The proportion of DEA-efficient units of non-985Usfirst decreases in 2007 and then
stays at 38% from that on. According to thefive-year mean data in the last column,
compared with the production frontier of full sample, the DEA-efficient proportion
of 985Us is lower than that of non-985Us, the difference is around 8.4%. Generally
speaking, both 985Us and non-985Us have a moderate DEA-efficient proportion,
and more than half of all universities need to improve their research efficiency.
Table4.13presents the means and SDs of TE scores for 985Us and non-985Us
in each year. In 2006–2010, the means of TE scores of 985Us and non-985Us are
0.77–0.84 and 0.80–0.82 respectively. There is no wide variation to non-985Us
across the years, while a wider variation appears to 985Us. Their means of TE
scores converge after 2008 (see Fig.4.6). According to the SD of TE scores, there
Table 4.12 Distribution of DEA-efficient units by university type (NEAM)
2006 2007 2008 2009 2010 Mean
985Us Number of
DEA-efficient units15 12 9 10 11 11.4
% DEA-efficient units 41% 32% 24% 27% 30% 30.8%
Non-985Us Number of
DEA-efficient units28 24 24 24 24 24.8
% DEA-efficient units 44% 38% 38% 38% 38% 39.2%Table 4.13 TE for NEAM disciplines by university type (2006–2010)
Types 2006 2007 2008 2009 2010
Mean
(SD)985Us 0.842
(0.185)0.833
(0.162)
0.779
(0.177)
0.814
(0.170)
0.801
(0.184)
Non-985Us 0.817
(0.227)0.800
(0.203)
0.801
(0.202)
0.811
(0.197)
0.809
(0.197)
Fig. 4.6 Changes of TE for
NEAM disciplines by
university type (2006–2010)
130 4 Evaluation on Research Efficiency of 211Us: The DEA Approach