Prediction of teachers’ and principals’ external and internal accountability
In this section, we present findings where we predict teachers' and principals' accountability dispositions by cultural values, perceived organizational support, and background variables. For the prediction of teacher accountability, we also added principals’ accountability disposition as an additional predicting variable. Whereas the teacher models without principals' accountability dispositions were calculated for eight countries, the models that included principals' dispositions were calculated for six countries (Canada and China did not provide principal data). We first describe the teacher models: external and internal accountability without and with principal accountability contribution, and audiencefocused external accountability referring to parents and school management. Then we present the principal models.
Prediction of teachers’ accountability
Relationships between model variables
We first ran correlations to explore the relations among the study variables: teachers' external and internal accountability dispositions, cultural values (individualism and collectivism), and perceived organizational support (Table 4.33). We also included personal background variables (gender, seniority) and school size.
It should be noted that although the correlation between the two teacher accountability dimensions was substantial (r=.504, p < .001), suggesting that the two shared common content (variance), these dimensions were still far from identical: 75% of the variance attested to differences between the two. Both external and internal accountability were each significantly and positively related to the two cultural values individualism and collectivism, with small and medium effect sizes, respectively. Organizational support correlated significantly and positively with the two accountability dimensions with a medium effect size. In regard to background variables, seniority was significantly and positively related to internal accountability with a rather small effect size. Gender was significantly but negatively related to internal accountability only, meaning that females were more internally
Table 4.33 Correlations Between Teacher Study Variables
1 
2 
3 
4 
5 
6 
7 
8 

(1) Gender 
 

(2) Seniority 
.003 
 

(3) External accountability 
.059** 
.029 

(4) Internal accountability 
.110** 
.068** 
.504** 

(5) Individualism 
.066** 
.047* 
.060** 
.124** 

(6) Collectivism 
.086** 
.061** 
.282** 
.369** 
.075** 

(7) Organizational support 
.021 
.078** 
.366** 
.298** 
.075** 
.347** 

(8) School size 
.046* 
.105** 
.040* 
.107** 
.035 
.154** 
.252** 
Note: * p < .05; ** p < .01; Gender: 0=Female, l=Male.
accountable than males. The relation of school size to internal accountability was significantly negative, again with a small effect size. These correlations paved the way to more advanced analyses of the relationships among the smdy variables by building multilevel multiple linear regression models for predicting teacher external and internal accountability.
Modeling teachers’ accountability
In Chapter 3  Study Methods (p. 32), we explained the need for a multilevel approach because our teacher data had a hierarchical structure in which teachers were nested within schools and within countries. Because of the small number of countries where teacher data were collected (eight) at the highest level, a twolevel model was chosen over a threelevel model. The final study model then included variables representing the individual teacher (first level) and those representing the school (teacher faculty) (second level).
To test for the need to consider the data's nested structure, we calculated the ICC between the two levels, i.e., the proportion of variance at the teacher and at the school levels. These ICCs showed that significant amounts of variance were located at the school level. For external accountability, 19.3% of the variance was located at the school level and for internal accountability, 20.4%. These results confirmed the leveled structure and the need for multilevel analyses.
We performed multilevel regression analyses including teacher variables and school variables as predictors. To create school scores, we aggregated the scores on the individual teacher level and used the average school teachers’ score. This procedure was performed for the accountability, cultural values, and organizational support variables. Based on preliminary tests, we also included interactions between the cultural values (individualism and collectivism) and organizational support. In regard to background variables, we included gender and seniority at the individual level and school size (student body size) at the school level. When the model included principals’ data (collected from six, not eight, countries), we considered principals’ accountability dispositions as a school level variable.
A relation between school level variables and each of the accountability dispositions should be interpreted at the school level. The teachers ’ school mean score of, e.g., schoollevel individualism can only predict the school mean of teachers' accountability dispositions. Regression coefficients are provided that give information about individual teachers compared to their school mean. So, positive coefficients mean that teachers with a higher score for that predictor showed a higher accountability disposition compared to their within school colleagues.
Table 4.34 shows the models for predicting external and internal accountability without principals' accountability dispositions. We discuss the statistical model of external accountability followed by the statistical model predicting internal accountability.
EXTERNAL ACCOUNTABILITY RESULTS
At the teacher level, teachers' collectivism (/?=0.12, /(2,540)=9.866, p < .001) and organizational (school) support (/?=0.16, /(2,540)=10.814, p < .001) were positive significant predictors of teachers' external accountability disposition. The more teachers adhered to collectivistic views and the more they experienced organizational support, the higher their external accountability disposition. All other variables did not predict teachers' external accountability. Of the two predictors at the teacher level, organizational support had a larger standardized coefficient than collectivism, meaning that organizational support had a slightly larger predictive value. There was no significant interaction effect between collectivism and organizational support; thus, the regression coefficients for collectivism were similar for teachers across the whole range of organizational support.
The other cultural value, individualism, did not predict teachers’ external accountability. Similarly, the interaction between individualism and organizational support did not predict teachers’ external accountability disposition. It should be noted that the pvalue of seniority at the individual level suggested a marginal positive relation (JS=0.02, t(2,540)=1.696, p=.09). We also observed a marginal positive teacher level interaction between
Table 4.34 Models for Predicting Teachers' External and Internal Accountability
External Accountability 
Internal Accountability 

