Hostname: page-component-5db6c4db9b-mcx2m Total loading time: 0 Render date: 2023-03-23T23:29:45.180Z Has data issue: true Feature Flags: { "useRatesEcommerce": false } hasContentIssue true

How and Why Do Judges Cite Academics? Evidence from the Singapore High Court

Published online by Cambridge University Press:  18 May 2022

Jerrold Tsin Howe SOH*
Singapore Management University
Yihan GOH
Singapore Management University
Corresponding author. E-mail:
Get access


Legal academics were once thought to be parasitic on the work of judges, so much so that citing academic work was said to weaken a judgment's authority. Recent times have however seen prominent academics appointed to the highest courts, and judicial engagement with academic materials appears to have increased. In this light, this article empirically studies academic citation practices in the Singapore High Court. Using a dataset of 2,772 first-instance High Court judgments, we show that citation counts have indeed increased over time. This increase was distributed across most legal areas, and was not limited to, though more pronounced in, judgments authored by judges with post-graduate law degrees. Books, not journal articles, have consistently accounted for the bulk of the court's citations. The study sheds new statistical light on the evolving relationship between judges and academics, particularly in the context of an Asian, first-instance court.

Copyright © The Author(s), 2022. Published by Cambridge University Press on behalf of the National University of Singapore

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)



Assistant Professor of Law, Yong Pung How School of Law, Singapore Management University.


Advocate & Solicitor (Singapore), Senior Counsel. Dean and Professor of Law, Yong Pung How School of Law, Singapore Management University, Singapore. The first-named author was responsible for a significant majority of this work. In particular, he collected, analysed, and tabulated all empirical data and results therein, as well as reviewed the qualitative literature. The second-named author assisted in the review of the qualitative literature. We thank Jeremy Chai for his research assistance as well as Goh Yu Xuan for assistance with the dataset. This research is funded by the Singapore Judicial College under its Empirical Judicial Research initiative.


1 See eg, Legal Profession (Professional Conduct) Rules 2015 (Singapore), r 9(3)(a).

2 See the section immediately following this for a literature review.

3 John Henry Merryman, ‘Towards a Theory of Citations: An Empirical Study of the Citation Practice of the California Supreme Court in 1950, 1960, and 1970’ (1978) 50 Southern California Law Review 381, 419.

4 John Henry Merryman, ‘The Authority of Authority: What the California Supreme Court Cited in 1950’ (1954) 6 Stanford Law Review 613; John Henry Merryman, ‘Towards a Theory of Citations: An Empirical Study of the Citation Practice of the California Supreme Court in 1950, 1960, and 1970’ (1978) 50 Southern California Law Review 381.

5 William M Landes & Richard A Posner, ‘Legal Precedent: A Theoretical and Empirical Analysis’ (1976) 19 Journal of Law and Economics 249. See also William M Landes, Lawrence Lessig & Michael E Solimine, ‘Judicial Influence: A Citation Analysis of Federal Courts of Appeals Judges’ (1998) 27 Journal of Legal Studies 271; Stephen J Choi & Mitu G Gulati, ‘Ranking Judges According to Citation Bias (As a Means to Reduce Bias)’ (2007) 82 Notre Dame Law Review 1279.

6 A statistical model supplied only with six political and ideological factors correctly predicted 75% of the US Supreme Court's affirm/reverse decisions. A team of legal experts predicting the same cases using conventional legal analysis were 59.1% accurate. Theodore W Ruger et al, ‘The Supreme Court Forecasting Project: Legal and Political Science Approaches to Predicting Supreme Court Decisionmaking’ (2004) 104 Columbia Law Review 1150. Network scientists have also shown that US Supreme Court justices tend to vote in blocks. See Roger Guimerà & Marta Sales-Pardo, ‘Justice Blocks and Predictability of U.S. Supreme Court Votes’ (2011) 6 Public Library of Science One e27188.

7 Frankenreiter, Jens, ‘The Politics of Citations at the ECJ – Policy Preferences of E.U. Member State Governments and the Citation Behavior of Judges at the European Court of Justice’ (2017) 14 Journal of Empirical Legal Studies 813CrossRefGoogle Scholar.

