1 See Bagaric, M. and Morss, J., ‘International Sentencing Law: In Search of a Justification and Coherent Framework’, (2006) 6 International Criminal Law Review 191, at 255Google Scholar; Endo, G., ‘Nullum Crimen Nulla Poena Sine Lege Principle and the ICTY and ICTR’, (2002) 15 Revue québécoise de droit international 205, at 205Google Scholar; Harmon, M. B. and Gaynor, F., ‘Ordinary Sentences for Extraordinary Crimes’, (2007) 5 Journal of International Criminal Justice 683, at 684Google Scholar.

2 UN Doc. S/RES 827 (1993).

3 Prosecutor v. Delalić, Mucić, Delić and Landžo, Judgement, Case No. IT-96–21-A, App. Ch., 20 February 2001, para. 756.

4 Statute of the International Criminal Tribunal for the Former Yugoslavia adopted 25 May 1993 by Resolution 827, supra note 2 (hereinafter ICTY Statute).

5 Rules of Procedure and Evidence of the International Criminal Tribunal for the Former Yugoslavia, UN Doc. IT/32/Rev. 37 (1994) (hereinafter Rules), Rules 100 and 101.

6 ICTY Statute, supra note 4, Art. 24; Rules, supra note 5, Rule 101.

7 ICTY Statute, supra note 4, Art. 24(3).

8 Ibid., Art. 24(2).

9 Prosecutor v. Delalić, Mucić, Delić and Landžo, Judgement, Case No. IT-96–21-T, T.Ch. II, 16 November 1998, para. 1225.

10 Prosecutor v. Josipović, Santić, Z. Kupreškić, M. Kupreškić, V. Kupreškić and Papić, Judgement, Case No. IT-95–16-T, T.Ch. II, 14 January 2000, para. 852.

11 See Prosecutor v. Haradinaj, Balaj and Brahimaj, Judgement, Case No. IT-04–84-T, T.Ch. I, 3 April 2008, para. 489.

12 In contrast, Rule 145 of the Rules of Procedure and Evidence of the International Criminal Court includes a demonstrative list of mitigating and aggravating factors. The mitigating and aggravating factors included in the list actually reflect the ICTY and ICTR case law.

13 Rules, supra note 5, Rule 101.

14 ICTY Statute, supra note 4, Art. 7(4).

15 Prosecutor v. Stakić, Judgement, Case No. IT-97–24-T, T.Ch. II, 31 July 2003, para. 911.

16 Prosecutor v. Kunarac, Kovač and Vuković, Judgement, Case No. IT-96–23&23/1-T, T.Ch. I, 22 February 2001, para. 911.

17 The SFRY sentencing practice is, however, one of the factors judges should take into account. If a sentence departs from limits set in this practice, reasons must be given for such a departure. See Prosecutor v. Dragan Nikolić, Judgement, Case No. IT-94–2-A, App. Ch., 4 February 2005, para. 69.

18 Prosecutor v. Aleksovski, Judgement, Case No. IT-95–14/1-A, App. Ch., 24 March 2000, para. 185.

19 Prosecutor v. Kordić and Čerkez, Judgement, Case No. IT-95–14/2-A, App. Ch., 17 December 2004, para. 1079.

20 Prosecutor v. Nikolić, Judgement, Case No. IT-02–60/1-T, T.Ch. I, 2 December 2003, paras. 58–60.

21 Prosecutor v. Nikolić, Judgement, Case No. IT-94–2-S, T.Ch. II, 18 December 2003, para. 139.

22 Sentence determination should only be based on legally relevant factors. It should never be affected by consideration of the so-called extra-legal factors such as the ethnicity, race, or gender of the offender. Cf. A. Ashworth, ‘Sentencing’, in M. Maguire, R. Morgan, and R. Reiner (eds.), The Oxford Handbook of Criminology (2007), 990 at 1003; see note 23, infra.

