OPINION and ORDER
GORDON, Judge:This action involves an administrative review conducted by the U.S. Department of Commerce (“Commerce”) of the anti-dumping duty order covering polyethylene terephthalate film, sheet, and strip from Taiwan. See Polyethylene Terephthalate Film, Sheet, and Strip from Taiwan, 76 Fed. Reg. 76,941 (Dep’t of Commerce Dec. 9, 2011) (final results admin, review) (“Final Results”)', see also Issues and Decision Memorandum, A-583-837 (Dep’t of Commerce Dec. 5, 2011), available at http://ia.ita.doe.gov/frn/summary/taiwan/ 2011-31695-l.pdf (last visited this date) (“Decision Memorandum”). Before the *1350court is the motion for judgment on the agency record of Plaintiff Nan Ya Plastics Corporation, Ltd. (“Nan Ya”) challenging Commerce’s assignment of a total adverse facts available (“AFA”) rate of 74.34 percent to Nan Ya. See Pl.’s Rule 56.2 Mot. for J. upon the Agency R., ECF No. 38 (“Pl.’s Br.”). The court has jurisdiction pursuant to Section 516A(a)(2)(B)(iii) of the Tariff Act of 1930, as amended, 19 U.S.C. § 1516a(a)(2)(B)(iii) (2006),1 and 28 U.S.C. § 1581(c) (2006). For the reasons set forth below, the court remands this action to Commerce for further consideration.
I. Standard of Review
For administrative reviews of anti-dumping duty orders, the court sustains Commerce’s determinations, findings, or conclusions unless they are “unsupported by substantial evidence on the record, or otherwise not in accordance with law.” 19 U.S.C. § 1516a(b)(l)(B)(i). More specifically, when reviewing agency determinations, findings, or conclusions for substantial evidence, the court assesses whether the agency action is reasonable given the record as a whole. Nippon Steel Corp. v. United States, 458 F.3d 1345, 1350-51 (Fed.Cir.2006). Substantial evidence has been described as “such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.” Dupont Teijin Films USA v. United States, 407 F.3d 1211, 1215 (Fed.Cir.2005) (quoting Consol. Edison Co. v. NLRB, 305 U.S. 197, 229, 59 S.Ct. 206, 83 L.Ed. 126 (1938)). Substantial evidence has also been described as “something less than the weight of the evidence, and the 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. Fed. Mar. Comm’n, 383 U.S. 607, 620, 86 S.Ct. 1018, 16 L.Ed.2d 131 (1966). Fundamentally, though, “substantial evidence” is best understood as a word formula connoting reasonableness review. 3 Charles H. Koch, Jr., Administrative Law and Practice § 9.24[1] (3d ed. 2012). Therefore, when addressing a substantial evidence issue raised by a party, the court analyzes whether the challenged agency action “was reasonable given the circumstances presented by the whole record.” Edward D. Re, Bernard J. Babb, and Susan M. Koplin, 8 West’s Fed. Forms, National Courts § 13342 (2d ed. 2012).
II. Background
On August 31, 2010, Commerce initiated an administrative review of mandatory respondents Shinkong Materials Technology Co., Ltd. (“Shinkong”) and Nan Ya. See Initiation of Antidumping and Countervailing Duty Administrative Reviews and Deferral of Initiation of Administrative Review, 75 Fed. Reg. 53,274, 53,275 (Dep’t of Commerce Aug. 31, 2010). Nan Ya cooperated in the prior review, and Commerce calculated an antidumping duty rate of 18.30 percent based on Nan Ya’s sales and cost data. See Polyethylene Terephthalate Film, Sheet, .and Strip From Taiwan, 76 Fed. Reg. 18,519, 18,520 (Dep’t of Commerce Apr. 4, 2011) (amended final results). In the present administrative review Nan Ya chose not to cooperate, failing to respond to Commerce’s request for information. Commerce therefore preliminarily assigned Nan Ya a total AFA rate of 99.31 percent derived from two transaction-specific margins that were calculated for Nan Ya during the prior administrate review. See Polyethylene Terephthalate Film, Sheet, and Strip from Taiwan, 76 *1351Fed. Reg. 47,540 (Dep’t of Commerce Aug. 5, 2011) (preliminary results); Decision Memorandum at 5.