B(SE) 
P 
B(SE) 
P 

Fixed Part Intercept 
3.91 (0.022) 
4.49(0.017) 

Gender 
0.03 (0.023) 
0.01 
0.04(0.018) 
0.02* 
Seniority 
0.002 (0.001) 
0.02* 
0.001 (0.001) 
0.01 
Individualism 
0.0003 (0.014) 
0.0003 
0.03 (0.014) 
0.02* 
Collectivism 
0.21 (0.021) 
0.12*** 
0.23 (0.018) 
0.13*** 
Organizational support 
0.20 (0.018) 
0.16*** 
0.08(0.014) 
0.07*** 
Org. support X individualism 
0.01 (0.018) 
0.005 
0.000003 (0.017) 
0.000002 
Org. support X collectivism 
0.04 (0.025) 
0.03* 
0.01 (0.022) 
0.005 
School size (student body/1,000) 
0.04 (0.034) 
0.03 
0 (0.025) 
0.002 
School mean individualism 
0.10(0.058) 
0.03 
0.15 (0.045) 
0.05** 
School mean collectivism 
0.10(0.084) 
0.03*** 
0.27 (0.077) 
0.07** 
School mean org. support 
0.27 (0.064) 
0.12 
0.16(0.042) 
0.07*** 
School mean individualism x school mean org. support 
0.17(0.136) 
0.03 
0.05 (0.081) 
0.008 
School mean collectivism x school mean org. support 
0.03 (0.11) 
0.01 
0.13 (0.073) 
0.03* 
Random Part 
Variance 
Explained Variance 
Variance 
Explained Variance 
€ 
0.228 
15% 
0.167 
13% 
2 a «_{0} 
0.045 
30% 
0.023 
53% 
Note:'p < .1, *p < .05, **p < .01, ***p< .001; Gender: 0=Female, l=Male.
teachers’ collectivism and organizational support (>5=0.03, t(2,540)=1.766, p=.08). This means that the higher teachers' collectivist views were, the stronger the relation between teachers’ organizational support and their external accountability.
At the teacher school level, similar to the individual level predictors, the teachers' school mean of organizational support (/?=0.13, f( 178)=5.452, p < .001) positively predicted the school mean of teachers’ external accountability: the more teachers on average felt organizational support at their work, the more they felt externally accountable. School mean of collectivism was a significant predictor of external accountability, but individualism was not, nor was there any interaction. School size had no relation to external accountability. These results showed that the teachers’ individualistic, collectivistic, and organizational support characteristics predicted external accountability to a stronger degree than personal characteristics.
The model explained 15% of the variance on the teacher level and 30% on the school level. The total explained variance of the model was 18%, so about onefifth of the variance between teachers' external accountability scores is explained by the predictors both at the teacher and school levels.
INTERNAL ACCOUNTABILITY RESULTS
Similar to the external accountability model, both teachers’ collectivism (/?=0.13, t(2,540)=12.739, p < .001) and organizational support (>5=0.07, f(2,540)=5.989,/> < .001) were positive significant predictors at the teacher level for internal accountability. The other cultural value, individualism 05=0.02, /(2,540)=2.028, p=.042) showed to be a positive significant predictor as well. Gender turned out to be the only significant background predictor (/?=0.02, T(2,540)=—2.376. />=.018). The negative sign meant that female teachers felt more accountable than their male counterparts. Both interactions between organizational support and the cultural values were not significant predictors.
At the school level, all variables predicted the school mean of teachers' internal accountability. The school mean of individualism (>5=0.05, /(178)=3.355,/>=.001), collectivism (/?=0.07, f(178)=2.515,/>=.001), and organizational support (>5=0.07, /(178)=3.714, p < .001) were all positive significant predictors. The pvalue of the interaction between the school mean of collectivism with the school mean of organizational support suggested a marginal negative relation (>5=0.03, f( 178)=—1.822, />=.07). School size had no predictive effect, just like in predicting internal accountability at the individual level.
The amount of explained variance at the teacher level was 13% and at the school level was 53%. Similar to the model of external accountability, the total explained variance of the internal accountability model was 21%, meaning that this set of predictors also predicted about onefifth of the variance between teachers' internal accountability scores.
Principals’ accountability! predicting teachers ’accountability
In order to explore the predictive power of principals' accountability disposition for their teachers’ accountability, we had to limit our teacher sample to that composed of the same six countries represented in the principal sample. To this reduced teacher sample, we added the principals' own accountability dispositions as predictors of teachers’ accountability.
We first explored the relations between teachers and principals on both external and internal accountability. The correlation between teachers’ external accountability score and the principals' external accountability score was significant (r( 1,894)=. 167, p < .001) but relatively small. We also found a significant correlation (r(l,894)=.2219, p < .001) between teachers’ and principals' internal accountability, slightly larger than the correlation for external accountability. The significant correlations allowed the inclusion of principals' external and internal accountability in the models as schoollevel predictors. These models are shown in Table 4.35.
PRINCIPALS’ EXTERNAL ACCOUNTABILITY AS A PREDICTOR OF TEACHERS'
EXTERNAL ACCOUNTABILITY
The most striking finding is the predicting effect at the school level of principals' external accountability on teachers' external accountability (/?=0.07, T( 109)=3.155, p=.002). We found that at the individual level, teachers’ collectivism (/?=0.11, f(l,879)=8.313, p < .001) and organizational support 05=0.15, f(l,879)=9.455, p < .001) significantly predicted teachers’ external accountability. Also, a positive marginal interaction between collectivism and organizational support was observed (/?=0.03, f(l,879)=1.690, p=.091). All other variables failed to predict external accountability at the teacher level. At the school level, organizational support predicted teachers' external accountability 05=0.11, i(109)=4.722, p < .001), and school size was a positive marginal predictor (jS=0.02, f(109)=l. 792, p=075).
Another notable finding is the 15% explained variance at the teacher level and 43% at the school level. Comparing the explained variance for the model without and with principals' external accountability as predictor in the sixcountry sample (see Appendix 3.5a, second part, columns M7 and M8), we saw no change at the teacher level and 8 percent points more explained variance at the school level (the level where the principal's score was added as a predictor). Considering the small change in other predictors, we may conclude that all extra variance was explained by principals’ external accountability scores.
External Accountability 
Internal Accountability 