23 The legal factors are those considerations that should play a decisive role in the sentence determination. They are derived from international law, the ICTY Statute, and the Rules, and include factors such as nature of the crime, degree of responsibility, or aggravating or mitigating circumstances. In contrast, the extra-legal factors are factors not regulated or permitted by law. They should not influence the sentence decision-making in any way. Examples of these legally irrelevant considerations are, e.g., ethnicity of the offender, professional background or personal characteristics of judges, or political factors.

24 Meernik, J. and King, K. L., ‘The Effectiveness of International Law and the ICTY – Preliminary Results of an Empirical Study’, (2001) 1 International Criminal Law Review 343Google Scholar; J. Meernik and K. L. King, ‘The Sentencing Determinants of the International Criminal Tribunal for the Former Yugoslavia: An Empirical and Doctrinal Analysis’, (2003) 16 LJIL 717; Meernik, J., ‘Victor's Justice or the Law? Judging and Punishing at the International Criminal Tribunal for the Former Yugoslavia’, (2003) 47 Journal of Conflict Resolution 140Google Scholar; Meernik, J., King, K. L., and Dancy, G., ‘Judicial Decision Making and International Tribunals: Assessing the Impact of Individual, National and International Factors’, (2005) 86 Social Science Quarterly 683Google Scholar.

25 Meernik, supra note 24, at 159; Meernik, King, and Dancy, supra note 24, at 701.

26 There is only one study where solely the interplay between sentence length and the legal factors is examined. However, the number of legal factors included in the study is rather limited. See Meernik, supra note 24; in all the other studies legal factors are combined with extra-legal, such as ethnicity, national background of judges, or judges’ gender. In all these studies, therefore, the final result may be affected by the inclusion of these extra-legal factors.

27 The data were collected up to August 2008.

28 This number includes all the final judgments and all cases where sentence was handed down by the trial chamber but the case is still pending on appeal.

29 Specific reasons for selecting these particular variables are provided further in the text; see section 2.2., infra.

30 Available at www.un.org/icty/cases-e/index-e.htm.

31 The mean sentence is the arithmetic mean of all sentences (mathematical average). The median sentence is the sentence which lies exactly in the middle of the sentence distribution – half of the issued sentences lies above the median and half below. As opposed to mean, median is not influenced by extreme sentences.

32 In statistics, loosely speaking, a result is significant when it is unlikely to occur by chance.

33 L. S. Meyers, G. Gamst, and A. J. Guarino, Applied Multivariate Research – Design and Interpretation (2006), at 147–50.

34 Normal distribution was checked by inspecting a histogram of the standardized residuals. Homoscedasticity was checked by inspecting a scatterplot of residuals versus predicted values of the sentence. Independence of residuals was tested by Durbin Watson statistic. Its value equalled 1.906, indicating no first-order autocorrelation given the number of predictors in the model and the sample size. Multicollinearity among variables was also checked, using variance inflation factor (VIF) – the highest value was less than 1.9, thus meeting tests of acceptability, which are normally set around 10. Ibid., at 198.

35 The category of war crimes includes crimes under Art. 2 of the Statute and crimes under Art. 3 of the Statute without distinguishing between these two. The category of crimes against humanity includes crimes under Art. 5 of the Statute. Genocide includes crimes under Art. 4 of the Statute.

36 Prosecutor v. Kayishema and Ruzindana, Judgement, ICTR-95–1-T, T.Ch. II, 21 May 1999, para. 9.

37 Prosecutor v. Furundzija, Judgement, Case No. IT-95–17/1-A, App. Ch., 21 July 2000, para. 243.

38 In contrast, in the early ICTY cases it was accepted that ‘an act committed as a part of a crime against humanity, [i]s all else being equal, a more serious offence than an ordinary war crime’. Prosecutor v. Tadić, Sentencing Judgement, Case No. IT-94–1-T, T.Ch. II, 14 July 1997, para. 73; Prosecutor v. Erdemović, Case No. IT-96–22-A, App. Ch., 7 October 1997, Joint Separate Opinion of Judges McDonald and Vohrah, paras. 20–22; This approach was subsequently condemned on appeal in Prosecutor v. Tadić, Judgement in Sentencing Appeals, Case No. IT-94–1-A, App. Ch., 26 January 2000, para. 69.

