What happened
The applicant, a citizen of Lebanon who arrived in Australia on 24 October 2010 on a student visa that expired on 30 August 2013, applied for a protection visa on 13 August 2013. He claimed a fear of harm arising from his rejection of the strict fundamentalist Sunni Islam in which he had been raised. He asserted that his parents had disowned him, ceased financial support, and that he feared forced conscription by extremists to fight in Syria as well as torture or beating by family and community members because of his apostasy-like views. The Minister’s delegate refused the application on 20 March 2014. The then Refugee Review Tribunal affirmed that decision on 23 February 2015.
Because the Tribunal decision was handed down on 23 February 2015, s 477(1) of the Migration Act 1958 (Cth) required any judicial review application in the Federal Circuit Court of Australia (FCCA) to be filed within 35 days, i.e. by 30 March 2015. The applicant filed on 13 April 2015, 14 days late, and therefore required an extension of time under s 477(2). The first return date was 14 May 2015. The primary judge (Judge Street) declined to make consent orders and instead set a timetable requiring the Minister to file a Court Book by 28 May 2015, the applicant to file an amended application and evidence by 11 June 2015, and listed the matter for final hearing on 29 July 2015.
On the morning of the hearing the applicant sought leave to file a further amended application together with two affidavits. One affidavit, sworn by Mr Victor Kline (editor of the Federal Court Reports and Federal Law Reports since 1990), contained a statistical analysis of every judgment delivered by Judge Street between his Honour’s appointment on 1 January 2015 and 19 June 2015. Mr Kline identified 286 decisions, 254 of which (88.81%) were immigration matters in which the Minister was respondent. In all 254 the judgments were or appeared to be ex tempore; the Minister succeeded in 252 (99.21%); the applicant succeeded in only two cases where error was conceded by the Minister. Judge Street had delivered 45.11% of all immigration judgments in the Sydney Registry during the period. The affidavit also compared the primary judge’s 0.79% set-aside rate with the 10.8% and 12.2% set-aside rates for MRT and RRT decisions reported in the most recent Annual Report.
Counsel for the applicant (Mr Jay Williams) submitted that these statistics, coupled with the frequent use of forceful phrases such as “no substance” (140 occasions) and “it is clear” (160 occasions), demonstrated that the primary judge was predisposed to dismiss migration applications and could not bring an open mind. The primary judge dismissed the recusal application in ex tempore reasons (ALA15 v Minister for Immigration and Border Protection [2015] FCCA 2047) at [11]-[12], accepting the Minister’s submission that statistical material was not evidence upon which a fair-minded observer might apprehend lack of impartiality.
The primary judge then refused the extension of time (ALA15 v Minister for Immigration and Border Protection (No 2) [2015] FCCA 2048). His Honour held that the explanation for delay—financial hardship—was inadequate because it consisted of “bald assertions” without detail of the applicant’s financial position or the source of funds eventually used. Although that was sufficient to refuse extension, the primary judge went on to assess the merits of the four proposed grounds of review directed at paragraphs [67]-[70] of the Tribunal’s reasons. Those grounds alleged failure to deal with a substantial clearly articulated claim, failure to complete the statutory task, legal unreasonableness, and failure to consider relevant considerations (including PAM 3 guidelines). The primary judge found each ground lacked substance and therefore declined to extend time.
In this Court the applicant commenced proceedings under s 39B of the Judiciary Act 1903 (Cth) challenging only the refusal to recuse. An initial application for leave to appeal the recusal judgment was filed but “withdrawn” by email at 4:23 pm on Friday 26 February 2016, the last business day before the Full Court hearing fixed for 29 February 2016. The Full Court (Allsop CJ, Kenny and Griffiths JJ) granted leave to rely on an amended s 39B application that joined the FCCA as a party but otherwise dismissed both the judicial review application and the formal application for extension of time and leave to appeal, ordering the applicant to pay the Minister’s costs in each case.