B(SE) 
fi 
B(SE) 
fi 

Fixed Part Intercept 
3.94 (0.024) 
4.51 (0.019) 

Gender 
0.03 (0.027) 
0.01 
0.04 (0.022) 
0.02* 
Seniority 
0.002 (0.001) 
0.02 
0.001 (0.001) 
0.01 
Individualism 
0.01 (0.016) 
0.01 
0.04 (0.016) 
0.03* 
Collectivism 
0.20 (0.024) 
0.11*** 
0.22 (0.021) 
0.12*** 
Org. Support 
0.20(0.021) 
0.15*** 
0.08 (0.015) 
0.06*** 
Org. Support x Individualism 
0.004 (0.02) 
0.003 
0.005 (0.021) 
0.003 
Org. Support x Collectivism 
0.06 (0.033) 
0.03* 
0.03 (0.035) 
0.02 
School size (Student Body/1000) 
0.09 (0.049) 
0.04* 
0.02 (0.038) 
0.004 
School mean Individualism 
0.04 (0.058) 
0.01 
0.16(0.051) 
0.06** 
School mean Collectivism 
0.025 (0.085) 
0.005 
0.26 (0.078) 
0.06** 
School mean Org. Support 
0.30 (0.064) 
on*** 
0.13 (0.052) 
0.04* 
School mean Individualism x School mean Org. Support 
0.09(0.145) 
0.01 
0.02 (0.118) 
0.004 
School mean Collectivism x School mean Org. Support 
0.33 (0.226) 
0.03 
0.06 (0.167) 
0.001 
Principal External Internal Accountability 
0.16(0.051) 
0.07** 
0.22 (0.044) 
0.07*** 
Random Part 
Variance 
Explained Variance 
Variance 
Explained Variance 
<7^{2} 
0.228 
15% 
0.164 
14% 
2 € 
0.031 
43% 
0.016 
58% 
Note:' p < .1, * p < .05, **p < .01, ***p < .001; Gender: 0=Female, l=Male.
PRINCIPALS' INTERNAL ACCOUNTABILITY AS A PREDICTOR FOR TEACHERS' INTERNAL ACCOUNTABILITY
Similar to the relation between principals' and teachers’ external accountability previously reported, within the model for teachers’ internal accountability with principal's internal accountability disposition, at the school level principals' internal accountability significantly predicted teachers’ internal accountability (/?=0.07, t(109)=3.846, p < .001). We also included in this model, at the individual level, teachers' gender (/?=0.02, i(l,879)=1.887,/>=059), individualism (>5=0.03, t(l,879)=2.286,/>=.022), collectivism (>5=0.12, r( 1,879)=10.511. p < .001), and organizational support (/7=0.06. /(l,879)=4.941,p < .001), which significantly predicted teachers’ internal accountability, though for gender only marginally. All other variables did not predict internal accountability at the teacher level. Again, collectivism and organizational support were the strongest predictors.
At the school level, next to principals’ internal accountability, the school means of individualism (>5=0.06, i(109)=3.656, />=.001), collectivism (/?=0.06, t( 109)=3.412, p=.001), and organizational support (/?=0.05, t(109)=2.308, p=.023) significantly predicted teachers’ internal accountability.
When looking at the explained variance, we see for principals' internal accountability a similar trend for the explained variance of the model as for external accountability. The amount of explained variance was at the teacher level 14% and 58% at the school level. The latter is in the sixcountry sample 20 percent points more than in the model without the principals’ internal accountability included (see Appendix 3.5b, second part, columns M7 and M8). Thus, and similar to the case of external accountability, adding the principals' accountability made the predictive value of the models substantially stronger.
Prediction of teachers’ accountability dispositions toward parents and school management
In order to examine differences between accountability audiences, we analyzed teachers' accountability dispositions toward parents and school management. As in the case of teachers' general external and internal accountability, hierarchical models were appropriate for the audience focused accountability data.
Relationships between model variables
We first ran correlations to explore the relations between the teachers' external accountability dispositions toward parents and school management with cultural values (individualism and collectivism) and perceived organizational support (Table 4.36). We also included personal background variables (gender, seniority) and school size. As in the case for correlations between external and internal accountability and teacher variables, we see many
Accountability 