Before the agency, Nan Ya argued that Commerce did not adequately corroborate the total AFA rate, and that Commerce should instead select Nan Ya’s total AFA rate from data available on the current administrative review, and more specifically, the transaction-specific data of the cooperating respondent, Shinkong. Nan Ya Admin. Case Br. at 7, PD 23 (Oct. 4, 2011).2 In the Final Results Commerce obliged, selecting the highest transaction-specific margin from among Shinkong’s data — 74.34 percent — as Nan Ya’s total AFA rate. See Memorandum from Gene H. Calvert to Mark Hoadley, Final Results in the Antidumping Duty Administrative Review of Polyethylene Terephthalate Film, Sheet, and Strip from Taiwan: Assignment of the Adverse Facts Available Rate for Nan Ya Plastics Corporation, Ltd. (Nan Ya), CD 27 (Dec. 5, 2011) (“Final AFA Memo ”). Commerce corroborated the 74.34 percent rate against Nan Ya’s own transaction-specific margins from the prior review, and found multiple transactions at or above the 74.34 percent rate. Id. at 3. Nan Ya now challenges the total AFA rate of 74.34 percent as “an unlawful aberrant outlier” and not reflecting Nan Ya’s “commercial reality albeit with some built in increase to induce compliance.” Pl.’s Br. at 3.
III. Discussion
In a total adverse facts available scenario like the one presented here, Commerce typically cannot calculate an anti-dumping rate for an uncooperative respondent because the information required for such a calculation (the respondent’s sales and cost information for the subject merchandise during the period of review) has not been provided. As a substitute, Commerce relies on various “secondary” sources of information (the petition, the final determination from the investigation, prior administrative reviews, or any other information placed on the record), 19 U.S.C. § 1677e(b) & (c), to select a proxy that should be a “reasonably accurate estimate of the respondent’s actual rate, albeit with some built-in increase intended as a deterrent to noncompliance.” F.lli De Ceceo Di Filippo Fara S. Martino S.p.A. v. United States, 216 F.3d 1027, 1032 (Fed.Cir.2000) (“De Ceceo ”). When selecting an appropriate total AFA proxy, “Commerce must balance the statutory objectives of finding an accurate dumping margin and inducing compliance.... ” Timken Co. v. United States, 354 F.3d 1334, 1345 (Fed.Cir.2004). The proxy’s purpose “is to provide respondents with an incentive to cooperate, not to impose punitive, aberrational, or uncorroborated margins.” De Ceceo, 216 F.3d at 1032. Although a higher AFA rate creates a stronger incentive to cooperate, “Commerce may not select unreasonably high rates having no relationship to the respondent’s actual dumping margin.” Gallant Ocean (Thailand) Co. v. United States, 602 F.3d 1319, 1323 (Fed.Cir.2010) (citing De Ceceo). “Commerce must select secondary information that has some grounding in commercial reality.” Id. at 1323-24.
As De Ceceo explained, these requirements are logical outgrowths of the statute’s corroboration requirement, 19 U.S.C. § 1677e(c), which mandates that Commerce, to the extent practicable, corroborate secondary information. See De Ceceo, 216 F.3d 1027, 1032. In practice “corroboration” involves confirming that secondary information has “probative value,” 19 C.F.R. § 351.308(d), by examining its “reliability and relevance.” Mittal *1352Steel Galati S.A. v. United States, 31 CIT 730, 734, 491 F.Supp.2d 1273, 1278 (2007) (citing Ball Bearings and Parts Thereof from France, Germany, Italy, Japan, Singapore, and the United Kingdom, 70 Fed. Reg. 54,711, 54,712-13 (Sept. 16, 2005) (final results)). More simply, to corroborate the selection of a total AFA rate, Commerce must (to the extent- practicable), “demonstrate that the rate is reliable and relevant to the particular respondent.” Yantai Xinke Steel Structure Co. v. United States, 36 CIT-, 2012 WL 2930182, 15 (July 18, 2012).