Why the court decided this way
The Full Court began by noting substantial agreement between the parties on the applicable apprehended bias test, which it restated at [35] by reference to the line of High Court authority running from R v Watson; Ex parte Armstrong (1976) 136 CLR 248 through Livesey v New South Wales Bar Association (1983) 151 CLR 288, Ebner v Official Trustee in Bankruptcy (2000) 205 CLR 337, Concrete Pty Ltd v Parramatta Design and Developments Pty Ltd (2006) 229 CLR 577, British American Tobacco Australia Services Ltd v Laurie (2011) 242 CLR 283 and Isbester v Knox City Council [2015] HCA 20. The test is whether a fair-minded and appropriately informed lay observer might reasonably apprehend that the judge might not bring a fair, impartial and independent mind to the determination on the merits.
Two mandatory steps from Ebner at [8] were emphasised: first, identification of the matter said to lead the judge to decide otherwise than on the merits; second, articulation of the logical connection between that matter and the feared deviation. The allegation of prejudgment was described as serious, carrying an imputation that the judge had failed to honour the judicial oath, and therefore “must be distinctly made and clearly proved” (Jia Legeng at [69], cited at [36]).
The Court rejected the applicant’s central contention that the hypothetical observer should be attributed with Mr Kline’s raw statistics “without any further analysis or attempt to go behind the raw statistics”. Five independent reasons were given.
First, raw statistics are meaningless to an informed observer unless placed in context by analysis of the individual reasons for judgment. Without that, the observer cannot know whether the outcomes reflected the strength of the cases, concessions, or error rather than prejudgment. Even erroneous decisions could be the product of “human frailty” rather than bias ([38]).
Second, the fact that one judge may be easier to persuade than another does not connote bias (Jia Legeng at [71], cited at [39]).
Third, the MRT-RRT Annual Report statistics were irrelevant because they related to 2013-2014 (before the primary judge’s appointment) and aggregated tribunal, FCCA, Federal Court and High Court outcomes, providing no valid comparator for FCCA judicial review decisions ([40]).
Fourth, the Court adopted Heerey J’s reasoning in Vietnam Veterans’ Association of Australia (New South Wales Branch Inc) v Gallagher (1994) 52 FCR 34 at [26] and [33]. Statistical evidence can at most show a propensity to decide a category of case a particular way; that does not equate to apparent bias. Judges are not “automatons”. The appeal system exists precisely because judges make errors of fact and law without thereby being biased. The Court expressly declined to distinguish or overrule Vietnam Veterans, holding that the statistical analysis there (51.28% appeal rate) was no different in character from the 99.21% success rate here; both were irrelevant absent contextual material ([41]-[43]).
Fifth, and independently, the statistics ended on 19 June 2015, the very day on which the Full Court delivered SZWBH and Shrestha, which were critical of the primary judge’s practice of summarily dismissing migration matters at the first court date without notice and, in one case, over the Minister’s opposition. The present FCCA proceeding was conducted on 29 July 2015, after those decisions. The primary judge had before him the applicant’s affidavits and further amended application, granted leave to amend, heard full argument on the recusal application, delivered separate ex tempore reasons on both recusal and extension, and addressed each of the four grounds on their merits rather than dismissing at the first return date. The pre-19 June 2015 conduct was therefore irrelevant to an allegation of apprehended bias in a post-SZWBH matter ([44]). Later single-judge decisions critical of the primary judge (AAV15, AEG15, AHT15, SZWCW, Sekigawa) all concerned pre-SZWBH decisions and took the matter no further ([45]).
The applicant’s reliance on frequent use of the phrases “no substance” and “it is clear” was dismissed because no evidence substantiated the claim beyond assertion, and even if true such language is not uncommon in judicial reasons and does not, without context, suggest a closed mind ([46]).
Because the bias ground failed, the consequential ground that the primary judge lacked jurisdiction to hear the extension application necessarily failed ([47]). The s 39B application was therefore dismissed. Costs followed the event; impecuniosity and the “public interest test case” characterisation were insufficient to displace the ordinary rule (Ruddock v Vadarlis (2001) 115 FCR 229 at [18]-[25] distinguished at [51]).