8 Wilson, Geoffrey, ‘English Legal Scholarship’ (1987) 50 Modern Law Review 818CrossRefGoogle Scholar.

9 Andrew Phang, Yihan Goh & Jerrold Soh, ‘The Development of Singapore Law: A Bicentennial Retrospective’ (2020) 32 Singapore Academy of Law Journal 804, 832.

10 Cheah Wui Ling & Goh Yihan, ‘An Empirical Study on the Singapore Court of Appeal's Citation of Academic Works: Reflections on the Relationship between Singapore's Judiciary and Academia’ (2017) 29 Singapore Academy of Law Journal 75

11 See eg, Caldeira, Gregory, ‘The Transmission of Legal Precedent: A Study of State Supreme Courts’ (1985) 79 American Political Science Review 179CrossRefGoogle Scholar.

12 Smyth, Russell, ‘What do Intermediate Appellate Courts Cite? A Quantitative Study of the Citation Practice of Australian State Supreme Courts’ (1999) 21 Adelaide Law Review 51, 53Google Scholar.

13 Wilson, Geoffrey, ‘English Legal Scholarship’ (1987) 50 Modern Law Review 818CrossRefGoogle Scholar.

14 Bridge, John, ‘The Academic Lawyer: Mere Working Mason or Architect?’ (1975) 91 Law Quarterly Review 488, 493Google Scholar.

15 Alexandra Braun, ‘Judges and Academics: Features of a Partnership’, in James Lee (ed), From House of Lords to Supreme Court: Judges, Jurists and the Process of Judging (Hart Publishing 2011).

16 Susan Kiefel CJ, ‘The Academy and the Courts: What Do They Mean to Each Other Today?’ (2020) 44 Melbourne University Law Review 447, 449.

17 ibid 449. However, even within civilian jurisdictions, there are differences. Thus, the stricter separation of powers in France mean that the courts should only be seen as enforcing the law, not expounding it. This has led to the exclusion of the citation of any secondary materials from the text of judgments.

18 Braun (n 15).

19 Keith Stanton, ‘Use of Scholarship by the House of Lords in Tort Cases’, in James Lee (ed), From House of Lords to Supreme Court: Judges, Jurists and the Process of Judging (Hart Publishing 2011).

20 Andrew Burrows, ‘Challenges for Private Law in the 21st Century’ (Oxford Legal Studies Research Paper 3/2016, 3 Jan 2016) 4 <> accessed 14 Jun 2021.

21 ibid.

22 Kiefel CJ (n 16) 457.

23 ibid 457.

24 Ulen, Thomas S, ‘The Unexpected Guest: Law and Economics, Law and Other Cognate Disciplines, and the Future of Legal Scholarship’ (2009) 79 Chicago-Kent Law Review 403, 414Google Scholar.

25 Asit K Biswas & Julian Kirchherr, ‘Prof, No One is Reading You’ (The Straits Times, 11 Apr 2015) <> accessed 7 Apr 2022. Note however that Biswas and Kirchherr do not cite a source for these numbers. The scientific literature on this is mixed. See Mark Ware & Michael Mabe, The STM Report: An Overview of Scientific and Scholarly Journal Publishing (International Association of Scientific, Technical and Medical Publishers 2012) 32 < > accessed 26 Aug 2021.

26 See eg, Kiefel CJ (n 16) 454.

27 Jack Beatson, ‘Legal Academics: Forgotten Players or Interlopers?’, in Andrew Burrows, David Johnston & Reinhard Zimmermann (eds), Judge and Jurist: Essays in Memory of Lord Rodger of Earlsferry (Oxford University Press 2013).

28 Patrick Devlin, ‘Statutory Offences’ (1958) 4 Journal of the Society of Public Teachers of Law 206.

29 Beatson (n 27).

30 Smyth, Russell, ‘Academic Writing and the Courts: A Quantitative Study of the Influence of Legal and Non-legal Periodicals in the High Court’ (1998) 17 University of Tasmania Law Review 164Google Scholar.

