Proceeding from first principles, I would say the answer is that the internal standard of review depends on an interpretation of the statutory scheme. Accordingly, a court should be deferential in reviewing an appellate decision-maker's choice of standard of review. Indeed, in Canada, deference would be presumed as long as the appellate decision-maker was interpreting its home statute. As for the appellate decision-maker, its determination of whether to conduct a de novo hearing or, alternatively, a deferential review, will depend on its appreciation of the statutory scheme.
The Alberta Court of Appeal has come to different conclusions on these questions in a line of cases, the most recent of which is Kikino Métis Settlement v Métis Settlements Appeal Tribunal, 2013 ABCA 151.
Here, the decision-making structure regulated the distribution of real property on Aboriginal lands. A settlement council took decisions in the first instance, but these were subject to appeal to the Tribunal.
At issue in the instant case was a distribution of land. The beneficiaries did not agree on the appropriate distribution, but the applicable rules allowed for the council or Tribunal to make a distribution where there was "substantial agreement" amongst the beneficiaries. Here, four of the six beneficiaries who participated in the hearing before the Tribunal (though not before the council) agreed. Was this "substantial agreement"?
Following Newton v. Criminal Trial Lawyers’ Association, 2010 ABCA 39, the Court held that the Tribunal's choice of the standard of review was to be reviewed on a standard of correctness. If the Tribunal did not get it right, the Court could intervene. (I don't see how this is consistent either with Dunsmuir, or first principles, but there is no question that Newton is nonetheless good law in Alberta.) In determining the appropriate relationship between the council and the Tribunal, regard was had to the following Newton factors:
(a) the respective roles of the tribunal of first instance and the appellate tribunal, as determined by interpreting the enabling legislation;
(b) the nature of the question in issue;
(c) the interpretation of the statute as a whole;
(d) the expertise and advantageous position of the tribunal of first instance, compared to that of the appellate tribunal;
(e) the need to limit the number, length and cost of appeals;
(f) preserving the economy and integrity of the proceedings in the tribunal of first instance; and
(g) other factors that are relevant in the particular context.
That conclusion is surprising. Four out of six is a two-thirds majority. Such a majority usually gets its way! Moreover, logic is not on Slatter J.A.'s side. Requiring five out of six to satisfy the substantial agreement standard seems wrong to me. Unanimity is not the standard, so five out of six must be a reasonable interpretation of the substantial agreement standard. On Slatter J.A.'s logic, five out of six is the only possible interpretation. This cannot be right: it means that there is only one answer, which is the antithesis of review for reasonableness. Accordingly, it must have been reasonable for the Tribunal to consider that four out of six satisfied the substantial agreement. In order to be a “substantial agreement” there must be an agreement between something approaching all of the beneficiaries, on something approaching all of the issues in the distribution. The clearest example would be where all the beneficiaries agree on the distribution, but there is some disagreement on minor details, or small items in the estate. Since each estate distribution will depend on its own facts, there may be “substantial agreement” where one (or perhaps more than one) of the beneficiaries is not in agreement with the substantial majority on an objectively fair distribution. There may also be a “substantial agreement” where all of the beneficiaries agree in general terms, but there are moderately important issues that remain outstanding. The Land Policy provides that whether there is a substantial agreement is “in the opinion of the body making the decision”, so that body clearly has a discretion to exercise. That discretion is not, however, unlimited, and it can be reviewed for reasonableness. Where 4 of 6 beneficiaries agree to totally exclude the other 2 from the estate, it is unreasonable to suggest that there is substantial agreement.
MacDonald J.A. concurred. He took the view that "very little deference" should be accorded by the Tribunal to the council (at para. 47):
"Firstly, unlike the situation regarding the Tribunal itself, there is no privative clause in the Act according any deference to a decision of Council. Secondly, while a right of appeal to the Tribunal is not a decisive factor, it does support the position that less, rather than greater, deference ought to be accorded to the decision of a Council...Furthermore, the appeal from a Council decision to the Tribunal is really in the nature of a hearing de novo. Unlike the situation in Newton, there is basically no restriction on the evidence that can be called before the Tribunal... The Tribunal has representatives from other Métis Settlements and that fact combined with its statutory mandate indicates that less, as opposed to more, deference is due to the decision of a Council at least as regards an issue involving desent of Metis Title under under the Land Policy. As argued by counsel for the Tribunal before this court, the role of the Tribunal includes bringing a more objective perspective to bear on matters that sometimes get clouded by local personalities and issues" (at paras. 52-54).This seems like a sensible approach to me, one which maps more closely onto the old "pragmatic and functional analysis" employed by the Supreme Court of Canada than onto the hodge-podge of factors laid out in Newton. MacDonald J.A.'s overriding concern was to identify legislative intent, which seems entirely proper.
He too found the Tribunal's decision unreasonable on the merits, though because of its failure to implement its decision. He wisely stayed out of the "substantial agreement" controversy.
This question, of course, is the dog that didn't bark in an important recent Quebec Court of Appeal case (blogged here): Autorité des marchés financiers c. Groupe SNC-Lavalin Inc., 2013 QCCA 204. If this case ever makes it the Supreme Court of Canada, we will presumably get some welcome clarification on the question of the internal standard of review.