Separately, the Court recorded serious concerns about the conduct of the applicant’s legal representatives who, despite multiple case management orders made by Robertson J on 23 September and 20 October 2015, filed late, over-length, repetitive and error-ridden submissions, failed to file a chronology or application book by the due dates, delivered a 555-page book and then a 750-page replacement on the Sunday before the Monday hearing, used “dropbox” despite the Australian Government Solicitor’s security restrictions, and supplied yet further “final submissions” and an outline of oral argument on the morning of hearing. While acknowledging the pro bono nature of the representation, the Court stated that such conduct was “most unsatisfactory”, caused inconvenience to the Court and the Minister, and that repetition might attract personal costs orders under ss 37N(5) and 43(3)(f) of the Federal Court of Australia Act 1976 (Cth) ([53]-[68]).
Before and after state of the law
Prior to this judgment the test for apprehended bias was already settled in the terms articulated in Ebner and confirmed in British American Tobacco and Isbester. What was less clearly settled was the precise role that statistical evidence of a judicial officer’s “strike rate” could play. Vietnam Veterans had rejected statistical evidence in an administrative context, but that decision had not been applied squarely to a judicial officer in the migration jurisdiction where docket management practices had come under scrutiny in SZWBH and Shrestha.
This Full Court authoritatively determined that raw statistical material, without a detailed contextual analysis of the individual reasons, is irrelevant to the hypothetical observer’s assessment. It elevated Vietnam Veterans from a single-judge administrative law decision to a principle applicable to judges, emphasising that the law does not treat judicial officers as “automatons”. The decision also clarified that criticism of a judge’s earlier case-management practices (as in SZWBH) resets the baseline; subsequent decisions that afford procedural fairness cannot be impugned by reference to pre-criticism statistics.
The judgment reinforces that an allegation of prejudgment carries a serious imputation and must be “distinctly made and clearly proved”. It confirms that mere frequency of robust language (“no substance”, “it is clear”) is not probative absent context. On costs, it reaffirms that personal circumstances and aspirational “public interest” labels do not readily displace the ordinary rule that costs follow the event.
After the decision, a party seeking to allege bias on the basis of workload statistics or success rates in a particular jurisdiction must now adduce evidence that analyses the underlying merits or concessions in each cited case. Bare percentages, even dramatic ones, will not suffice. The decision also signals that courts will scrutinise last-minute “withdrawals” of leave applications and non-compliance with case-management orders more rigorously, with the possibility of personal costs orders against practitioners.
Key passages with plain-English translation
Paragraph [38]: “for such raw statistical material to be attributed to the hypothetical observer, it normally would need to be accompanied by a relevant analysis of the individual immigration judgments determined by the primary judge in order that the statistics were placed in a proper context. Absent such analysis, the hypothetical observer would not be able to make an informed assessment of the significance of the raw statistics. It may be, for example, that a close analysis of some, many, or all of the relevant judgments reveal that they had been determined on a reasonable and plausible basis. And, even if some or all of the judgments were wrongly decided, that may be the consequence of human frailty on the part of the judge and not prejudgment…”
Plain English: You cannot throw a pile of win-loss numbers at a judge and say “see, he always rules for the government”. A fair-minded person would want to know why each case was decided that way. Maybe the applicants simply had weak cases. Even if the judge got some wrong, that does not prove he had closed his mind before hearing the current case; judges, like everyone else, make honest mistakes.
Paragraph [41] (adopting Heerey J in Vietnam Veterans): “All such evidence could show is that, because a decision-maker has decided a particular kind of case in a particular way in the past, he or she is likely to decide a case of the same nature in the same way in the future. Even if that be accepted as a conclusion of fact, it does not make out a case of apparent bias. The law is not so ignorant or disdainful of human nature as to assume that judges or quasi-judicial decision-makers are automatons…”
Plain English: Predicting that a judge who has ruled against asylum seekers many times will probably do so again is just common sense about human patterns. That prediction does not mean the judge is biased. The legal system knows judges are human beings who form habits and preferences; that does not disqualify them.