Commerce has in the past demonstrated the reliability and relevance of an AFA rate for particular respondents by analyzing the uncooperative respondents’ transaction-specific margins from prior proceedings, if available. See, e.g., PAM, S.p.A. v. United States, 582 F.3d 1336, 1340 (Fed. Cir.2009); Ta Chen Stainless Steel Pipe, Inc. v. United States, 298 F.3d 1330, 1339-40 (Fed.Cir.2002); Fujian Lianfu Forestry Co. v. United States, 34 CIT-, 700 F.Supp.2d 1361, 1363 (2010). Commerce did that here, analyzing Nan Ya’s transaction-specific margins from the prior administrative review and finding multiple transactions above the 74.34 percent rate. Final AFA Memo at 3. Commerce therefore appeared to corroborate Nan Ya’s AFA rate; Commerce demonstrated that the rate is reliable and relevant to Nan Ya by analyzing Nan Ya’s own transaction-specific data (albeit without affording Nan Ya the opportunity to comment upon the prior review data that Commerce used).
That is not all Commerce did. Commerce also analyzed whether the rate was “unusual ... or ... aberrational” by analyzing Shinkong’s sales data. Decision Memorandum at 6. Commerce examined Shinkong’s [Confidential Data Deleted \ transaction sales margins for this administrative review in ascending or descending order. It found that there were “no specific gaps between the transaction specific margins,” and the 74.34 percent AFA rate was “within a range of margins.” Final AFA Memo at 3; Decision Memorandum at 6. Moreover, Commerce analyzed the underlying U.S. sales transaction that resulted in the 74.34 percent margin and found the sale was neither aberrational nor an outlier because the sales quantity of [Confidential Data Deleted ] kg was within [Confidential Data Deleted ] percent of the average quantity of all of Shinkong’s U.S. sales. Final AFA Memo at 3. Commerce, therefore, concluded the sale was “representative of the company’s sales practice,” because [Confidential Data Deleted] of Shinkong’s [Confidential Data Deleted] transactions had quantities above [Confidential Data Deleted ] kg and [Confidential Data Deleted ] of the transactions had quantities below that amount. Final AFA Memo at 3. Based on these facts, Commerce determined the selected margin was not aberrational or an outlier.
Nan Ya contests Commerce’s finding that the AFA rate is non-aberrational by presenting a robust statistical argument. First, Nan Ya explains that the 74.34 percent AFA rate reflects only [Confidential Data Deleted] percent of Shinkong’s total sales and only [Confidential Data Deleted] of [Confidential Data Deleted ] U.S. sales. Additionally, it explained that the underlying transaction for the rate involved a rare product, which was sold in only [Confidential Data Deleted ] of Shinkong’s [Confidential Data Deleted] U.S. sales. Pl.’s Br. at 6 (citations omitted). Nan Ya also argues that
Commerce when calculating the dumping margin on Shinkong’s U.S. sales, on average the U.S. sales are each individually compared to [Confidential Data Deleted] home market sales, where the average total quantity of the home market sales used in the dumping margin *1353calculation of each particular U.S. sale was [Confidential Data Deleted ] kg. In stark contrast, the Shinkong U.S. sale with the 74.34% dumping margin was based on a comparison to [Confidential Data Deleted ] home market sale, whose total quantity is [Confidential Data Deleted] kg — i.e., [Confidential Data Deleted ]% of the average quantity of total home market sales used in dumping margin calculation as to the other Shinkong U.S. sales for which a dumping margin was calculated.
Id. at 7 (citations omitted).
