. 3.3. The control group The revised structure of the ADMQ, found in the experimental group, was tested in the control group. Comparable results of the PCA in both the experimental and the control group indicate that the revised structure of the ADMQ is not based on chance. The PCA in the control group indeed resembles very much the experimental group (Table 4). Also the internal

J. Tuinstra et al. / Personality and Individual Di€erences 28 (2000) 273±285

281

consistency and the mean inter-item correlations for the revised ADMQ subscales were very similar in the experimental and control group (Table 4). An SCA con®rms that the revised structure of the ADMQ is satisfactory: 41.8% of the possible maximal 43.1% explained variance is explained by the revised structure. Looking at the cumulative percentages of explained variance (Table 4), the revised structure of the ADMQ increases with 3.5 and 3.2% for, respectively, the experimental and the control group compared with the original structure (43.4 and 43.1% vs 39.9%). Also the inter-item correlations and the Cronbach's a increased in the revised ADMQ. 3.4. Convergent validity The convergent validity was examined for the total sample. First of all, Table 5 shows the mean scores of the subscales of the revised ADMQ and of the instrument used for testing the convergent validity, peer group pressure. In addition, Table 5 shows the correlations between the four decision making styles and peer group pressure. The correlations are found as expected. The strongest correlation is between peer group pressure and avoidance. Also, between peer group pressure on the one hand and panic and impulsiveness on the other hand, we found positive correlations. There is no correlation between self-con®dence and peer group pressure. 3.5. Discriminative validity Table 6 shows the e€ects of sex and educational level on the four revised subscales of the ADMQ. For all four subscales we see signi®cant di€erences in scores between male and female adolescents and between educational levels. Male adolescents score, as expected, higher on self-con®dence and lower on panic than female adolescents. Male adolescents also scored higher than females on the two other subscales, avoidance and impulsiveness. For the original subscale complacency, Friedman and Mann (1993) found higher scores for male adolescents. Partly the complacent way of decision making can be recognised in the new subscale impulsiveness, thoughtlessness. Higher scores of male adolescents are a consequence in this respect. We also found e€ects for educational level in the expected directions. Adolescents in the highest Table 5 Subscale scores and distribution characteristics of the revised ADMQ and peer group pressure (N = 1642), and their correlations No. of items Mean2 SD Median Range Avoidance Self con®dence Panic Impulsive Peer group pressure 

P < 0.01;



6 6 5 5 5

P < 0.001.

1.712 0.41 2.642 0.44 1.742 0.44 1.892 0.46 1.372 0.40

1.67 2.67 1.80 1.80 1.20

Avoidance Self con®dence Panic

1.00±3.50 1.00 1.33±4.00 ÿ0.33 1.00±3.60 0.41 1.00±4.00 0.24 1.00±3.20 0.27

1.00 ÿ29 ÿ17 ÿ0.03

Impulsive

1.00 0.08 1.00 0.19 0.14

282

J. Tuinstra et al. / Personality and Individual Di€erences 28 (2000) 273±285

Table 6 ADMQ scores per subscale by sex and educational level (N = 1642) Low Avoidance Male Female Self-con®dence Male Female Panic Male Female Impulsive Male Female

1

2

3

High 4

Total

E€ect

F

P

1.81 1.79

1.74 1.63

1.70 1.71

1.72 1.62

1.74 1.68

Sex Educ. level

7.44 7.63

0.006 0.000

2.58 2.49

2.64 2.61

2.74 2.59

2.76 2.66

2.68 2.59

Sex Educ. level

19.55 10.65

0.000 0.000

1.75 1.91

1.65 1.77

1.66 1.81

1.65 1.71

1.67 1.80

Sex Educ. level

34.30 8.00

0.000 0.000

1.97 1.90

1.97 1.85

1.91 1.86

1.90 1.73

1.94 1.83

Sex Educ. level

21.38 5.42

0.000 0.001

educational level score, as expected, higher on the competent decision making style self-con®dence and lower on the maladaptive decision making styles avoidance, panic and impulsiveness.