39 See Carcano, A., ‘Sentencing and the Gravity of the Offence in International Criminal Law’, (2002) 51 International and Comparative Law Quarterly 583, at 607–9Google Scholar; Olusanya, O. A., ‘Do Crimes against Humanity Deserve a Higher Sentence than War Crimes?’, (2004) 4 International Criminal Law Review 431Google Scholar.

40 M. Frulli, ‘Are Crimes against Humanity More Serious than War Crimes?’, (2001) 12 EJIL 329, at 333–8.

41 Ibid.

42 Prosecutor v. Krstić, Case No. IT-98–33.

43 Del Ponte, C., ‘Prosecuting the Individuals Bearing the Highest Level of Responsibility’, (2004) 2 Journal of International and Comparative Law 516Google Scholar.

44 Cf. M. A. Drumbl, Atrocity, Punishment and International Law (2007), at 25; A. Smeulers, ‘Perpetrators of International Crimes: Towards a Typology’, in A. Smeulers and R. Haveman (eds.), Supranational Criminology: Towards a Criminology of International Crimes (2008), at 233.

45 Prosecutor v. Babić, Judgement, Case No. IT-03–72-A, App. Ch., 18 July 2005, para. 80.

46 The classification of JCE as a principal or derivative mode of responsibility is contentious in theory and not really clear and settled in the case law. This fact relates to its combined common/civil law origin. It can also be argued that participants in the first and second category of JCE can be seen as co-perpetrators falling under the category of primary modes of responsibility. For an interesting discussion thereof see van Sliedregt, E., ‘Joint Criminal Enterprise as a Pathway to Convicting Individuals for Genocide’, (2006) 5 Journal of International Criminal Justice 184CrossRefGoogle Scholar.

47 E. van Sliedregt, The Criminal Responsibility of Individuals for Violations of International Humanitarian Law (2003).

48 Werle, G., ‘Individual Criminal Responsibility in Article 25 ICC Statute’, (2007) 5 Journal of International and Comparative Law 953, at 955Google Scholar.

49 The only low-ranking offender who was convicted for instigation by the ICTY was Zoran Žigić, a guard at the Keraterm camp, who was convicted inter alia for committing, instigating, and aiding and abetting serious crimes in Keraterm and Trnopolje camps. See Prosecutor v. Kvočka et al., Judgement, Case No. IT-98–30/1-T, T.Ch. I, 2 November 2001, para. 747. All the other offenders held guilty of the planning, ordering, or instigation of crimes were indeed middle- or high-ranking individuals.

50 S. M. Sayers, ‘Defence Perspectives on Sentencing Practice in the International Criminal Tribunal for the Former Yugoslavia’, (2003) 16 LJIL 751, at 768.

51 This number includes also the appeal proceedings following acquittals of Ramush Haradinaj, Idriz Balaj, and Ljube Boskoski on trial. See Prosecutor v. Haradinaj, Brahimaj and Balaj, Judgement, Case No. IT-04–84, T.Ch. I, 3 April 2008; Prosecutor v. Boskoski and Tarculovski, Judgement, Case No. IT-04–82, T.Ch. II, 10 July 2008.

52 In this article all the numbers have been rounded to one decimal place.

53 See note 28, supra; these numbers include all the final judgments and all cases where sentence was handed down by the trial chamber but the case is still pending on appeal. For the purposes of statistical analysis life sentences were recoded to reflect the expected duration of the imprisonment term in each individual case, primarily taking into account the age of a sentenced person (a life sentence has been handed down to two defendants so far – to Milomir Stakić at trial and Stanislav Galić on appeal; it has been recoded in the following way: Stakić: life →30 years; Galić: life → 20 years).

54 See Prosecutor v. Hadzihasanović and Kubura, Judgement, Case No. IT-01–47-A, App. Ch., 22 April 2008; Prosecutor v. Orić, Case No. IT-03–68; Prosecutor v. Stakić, Case No. IT-97–24; Prosecutor v. Galić, Case No. IT-98–29.

