Opinion of the Court filed by Circuit Judge SCALIA.
Concurring opinion filed by Chief Judge WALD.
SCALIA, Circuit Judge:Petitioner, Consumers Union of U.S., Inc., challenges the Federal Trade Commission’s “Used Car Rule,” a Trade Regulation Rule governing the sale of used motor vehicles. See 16 C.F.R. Part 455 (1986). Petitioner objects to the Commission’s decision to omit from the final rule the so-called “known-defects provision,” which would have required used-car dealers to list on the window sticker affixed to any used car offered for sale the presence of certain mechanical defects of which they have knowledge. The principal issues addressed are the legality of the procedures by which the Commission withdrew the known-defects provision following remand of the initial rule by a reviewing court, and the substantive validity of the withdrawal.
*205I
The tumultuous history of the Used Car Rule began in 1975, with the enactment of the Magnuson-Moss Trade Commission Improvement Act (“Magnuson-Moss Act”), Pub.L. No. 93-637, 88 Stat. 2183 (1975) (codified at 15 U.S.C. §§ 2301-2312 (1982) and other scattered sections of 15 U.S.C.), which, in § 109(b), directed the Commission to promulgate a rule regulating “warranties and warranty practices in connection with the sale of used motor vehicles,” 15 U.S.C. § 2309(b). In 1981, relying upon § 109(b) and upon § 18(a)(1) of the Federal Trade Commission Act (“FTC Act”), ch. 311, 38 Stat. 717 (1914), as added by the Magnuson-Moss Act, § 202, 88 Stat. 2193 (codified as amended at 15 U.S.C. § 57a(a)(1) (1982)), which grants the Commission rulemaking authority to regulate “unfair or deceptive” trade practices, the Commission promulgated a rule that required used-car1 dealers to post on a standard window sticker several consumer warnings, the terms of any warranty, and a list of certain specific mechanical defects known to the dealer. 46 Fed.Reg. 41,328 (1981) (“Initial Rule”).
Timely petitions for review of the Initial Rule were filed in the United States Court of Appeals for the Second Circuit, pursuant to § 18(e) of the FTC Act, 15 U.S.C. § 57a(e), attacking primarily the known-defects provision. Miller Motor Car Corp. v. FTC, Nos. 81-4144, 81-4172, 81-4216. A lengthy delay interrupted the case, during which Congress vetoed the Initial Rule, see 128 Cong.Rec. 10,343-44 (1982); 128 Cong. Rec. H2856-58 (daily ed. May 26,1982), and the Supreme Court, in turn, declared the legislative veto unconstitutional, see United States Senate v. FTC, 463 U.S. 1216, 103 S.Ct. 3556, 77 L.Ed.2d 1402, 1403, 1413 (1983); United States House of Representatives v. FTC, 463 U.S. 1216, 103 S.Ct. 3556, 77 L.Ed.2d 1402, 1403, 1413 (1983). While Miller was pending, the Commission announced its intention to consider modifying the Initial Rule. 48 Fed.Reg. 36,096 (1983). Shortly thereafter, alleging that the Commission had never afforded them the requisite opportunity to comment on the known-defects provision, the Miller petitioners moved the court to remand the case to the Commission pursuant to § 18(e)(2) of the FTC Act, 15 U.S.C. § 57a(e)(2), for additional oral and written presentations on the Initial Rule. By stipulation of the parties, in which the Commission disputed the alleged procedural irregularity, the court remanded the case pursuant to § 18(e)(2), retaining jurisdiction.
After soliciting further evidence and comments on the Initial Rule, and particularly comments on the know-defects provision, 48 Fed.Reg. 55,874 (1983), the Commission (in addition to making minor changes not relevant here) revised the Initial Rule to delete the known-defects provision. 49 Fed.Reg. 45,692 (1984) (“Revised Rule”).
Petitioner sought review of the Revised Rule in this court, challenging both the Commission’s decision to omit the known-defects provision and the procedures it followed in arriving at that decision. The Second Circuit transferred Miller to this court and we designated the National Automobile Dealers Association (“NADA”) and the National Independent Automobile Dealers Association, which were petitioners in Miller, as intervenors in this case.
