Results
Using the overall comfort questionnaire data from the 10 participating countries, the mean values and the distribution of the values for the four versions of the combined comfort index did not differ greatly (Table 12).
TABLE 12 Mean values and distribution of values of the four CCIs

Mean 
Standard 
Median 
Min. 
Max. 
10th 
25th 
75th 
90th 
CCI1 
3.64 
0.64 
3.75 
1.0 
5.0 
2.75 
3.25 
4.0 
4.5 
CCI2 
3.73 
0.62 
3.8 
1.0 
5.0 
2.9 
3.4 
4.1 
4.4 
CCI3 
2.64 
0.77 
2.5 
1.0 
6.0 
1.75 
2.0 
3.25 
3.75 
CCI4 
2.57 
0.79 
2.5 
1.0 
6.0 
1.7 
2.0 
3.0 
3.7 
To test the performance (informative value) of the CCIs we compared the CCI values with children’s responses to the question “Do you like your classroom?” (B1).
Table 13 shows the significant correlation coefficients found using Spearman’s correlations between the scores for question BI and the CCIs.
TABLE 13 Spearman’s correlation between scores for B1 and the CCIs
Correlation coefficient with B1 scores 

CCI1 
0.1242 
CCI2 
0.2693 
CCI3 
0.1128 
CCI4 
0.2540 
According to the results of the regression analysis, all CCIs were significantly associated with the scores for this question (even if with a low adjusted R2) (Table 14).
TABLE 14 Association of CCIs with question B1

Coefficient 
Standard error 
adjR2 
P>(t) 
95% conf. intervals 
CCI1 
0.39416 
0.02696 
0.0312 
0.0000 
0.34131 0.44702 
CCI2 
0.68242 
0.02599 
0.0945 
0.0000 
0.63147 0.73336 
CCI3 
0.28879 
0.02031 
0.0296 
0.0000 
0.32861 0.24899 
CCI4 
0.50733 
0.01967 
0.0915 
0.0000 
0.54589 0.46876 
The following two conclusions can be drawn:
 The weighted indexes performed better than the unweighted ones.
 The original scoring system (CCI1 and CCI2) was better than the scoring weighted for extreme perceptions.
At the meeting of Italian and Hungarian experts held in Budapest on March 4, 2013, it was decided that extreme values give an unclear picture, thus only CCI1 and CCI2 were considered in the following procedures.
The following observations were also made:
 Statistical analysis showed that both CCI1 and CCI2 scores were significantly different according to gender and age.
 There was significant heterogeneity between CCI1 and CCI2 among countries.
 Both CCI1 and CCI2 were significantly associated with headaches among children (Question B7b). There was not much difference in the correlation coefficients between the two CCIs (CCI1: 0.1245; CCI2: 0.1353)
 Spearman’s correlation analysis between CCIs and the measured values for temperature, relative humidity and CO_{2} concentrations are shown in Table 15. Coefficients were highest for (and statistically significant only in the case of) relative humidity.
TABLE 15 Spearman’s correlation between CCIs and measured values

Temperature (°C) 
Relative humidity (%) 
CO_{2} (ppm) 
CCI1 
0.0314 
0.0908*** 
0.0337 
CCI2 
0.0063 
0.1346*** 
0.0073 
*** p<0.001