DESIDOC Journal of Library & Information Technology, Vol. 31, No. 5, September 2011, pp. 371-376 © 2011, DESIDOC Simple Technique to Normalise Impact Factor of Journals K.C. Garg, Suresh Kumar, and Bharvi Dutt CSIR-National Institute of Science, Technology and Development Studies, New Delhi-110 012 E-mail: gargkc022@gmail.com ABSTRACT Various methods have been suggested in the literature to normalise the impact factor of journals. However, these methods have their own limitations. Present communication suggests a simple alternative method to normalise the impact factor of journals based on average impact of journals. Keywords: Impact factor, normalised impact factor, standardised impact factor 1. INTRODUCTION The impact factor of a journal is basically a ratio between citation and citable items published in a journal and indicate the relative standing and influence of the journal within its disciplinary boundaries. It is calculated by dividing the number of citations a journal receives for papers published during the last two years by the number of articles this journal published during the same time. However, Moed1, et al. and van Leeuwen & Moed2 have questioned the accuracy and validity of the impact factor of journal. The methods suggested by Sen5 & Fromter6 exclude review journals, while Marshakova-Shaikevich7 takes into consideration five journals with highest impact factor of the specialty in the calculation of the NIF. According to Pudovkin & Garfield,8 the methodologies suggested by Sen5 and Marshakova-Shaikevich7 are not quite satisfactory, as these involve either the maximal impact factor or a few of the highest impact factors in a speciality. The champion values are not always characteristics of impact factor values of the majority of journals within the specialty, and thus, introduce fortuitous elements in the NIF. Due to diversity of citing behaviour in different disciplines, Balaban3 and Makino4 argued that direct comparison between impact factor of journals dedicated to different disciplines is inadequate. Using the same rational it is not advisable to use impact factor of journals for an inter-institutional or inter-country assessment of research performance in different disciplines. To overcome this problem, use of normalised impact factor (NIF) has been suggested. Different authors5-11 have suggested different methods to normalise the impact factor of the journals. Moed9, et al. suggested normalised impact factor of journal, that takes into account both the citation characteristics in the sub-fields covered by a journal, as well as the composition of the journal in terms of types of documents, particularly ‘normal’ research articles, notes, and review articles. In the methodology suggested by Moed9, et al. one has to examine the citations of all the documents published in a journal X under the category C for a particular year. Examining the citations of all documents published in a journal is not only cumbersome but also a time-consuming process and necessitates the use of citation index. According to Sen5 the impact factor of the publishing journal is divided by the highest impact factor of the journal within the sub-field excluding review journals, which is then multiplied by a constant number. Fromter6 suggests another method, according to which the impact factor of the publishing journal is divided by the arithmetic mean of all impact factors of the category excluding review journals. Pudovkin & Garfield8 have suggested the use of ranknormalised impact factor to compare journal performance across subject categories. Egghe & Rousseau10 also suggested relative impact factor to compare the impact of journals belonging to different fields. The methods suggested by Pudovkin & Garfield8 and Egghe & Rousseau10 are simple to use but have practical difficulties. The calculation of rank-normalised impact 371 factor requires the use of Journal Citation Report (JCR) for recording the rank of the journal in the specialty. Similarly, for calculating the relative impact factor, one needs JCR for recording the citations and the source items for a specified journal in the field. In view of the above limitations in the methods suggested by different authors, a simple