II
We address first petitioner’s contention that the Commission, in promulgating the Revised Rule, neglected to provide statutorily required procedures. The essence of the dispute is this: Section 18(e)(2) of the FTC Act, pursuant to which the Second Circuit remanded the Initial Rule to the Commission, provides that the reviewing court may, at the request of either the petitioner or the Commission, order the Commission to provide opportunity for “additional oral submissions or written presentations” if “there were reasonable grounds for ... failure to make such submissions *206and presentations in the proceeding before the Commission.” 15 U.S.C. § 57a(e)(2). It continues:
The Commission may modify or set aside its rule or make a new rule by reason of the additional submissions and presentations and shall file such modified or new rule, and the rule's statement of basis of [sic] purpose, with the return of such submissions and presentations. The court shall thereafter review such new or modified rule.
Id. By contrast, § 18(d)(2)(B) of the FTC Act, 15 U.S.C. § 57a(d)(2)(B), requires that “[a] substantive amendment to, or repeal of, a rule promulgated under subsection (a)(1)(B) of this section [as was the Initial Rule in the present case] shall be prescribed ... in the same manner as a rule prescribed under such subsection.” That manner of prescription includes not merely the opportunity for oral presentation and written submission, 15 U.S.C. § 57a(c)(2)(A), but also opportunity “to present such rebuttal submissions and to conduct (or have conducted ...) such cross-examination of persons as the Commission determines (i) to be appropriate, and (ii) to be required for a full and true disclosure” with respect to disputed issues of material fact, 15 U.S.C. § 57a(c)(2)(B). Rebuttal submissions were permitted in the present case. The Commission did not, however, either permit cross-examination or determine that it was not “appropriate” or not “required for a full and true disclosure” with respect to disputed issues of material fact. It is this failure petitioner complains whether the complaint is justified depends upon (1) whether § 18(e)(2) requires such procedures, and (2) if not, whether § 18(e)(2) or rather § 18(d)(2)(B) governed the proceedings on remand.
The first point is not contested. As odd as it might seem to allow supplementation of the rulemaking record in a fashion different from that by which the original record itself was compiled, that is precisely what § 18(e)(2) prescribes, in the limited circumstances in which it is applicable. And there is some reason for the oddity. The § 18(e)(2) procedure, while not as demanding as that of § 18(d)(2)(B) (or § 18(c)) is fully as demanding as that of normal rulemaking under the Administrative Procedure Act (“APA”). See 5 U.S.C. § 553(b), (c) (1982). In fact, it is somewhat more demanding, since it permits the remanding court to require receipt of not merely written comments but oral presentations. Congress evidently thought that this compromise procedure was a reasonable accommodation to the need for expedition where a lengthy completed § 18(c) proceeding needs to be supplemented.
The issue reduces itself, therefore, to whether § 18(e)(2) governed the remand. Petitioner’s principal argument that it does not is based upon the simple (and simply rejected) premise that § 18(d)(2)(B) categorically requires § 18(c) procedures for “substantive amendment to ... a rule promulgated under subsection (a)(1)(B).” It does indeed contain that requirement, but the requirement cannot possibly be categorical, since it must be reconciled with § 18(e)(2), which explicitly authorizes the Commission to “modify or set aside its rule or make a new rule by reason of the additional submissions and presentations.” The manner of reconciliation is obvious: the traditional principle of construction that the more specific prevails over the more general. It is § 18(e)(2), and not § 18(d)(2)(B), that governs the narrow category of case in which a reviewing court authorizes reopening of the record under the former section. There is no question that is what occurred here. Though petitioner at one point puzzlingly asserts “the impropriety of treating the proceedings used by the Commission here os if they were made pursuant to a section [18(e)(2)] remand,” Reply Brief for Petitioner at 14 (emphasis added), it is entirely beyond doubt that that is precisely how they were “made.” The Second Circuit remanded the Initial Rule to the Commission expressly “[pjursuant to [§ 18(e)(2),] 15 U.S.C. § 57a(e)(2),” Miller Motor Car Corp. v. FTC, Nos. 81-4144, etc., slip op. at 1 (2d Cir. Sept. 14, 1983) (Order), and directed the Commission to “reopen the record with respect to” the *207known-defects provision and “provide petitioners and all other interested persons with notice and an opportunity to submit comments and rebuttal thereto,” leaving it to the Commission’s discretion whether to allow oral presentations or any “other proceedings.” Slip op. at 4 (Stipulation ¶ 2, incorporated by reference into remand order).