31 C de B Murray 1950 SLT 1, 2, cited by Kenneth Reid, ‘The Third Branch of the Profession’, in Hector MacQueen & William Wilson (eds), Scots Law into the 21st Century: Essays in Honour of WA Wilson (Sweet and Maxwell 1996).

32 Robert Goff, ‘The Search for Principle’ (Maccabaean Lecture in Jurisprudence (5 May 1983), The British Academy 1984).

33 [1987] AC 460.

34 Spiliada Maritime Corp v Cansulex Ltd [1987] AC 460, 488.

35 Birks, Peter, ‘The Academic and the Practitioner’ (1998) 18 Legal Studies 387, 399CrossRefGoogle Scholar.

36 Kiefel CJ (n 16) 454.

37 ibid 456.

38 Smyth, Russell, ‘Other Than ‘Accepted Sources of Law’?: A Quantitative Study of Secondary Source Citations in the High Court’ (1999) 22 University of New South Wales Law Journal 19, 22Google Scholar.

39 ibid 22.

40 ibid 22.

41 Beatson (n 27).

42 [1995] 2 AC 145.

43 Henderson v Merrett Syndicates Ltd [1995] 2 AC 145, 184.

44 Smyth, ‘Other Than ‘Accepted Sources of Law’? (n 38) 23.

45 ibid 23.

46 Beatson (n 27).

47 Cordell v Second Clanfield Properties Ltd [1969] 2 Ch 9, 16.

48 Beatson (n 27).

49 Russell Smyth, ‘Other Than ‘Accepted Sources of Law’?’ (n 38) 23.

50 Lord Neuberger, ‘Judges and Professors – Ships Passing in the Night?’ (Lecture at the Max Planck Institute, Hamburg, 9 Jul 2012) <> accessed 14 Jun 2021.

51 Smith, John, ‘An Academic Lawyer and Reform’ (1981) 1 Legal Studies 119, 120121CrossRefGoogle Scholar.

52 Duxbury, Neil, ‘When We Were Young: Notes in the Law Quarterly Review’ (2000) 116 Law Quarterly Review 474Google Scholar.

53 Newark, FH, ‘The Boundaries of Nuisance’ (1949) 65 Law Quarterly Review 480Google Scholar.

54 Keith Stanton, ‘Use of Scholarship by the House of Lords in Tort Cases’, in James Lee (ed), From House of Lords to Supreme Court: Judges, Jurists and the Process of Judging (Hart Publishing 2011).

55 Smyth, ‘Other Than ‘Accepted Sources of Law’?’ (n 38) 24.

56 ibid 24.

57 Singapore Courts, ‘Role and structure of the State Courts’ <> accessed 1 Apr 2022.

58 Singapore Courts, ‘Media Release: Structural reforms to the High Court and appointment of Judges of the Appellate Division from 2 January 2021’ (18 Dec 2020) <> accessed 1 April 2022.

59 See generally, Leong Sze Hian v Lee Hsien Loong [2019] 2 SLR 591 paras 6–7.

60 [2020] SGHC 63 para 143.

61 Canada (Attorney General) v Bedford [2013] 3 SCR 1101.

62 Wong Hong Toy v PP [1985–1986] SLR(R) 656 para 11; Attorney-General v Shadrake Alan [2011] 2 SLR 445.

63 Lord Neuberger, ‘Some thoughts on the post-LASPO civil judge's role before and during trial’ (Address to the Manchester Law Society and Northern Circuit Commercial Bar Association, 22 Jan 2015) para 5 <> accessed 7 Apr 2022.

64 Smyth, ‘What do Intermediate Appellate Courts Cite?’ (n 12) 69.

65 ibid 69.

66 ibid 69, citing Judith Kaye, ‘One Judge's View of Academic Law Review Writing’ (1989) 39 Journal of Legal Education 313, 319.