4. Conclusion and discussion In this study we examined three questions concerning the reliability, validity and structure of the Adolescent Decision Making Questionnaire. First, we investigated whether the structure of ®ve subscales of the ADMQ could be recognised in a population of Dutch adolescents. The relatively low Cronbach's a and low inter-item correlations of the subscales self-con®dence, vigilance, panic, evasiveness and complacency indicated that the subscales were not clearly present in our data. Both a Principal Component Analysis (PCA), Varimax rotated, on ®ve components and a Simultaneous Component Analysis supported this idea. Eventually, a revised ADMQ was proposed, with four components and 22 items. We omitted eight items of the original ADMQ, based on their content, their low loadings and the increase of Cronbach's a by omitting the particular item. Second, we evaluated this revised structure of the ADMQ by a cross validation study. Similar psychometric properties for both the experimental and the control group evidenced that the revised structure of the ADMQ is a valid one. Three results con®rmed the improvement of the revised ADMQ. The ®rst and most important result concerns the content of both the subscales and `their' separate items. Each item ®ts in the intended subscale, regarding to content and loading. The second result is the increase of the internal consistency of the revised subscales. Both the Cronbach's a and the inter-item correlations of the revised ADMQ subscales are higher than the originals. The third result that supports the improvement

J. Tuinstra et al. / Personality and Individual Di€erences 28 (2000) 273±285

283

of the revised ADMQ is the increase of the cumulative explained variance with 3.5 and 3.2% for the experimental and the control group, respectively. The structure of the revised ADMQ subscales partly changed the focus of the underlying concepts of decision making. The ®ve original concepts were reduced into four concepts. Two of the revised concepts maintained the essence of two original concepts, namely, self-con®dence and panic. The other two revised concepts, avoidance and impulsiveness were composed of items from the original concepts evasiveness, vigilance and complacency. These ®nal two original concepts vigilance and complacency showed many mis®tting items in the ®rst analysis. The third question in this paper concerned the convergent and discriminative validity of the revised ADMQ. Our results con®rmed that the revised ADMQ is a valid instrument. Social pressure from the peer group correlated, as expected, positively with avoidance. Also panic and an impulsive, thoughtless way of decision making was positively correlated with peer group pressure. We found no correlation between peer group pressure and self-con®dence. Di€erences between sex and educational level were also found as expected. Male adolescents outscored females on self-con®dence, avoidance and impulsiveness. They scored lower on panic than female adolescents. Adolescents in the highest educational level score higher on the competent decision making style self-con®dence and lower on the less competent decision making styles avoidance, panic and impulsiveness. However, although the di€erences were statistically signi®cant, their e€ect size is `small' in terms of Cohen's assessment (1992). The e€ect sizes for the di€erences between males and females, for example, are between 0.20 and 0.50 for self-con®dence, panic and impulsiveness. For avoidance the e€ect size is even lower, 0.15. In conclusion, our study showed that we improved the original ADMQ into a revised ADMQ. Given the 30 available items, the revised 22 item version of the ADMQ improved for structure, reliability and validity. Further development in instruments measuring decision making in adolescents might consider whether theoretical concepts are lacking and thus items should be added.

Acknowledgements We would like to thank Henk Kiers for his valuable explanation and assistance with the Simultaneous Component Analysis. Appendix A A.1. The items of the ADMQ

Self-con®dence SELFC1 I feel con®dent about my ability to make decisions. I am not as good as most people in making decisions. SELFC2

284

J. Tuinstra et al. / Personality and Individual Di€erences 28 (2000) 273±285

SELFC3 SELFC4 SELFC5 SELFC6

It is easy for other people to convince me that their decision is the correct one. I feel so discouraged that I give up trying to make decisions. The decisions I make turn out well. I think that I am a good decision maker.

Vigilance VIGIL1 VIGIL2 VIGIL3 VIGIL4 VIGIL5 VIGIL6

I take a lot of care before make my choice. Once I have made a decision then I don't change my mind. I like to think about a decision before I make it. When I make a decision, I feel that I made the best one possible. I like to make decisions myself. When I decide to do something, I get right on with it.