Petitioner suggests, however, that a § 18(e)(2) remand was improper because it is only available where evidence has been improperly excluded in the original proceeding. That limitation has absolutely no foundation in the text of the statute, and we decline to limit the plain meaning of the text by resort to bits of legislative history that describe some, but not necessarily all, of the purposes that the provision serves. See Block v. Meese, 793 F.2d 1303, 1310 (D.C.Cir.1986). Cf. Vargas-Gonzalez v. INS, 647 F.2d 457 (5th Cir.1981) (interpreting 28 U.S.C. § 2347(c), a remand provision similar to § 18(e)(2)). We do not fear that reading the text as it is written will, as petitioner threatens, enable the Commission to write the revocation provision of § 18(d)(2)(B) out of existence whenever it wishes to “revok[e] a rule while a protracted judicial review of the rule is still pending.” Reply Brief for Petitioner at 13. The Commission can only make use of the § 18(e)(2) procedures by order of the reviewing court. Cf. Exxon Corp. v. Train, 554 F.2d 1310, 1316 n. 12 (5th Cir.1977) (interpreting 33 U.S.C. § 1369(c), similar remand provision of Federal Water Pollution Control Act Amendments of 1972); Greater Boston Television Corp. v. FCC, 463 F.2d 268, 283 (D.C.Cir.1971) (interpreting 28 U.S.C. § 2347(c)), cert. denied, 406 U.S. 950, 92 S.Ct. 2042, 32 L.Ed.2d 338 (1972).
Petitioner further suggests that the § 18(e)(2) remand was unlawful because the Second Circuit “held no hearing to determine the propriety of the remand,” Reply Brief for Petitioner at 15, and made no specific finding of the requisite “reasonable grounds for ... failure to make” the relevant submissions and presentations in the original proceeding, 15 U.S.C. § 57a(e)(2). But the statute does not require those formalities. Finally, petitioner’s assertion that “there is no indication that the Second Circuit was aware of [§ 18(d)(2)(B)], or that the court analyzed the propriety of the statutory remand in light of that section,” Reply Brief for Petitioner at 15, is patently incredible. The Miller petitioners had specifically directed the court’s attention to the cumbersome FTC Act rulemaking procedures: it was the Commission’s alleged failure to follow them that was the very basis of their procedural attack. The court could not have missed the statutory directive, codified a mere subsection away from § 18(e)(2), that applied the same procedures to the amendment or repeal of an FTC rule. And even if it were unaware of that provision it would have assumed the same result in any event, since procedural equation of revocation and promulgation is the general (indeed, as far as we are aware, the invariable) rule. See 5 U.S.C. § 551(5) (1982); Motor Vehicle Manufacturers Association v. State Farm Mutual Automobile Insurance Co., 463 U.S. 29, 41, 103 S.Ct. 2856, 2865, 77 L.Ed.2d 443 (1983) (“State Farm”); see also 15 U.S.C. § 2058(h) (1982) (Consumer Product Safety Act); 21 U.S.C. § 371(e)(1) (1982) (Food, Drug, and Cosmetic Act).
We are unable, on the record before us, to identify the particular reasons why the Second Circuit found “reasonable grounds for the ... failure to make” the submissions that were the subject of its § 18(e)(2) remand in the original proceeding. That it made such a remand, however, is unquestionable, and that decision is, at the very least, the law of this case if not an entirely separate determination of another court immune from collateral attack here. Even if it is merely the former, we cannot reconsider it absent a finding of clear error causing “manifest injustice.” See Laffey v. Northwest Airlines, Inc., 740 F.2d 1071, 1082-83 & n. 18 (D.C.Cir.1984) (per curiam), cert. denied, 469 U.S. 1181, 105 S.Ct. 939, 83 L.Ed.2d 951 (1985); City of Cleveland v. FPC, 561 F.2d 344, 346 (D.C.Cir.1977) (law *208of case binding on administrative agency). We can find no “manifest injustice” produced either by the protraction of these rulemaking proceedings or by the deprivation of cross-examination (which is standard in the vast majority of rulemakings).