67 Kiefel CJ (n 16) 455.

68 Singapore Courts, ‘One Judiciary Annual Report 2019’ <> accessed 1 Apr 2022.

69 Kiefel CJ (n 16) 455.

70 Richard Posner, ‘An Economic Analysis of the Use of Citations in the Law’ (2000) 2 American Law and Economics Review 381, 383–386.

71 These include only judgments identified by LawNet as originating from the High Court of Singapore, the High Court of Singapore (Family Division), and the Court of Three Judges.

72 A 2017 end date was adopted for primarily practical reasons: because data collection for this work began in (late) 2018, and given the typical lags between a court's decision the judgment publication, only a 2017 cut-off would let us to capture a full year's worth of judgments.

73 Given the focus on the relationship between legal academia and the judiciary, ‘academic citations’ as used here broadly excludes professional publications, committee reports, command papers, the Halsbury's volumes, encyclopaedias, dictionaries, though it is acknowledged that many are written or co-written by legal academics. It further excludes all non-legal materials, including non-legal academic journals (such as medical texts sometimes referenced in medical cases).

74 The first-named author is a co-founder of the firm and personally worked on the data collection. The second-named author worked closely with the firm throughout the process. The first-named author's interests in the firm, as well as the firm's engagement for this research project, pre-dates the first-named author's involvement in authoring this article and indeed their appointment to academia. The firm has worked on data collection for projects such as Hans Tjio, ‘An Empirical Look at the Consequences of Oppression in Singapore’ (2017) 17 Journal of Corporate Law Studies 405; Hans Tjio, ‘Restructuring the Bond Market in Singapore’ (2019) 14 Capital Markets Law Journal 16; Simon Chesterman, ‘Do Better Lawyers Win More Often? Measures of Advocate Quality and Their Impact in Singapore's Supreme Court’ (2020) 15 Asian Journal of Comparative Law 250.

75 Rules-based extraction of legal citations remains a research problem. Some issues, relevant literature, and proposed approaches are outlined in Akshita Gheewala, Chris Turner & Jean-Rémi de Maistre, ‘Automatic Extraction of Legal Citations using Natural Language Processing’ (15th International Conference on Web Information Systems and Technologies, Austria, Sep 2019) 202–209.

76 For instance, consider the following paragraph from Spandeck Engineering (S) Pte Ltd v Defence Science & Technology Agency [2007] 4 SLR(R) 100 para 67:

In relation to the reasons traditionally given as justification for a different approach, Robby Bernstein points out in Economic Loss (Sweet & Maxwell, 2nd Ed, 1998) (at p 21) that it is important to note that it is not anything inherent in the type of the damage itself that necessitates a different approach…

(italics in original)

It is far from trivial to devise a rule capable of identifying where the citation begins (and ends). In particular, the bridging words ‘points out in’ mean that a program which treats the style guide as an absolute rule would fail to extract this citation.

77 The model's architecture and inputs closely follow that of Jason Chiu & Eric Nichols, ‘Named Entity Recognition with Bidirectional LSTM-CNNs’ (2016) 4 Transactions of the Association for Computational Linguistics 357. Scholars interested to replicate the data construction process will find this paper instructive. Scholars interested to replicate only our analysis should contact the first-named author for the raw dataset.

78 Specifically, half the dataset (100 cases) was set aside as a holdout set. The remaining 100 cases were sub-divided into training (80% of the data) and cross-validation sets (the other 20%). The model was trained solely using the training set, and model hyper-parameters were then tweaked until it achieved satisfactory scores on the cross-validation set. In particular, the data vendor adjusted only the learning rate as well as and the number of epochs the model was trained for. The tuned model was then trained on the full training set of 100 cases and used to extract citations from the holdout set. After confirming the model's satisfactory performance, Thereafter, the model was re-trained on the entire set of 200 cases before being used to annotate the remaining 2,572 cases.

79 For specialists and interested observers, the specific accuracy metrics are as follows. The final model used for annotating the 2,552 residual cases successfully detected at least 50% or more of the characters in 116 of all 142 references (81.69%) present in the holdout test set. Of the characters identified by the model as belonging within academic references, 10765 were correctly identified as such while 1481 were wrongly identified. This translates to a character level precision, recall, and F1 score of 87.91%, 84.07%, and 85.94% respectively. Precision measures how likely the model's positive predictions are correct, recall measures how likely the model successfully identifies positives (to see how recall differs from precision, notice that a trivial way to identify all positives is to identify everything as positive, but this would lead to very poor precision), while the F1 score is an average between precision and recall.