Paragraph [44]: “We accept the Minister’s submission that the primary judge’s conduct prior to 19 June 2015 is irrelevant to an allegation of apprehended bias in a case which post-dates the Full Court’s decisions [in SZWBH and Shrestha]. … These circumstances are very different from those which arose in cases such as SZWBH and Shrestha.”
Plain English: Once the Full Court had publicly criticised the judge for rushing cases at the first hearing date, any later case where the judge actually reads the papers, hears argument and gives reasoned decisions cannot be attacked using the old statistics. The environment had changed.
Paragraph [51]: “The proceedings were primarily designed to advance the applicant’s personal rights and interests under migration law.”
Plain English: Calling this a “test case” to protect public confidence in the judiciary does not change the fact that the applicant was simply trying to stay in Australia. That does not justify making the Minister pay the costs.
What fact patterns trigger this precedent
This precedent is triggered whenever a litigant, particularly in high-volume jurisdictions such as migration, seeks to rely on aggregate success rates, percentages of ex tempore judgments, or frequency of particular judicial phrases to ground an apprehended bias application. It applies where the allegation is one of systemic prejudgment rather than any specific statement or conduct in the instant proceeding. It is engaged when raw statistical affidavits (such as Mr Kline’s) are tendered without accompanying detailed analysis of the merits or concessions in each cited case. The decision is also relevant where a party attempts to rely on statistics that pre-date a Full Court’s public criticism of a judge’s case-management practices, or where the current proceeding involved consideration of evidence and full argument rather than summary dismissal at the first return date.
The costs aspects are triggered whenever an impecunious migration applicant or practitioner characterises proceedings as a “public interest test case” in an attempt to resist an adverse costs order. The case-management strictures are engaged by repeated failure to comply with timetabling orders, late filing of over-length or repetitive submissions, last-minute changes of position on the eve of hearing, and use of electronic file-sharing services that opposing parties cannot access.
How later courts have treated it
Although the present judgment is the source judgment, it expressly treats Vietnam Veterans as correctly decided and applies it to the judicial (rather than purely administrative) context, thereby strengthening its authority. The Court distinguished SZWBH and Shrestha on their facts, confirming that those decisions criticised only a particular pre-June 2015 practice of summary dismissal and did not create a permanent taint. It followed Jia Legeng, Ebner, British American Tobacco and Johnson without reservation, demonstrating continuity rather than innovation in the High Court’s apprehended bias jurisprudence. The judgment’s treatment of Ruddock v Vadarlis confirms that the “public interest” costs exception remains narrowly confined. The detailed criticism of the applicant’s legal representatives’ conduct signals that later courts may treat repeated non-compliance with case-management orders as warranting personal costs orders under the Federal Court of Australia Act, a course expressly left open.
Still-open questions
The judgment leaves open what degree of contextual analysis would be sufficient to render statistical material relevant. It states that “normally” such analysis would be required, but does not prescribe whether a sample, a statistical regression controlling for case strength, or some other methodology would suffice. It does not decide whether, in a case where detailed analysis showed a pattern of legally unsupportable reasoning, that pattern could ever rise to the level of apprehended bias. The Court expressly leaves to another day the question whether personal costs orders under ss 37N(5) and 43(3)(f) should be made against practitioners who engage in the criticised conduct. It does not address whether the same principles apply to tribunal members rather than judges, although the adoption of Vietnam Veterans (itself an administrative case) suggests they may. Finally, the judgment assumes but does not explore the outer limits of the “informed” observer’s knowledge: how much of the broader docket pressures, ministerial concession rates, or prevalence of unmeritorious migration claims the observer is taken to know remains unarticulated. These questions will require future litigation to resolve.