Ill
Before turning to the merits, we pause to consider our standard of review. We may not set aside the Revised Rule unless it is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law,” see 15 U.S.C. § 57a(e)(3) (1982) (incorporating by reference 5 U.S.C. § 706(2)(A) (1982)), or the factual determinations on which it is based are “not supported by substantial evidence in the rule-making record ... taken as a whole,” 15 U.S.C. § 57a(e)(3)(A). See American Financial Services Association v. FTC, 767 F.2d 957, 985 (D.C.Cir.1985), cert. denied, - U.S. -, 106 S.Ct. 1185, 89 L.Ed.2d 301 (1986). Although this court flirted briefly with the notion that the “substantial evidence” standard of the FTC Act called for a more intensive review than the “arbitrary and capricious” standard applicable to rulemaking under the APA, see American Optometric Association v. FTC, 626 F.2d 896, 904-05 (D.C.Cir.1980), our more recent pronouncements are to the contrary. We have held that in the context of the APA, the substantial evidence test (which is applied only to formal adjudication and formal rulemaking, see 5 U.S.C. § 706(2)(E)) and the arbitrary and capricious test (which governs review of all proceedings, see 5 U.S.C. § 706(2)(A)) “are one and the same” insofar as the requisite degree of evidentiary support is concerned. Association of Data Processing Service Organizations, Inc. v. Board of Governors of the Federal Reserve System, 745 F.2d 677, 683 (D.C.Cir.1984). We have noted, moreover, that the APA standard of review must be deemed unaltered unless “Congress’s intent to make a substantive change [is] clear” on the statute’s face. Id. at 686. As application of these principles would require, our most recent review of a trade rule described the substantial evidence test of § 57(e)(3)(A) in the same fashion as the substantial evidence test of the APA is described, i.e., as demanding “ ‘such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.’ ” See American Financial Services Association v. FTC, 767 F.2d at 985 (citations omitted). Compare Consolo v. FMC, 383 U.S. 607, 619-20, 86 S.Ct. 1018, 1026-27, 16 L.Ed.2d 131 (1966). This is, as we have said, the same degree of evidentia-ry support needed to satisfy the arbitrary and capricious standard.
The Commission asserts that the gravamen of the present appeal is simply its “decision not to engage in rulemaking,” see Brief for Respondent at that presumably our review would be governed by the deferential standard applicable to an agency refusal to adopt a rule not required by law. See Professional Drivers Council v. Bureau of Motor Carrier Safety, 706 F.2d 1216, 1223 n. 24 (D.C.Cir.1983) (“[Wjhile an agency may possess the authority to regulate an area, substantial discretion will be accorded an agency’s decision not to promulgate regulations.”); id. at 1220-21, 1223; WWHT, Inc. v. FCC, 656 F.2d 807, 818 (D.C.Cir.1981) (“It is only in the rarest and most compelling of circumstances that this court has acted to overturn an agency judgment not to institute rulemaking.”); id. at 816-17. We must reject this contention. When an agency has declined entirely to enter a particular arena of regulation its discretionary judgment as to where its administrative and enforcement resources should best be expended is prominently involved. That is not the case where the agency merely fails to include in a final rule a proposed provision that is closely related to the provisions that are adopted. The latter case, which is what we have here, has never been thought to present an issue of agency refusal to adopt a rule but rather an issue of the rationality of the rule it did adopt, given the alternatives that were suggested and rejected. See Vermont Yankee Nuclear Power Corp. v. NRDC, Inc., 435 U.S. 519, 551, 98 S.Ct. 1197, 1215, 55 L.Ed.2d 460 *209(1978); American Financial Services Association v. FTC, 767 F.2d at 988; cf. Public Citizen v. Steed, 733 F.2d 93, 103-04 (D.C.Cir.1984) (failure to consider alternatives to rescission).