Toward Parents 
Toward Management 

(1) Gender 
.017 
.045* 
(2) Seniority 
.040* 
.012 
(3) External accountability 
_{464}** 
.543** 
(4) Internal accountability 
331** 
.475** 
(5) Individualism 
.007 
.054** 
(6) Collectivism 
.257** 
.289** 
(7) Organizational support 
152** 
.314** 
(8) School size 
.068** 
.078** 
Note: * p < .05; ** p < .01; Gender: 0=Female, l=Male.
significant correlations and thus the more advanced analyses of the relationships between the study variables with multilevel multiple linear regression models for predicting teachers’ accountability toward parents and school management were appropriate.
In Table 4.37, we present the models for teachers' external accountability toward parents and school management. As we did earlier in regards to teachers’ general external accountability, as a last step in the analysis, we included principals' own external accountability toward the respective audiences as a predictor for teachers' accountability toward these two audiences (Table 4.38).
In regard to accountability toward parents, calculation of the ICC showed that 12% of the total variance was located at the school level, where for accountability toward school management the variance at the school level was 25%. The amount of variance at the school level within the accountability model toward school management was surprisingly high.
Two more features of the models should be noted. When looking at the models, it appears that the variables used to predict general external accountability (see Table 4.34) are about equally successful in predicting the accountability dispositions toward the two audiences. Seniority predicted accountability toward parents but not toward school management. In other words, teachers who had more years of work experience in school tended to feel more accountable to parents than teachers with less work experience, but no difference was found for school management in this regard. At the school level, we see in both audiencespecific accountability disposition models that the school mean of teachers' collectivism plays a role in predicting aggregated school accountability, which was not a predictor in the general external accountability model (see Table 4.34). Noteworthy, further,
Table 4.37 Prediction of Teachers’ External Accountability Toward Parents and School Management
Accountability’ Toward 

Parents 
School Management 

B(SE) 
P 
B(SE) 
P 

Fixed Part Intercept 
3.66 (0.028) 
4.07 (0.03) 

Gender 
0.01 (0.034) 
0.003 
0.003 (0.023) 
0.001 
Seniority 
0.003 (0.002) 
0.03* 
0.001 (0.001) 
0.01 
Individualism 
0.002 (0.025) 
0.002 
0.01 (0.018) 
0.005 
Collectivism 
0.28 (0.034) 
0.16*** 
0.25 (0.023) 
0.14*** 
Org. support 
0.12 (0.026) 
0.09*** 
0.17(0.022) 
0.13*** 
Org. support X individualism 
0.02 (0.027) 
0.02 
0.004 (0.024) 
0.002 
Org. support X collectivism 
0.03 (0.032) 
0.01 
0.04 (0.028) 
0.02 
School size (size/1,000) 
0.003 (0.023) 
0.002 
0.001 (0.022) 
0.0004 
School mean individualism 
0.15 (0.078) 
0.05* 
0.11 (0.074) 
0.04 
School mean collectivism 
0.61 (0.098) 
0.16*** 
0.30(0.130) 
0.08* 
School mean ors. support 
0.05 (0.077) 
0.02 
0.24 (0.079) 
0.11** 
School mean individualism x school mean org. support 
0.12(0.184) 
0.02 
0.12 (0.143) 
0.02 
School mean collectivism x school mean org. support 
0.09 (0.131) 
0.02 
0.07 (0.105) 
0.01 
Random Part 
Variance 
% Explained Variance 
Variance 
% Explained Variance 
2 G 0 
0.521 
7% 
0.307 
11% 
0 
0.061 
23% 
0.082 
28% 
Note:' p< .1, *p < .05, **p < .01, ***p < .001; Gender: Female=0, Male=l.
is the absence of predictive value of the aggregated scores of organizational support at the school level within the model that predicts teachers' disposition toward parents. After these general observations, we now focus on the separate models for the two audiences.
Table 4.3S Predictive Model for Teachers’ External Accountability Toward Parents and School Management With Principals’ Parents or School Management Accountability Scores as Predictors
Toward Parents 
Toward School Management 

B(SE) 
/5 
B(SE) 
/5 

Fixed Part Intercept 
3.63 (0.032) 
4.08 (0.035) 