80 The list of words was devised from studying the extracted references. For court filings, the words are ‘notes’, ‘bundle’, ‘affidavit’, ‘pleading’, ‘exhibit’, ‘form’, ‘plaintiff’, ‘defendant’, ‘respondent’, ‘appellant’, and ‘petitioner’. For scientific texts, the words are ‘neurology’, ‘psychology’, ‘psychological’, ‘psychiatric’ ‘medical’, ‘clinical’, ‘neurosurgery’, ‘surgical’, ‘surgery’, ‘pathology’, ‘oncology’, ‘cancer’, ‘biology’, ‘disease’, ‘mental’, ‘medicine’, ‘medical’, ‘diagnosis’, ‘treatment’, ‘pharmacology’, ‘toxicology’, ‘trauma’, ‘microsurgical’, ‘urology’, ‘fracture’, ‘fractures’, ‘lymphatic’, ‘testis’, ‘orthopaedic’, ‘accountancy’ ‘architecture’. Note that these words may over-inclusively flag out legal academic references to medical law texts.

81 The full algorithm used is rather involved, and interested readers are welcome to obtain its source code by contacting the first-named author. Briefly, the algorithm first tests if the citation is to a journal by searching for the specific year-volume-journal-page format that only journal citations have. Failing which, it tests if the citation is to a collection by looking for an article or chapter citation expressed as appearing ‘in’ another title. Further failing which, it attempts to identify book citations by searching for the title-publisher-and-editor-in-brackets format unique to books, taking reference from a manually compiled list of legal publishers (eg, Sweet & Maxwell, Hart Publishing, etc). A citation not fulfilling any of the above is classified as ‘other’.

82 The current authoritative version of the Singapore Academy of Law Style Guidance dates back to 2004, which is close to the beginning of our study period.

83 Phang, Goh & Soh (n 9).

84 Cheah & Goh (n 10) 112–115.

85 See Annex A for a tabulation of how the raw subject areas were mapped over.

86 Standard deviation.

87 Statistics in this row were calculated excluding subjects assigned to ‘Others’.

88 Standard deviation.

89 Statistics here were calculated including the ‘Other’ subjects. Only two judgments in the dataset were not assigned any subjects. Both were procedural, one-paragraph-long judgments: Kempinski Hotels SA v PT Prima International Development [2011] 4 SLR 669; Kempinski Hotels SA v PT Prima International Development [2011] 4 SLR 670.

90 This excludes two judges who hold honorary doctorates.

144 For brevity, only subjects statistically significant at the five per cent or lower level in at least one specification are shown.

145 Adjusted r-squared values are presented for the OLS regressions; pseudo-r-squared values are presented for the Poisson regressions.

91 Only four judges had at least one judgment per year for the entire 17-year period.

92 Except to a limited extent in analysis below that examines whether certain individual judges are driving results for their education group.

93 This was Public Prosecutor v Lam Leng Hung and Ors [2017] 4 SLR 474, a significant litigation involving criminal breach of trust by the leaders of one of Singapore's largest churches. Multiple defendants and senior counsel were involved. Three judges sat on the High Court coram.

94 This was Kempinski Hotels SA v PT Prima International Development [2011] 4 SLR 669. See also (n 89) above.

95 AEL and others v Cheo Yeoh & Associates LLC and another [2014] 3 SLR 1231; Giant Light Metal Technology (Kunshan) Co Ltd v Aksa Far East Pte Ltd [2014] 2 SLR 545; Goi Wang Firn (Ni Wanfen) and others v Chee Kow Ngee Sing (Pte) Ltd [2015] 1 SLR 1049; Public Prosecutor v Low Kok Heng [2007] 4 SLR(R) 183; Wu Yang Construction Group Ltd v Zhejiang Jinyi Group Co, Ltd and others [2006] 4 SLR(R) 451.