In examining Shinkong’s [Confidential Data Deleted ] U.S. transaction sales margins, Nan Ya argues that the [Confidential Data Deleted ] highest dumping margins are distinct from the rest of the margins in terms of their incremental increases in percentage. It compared the gap between the dumping margins of Shinkong’s sales that exclude the [Confidential Data Deleted ] highest margins with the gap between the [Confidential Data Deleted] highest margins themselves. Nan Ya found the [Confidential Data Deleted ] highest margins to have an average (mean) gap of [Confidential Data Deleted ] percent, a median gap of [Confidential Data Deleted ] percent, and a highest observed gap of [Confidential Data Deleted ] percent. PL’s Br. at 8 (citations omitted). In contrast, the sales that exclude the [Confidential Data Deleted ] highest margins have an average gap of [Confidential Data Deleted ] percent, a median gap of [Confidential Data Deleted ] percent, and a highest observed gap of [Confidential Data Deleted ] percent. Id. From this data, Nan Ya deduced that the average dumping margin gap for Shinkong’s top [Confidential Data Deleted] U.S. sales is over [Confidential Data Deleted ] times more than the gaps for the remaining sales. Id. at 9.
In addition, Nan Ya relies upon the U.S. Internal Revenue Service (“IRS”) in making another statistical argument that the 74.34 percent AFA rate is aberrational. Nan Ya contends that the 74.34 percent AFA rate is an outlier by referencing the IRS’ method of evaluating the interquartile range from a data set to determine a typical versus aberrational result. Id. at 8-9 (citing Intercompany Transfer Pricing Regulations Under Section J¡,82, 59 Fed. Reg. 34,971, 34,995 (Internal Revenue Service July 8, 1994) (Treas. Reg. 1.482-l(e)(2)(iii)(B) & (C))). Nan Ya explains,
[t]he interquartile range comparison (applied here) is the average dumping margin for the 25% of sales with the lowest dumping margin compared to the average dumping margin for the 25% of sales with the highest dumping margin.
For Shinkong, those dumping margin figures are [Confidential Data Deleted]% and [Confidential Data Deleted]%, respectively. The 74.34% AFA rate is well outside the rate that this widely accepted statistical methodology used by the IRS as to what is non-aberrational (representative) and rather indicates that it is in fact aberrational. The Final Results’ 74.34% AFA rate is [Confidential Data Deleted]% higher than the upper interquartile dumping margin range; [Confidential Data Deleted ]% percentage points higher.
PL’s Br. at 10 (citations omitted).
Finally, Nan Ya argues the 74.34 percent AFA rate is an outlier by analyzing Shinkong’s data and the standard deviation of the [Confidential Data Deleted ] sales, excluding the values that are more than one or two standard deviations from the average. Id. at 10-11. Nan Ya emphasizes that this method of statistical analysis is consistent with Commerce’s standard deviation analysis in other contexts, such as determining whether there has been targeted dumping. Id. at 11 (citing Certain Steel Nails from China, 73 Fed. Reg. 33,977 (Dep’t of Commerce June *135416, 2008) (final determ.), Issues and Decision Memorandum at Cmt. 3; Multilayered Wood Flooring from China, 76 Fed. Reg. 64,318 (Dep’t of Commerce Oct. 18, 2011) (final determ.), Issues and Decision Memorandum at Cmt. 4; High Pressure Steel Cylinders from China, 76 Fed. Reg. 77,964, 77,968 (Dep’t of Commerce Dec. 15, 2011) (prelim, determ.); Certain Steel Nails from UAE, 76 Fed. Reg. 68,129, 68,133 (Dep’t of Commerce Nov. 3, 2011) (prelim, determ.)). Nan Ya calculates Shinkong’s mean dumping margin to be [Confidential Data Deleted ] percent and the standard deviation to be [Confidential Data Deleted]. Therefore, the dumping margins within one standard deviation of the mean dumping margin are [Confidential Data Deleted ] percent, and the dumping margins within two standard deviations of the mean are [Confidential Data Deleted] percent. Pl.’s Br. at 11. Nan Ya further explains,
Statistics methodology supports that there is a 95% probability that actual dumping margins are. within two standard deviations of the mean, which here means a dumping margin under [Confidential Data Deleted]%. Even at the upper limit of the two (2) standard deviation range (i.e., [Confidential Data Deleted ]), the 74.34% AFA rate is over [Confidential Data Deleted ] percentage points [Confidential Data Deleted]. In fact, the AFA rate is [Confidential Data Deleted] standard deviations more than the mean.