At the other end of the spectrum, we reject petitioner’s argument that our review must be particularly strict and probing in the present case because the agency has revoked a preexisting rule. Reply Brief for Petitioner at 17-22. We have already disapproved that characterization of what has occurred. Section 18(e)(2) furnishes the Commission an opportunity to promulgate the ultimate rule with the benefit of the new evidence that § 18(e)(2) accommodates, not merely an opportunity to repeal the old rule if repeal can be substantiated on the basis of that new evidence. The proceedings conducted after remand in the present case were part of the same rulemaking proceeding as that which produced the Initial Rule; and any revisions made during those proceedings are to be treated like any other agency decision to alter its course during an ongoing rule-making is, the Commission “is clearly free to reject an early conclusion after further study and reflection. All that is important is ... that the final [action is] supported by substantial evidence.” Marathon Oil Co. v. EPA, 564 F.2d 1253, 1272 (9th Cir.1977) (citing Consolo v. FMC, 383 U.S. at 620, 86 S.Ct. at 1026).
IV
The Commission rejected the known-defects disclosure alternative because in its view “the record does not support ... a conclusion that [its] benefits ... outweigh its costs.” 49 Fed.Reg. at 45,712. We address each side of the Commission’s balance, in turn.
Benefits
The Commission correctly reasoned that the benefits of the provision would depend upon (1) the extent to which dealers operating under it would know of specific mechanical defects in the vehicles they sell, and (2) the extent to which the stickers would convey this knowledge in an accurate and unconfusing way. 49 Fed.Reg. at 45,712-13. It found inadequate support to establish that either would be considerable.
As to the former factor, it must be borne in mind that the rule does not require dealers to inspect their cars before sale, nor does it include the “optional inspection” provision that had been suggested, which would have established a standard inspection and would have required the sticker to show, along with the known defects, whether that inspection had been conducted. See 49 Fed.Reg. at 45,718-20 (Revised Rule rejecting mandatory and optional inspection); 46 Fed.Reg. at 41,375-76 (Initial Rule rejecting mandatory and optional inspection). Without one or the other of these provisions, whose omission petitioner does not contest, there would obviously be no incentive for dealers to inform themselves of specific mechanical defects (as opposed to the general operability of the vehicle which would be apparent to a purchaser). As the Commission said:
[H]onest dealers who learn of defects must reveal their knowledge on the disclosure portion of the window sticker, whereas dealers who avoid gaining this knowledge may honestly leave the sticker blank. ... Thus, a dealer who regularly inspects and honestly discloses all “known defects” may be put at a competitive disadvantage relative to dealers who do not inspect. This fáctor may then have the unintended and perverse effect of discouraging, rather than encouraging, inspections and disclosure of defects.
49 Fed.Reg. at 45,713.
It was reasonable for the Commission to conclude that the degree of information dealers would have concerning specific defects under the disclosure provision would be no more, though it might well be less, than what they currently possess. And there was substantial evidence to support the related finding that it is uncertain whether, currently, dealers “ordinarily know about specific defects,” 49 Fed.Reg. at 45,712, since the walk-around inspections *210and test drives that they customarily perform before purchasing cars convey knowledge only of the car’s general condition, not of specific defects. See, e.g., Record Transcript (“Tr.”) 1557-58, 1576 (even experienced mechanic may not notice certain defects); Tr. 4077 (independent dealers lack repair facilities and trained personnel to inspect certain systems such as brakes and transmission); Tr. 4107 (defects not always apparent from test drive); Record at D-10, p. 3 (same); id. at D-19, p. 3 (dealers have no way of knowing exact defect information even after inspection); id. at D-56, p. 2 (dealers generally unaware of specific defects after car renovated). There is no reason for dealers to engage in a more probing inspection because most buyers “are likely to perform no more than a similarly superficial examination.” 49 Fed. Reg. at 45,713 (footnote omitted); see, e.g., Bureau of Consumer Protection, FTC, Sale of Used Motor Vehicles, Final Staff Report 84-87 (1978) (“Final Staff Report”); Tr. 3982-83 (test drive around block uninformative); Tr. 5213-14 (state official testifies that consumers often perform ritualistic inspection under hood without knowing what to look for); Tr. 5422 (consumers’ test drives are generally less than a mile). Instead, dealers spend more time and money “detailing” the car, i.e., reconditioning the car’s appearance, catering to customers’ perceptions that a “ ‘good looking’ car is also mechanically sound.” 49 Fed.Reg. at 45,713 (footnote omitted); see, e.g., Final Staff Report at 98 & n. 94; Tr. 265-66, 1585-86, 1785, 4103, 4431-32; Record at D-7, p. 2.