alternative method for calculating NIF is suggested. Table 2. Calculation of average impact facor Category {r} Number of journals Total impact factor (IF) Average impact factor (AIF) 1 86 18.163 18.163 / 86 = 0.211 2 34 25.524 25.524 / 34 = 0.751 3 19 29.932 29.932 / 19 = 1.575 2. METHODOLOGY where, PNIFj denotes the pre-normalised impact factor of the journal j; and Max (IFr) denotes maximum impact factor of corresponding category. In the suggested procedure, average impact factor of the journals in the sub-disciplines is calculated. Based on the average impact factor of the journals, the journals in the sub-discipline are divided into three categories as low impact factor journals, medium impact factor journals, and high impact factor journals. Piecewise linear mapping technique has been applied to calculate the normalised impact factor. Stepwise procedure for computations of NIF by the suggested procedure is as follows: In case of r = 1, i.e., the Ist category, the value of IFr-1 would assume the value of zero, since there is no category prior to the Ist category. Hence, in case of Ist category the values of AIFr-1 and Max (IFr-1) would be zero. Calculation for PNIF for different categories of journals given in the Appendix 1 are:  ( AIFr − AIFr −1 ) ( IF j − Max ( IFr −1 ) )   PNIFj = AIFr −1 +   Max ( IFr ) − Max ( IF r −1 )  Step 1: Arrange the journals within a sub-discipline in ascending order of impact factor (column 2 of Table 1). Step 2: Calculate the average impact factor of the journals in the sub-discipline using the formula n 1 n IF ( ) ∑ j , where, IF denotes the impact factor of j =1 the journals and n denotes the number of journals in the sub-discipline. In the set of journals given in n 1 73.619 . n IF = ( ) ∑ j the Appendix 1, n=139 and j =1 Average impact factor is thus 73.619 /139 = 0.530. Step 3: Based on the average impact factor of the journals, divide the journals into three categories as low impact factor journals, medium impact factor journals, and high impact factor journals. In the present case, low impact factor journals are those whose impact factor is ≤ average impact factor of the sub-discipline, i.e., ≤ 0.530; medium impact factor journals are those whose impact factor is more than average impact factor but less than or equal to twice the average impact factor, i.e., > 0.530 ≤ 1.06; and high impact factor journals are those whose impact factor is more than twice the average impact factor, i.e.>1.06. Step 4: Calculate the average impact factor for each category of the journals as illustrated in Table 2 for the list of journals given in the Appendix 1. Step 5: Calculate pre-normalised impact factor (PNIF) values for each journal by piecewise linear mapping technique using the following formula: ( AIFr − AIFr −1 ) ( IFj − Max ( IFr −1 ) )   PNIFj = AIFr −1 +   Max ( IFr ) − Max ( IF r −1 )  372 Example for calculation of PNIF for different categories: st ( 0.211 − 0 )( 0.012 − 0 )  (1 category ) PNIF6 = 0 + [0.514 − 0] = 0.005 Step 6: Normalise the PNIF to the scale of 10 using the following formula: PNIFj   NIFj =   *10 Max PNIF ( )   3. ADVANTAGES The advantages of using the suggested method for calculating the NIF are: (a) It does not exclude review journals or any other high impact factor journal in the calculation of NIF as suggested by Sen5 and Fromter6, and thus takes into consideration the wide variations in the range of impact factors. (b) It does not take into consideration only the journals with highest impact factor as suggested by Marshakova-Shaikevich7. (c) It does not require the tools like JCR for obtaining the ranks of the journals in a speciality as suggested by Pudovkin & Garfield8 and data on source items and number of citations as suggested by Egghe & Rousseau10 as well as the Science Citation Index for examining citations for different types of citable items as pointed out by Moed9, et al. ACKNOWLEDGEMENTS Authors are thankful to Dr Gangan Prathap, Director, National Institute of Science Communication and DESIDOC J. Lib. Inf. Technol., 2011, 31(5) Information Resources, New Delhi, for his valuable inputs and advice that has been of immense help in the preparation of the present paper. 