Gender 
0.01 (0.041) 
0.01 
0.003 (0.029) 
0.001 
Seniority 
0.003 (0.002) 
0.03^{1} 
0.001 (0.001) 
0.01 
Individualism 
0.01 (0.029) 
0.005 
0.01 (0.02) 
0.004 
Collectivism 
0.26 (0.037) 
0.14*** 
0.24 (0.025) 
0.14*** 
Org. support 
0.12(0.029) 
0.09*** 
0.15 (0.025) 
0.11*** 
Org. support x individualism 
0.02 (0.033) 
0.01 
0.005 (0.027) 
0.004 
Org. support x collectivism 
0.005 (0.042) 
0.003 
0.03 (0.037) 
0.02 
School size (size/1.000) 
0.09 (0.063) 
0.04 
0.1 (0.083) 
0.05 
School mean individualism 
0.22(0.091) 
0.07* 
0.14(0.09) 
0.04 
School mean collectivism 
0.45 (0.109) 
0.10*** 
0.23 (0.153) 
0.05 
School mean org. support 
0.01 (0.095) 
0.004 
0.18(0.104) 
o.or 
School mean individualism x school mean org. support 
0.58(0.273) 
0.07* 
0.44 (0.221) 
0.05’ 
School mean collectivism x school mean org. support 
0.01 (0.395) 
0.001 
0.02 (0.374) 
0.002 
Principal external accountability toward parents/ school management 
0.11 (0.035) 
0.08** 
0.16(0.046) 
0.11** 
Random Part 
Variance 
Explained Variance 
Variance 
Explained Variance 
2 <7 
0.531 
6% 
0.317 
9% 
9 a € 
0.056 
33% 
0.078 
29% 
Note:^{r} p < .1, * p < .05, **p < .01, ***p < .001; Gender: 0=Female, l=Male.
Teachers ’ external accountability toward parents
MODEL WITHOUT PRINCIPALS'ACCOUNTABILITY
At the first level, teachers’ collectivism (/?=0.16, i(2,540)=8.343,/> < .001) and organizational support (>5=0.09, i(2,540)=4,540,p < .001) were positive significant predictors of teachers’ accountability disposition toward parents. The more teachers adhered to collectivistic views and the more they experienced organizational support, the higher their accountability disposition toward parents. Additionally, at the teacher level, teachers' seniority was a significant positive predictor of teachers' accountability disposition toward parents (>5=0.03, f(2,540)=8.343, p=.045), meaning that the more work experience teachers had, the higher their accountability disposition toward parents. All other variables at the teacher level did not predict teachers’ accountability toward parents. Of the two predictors at the teacher level, collectivism had a larger standardized coefficient than the regression coefficient for organizational support, meaning that collectivism had a slightly higher predictive value. There was no significant interaction effect between collectivism and organizational support, meaning that the regression coefficients for collectivism were similar for teachers across the whole range of organizational support. The second cultural value, individualism, as well as the interaction between individualism and organizational support, did not predict teachers’ external accountability toward parents.
At the second level, school, the school mean of teachers' collectivism (>5=0.16, t(178)=6.267, p < .001) positively predicted the school mean of teachers’ accountability toward parents: the more teachers on average had collective feelings, the more they felt externally accountable toward parents (similar to the teacher level prediction by teacher collectivism). The school mean of teachers’ organizational support was not a significant predictor of teachers’ accountability toward parents at the school level, and the school mean of teachers' individualism was a marginally negative significant predictor (>5=—0.05, T( 178)=— 1.895. p=.059). None of the interactions at the school level of the cultural values and organizational support were predictors of accountability toward parents. These results showed that the teachers’ individualistic and collectivistic values predicted external accountability toward parents to a stronger degree than their background characteristics.
At the teacher level, the model explained 7% of the teacher variance and at the school level, it explained 23% of the school faculty variance. The total explained variance of the model is 9%, so about onetenth of the variance between teachers in external accountability scores toward parents is explained by the cultural values and organizational support predictors together at the teacher and the school level.
CONTRIBUTION OF PRINCIPALS' EXTERNAL ACCOUNTABILITY TOWARD PARENTS TO TEACHERS' EXTERNAL ACCOUNTABILITY TOWARD PARENTS
In light of the findings in regard to differences between teachers' and principals' accountability dispositions toward parents (Table 4.15), we set out on an analysis to examine the contribution of principals' accountability disposition to the teachers’ accountability toward parents in the same schools. In Table 4.38, we see that principals' external accountability predicted teachers’ accountability toward parents.
Of the variance at the teacher level, 6% is explained and at the school level, 33%. Compared with the explained variance for the model without external accountability scores as predictors in the sixcountry sample (see Appendix 3.6a, second part, columns M7 and M8), we see almost no change at the teacher level, and 6 percent points rise in explained variance at the school level (the level of the included predictor). We may conclude that a high amount of extra variance was explained by adding principals' accountability toward parents in the model. The total model did explain a somewhat similar amount of variance (9.5%) as the model without the principals’ scores included as predictor.
Teachers’ external accountability toward school management
MODEL WITHOUT PRINCIPALS' ACCOUNTABILITY
Similar to the accountability model for accountability toward parents, both teachers' collectivism (/?=0.14, f(2,540)=10.557,/> < .001) and experienced organizational support (/?=0.13, f(2,540)=7.536, p < .001) were positive significant predictors at the teacher level (see Table 4.38). The other cultural value, individualism, showed not to be a significant predictor, and no background predictors were found to predict teachers' external accountability toward school management. Similar to the external model, both interactions between school support and the cultural values were not significant predictors of teachers’ accountability toward school management.
At the second (school) level, the school means of teachers' collectivism (/?=0.08, /(178)=2.282, p=.024) and organizational support (/3=0.11, /(178)=3.008, p=.003) were significant positive predictors for the school mean of teachers' external accountability toward school management. School size and the interactions between the school means of teachers' individualism and collectivism with the school mean of teachers' organizational support did not predict the school mean of teachers’ accountability toward school management.
The amount of explained variance at the teacher level was 11% and at the school level was 28%. The total explained variance of the model predicting teachers' external accountability toward school management was 15%, which is a bit more than in the model predicting teachers' accountability toward parents.
CONTRIBUTION OF PRINCIPALS ’ EXTERNAL ACCOUNTABILITY TOWARD SCHOOL MANAGEMENT TO TEACHERS’ EXTERNAL ACCOUNTABILITY TOWARD SCHOOL MANAGEMENT
Similar to including principals’ accountability disposition toward parents for predicting the teachers' accountability toward parents in the same schools, we included principals' accountability toward school management as a predictor for teachers' accountability toward school management (see Table 4.38). Principals' external accountability toward school management significantly predicted teachers' external accountability to this audience (/?=0.11. /(109)=3.567, />=.001), as did teachers' collectivism and organizational support. At the second level, the prediction by contribution of organizational support to the model including principal contribution was marginally significant (/?=0.07, /(109)= 1.784, p=.0H), as was the interaction of collectivism with organizational support (>5=0.05, /(109)= 1.978, p=.050).
Of the variance, 9% is explained at the teacher level and 29% at the school level. Compared with the explained variances for the model without principals’ external accountability scores as predictor in the sixcountry sample (see Appendix 3.6b, second part, columns M7 and M8), we saw a small change at the teacher and again a large change at the school level, the level of the included predictor: a 12 percent point increase.
Prediction of principals ’ accountability dispositions
Relationships between modeI variables
In this section, we report on our findings on the prediction of principals’ accountability dispositions by their cultural values and their organizational (board) support. We also included principals' background variables (gender and seniority) in the analytical model and school size as school characteristic. Table 4.39 shows the results of running correlations for the principal study variables: culniral values, accountability dispositions, and principals' and schools' background variables. The many significant correlations allowed for building predicting models.
Table 4.39 Correlations Between Principal Study Variables
1 
2 
3 
4 
5 
6 
7 