In addition to being unable to conclude that dealers would generally have knowledge of specific mechanical defects, the Commission judged that the sticker system would fail to convey that knowledge accurately and might even mislead potential purchasers. See 49 Fed.Reg. at 45,713-16. It based this judgment on two studies, and on a common-sense prediction of a window sticker’s effect, informed by record evidence of dealer and purchaser behavior.
The first of the studies, prepared for the FTC by the Center for Public Representation, Madison, Wisconsin, entitled An Investigation of the Retail Used Motor Vehicle Market: An Evaluation of Disclosure and Regulation (1977) (“Wisconsin Study”), had been cited by the Commission in promulgating the Initial Rule as support for the proposition that the known-defects provision would be effective. 46 Fed.Reg. at 41,342 n. 255. “[U]pon close review,” however, the Commission discerned certain contradictory indications within the study that rendered it, at best, “inconclusive.” 49 Fed.Reg. at 45,714. The study compared Wisconsin purchasers’ awareness of the mechanical condition of used cars after enactment of a Wisconsin known-defects law with Wisconsin purchasers’ awareness before that enactment, and with the awareness of purchasers in Minnesota, which had no such law.2 It showed that fewer Wisconsin buyers knew about the defects of the cars they bought than Minnesota buyers. See Final Staff Report at 121 (38.7% in Wisconsin, compared to 40.7% in Minnesota). Moreover, the Wisconsin known-defects law did not appear to alter significantly purchasers’ perception that they lacked information on the ear’s condition, see id. at 116 (31.7% pre-law, compared to 28.5% post-law), or that their dealer had accurately apprised them of the car’s condition, id. at 117 (62.6% pre-law, compared to 62.8% post-law).
The second study on which the Commission relied, prepared by the Bureau of Social Science Research, Washington, D.C., was entitled A Report on a National Survey of Private Buyers and Sellers of Used Automobiles (1982). This study, the so-called “Baseline Survey,” submitted to the Commission after it promulgated the Initial Rule, was a nationwide survey of car buy*211ers and sellers designed as a pre-Rule yardstick against which to gauge the effectiveness of the Used Car Rule. It revealed that before purchasing a used car, Wisconsin residents were no more likely to know about defects, and were no more likely to learn about defects from dealers, than were their out-of-state counterparts. Nor were they any more likely to be “very satisfied” with their cars; indeed, they were significantly more likely to be “very dissatisfied.” See 49 Fed.Reg. at 45,715 (62% of Wisconsin residents “very satisfied” versus 63% elsewhere; 12% “very dissatisfied” versus 5% elsewhere).
The generalizability of field studies is a matter that “rests within the expertise of [the agency], and upon which a reviewing court must be most hesitant to intrude.” State Farm, 463 U.S. at 53, 103 S.Ct. at 2872. We see no basis for disturbing the Commission’s judgments here.3
Finally, the Commission was concerned that the window sticker would affirmatively mislead potential purchasers, leading them to believe that a car with no known defects was actually defect-free, or that a car with no known defects was necessarily better than a car in which several defects were so obvious as to surface upon the dealer’s cursory inspection. See 49 Fed.Reg. at 45,715-16. “[T]he extensive evidence of dealer misrepresentations concerning the mechanical condition of their cars,” id. at 45,716, gave it little cause to hope that dealers would correct this misimpression. This is the type of prediction of commercial behavior and consumer perception that is uniquely within the Commission’s expertise and for which statistical support is neither readily available nor necessary. See FCC v. National Citizens Committee for Broadcasting, 436 U.S. 775, 813-14, 98 S.Ct. 2096, 2121-22, 56 L.Ed.2d 697 (1978); American Home Products Corp. v. FTC, 695 F.2d 681, 686 (3d Cir.1982).