7. Marshakova-Shaikevich, I. The standard impact factor as an evaluation tool of science fields and scientific journals. Scientometrics, 1996, 35, 283-90. REFERENCES 8. Pudovkin, A.I. & Garfield, E. Rank-normalised impact factor: A way to compare journal performance across subject categories. In American Society for Information Science and Technology Annual Meeting, 17 November 2004. http://www.garfield.library.upenn .edu/papers/asistranknormalization2004.pdf 1. Moed, H.F.; van Leeuwen, Th.N. & Reedijk, J. Towards appropriate indicators of journal impact. Scientometrics, 1999, 46, 575-89. 2. van Leeuwen, Th.N. & Moed, H.F. Development and application of journal impact measures in the Dutch science system. Scientometrics, 2002, 53, 249-66. 3. Balaban, A.T. How should citations to articles in highand low-impact journals be evaluated, of what is a citation worth? Scientometrics, 1996, 37, 495-98. 4. Makino, J. Productivity of research groups–Relation between citation analysis and reputation within research communities. Scientometrics, 1998, 43, 8793. 5. Sen, B.K. Normalized impact factor. Journal of Documentation, 1992, 48, 318-25. 6. Fromter, E., et al. Evaluierung publizierter forschungsbeitrage in der medizin. Deutsche medizinische wochenschrift,1999, 124, 910-15. In Citation rates, knowledge export and international visibility of dermatology journals listed and not listed in the Journal Citation Reports, edited by J. Stegmann & G. Grohmann. Scientometrics, 2001, 50, 483-02. DESIDOC J. Lib. Inf. Technol., 2011, 31(5) 9. Moed, H.F.; van Leeuwen, T.N. & Reedijk, J. A new classification system to describe the aging of scientific journals and their impact factors. Journal of Documentation, 1998, 54, 387-19. 10. Egghe, L. & Rousseau, R. A general framework for relative impact indicators. Canadian J. Infor. Libr. Sci., 2002, 27, 29-48. 11. Ramirez, A.M.; Garcia, E.O. & Del Rio, J.A. Renormalised impact factor. Scientometrics, 2000, 47(1), 3-9. About the Author Dr K.C. Garg presently holds the position of Scientist ‘G’ at CSIR-National Institute of Science Technology and Development Studies (NISTADS), Council of Scientific and Industrial Research, New Delhi. Before joining NISTADS in 1983, he worked at Defence Science Library, DESIDOC from 1975 to 1983. He is working in the area of scientometrics for more than 20 years and has published more than 50 papers on various aspects of scientometrics in national and international journals. 373 Appendix 1 S. No. 374 Journal IF PNIF NIF 1. Sharp Tech J 0.000 0.000 0.000 2. Izv Vuz Radioelectr 0.000 0.000 0.000 3. Electronics 0.003 0.001 0.006 4. Electron Prod 0.007 0.003 0.019 5. IFIP Trans C 0.010 0.004 0.025 6. Electron Eng 0.012 0.005 0.032 7. Siemens Rev 0.014 0.006 0.038 8. Int J Elec Eng Edu 0.014 0.006 0.038 9. Electronica 0.015 0.006 0.038 10. Control Eng 0.016 0.007 0.044 11. Telecomm Radio Eng 0.017 0.007 0.044 12. Comput Des 0.018 0.007 0.044 13. Onde Elect r 0.019 0.008 0.051 14. Electron World Wirel 0.021 0.009 0.057 15. Electron Inform Plan 0.024 0.010 0.063 16. NEC Res Dev 0.025 0.010 0.063 17. EDN 0.027 0.011 0.070 18. Electron Des 0.031 0.013 0.083 19. Electr Commun 0.035 0.014 0.089 20. Brit Telecommun Eng 0.037 0.015 0.095 21. Mic rowave Rf 0.038 0.016 0.102 22. Electr Pow Syst Res 0.044 0.018 0.114 23. Electr Mach Pow Syst 0.049 0.020 0.127 24. Fujitsu Sci Tech J 0.050 0.021 0.133 25. Hewlett Packard J 0.056 0.023 0.146 26. IEE Rev 0.062 0.025 0.159 27. IEICE T Fund Elect r 0.088 0.036 0.229 28. NTT Review 0.091 0.037 0.235 29. Int J Elect Power 0.093 0.038 0.241 30. Compel 0.101 0.041 0.260 31. IEEE T Educ 0.104 0.043 0.273 32. Eur T Elect r Pow 0.119 0.049 0.311 33. Comput Electr Eng 0.133 0.055 0.349 34. J Micriwave Power Ee 0.145 0.060 0.381 35. Mic roprocess Mic ros y 0.149 0.061 0.387 36. Mic roelectron Reliab 0.152 0.062 0.394 37. GEC – J Res 0.156 0.064 0.406 38. IEE P Commun 0.167 0.069 0.438 39. IEICE T Electron 0.170 0.070 0.444 40. Radiotekh Elektron 0.173 0.071 0.451 41. Arch Electrotech 0.181 0.074 0.470 42. J Electrostat 0.184 0.076 0.483 43. Contr Theor Adv Tec h 0.185 0.076 0.483 44. Frequenz 0.190 0.078 0.495 45. IEEE T Broadcas t 0.194 0.080 0.508 46. Int J Adapt Cont rol 0.211 0.087 0.552 47. Mic rowave J 0.213 0.088 0.559 48. Appl Artif Int ell 0.217 0.089 0.565 49. Electron Commun Eng 0.239 0.098 0.622 DESIDOC J. Lib. Inf. Technol., 2011, 31(5) 50. Analog Integr Circ S 0. 