55 Prosecutor v. Galić, Case No. IT-98–29; Prosecutor v. Aleksovski, Case No. IT-95–14/1; Prosecutor v. Krnolejac, Case No. IT-97–25.

56 In the following analysis only the finalized cases have been included and compared, including cases of appeal acquittals (trial N = 60, appeal N = 56). If cases pending on appeal had also been included, the comparisons might have been biased. In cases of the life sentences handed down by the ICTY to two defendants so far (Milomir Stakić and Stanislav Galić), the sentence length was recoded. See note 53, supra.

57 For the relationship between the rank of the offender and these modes of responsibility see note 49 and accompanying text, supra.

58 Sliedregt, supra note 46, at 196.

59 Sliedregt, supra note 47, at 106.

60 Cf. Prosecutor v. Vasiljević, Judgement, Case No. IT-98–32-A, App. Ch., 25 February 2004; Prosecutor v. Krstić, Judgement, Case No. IT-98–33-A, App. Ch., 19 April 2004; Prosecutor v. Simić, Tadić and Zarić, Judgement, Case No. IT-95–9-A, App. Ch., 28 November 2004.

61 See section 2.1 ‘Methods’, supra; we are aware of the fact that, statistically, using 14 predictors for 63 cases is stretching the technique. On one hand, due to obvious practical limitations – ICTY has issued sentences to only 63 individuals so far – we were not able to enlarge our sample; in fact, we are analysing the entire population of ICTY sentenced offenders. On the other hand, we wanted to include all possible important legal determinants of sentence length in order to provide the reader with the complete picture (e.g. the category of mode of responsibility in itself consists of seven separate variables – it is impossible to leave out any of them). Reassuringly, despite the limited number of cases compared with the number of included predictors, the resulting model turned out to be stable across different sets of predictors.

62 The offenders were divided into categories on the basis of the ‘most serious crime’ of which they were convicted. Therefore the category of crimes against humanity includes also those found guilty of multiple acts under a combination of war crimes and crimes against humanity.

63 Genocide as a separate category was not included in the analysis because so far the ICTY has sentenced only one person for genocide.

64 So far the ICTY has finalized 29 cases, with defendants convicted either for war crimes or crimes against humanity. When the mean sentences are compared, those convicted solely of crimes against humanity are sentenced to longer terms of imprisonment (N = 16, M = 11.31, SD = 5.35) than those found guilty of war crimes (N = 13, M = 9.08, SD = 4.55).

65 All the other legal factors analysed have been excluded from the analysis as not being significant predictors of sentence length.

66 The word ‘trend’ is emphasized because the coefficient for these predictors just missed the level of statistical significance.

67 Cf. Brantingham, P. L., ‘Sentencing Disparity: An Analysis of Judicial Consistency’, (1985) 1 Journal of Quantitative Criminology 281Google Scholar; for brief overview of similar studies see Hofer, P. J., Blackwell, K. R., and Ruback, R. B., ‘The Effect of the Federal Sentencing Guidelines on Inter-judge Sentencing Disparity’, (1999) 90 Journal of Criminal Law and Criminology 239, at 242–3Google Scholar; B. J. Ostrom, C. W. Ostrom, R. A. Hansom, and M. Kleiman, ‘Assessing Consistency and Fairness in Sentencing: A Comparative Study in Three States’, National Center for State Courts, available at www.ncsconline.org/images/PEWExecutiveSummaryv10.pdf.

68 Cf. Ostrom et al., supra note 67.

69 See note 1, supra.

70 Cf. Harmon and Gaynor, supra note 1, at 710; Sloane, R. D., ‘The Expressive Capacity of International Punishment: The Limits of the National Law Analogy and the Potential of International Criminal Law, (2007) 43 Stanford Journal of International Law 39, at 89, 91Google Scholar; Cassese, A., ‘The ICTY: A Living and Vital Reality, (2004) 2 Journal of International Criminal Justice 585, at 596Google Scholar.