The results of TIF and SEM showed that measurement
accuracy of both versions was not similar. The amount of
information in the Japanese version decreased sharply
at the level of less than -2 (Figure 2), meaning that the
Japanese version had difficulty in differentiating respondents
with extremely low work engagement. On the
other hand, the amount of information in the original
Dutch version decreased gradually at the level of more
than 1 (Figure 3), meaning that the original version had
difficulty in differentiating respondents with high work
engagement.
Figure 1 Comparison of UWES-9 scores between Japan and 15 countries. Note. All comparisons were significant at the 0.1% level
(Bonferroni correction for multiple comparison was applied).
Shimazu et al. BioPsychoSocial Medicine 2010, 4:17
http://www.bpsmedicine.com/content/4/1/17
Page 4 of 6