efficiency and productivity, have their own applicability and certain limitations. For
example, it’s not necessary for DEA to consider the relative importance of input and
output indicators, while DEA still highly requires sample homogeneity, and its
results are sensitive to some outliers. Compared with DEA, SFA is much more
robust even though existing the outliers, but it’s less able to deal with production
efficiency in case of multiple outputs.
Second, the efficiency rankings obtained by different quantitative methods are
not always consistent, and there is no recognized approach or standard to judge
whether DEA or SFA is better to a specific sample (De Borger and Kerstens 1996 ).
Third, data quality of input and output is every essential to the calculations of
DEA and SFA results, but the efficiency scores are very sensitive to the indicator’s
measurement error. If these measurement errors are taken for granted, some
abnormal or counter-intuitive results might be reported. Moreover, DEA allows for
no missing data.
Forth, there is no guarantee for quantitative approach to be equitable. For
example, it’s hard for proxy indicators such as publications in SCI or SSCI journals
(a measure of research quality), or a journal’s impact factor (a measure of publi-
cation quality), to give an equitable measurement to the research performance of a
university or its faculty. The reason is that it’s impossible for every journal indexed
by SCI or SSCI to be held on to the same academic level, and even to those articles
published on the same journal, there still exist certain academic gaps. It also might
not be reasonable to take the level of awarded achievements as the indicator of
research quality, since“relationship”usually plays an important role in reviewing
these awards in China, and the level of award is not an equal interval variable,
which will result into estimation bias when adding it into econometric models or
DEA, SFA models. Instead, it would be better authority and equity, if the research
level of all targeted subjects is assessed by the experts from the samefield, i.e.
judging the research level by assessing the representative achievements submitted
by a university or its faculty.
It’s well recognized that quantitative approach and qualitative approach both
hold their own features and strengths in thefield of educational research. Currently,
a mixed research method, or the third research paradigm (Johnson and
Onwuegbuzie 2004 ), which combines quantitative and qualitative methods all
together, has become more and more popular in the international academia.
Quantitative approach based on parametric and non-parametric methods are still the
major instruments to analyze efficiency and productivity of university research
activities, but some qualitative approach should be utilized to do comparative
analysis of the typical university cases showing technical efficiency (inefficiency),
scale efficiency (inefficiency), or allocative efficiency (inefficiency). At the same
time, many other methods like interviewing university administrators, faculty, and
research fellows, doing textual analysis to internal and external policies of uni-
versity research evaluation, analyzing the micro-mechanism of the process of
university research resource allocation, exploring institutional factors’impact on
university research production process, such as personnel management, research
management, student training, property management,financial management, all
26 2 Evaluation on University Research Efficiency and Productivity...