Costs
Balanced against this finding of (at best) indeterminate benefits were the costs of the proposed provision. Petitioner attempts to trivialize these by focusing only on the dealer’s minimal expense of listing known defects on a window sticker. Brief for Petitioner at 23. This ignores entirely the provision’s massive enforcement costs, and its likely frustration of a central objective of the Used Car Rule. 49 Fed.Reg. at 45,716-18.
To enforce the known-defects provision, the Commission would have to prove, among other things, that a defect was present at the time of purchase (not just that it manifested itself shortly after purchase), and that the dealer knew of the defect. The Commission noted how difficult such proof had been in cases alleging known defects in new cars:
In most of these cases there have been thousands of vehicles with the relevant manufacturing defect, manufacturers have had regular reporting systems to identify problems, and engineering analy-ses have been able to establish the cause of the problem. Even so, establishing a manufacturer’s knowledge of a defect has been one of the most difficult aspects of these cases. Establishing that an individual used car dealer has knowledge of a particular defect in a series of sales of individual cars of different makes, models and ages would be a very difficult, resource-intensive undertaking.
49 Fed.Reg. at 45,717. The Commission concluded that the only way it could be done would be to field two battalions of enforcement personnel, the first of which would pose as potential shoppers, have the car inspected, and report to the dealer the defects found, and the second of which would return to the scene, ask to see the same car, and catch the dealer in a failure *212to disclose any of those defects. Id. at 45,716. Given the number of used-car dealers, even an effective “spot check” enforcement through a method such as this would be enormously time-consuming and expensive.
Perhaps an even greater cost, in the Commission’s view, would be the provision’s effect of frustrating an overarching purpose of the Used Car Rule. 49 Fed.Reg. at 45,718. The Revised ing dealers to post a window sticker that discloses the terms of any warranty, warns purchasers not to rely on dealers’ oral promises, lists a car’s mechanical and safety systems and their potential major defects, and encourages consumers to obtain a “pre-purchase aimed at inducing consumers not to rely upon dealer-controlled information. The Commission believed that objective, and in particular the objective of encouraging third-party inspections, would be compromised by including on the sticker itself & dealer representation concerning mechanical condition, as though that were reliable, and by the dealer’s ability to assure the buyer that “If we knew of any problems, we’d have to tell you about them.” 49 Fed.Reg. at 45,716. This predicted effect seems to us plausible, and we have no basis for disregarding it, even if it lacks supporting statistical data, see NRDC, Inc. v. SEC, 606 F.2d 1031, 1052 (D.C.Cir.1979); American Home Products Corp. v. FTC, 695 F.2d at 686.
In sum, we find the Commission’s assessment of the costs and benefits of the proposal to be a reasonable one, supported by the requisite substantial evidence; and we likewise find reasonable the conclusion that there is inadequate reason to believe that the benefits predominate. This is not to say, of course, that a reasonable person might not reach the opposite conclusion on both many of the subsidiary points and on the ultimate balancing the Commission itself originally did, when it issued the Initial Rule. “[T]he possibility of drawing two inconsistent conclusions from the evidence does not prevent an administrative agency’s finding from being supported by substantial evidence.” Consolo v. FMC, 383 U.S. at 620, 86 S.Ct. at 1026 (citations omitted). Nor does the fact that the Commission at one point reached one reasonable conclusion bar it from reaching a contrary reasonable conclusion upon further reflection. See Marathon Oil Co. v. EPA, 564 F.2d at 1272.