96 [2015] 5 SLR 122.

97 [2009] 4 SLR(R) 788.

98 Supreme Court Practice Direction No 1 of 2008.

99 If these correlations were calculated using cases decided on and after 2009 only, all values remain the same except for the same two which fall to 0.29 and 0.33 respectively.

100 These are calculated by bootstrapping following the default provided by the Seaborn Python package. See Michael Waskom, ‘Seaborn: Statistical Data Visualization’, (2021) 6 Journal of Open Source Software 3021.

101 National University of Singapore Faculty of Law, ‘History & Milestones’ (1 May 2009) <> accessed 30 Aug 2021.

102 Future work could more specifically study how far practitioners’ versus academic-oriented texts may be treated differently by the courts, if at all. The present study's scope precludes detailed examination into how texts may be separated into practitioner versus academic-oriented. As the line is not always clear, this was not something which we were confident leaving to a machine learning algorithm to perform.

103 These counts are approximate for two reasons. First, because we retrieved citation strings verbatim, there were often subtle differences between different citations to the same book. For example, Walter Woon, Walter Woon on Company Law (Tan Cheng Han gen ed, rev 3rd edn, Sweet & Maxwell 2009) is also referenced as ‘Walter Woon's Company Law’ and ‘Walter Woon's book Company Law’. These counts represent only the number of exact matches for the phrases in its left column name and may thus undercount the true number of citations. Second, recall the limitations in the automated citation extraction process explained above. It is possible to obtain exact counts manually, but this would be impractical and indeed undo the cost savings achieved by automating citation extraction in the first place.

104 Lee Zhe Xu et al, ‘The Use of Academic Scholarship in Singapore Supreme Court Judgments: 2005–2014’ (2015) 33 Singapore Law Review 25, 43.

105 As many books are long-running titles, and titles may have changed over the years, the date of first publication is not always clear. Where these are not readily available, we have offered approximate dates based on the oldest available authoritative source. In any event, the key consideration is simply whether these titles were in circulation before 2000. Sources published after 2000 are bolded.

106 National Library of Australia Catalog, ‘Chitty's Treatise on the law of contracts’ <> accessed 12 Aug 2021.

107 WorldCat, ‘Singapore civil procedure’ <> accessed 12 Aug 2021.

108 Chen Siyuan & Eunice Chua Hui Han, Civil Procedure in Singapore (2nd edn, Wolters Kluwer 2018) 46 fn 177.

109 National Library Board Singapore Catalog, ‘Company Law / Walter Woon’ <> accessed 12 Aug 2021.

110 We could not find an exact date of first publication, but the twelfth edition was published in 1961, significantly pre-dating the start of our analysis period. WorldCat, ‘Mayne and McGregor on Damages’ <> accessed 12 Aug 2021.

111 National Library of Australia Catalog, ‘The law of torts / by J.F. Clerk and W.H.B. Lindsell’ <> accessed 12 Aug 2021.

112 Singapore Management University Institutional Knowledge Database, ‘The Law of Torts in Singapore’ <> accessed 12 Aug 2021.

113 This is the old title for Robert Goff & Gareth Jones, The Law of Unjust Enrichment (9th edn, Sweet and Maxwell 2016), used for the seventh and earlier editions.

114 Jack Beatson, ‘Robert Goff’ (2019) 18 Biographical Memoirs of Fellows of the British Academy 241, 243.

115 Wildy & Sons, ‘Bennion on Statutory Interpretation 5th ed’ <> accessed 12 Aug 2021.

116 Zhuang-Hui Wu, ‘Tan Sook Yee's Principles of Singapore Land Law 3rd ed by Hang Wu Tang, Kelvin F.K. Low’ [2010] Singapore Journal of Legal Studies 222, 222.

117 BerkeleyLaw Library Catalog, ‘Ethics and professional responsibility : a code for the advocate and solicitor / Jeffrey Pinsler’ <> accessed 12 Aug 2021.