(1) Gender 
— 

(2) Seniority 
.053 

(3) External accountability 
.007 
.433** 

(4) Internal accountability 
.192’ 
.035 
001 

(5) Individualism 
.125 
.057 
.017 
.646** 

(6) Collectivism 
.187* 
.002 
.062 
.105 
.024 

(7) Organizational support 
.155 
.108 
.447** 
.276** 
.312** 
.201* 

(8) School size 
.036 
.045 
.317** 
.284** 
.217* 
.104 
.655** 
Note: * p < .05, **p < .01; Gender: 0=Female, l=Male.
Because of the small number of principals within schools and the small number of countries (six) where principal data were collected, hierarchical analysis like HLM was not possible (see Chapter 3 on the study methods for a more elaborated discussion). Therefore, a stepwise multiple regression model was developed for each of the accountability dispositions. First, we included principals' background variables and secondly the cultural values and organizational support. At the third step, similar to the teacher models, we included interactions between the cultural values and organizational support. Then, school size was added to the model. As a final step, we included dummy variables for each country in order to acquire insight into the possible predictive value of the principals' respective countries. We selected Hungary as a reference group since the country mean of Hungary> was closest to the general sample means of the principals' accountability dispositions. Table 4.40 shows the models for predicting principals' external and internal accountability.
Prediction ofprincipals’ external accountability
Our regression analysis included two models: the first included background variables, cultural variables, and organizational support. In the second model, we added the country dummy variables. The significant predictors in the first model were principals' gender (/?=0.20, f( 123)=—2.44. />=.016), individualism (/?= 0.28, ?(123)=3.30, />=.001), collectivism (>5=0.34, f(123)=2.63,/>=.010), and the interaction between individualism and organizational support (>5=0.19, t( 123)=2.30, />=.023). In regard to the gender effect, the minus sign of the [3 implied that male principals felt less externally accountable than their female counterparts. Furthermore, the more
Table 4.40 Regression Models for Predicting Principals’Accountability Without and With Dummy Variables for Countries
External Accountability 
Internal Accountability 

Without Dummies 
With Dummies 
Without Dummies 
With Dummies 

В (SE) 
P 
В (SE) 
P 
В (SE) 
P 
В (SE) 
P 

(Constant) 
2.67 (0.38) 
2.70(0.41) 
3.34(0.36) 
3.27(0.36) 

Seniority 
0.01 (0.004) 
0.13 
0.01 (0.004) 
0.12 
0.01 (0.004) 
0.14 
0.005 (0.004) 
0.12 
Gender 
0.19(0.08) 
0.20** 
0.09 (0.08) 
0.10 
0.07 (0.07) 
0.09 
0.06(0.07) 
0.07 
Individualism 
0.14 (0.04) 
0.28*** 
0.13 (0.04) 
0.26*** 
0.06 (0.04) 
0.14 
0.05 (0.04) 
0.10 
Collectivism 
0.21 (0.08) 
0.34** 
0.19(0.08) 
0.31* 
0.26 (0.07) 
0.48*** 
0.27(0.07) 
0.50*** 
Organizational support 
0.09 (0.05) 
0.18 
0.07(0.05) 
0.14 
0.02 (0.05) 
0.03 
0.01 (0.05) 
0.03 
Individualism x org. support 
0.10(0.04) 
0.19* 
0.11 (0.04) 
0.22* 
0.02 (0.04) 
0.05 
0.03 (0.04) 
0.07 
Collectivism x org. support 
0.08 (0.05) 
0.16 
0.02 (0.05) 
0.05 
0.06 (0.05) 
0.15 
0.01 (0.04) 
0.01 
School size' 
0.04 (0.07) 
0.04 
0.08 (0.09) 
0.09 
0.08 (0.07) 
0.10 
0.08 (0.08) 
0.10 
Being from Israel^{1} 
0.08 (0.12) 
0.07 
0.05 (0.11) 
0.05 