Panic PANIC1 PANIC2 PANIC3 PANIC4 Evasiveness EVASIV1 EVASIV2a EVASIV3 EVASIV4 EVASIV5a EVASIV6

I panic if I have to make decisions quickly. I feel as if I'm under tremendous time pressure when making decisions. I can't think straight if I have to make a decision in a hurry. The possibility that some small thing might go wrong causes me to immediately change my mind about what I'm going to do. I avoid making decisions. I put o€ making decisions. I'd rather let someone else make a decision for me so that it won't be my problem. I prefer to leave decisions to others. When I have to make a decision, I wait a long time before starting to think about it. I don't like to take responsibility for making decisions.

Complacency COMPLA1 When faced with a decision, I go along with what others suggest. COMPLA2 Whenever I get upset by having to make a decision, I choose on the spur of the moment. COMPLA3 I put a little e€ort into making decisions. COMPLA4 When I'm forced to make a decision, I couldn't care which way I choose. a I choose on the basis of some small detail. COMPLA5 COMPLA6 I tend to drift into decisions without thinking about them. COMPLA7 When making decisions I tend to choose the ®rst alternative that comes to mind. COMPLA8 I prefer to do what others choose because I don't like to be di€erent. a

Item omitted in revised version ADMQ.

J. Tuinstra et al. / Personality and Individual Di€erences 28 (2000) 273±285

285

References Cattell, R. B. (1966). The scree test for number of factors. Multivariate Behavioral Research, 1, 245±276. Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155±159. Friedman, I. A., & Mann, L. (1993). Coping patterns in adolescent decision making: an Israeli-Australian comparison. Journal of Adolescence, 16, 187±199. HaÈnninen, V., & Aro, H. (1996). Sex di€erences in coping and depression among young adults. Social Science and Medicine, 43(10), 1453±1460. Janis, I. L., & Mann, L. (1977). Decision making: a psychological analysis of con¯ict, choice and commitment. New York: Free Press. Mann L. (1982). Decision Making Questionnaires I and II. Unpublished scales. Flinders Decision Workshop, the Flinders University of South Australia. Mann, L., Burnett, P., Radford, M., & Ford, S. (1997). The Melbourne Decision Making Questionnaire: an instrument for measuring patterns for coping with decisional con¯ict. Journal of Behavioral Decision Making, 10, 1±19. Mann, L., Harmoni, R., & Power, C. (1989). Adolescent decision making, the development of competence. Journal of Adolescence, 12, 265±278. Nunally, J. C. (1978). Psychometric theory. New York: McGraw-Hill. OlaÂh, A. (1995). Coping strategies among adolescents: a cross-cultural study. Journal of Adolescence, 18, 491±512. Ormond, C., Luszcz, M. A., Mann, L., & Beswick, G. (1991). A metacognitive analysis of decision making in adolescence. Journal of Adolescence, 14, 275±291. Plancherel, B., & Bolognini, M. (1995). Coping and mental health in early adolescence. Journal of Adolescence, 18, 459±474. Poikolainen, K., Kanerva, R., & LoÈnnqvist, J. (1995). Social class and defence styles among adolescents. Journal of Adolescence, 18, 669±677. SchoÈnp¯ug, U., & Jansen, X. (1995). Self-concept and coping with developmental demands in German and Polish adolescents. International Journal of Behavioral Development, 18(3), 385±405. Schuyt C. J. M. (1995). Kwetsbare jongeren en hun toekomst. Beleidsadvies gebaseerd op literatuurverkenning. (Vulnerable youth and their future. Advice for policy based on literature). Rijswijk: Ministry of Public Health. Stevens, J. (1992). Applied multivariate statistics for the social sciences. New Jersey: Lawrence Erlbaum Associates Inc. Publishers. van der Linden F. J., Dijkman T. A. (1989). Jong zijn en volwassen worden in Nederland. Een onderzoek naar het psychosociaal functioneren in alledaagse situaties van de Nederlandse jongeren tussen 12 en 21 jaar. (Growing up in the Netherlands. A study into psychosocial functioning in everyday situations of the Dutch youth, aged 12 to 21.). Hoogveld Instituut, Nijmegen. Wills, T. A., McNamara, G., & Donato, V. (1995). Parental education related to adolescent stress-coping and substance use: development of a mediational model. Health Psychology, 14(5), 464±478.