118 Disa Sim, ‘Evidence, Advocacy and the Litigation Process (2nd ed.) by Jeffrey Pinsler’ [2004] Singapore Journal of Legal Studies 254, 254

119 WorldCat, ‘Law of intellectual property of Singapore’ <> accessed 12 Aug 2021.

120 Martin George, ‘Publication: Dicey, Morris & Collins on the Conflict of Laws’ (ConflictofLaws.Net, 14 Oct 2006) <> accessed 12 Aug 2021.

121 Andrew Ashworth, Sentencing and Criminal Justice (9th edn, Cambridge University Press 2018).

122 ibid iii.

123 WorldCat, ‘Principles of civil procedure’ <> accessed 12 Aug 2021.

124 RJ Smith, ‘Elements of Land Law by K. J. Gray’ (1990) 10 Oxford Journal of Legal Studies 260, 260.

125 Robert Goff & Gareth Jones, The Law of Unjust Enrichment (9th edn, Sweet and Maxwell 2016).

126 This by no means diminishes the importance and prestige of honorary qualifications. The intention is merely to sharpen the focus on a judge's academic background. In any event, our dataset records only two judges with honorary degrees. Chan Sek Keong CJ (as he then was) holds doctorates from both NUS (National University of Singapore, ‘NUS confers Honorary Doctor of Laws to CJ Chan Sek Keong’ (6 Jul 2010) <> accessed 22 Feb 2022) and SMU (Singapore Management University, ‘SMU Commencement 2013: Honorary Degree Citation’ (25 Jul 2013) <> accessed 31 Mar 2022). LP Thean JA (as he then was) holds an honorary doctorate from the University of Bristol: Law Gazette, ‘In Practice – AG's Tribute to LP Thean’ <> accessed 22 Feb 2022. An alternative analysis (on file) which includes honorary doctorates produces similar results.

127 We could not definitively identify the terminal degrees of S Rajendran J, Phang Kang Chau J, and Quentin Loh J.

128 There were 13, 15, 16, 6, and 10 PhD-holder-authored judgments in each year respectively.

129 In our dataset, 64.39% of cases had only one unique narrow subject attributed to it. 26.01% had only two; only 0.078% had three.

130 For brevity, only subjects statistically significant at the five per cent or lower level in at least one specification are shown.

131 Adjusted r-squared values are presented for the OLS regressions; pseudo-r-squared values are presented for the Poisson regressions.

132 We use this column because OLS coefficients may be directly interpreted (while Poisson coefficients cannot) and because column 3 includes the full set of controls.

133 Cheah & Goh (n 10) 110.

134 ibid 116

135 ibid 116.

136 Cheah & Goh (n 10) 113.

137 Lee et al (n 104) 38.

138 ibid 38.

139 ibid 38.

140 Cheah & Goh (n 10) 119.

141 Panel effects occur, broadly speaking, when a judge's behaviour in deciding a case is influenced by other judges sitting on the same panel. This is well-studied in US empirical literature. See eg, Cass R Sunstein et al, Are Judges Political? An Empirical Analysis Of The Federal Judiciary (Brookings Institute Press 2006); Kim, Pauline, ‘Deliberation and Strategy on the United States Courts of Appeals: An Empirical Exploration of Panel Effects’ (2009) 157 University of Pennsylvania Law Review 1319Google Scholar.

142 Posner, Richard, ‘An Economic Analysis of the Use of Citations in the Law’ (2000) 2 American Law and Economics Review 381, 387CrossRefGoogle Scholar.

143 For examples of empirical analysis on legal briefs, see Lance Long & William Christensen, ‘Clearly, Using Intensifiers is Very Bad – Or Is It’ (2008–2009) 45 Idaho Law Review 171; Lance Long & William Christensen, ‘Does the Readability of Your Brief Affect Your Chance of Winning an Appeal?’ (2011) 12 Journal of Appellate Practice & Process 145; Jerrold Soh, ‘Causal Inference with Legal Texts’ (MIT Computational Law Report, 15 Jul 2021) <> accessed 1 Sep 2021.