Being from the Netherlands 
0.42 (0.14) 
0.32*** 
0.42 (0.12) 
0.36*** 

Being from South Africa 
0.01 (0.15) 
0.01 
0.28 (0.13) 
0.22* 

Being from Spain 
0.11 (0.22) 
0.08 
0.18 (0.19) 
0.15 

Being from Zimbabwe 
0.15 (0.15) 
0.12 
0.18 (0.13) 
0.15 

R Square 
0.18 
0.29 
0.11 
0.32 
Note: ' p< 1, * p < .05, ** p < 01, *** p < 001; Gender: Female=0, Male=l; ‘Student body/1,000; * Hungary is used as reference group principals adhered to individualistic or collectivistic values, the higher their external accountability dispositions. The negativesign interaction between individualism and organizational support indicated that the more principals adhered to individualistic views, the weaker the relation between principals’ organizational support and their external accountability. All other variables did not predict principals' external accountability. Overall, the model without country dummy variables explained 18% of the variance of principals’ external accountability scores.
When country dummy variables were included, the gender of principals lost its predictive value. Individualism (5=0.26, f(118)=2.92,/>=.004), collectivism (5=0.31, t(118)=2.27, />=.025), and the interaction between individualism and organizational support (5=0.22, f(118)=2.67,/>=.009) still predicted principals' external accountability in the same way as without dummy variables. Additionally, being from the Netherlands (/?=—0.32, /(118)=—3.03, p=.003) had a negative influence on principals’ external accountability score, meaning that Dutch principals were less externally accountable than principals from other countries. Most notable is the extra amount of variance explained when including the country dummies. The model including the country variables explained 29% of the variance in principals' external accountability score, 11 percent points more than without country dummies.
Prediction ofprincipals’internal accountability
The number of variables that predicted principals' internal accountability was lower than in the model predicting principals’ external accountability. Within the model without country dummy variables, only collectivism (5=0.48, /(123)=3.55, />=.001) significantly predicted principal internal accountability. In other words, the more principals maintain collectivistic views, the higher their internal accountability disposition. All other variables failed to predict principals’ scores and the model explained 11% of the variance in the principals’ internal accountability.
When country dummy variables were included, collectivism still predicted principals' internal accountability (5=0.50, /(118)=3.65, p < .001). Additionally, being a Dutch principal (5=0.36, /(118)=—3.52,/z=.001) and being a principal from South Africa (5=0.22, /(118)=2.14. />=.035) proved also to be significant predictors. The positive coefficient for South African principals meant that principals from South Africa scored higher on internal accountability. For principals from the Netherlands, the opposite was true: being Dutch resulted in a lower score for principals' internal accountability. Similar to the external accountability model, the predictive value of the internal accountability model was higher when country dummies were included: without these sets of variables, the explained variance was 11%, while with them explained variance rose to 32%. In both cases, this was due only to the scores of principals in a small number of countries.
Prediction of principals’ accountability dispositions toward parents and school management
Relationships between model variables
We first ran correlations to explore the relations between the principals' accountability dispositions toward parents and school management with cultural values (individualism and collectivism) and perceived organizational support (Table 4.41). We also included personal background variables (gender, seniority) and school size. We concluded that the many significant correlations allowed for building predictive models similar to the models for external and internal accountability with stepwise regression. Again, we included as a last step the dummy variables for countries, with Hungary as the reference country. Table 4.42 shows the models for predicting principals' accountability toward parents and school management both with and without country dummies.
Prediction ofprincipals’ accountability toward parents
The significant predictors in the model for principals' accountability toward parents without country dummies were principals’ gender (/?=—0.20, ^( 123)=—2.42, />=.017), collectivism (>5=0.54, /(123)=4.20, p < .001), and
Table 4.41 Correlations Between Principal Variables and Accountability Toward Parents and School Management
Accountability 

Toward Parents 
Toward School Management 

(1) Gender 
.224** 
.179* 
(2) Seniority 
.135 
.019 
(3) External accountability 
.557** 
.572** 
(4) Internal accountability 
.425** 
—.499** 
(5) Individualism 
.032 
.009 
(6) Collectivism 
.135 
.098 
(7) Organizational support 
.025 
.090 
(8) School size 
.072 
.145 
Note: * p < .05, ** p < .01; Gender: 0=Female, l=Male.
Table 4.42 Regression Models for Predicting Principals’ Accountability Toward Parents and School Management Without and With Dummy Variables for Countries
Accountability Toward Parents 
Accountability Toward School Management 

Without Country Dummies 
With Country Dummies 
Without Country Dummies 
With Country Dummies 

В (SE) 
A 
В (SE) 
A 
В (SE) 
A 
В (SE) 
A 

(Constant) 
1.92 (0.55) 
1.31 (0.59) 
2.68 (0.56) 
2.35 (0.60) 