V
Petitioner raises for the first time on appeal a challenge to the impartiality of Commission Chairman Miller, based upon two of his statements. The first, allegedly made at a press conference shortly after Congress vetoed the Initial Rule, contained the assertion that “[t]his rule [without the known-defects provision] I should stress is the rule that I think best.” Reply Brief for Petitioner at A-1 (reprinting one page of uncited transcript). The second, allegedly made to a reporter near the end of the period for rebuttal comments,4 asserted that the Revised Rule would not contain a known-defects provision, id. at A-2 (reproducing uncited newspaper article), and explained that the provision was unnecessary because “most dealers are not mechanics, anyway,” id. at A-3 (affidavit of Bruce Ramsey paraphrasing Chairman Miller). At least as to the first of these statements, we have great doubts whether petitioner’s objection is timely. See United States v. L.A. Tucker Truck Lines, Inc., 344 U.S. 33, 37, 73 S.Ct. 67, 69, 97 L.Ed. 54 (1952) (“[C]ourts should not topple over administrative decisions unless the administrative *213body has not only erred but has erred against objection made at the time appropriate under its practice.”); Sanderlin v. United States, 794 F.2d 727, 734 (D.C.Cir.1986). Both objections are so insubstantial, however, that they are more readily disposed of on their merits.
Assuming that both statements were made as alleged, neither approaches the “clear and convincing evidence” that must be produced to prove that Chairman Miller had “an unalterably closed mind on matters critical to the disposition of the proceeding.” Association of National Advertisers, Inc. v. FTC, 627 F.2d 1151, 1170 (D.C.Cir.1979), cert. denied, 447 U.S. 921, 100 S.Ct. 3011, 65 L.Ed.2d 1113 (1980). The first of them indicated that he then favored deletion of the known-defects provision, but in no way demonstrated that that view could not be changed by the rulemaking proceedings that were to follow. The second of them likewise represents not an indication of unalterable prejudgment, but a prediction of Commission action and (implicitly) an announcement of Chairman Miller’s own considered position. It was inappropriate to provide either before all the relevant material was indeed before the Commission’s formal opinion issued. But that impropriety (still assuming it occurred) gives no indication of a mind that has been closed to the evidence in the past or that would disregard any significant new material subsequently introduced. This court has declined to find invalidating prejudice (in rulemaking cases) on substantially greater evidence of predisposition. See id. at 1172-74; United Steelworkers of America, AFL-CIO-CLC v. Marshall, 647 F.2d 1189, 1209-10 (D.C.Cir.1980), cert. denied, 453 U.S. 913, 101 S.Ct. 3148, 69 L.Ed.2d 997 (1981).
The petition for review is
Denied.
. Both versions of the Used Car Rule have applied not only to used cars, but also to used light-duty trucks. See 16 C.F.R. § 455.1(c)(1); 46 Fed.Reg. 41,328, 41,359 (1981).
. The Wisconsin law, unlike the Initial Rule’s known-defects provision, required dealers to inspect the cars, but, at the time of the studies, did not require that the disclosures be made on a window sticker. Instead, a defect-disclosure statement had to be presented to the purchaser before the sales contract was signed. See 49 Fed.Reg. at 45,714 n. 292. The Commission concluded that the latter distinction was not critical to the reliability of the comparison. Id.
. Petitioner’s objection that the Baseline Survey’s data are incompetent for any purpose other than their original purpose is based upon a misunderstanding of an introductory remark by the study’s author. What petitioner interprets as a "warning" not to use the Wisconsin data for purposes of comparison to other states, Brief for Petitioner at 39, was no more than an explanation of the reason for oversampling the Wisconsin population. See Baseline Survey at 2-3.
. The notice-and-comment period was 46 days. Chairman Miller’s statement was allegedly made eight days before the end of the 20-day period for rebuttal comments, and at a time when that was the last scheduled comment period. 49 Fed.Reg. 7835 (1984). On the last day of the period for rebuttal comments, however, NADA submitted a cost-benefit study of the known-defects provision, prompting the Commission to open an additional rebuttal period for responses to that document. 49 Fed.Reg. 17,517 (1984). Still later, when the Baseline Survey was made part of the rulemaking record, yet another comment period was announced for submissions on that document. 49 Fed.Reg. 30,511-12 (1984).