Finally, Shinkong’s weighted-average dumping margin for this POR is 6.98%. Shink[o]ng’s 74.34% highest dumping margin that Commerce used as AFA, which is almost eleven times more than 6.98% average, is again clearly an outlier (i.e., aberrant).
Pl.’s Br. at 12 (citations omitted).
Because Commerce changed the AFA rate from the preliminary results to the final, Nan Ya’s first opportunity to challenge the total AFA rate was in its brief before the court. This means that the agency has not had the opportunity to consider these arguments in the first instance. Defendant’s response presents the post hoc rationalizations of agency counsel to which the court may not defer. See Burlington Truck Lines, Inc. v. United States, 371 U.S. 156, 168-69, 83 S.Ct. 239, 9 L.Ed.2d 207 (1962) (“The courts may not accept ... counsel’s post hoc rationalizations for agency action; ... an agency’s discretionary order [must] be upheld, if at all, on the same basis articulated in the order by the agency itself.”). The court believes a remand is appropriate for the agency to address these issues in the first instance. Nan Ya has presented what appear to be good and compelling statistical arguments that test the reasonableness of Commerce’s total AFA rate. Commerce needs to address them.
The case also presents an interesting issue about corroboration. Defendant explains Commerce’s view that because the 74.34 percent rate was “obtained in the course of ... [the] review,” 19 U.S.C. § 1677e(c), it “is not secondary information” that Commerce must corroborate. Def.’s Resp. to PL’s R. 56.2 Mot. at 11, ECF No. 48. The statute provides that “when [Commerce] ... relies on secondary information rather than on information obtained in the course of [a] ... review, [Commerce] ... shall, to the extent practicable, corroborate that information from independent sources that are reasonably at [Commerce’s] disposal.” 19 U.S.C. § 1677e(c). The AFA rate in this case was derived from the cooperative respondent’s data obtained in the course of the review, which, according to Commerce, discharges its need to corroborate.
This argument raises a number of issues. First, on a practical level, the court *1355is left wondering why Commerce analyzed Nan Ya’s transaction-specific data from the prior review, effectively corroborating the rate. In other words, why corroborate if no corroboration is required? Second, and more important, Nan Ya’s entire case is predicated on AFA standards that emanate from the statute’s corroboration requirement. See PL’s Br. at 3 (Commerce’s selected AFA rate does not reflect Nan Ya’s “commercial reality albeit with some built-in increase to induce compliance”); see also De Ceceo, 216 F.3d 1027, 1032 (“It is clear from Congress’s imposition of the corroboration requirement in 19 U.S.C. § 1677e(e) that it intended for an adverse facts available rate to be a reasonably accurate estimate of the respondent’s actual rate, albeit with some built-in increase intended as a deterrent to non-compliance.”). If corroboration is inapplicable, what happens to the De Ceceo standard on which Nan Ya’s case depends, and on which the Court of International Trade and the Court of Appeals for the Federal Circuit currently evaluate the reasonableness of Commerce’s selection of AFA rates? Commerce and the parties need to address this.
IV. Conclusion
Accordingly, it is hereby
ORDERED that this action is remanded to Commerce to respond to Nan Ya’s statistical arguments challenging the reasonableness of the total AFA rate, as well as to provide a further explanation of the supposed inapplicability of the corroboration requirement with a detailed explanation of what statutory standards govern Commerce’s selection of a total AFA rate if the De Ceceo corroboration standard does not apply; it is further
ORDERED that Commerce shall file its remand results on or before March 28, 2013; and it is further
ORDERED that, if applicable, the parties shall file a proposed scheduling order with page limits for comments on the remand results no later than seven days after Commerce files its remand results with the court.
. Further citations to the Tariff Act of 1930, as amended, are to the relevant provisions of Title 19 of the U.S.Code, 2006 edition.
. "PD” refers to a document contained in the public administrative record. "CD” refers to a document contained in the confidential record.