vol. 14 no. 4, December, 2009  
The publication activity in a Swedish region 19982006 has been analysed in a recent publication (Jarneving 2009). The article provides a clear description of the methods used. However, whole counting has been used for publications and citations without due regard to the properties of this method. The publication also contains additional smaller methodological problems. Therefore, some of the results and conclusions in the publications are incorrect. This criticism is based on the following analysis.
Under the heading Counting Schemes it is stated that there are three standard counting methods.
with reference to Gauffriau et al. (2007). However, this is not in agreement with the results described in this reference. There are five, not three fundamental counting methods. In addition for each method the object of study and the basic unit of analysis must be stated. Both can for example be persons, institutions, regions and countries.
Jarneving adds: 'Fractional counting implies a difficulty as there is no clear correspondence between authors and corporate addresses'. This is correct but is only relevant if fractionation based on author numbers is considered. Fractional counting based on numbers of institutions or countries presents no difficulties.
Whole counting is the method used in the publication. Both the objects of study and the basic units of analysis are institutions, regions and countries (all the basic units are derived from the addresses in the publications). The choice of whole counting is supported by a reference to the European Commission (2003). However, OECD from 2001 has been using fractional counting in its R&D statistics. The OECD statistics depend on data from National Science Foundation (2000), also based on fractional counting.
It is stated, 'Furthermore, empirical experiences have shown that at the macro level, all three methods may yield satisfactory results, but a lower levels of aggregation, the whole counting scheme is more appropriate', with reference to Glänzel (1996). This is a distortion of the reference. The reference contains no empirical evidence whatsoever and, on counting methods, states: 'Whilst at the macrolevel all three methods may yield satisfactory results, at lower levels of aggregation firstaddress counts proved to be inappropriate and especially at the microlevel full [whole] counts should be preferred to fractional counts'. The choice of counting methods must be based on the questions addressed and different methods can provide widely different results (Gauffriau et al. 2008).
Of special importance is the fact that results obtained by whole counting are nonadditive. This fact was shown already by Anderson et al. in 1988 (compare Gauffriau et al. 2007). This means that the counting numbers for the institutions in a region cannot be added to provide the value for the region and that the counting numbers for the regions in a country cannot be added to provide the value for the country. Because of multi authorship and cooperation between institutions and regions the sums will always be larger than the values for the corresponding regions or countries.
The activity index defined by Jarneving (the country's share in world's publication output in the given field divided by the country's share in world's publication output in all science fields) is problematic with the counting method chosen. Only if the region studied has exactly the same extent of cooperation with scientists outside the region in all fields will the values be meaningful. However, in the section Research Collaboration this is shown not to be the case.
In the section Data sources and data processing 24,823 records from 19982006 are given for the region studied from 19982008 whereas the number for Sweden is 131,544 records.
In the section The regional article production it is stated that during the observation there was a linear growth for both the region and the nation. This statement is surprising and not documented. Exponential growth rate might be expected as already shown is Price's pioneering work (1961, 1963).
In the section The regional article production it is stated that the region contributes approximately 19% yearly (to Swedish scientific publication) (24,823 divided by 131,823 and multiplied by 100). However, with the counting method used the sum for all Swedish regions must exceed 100%. On the other hand it would be correct to state that the region contributed to or participated in 19% of all scientific publications with Swedish participation.
Figure 1 records the national shares in 11 broad fields and for the national total. It is written that 'The shares are based on the cumulated number of articles. Multiple counting was applied for the different broad fields while a single count was applied when computing shares of the national total'. This must mean that if a paper was counted in two fields it was only counted once for the total. In any event, again, the shares for the Swedish regions counted with the method used must add up to more than 100%.
Figure 1 shows that the share for the Region Västra Götaland of the national total is increasing slightly during the period studied. This may be caused only by increased cooperation with other regions and countries. In fact the counting method used makes it possible for all regions to increase their shares even if the total number of publications is constant (Anderson et al. 1988).
In Table 1 the percentages of production in different fields for the different institutions in the region are recorded. The problem about nonadditivity is clearly displayed by the values for Physics and Astronomy and for Mathematics and Statistics. In Physics and Astronomy two institutions are indicated to have percentages of 85 and 39%, together 114%. For Mathematics and Statistics the values are 82 and 35%, together 117%. The values are used for rankings. Again, it will be possible for all institutions to increase their shares over time by increasing cooperation, even if the total output of publications is constant.
Support from the Carlsberg Foundation is gratefully acknowledged.
Find other papers on this subject  
© the author, 2009. Last updated: 4 November, 2009 