239 51. IEEE T Energy Conser 0.243 0.098 0.100 0.622 0.635 52. IEICE T Commun 0.247 0.101 0.641 53. Mechatronics 0.250 0.103 0.654 54. IEEE T Consum Electr 0.252 0.104 0.660 55. Int J Elecron 0.258 0.106 0.673 56. Electromagnet ics 0.260 0.107 0.679 57. IEEE Circuit Devic 0.274 0.113 0.717 58. IEE P- Elect Pow Appl 0.291 0.120 0.762 59. IEEE T Ind Appl 0.292 0.120 0.762 60. IEE P- Gener Transm D 0.310 0.127 0.806 61. Int J Microwave Mill 0.318 0.131 0.832 62. Mic row Opt Techn Let 0.320 0.131 0.832 63. IEEE T Power Deliver 0.346 0.142 0.902 64. Circ Syst signal Pr 0.357 0.147 0.933 65. Expert Syst Appl 0.366 0.150 0.952 66. Bt Technol J 0.370 0.152 0.965 67. AEU- Arc h Elektron Ub 0.374 0.154 0.978 68. IEEE T Instrum Meas 0.402 0.165 1.048 69. IEE P – Sci Meas Tech 0.403 0.166 1.054 70. Kvant ovaya Elektron 0.409 0.168 1.067 71. Mic roelctron Eng 0.414 0.170 1.079 72. Multidim Syst Sign P 0.419 0.172 1.092 73. Int J Soft w Eng Know 0.420 0.173 1.098 74. IEE P- Circ Dev Syst 0.424 0.174 1.105 75. Signal Proces s 0.440 0.181 1.149 76. Int J I nfrared Milli 0.442 0.182 1.156 77. J Mat er Sci- Mater El 0.443 0.182 1.156 78. IEEE T Compon Hybr 0.447 0.184 1.168 79. IEEE T Reliab 0.450 0.185 1.175 80. IEEE T Aero Elec Sys 0.459 0.189 1.200 81. IEEE T Knowl Data En 0.461 0.189 1.200 82. IEEE T Ind Electron 0.471 0.194 1.232 83. Comput Networks ISDN 0.479 0.197 1.251 84. IEE P Cont r Theor Ap 0.500 0.205 1.302 85. Concurrency Pract Ex 0.500 0.205 1.302 86. J Electromagnet Wave 0.514 0.211 1.340 87. IEEE T Circuits II 0.540 0.240 1.524 88. IEEE T Elect ronagn C 0.549 0.250 1.587 89. Solid St ate Technol 0.571 0.274 1.740 90. IEEE T Power Syst 0.577 0.280 1.778 91. IEEE T Oceanic Eng 0.577 0.280 1.778 92. IEEE T Semiconduct M 0.581 0.285 1.810 93. Math Control Signal 0.595 0.300 1.905 94. Image Vision Comput 0.602 0.308 1.956 95. IEEE Spectrum 0.623 0.331 2.102 96. Int J Circ Theor App 0.627 0.335 2.127 97. IEEE Expert 0.629 0.337 2.140 98. IEEE T Syst Man Cyb 0.649 0.359 2.279 99. J Supercomput 0.656 0.367 2.330 DESIDOC J. Lib. Inf. Technol., 2011, 31(5) 375 100. Pattern Recogn 0.691 0.405 2.571 101. Sensor Actuat A- Phys 0.704 0.420 2.667 102. IEE P – Optoelectron 0.727 0.445 2.825 103. IEEE T Circuits –I 0.732 0.450 2.857 104. Radio sci 0.753 0.473 3.003 105. IEEE T Magn 0.758 0.479 3.041 106. Solid St ate Electron 0.759 0.480 3.048 107. IEEE T Elect r Insul 0.776 0.498 3.162 108. IEEE T Veh Tec hnol 0.796 0.520 3.302 109. IEEE T Ant enn Propag 0.806 0.531 3.371 110. IEEE Commun Mag 0.840 0.569 3.613 111. IEEE T Aut omat Cont r 0.867 0.598 3.797 112. IEEE J Solid St Circ 0.903 0.638 4.051 113. IEEE T Comput 0.904 0.639 4.057 114. IEEE T Parall Distr 0.905 0.640 4.063 115. IEEE t Ultrason Ferr 0.927 0.664 4.216 116. Adv Mater Opt Electr 0.957 0.697 4.425 117. IEEE J Sel Area Comm 0.964 0.705 4.476 118. IEEE T Commun 0.969 0.710 4.508 119. IEEE T Mirowav e Theory 1.004 0.749 4.756 120. IEEE T Robotic Autom 1.006 0.751 4.768 121. Sensor Actuat B- Chem 1.074 0.785 4.984 122. IEEE T Sof tware Eng 1.117 0.807 5.124 123. Electron Lett 1.159 0.829 5.263 124. IEEE T Nucl Sci 1.183 0.841 5.340 125. Network Comp Neural 1.196 0.848 5.384 126. IEEE t Signal Proces 1.234 0.867 5.505 127. J Electron Mater 1.238 0.869 5.517 128. Opt Quant Electron 1.303 0.903 5.733 129. IEEE T Geosci Remote 1.356 0.930 5.905 130. Semicond Sci Tech 1.389 0.947 6.013 131. P IEEE 1.494 1.000 6.349 132. IEEE Electr Dev ice L 1.610 1.060 6.730 133. IEEE T Elect ron Dev 1.630 1.070 6.794 134. Prog Quant Electr 1.818 1.166 7.403 135. IEEE T Neural Network 1.941 1.229 7.803 136. IEEE T Inform Theory 1.971 1.244 7.898 137. IEEE T Pat tern Anal 2.006 1.262 8.013 138. IEEE J Quantum Electr 2.595 1.564 9.930 139. Semiconduct Semimet 2.618 1.575 10.000 Total 73.619 AIF1 = 0.211 for journals at S.No. AIF2 = 0.751 for journals at S.No. 1- 86 (86 journals) 87-120 (34 journals) AIF3 = 1.575 for journals at S.No. 121- 139 (19 journals) Source: Science Citation Index Journal Subject Category listing 1994 (Electrical and Electronic Engineering) 376 DESIDOC J. Lib. Inf. Technol., 2011, 31(5)