Seniority 
0.01 (0.006) 
0.12 
0.01 (0.006) 
0.08 
0.002 (0.007) 
0.03 
0.003 (0.006) 
0.04 
Gender 
0.29(0.12) 
0.20* 
0.21 (0.12) 
o.is^{1} 
0.25 (0.12) 
0.17* 
0.12 (0.12) 
0.09 
Individualism 
0.09 (0.07) 
0.12 
0.02 (0.07) 
0.03 
0.09 (0.07) 
0.11 
0.008 (0.07) 
0.01 
Collectivism 
0.49 (0.12) 
0.54*** 
0.63 (0.13) 
0.69*** 
0.31 (0.12) 
0.35* 
0.39 (0.13) 
0.43** 
Organizational support 
0.08 (0.08) 
0.10 
0.02 (0.08) 
0.03 
0.04 (0.08) 
0.06 
0.06 (0.08) 
0.08 
Individualism x org. support 
0.10(0.06) 
0.13 
0.06 (0.06) 
0.08 
0.09 (0.06) 
0.11 
0.08 (0.06) 
0.11 
Collectivism x org. support 
0.32 (0.08) 
0,44*** 
0.19(0.05) 
0.27** 
0.31 (0.08) 
0,44*** 
0.19(0.08) 
0.27* 
School size” 
0.09 (0.11) 
0.07 
0.10(0.09) 
0.08 
0.19 (0.11) 
0.14 
0.02 (0.13) 
0.12 
Being from Israel^{b} 
0.01 (0.18) 
0.003 
0.23 (0.18) 
0.14 

Being from the Netherlands 
0.49 (0.21) 
0.25* 
0.58 (0.21) 
0.30** 

Being from South Africa 
0.09 (0.22) 
0.04 
0.001 (0.22) 
0.001 

Being from Spain 
0.82 (0.33) 
0.42* 
0.65 (0.33) 
0.33^{f} 

Being from Zimbabwe 
0.33 (0.22) 
0.16 
0.14 (0.23) 
0.07 

R Square 
0.24 
0.36 
0.19 
0.33 
Note: ' p< .1, * p < .05, ** p < .01,*** p < .001; Gender: Female=0, Male=l; * Student body/1,000;^{b} Hungary is used as reference group.
the interaction between collectivism and organizational support (/7=0.44, i(123)=4.17,/> < .001). Male principals felt less accountable toward parents than their female counterparts. The more principals adhered to collectivis tic values, the higher their accountability toward parents. The interaction between collectivism and organizational support indicated that the more principals held to collectivistic views, the stronger the relationship between principals’ organizational support and accountability toward parents. None of the other variables predicted principals' accountability toward parents. Overall, the model without country dummy variables explained 24% of the variance of principals' accountability toward parents.
When country dummy variables were included, the predictive value of gender of principals became marginally significant (/?=0.15, t{ 118 )=1.77, />=.079). Collectivism (/7=0.69, /(118)=5.01. p < .001) and the interaction between collectivism and organizational support (/7=0.27, /(118)=2.52, />=.013) still predicted principals' accountability toward parents in the same way as without dummy variables. Additionally, being from the Netherlands (/7=0.25, ft 118)=—2.36. />=.020) had a negative influence on principals' external accountability score, meaning that Dutch principals were less accountable toward parents than principals from other countries. Spanish principals felt significantly more accountable toward parents than their colleagues in other countries (/7=0.42, t(118)=2.48, />=.015). Most notable is the extra amount of variance explained when including the country dummies. The model including the country variables explained 36% of the variance in principals’ accountability toward parents, 12 percent points more than in the model without country dummies.
Prediction of principals ’ accountability toward school management
In the model for principals' accountability toward school management without country dummies, we saw the same predictors and in the same direction as for accountability toward parents: principals' gender (/?=0.17, 123)=—2.06, />=.041), collectivism (>5=0.35, i(123)=2.60, />=.011), and the interaction between collectivism and organizational support (/7=0.44, t(123)=3.97,/> < .001). Male principals felt less accountable toward school management than their female counterparts. The more principals adhered to collectivistic values, the higher their accountability toward school management. The interaction between collectivism and organizational support indicated that the more principals adhered to collectivistic values, the stronger the relationship between principals' organizational support and their accountability toward school management. All other variables did not predict principals’ accountability toward school management. Compared to the general external accountability model (Table 4.40) of the predictors, only collectivism had a similar role. Overall, the model without country dummy variables explained 19% of the variance of principals' accountability toward school management, a bit less than for accountability toward parents.
When country dummy variables were included, the predictive value of principals' gender disappeared. Collectivism (J3=0.43, /(118)=3.04. p=.003) and the interaction between collectivism and organizational support (fi=0.27, t(118)=2.45,/>=.016) still predicted principals' accountability toward parents in the same way as without dummy variables.
Additionally, again, being from the Netherlands (/?=0.30, ?(118)=—2.78, />=.006) had a negative influence on principals' external accountability score, whereas the positive relationship for Spanish principals was marginally significant (/?=0.33 t(118)=1.94, />=.055). Dutch principals were less accountable toward school management than principals from other countries. Again, there was a considerable extra amount of variance explained when including the country dummies: 33% of the variance in principals’ accountability toward school management, 14 percent points more than in the model without country dummies.