No. 24-4063

D.C. No. 3:23-cv-00707AN

Appeal from the United States District Court for the District of Oregon Adrienne C. Nelson, District Judge, Presiding

Argued and Submitted August 19, 2025 Portland, Oregon

Filed March 18, 2026

Before: Consuelo M. Callahan and Salvador Mendoza, Jr., Circuit Judges, and G. Murray Snow, District Judge. *

Opinion by Judge Callahan

Page 2SUMMARY **

The panel reversed the district court’s decision affirming the denial of a claimant’s applications for disability insurance benefits and supplemental security income under Title II and XVI of the Social Security Act, and remanded to the agency for further proceedings to address discrepancies and inconsistencies concerning job numbers.

Step five of the sequential evaluation process to determine whether a claimant is disabled requires the administrative law judge ('ALJ') to decide if there is a significant number of jobs in the national economy that a person with a claimant's limitations, age, education, and experience can perform. To make this decision, an ALJ may seek guidance from a vocational expert (“VE”).

The panel held that the ALJ erred by failing to address and resolve the inconsistencies between the claimant’s job -number evidence and that of the VE. The panel applied the rule articulated in Wischmann v. Kijakazi , 68 F.4th 498, 505 (9th Cir. 2023), and clarified the analysis used to determine whether evidence is “probative.” First, the panel held that claimant's contrary job -number evidence was significant and probative. Next, the panel held that the ALJ was required to both address the discrepancy and resolve the inconsistency between the job-number estimates provided by the claimant and the VE. Here, the ALJ failed to properly address the inconsistencies between the data. Accordingly, Page 3the panel reversed the district court’s decision with instructions to remand to the agency.

Alyson R. Young (argued) and Kevin Kerr, Kerr Robichaux & Carroll, Portland, Oregon, for Plaintiff-Appellant.

Edmund J. Darcher (argued), Special Assistant United States Attorney; Mathew W. Pile, Associate General Counsel; Office of Program Litigation, Social Security Administration, Baltimore, Maryland; Kevin C. Danielson, Executive Assistant United States Attorney; Natalie K. Wight, United States Attorney; Office of the United States Attorney, United States Department of Justice, Portland, Oregon; for Defendant-Appellee.

CALLAHAN, Circuit Judge:

In reviewing a claimant's applications for disability insurance benefits and supplemental security income, an Administrative Law Judge ('ALJ') employs a five-step sequential evaluation process to determine whether the claimant is disabled. See 20 C.F.R. §§ 404.1520(a), 416.920(a). Step five of that process requires the ALJ to decide if there is a significant number of jobs in the national economy that a person with a claimant's limitations, age, education, and experience can perform. Id.

Page 4§§ 404.1520(a)(4), 416.920(a)(4). To make this decision, an ALJ ma y seek guidance from a vocational expert (“VE”). Id. §§ 416.960(b), 416.966(e).

Should a claimant disagree with a VE as to the number of jobs available in the national economy that they can perform, a claimant may challenge th e VE’s testimony . For instance, a claimant may produce counter job-number evidence to rebut the VE’s conclusions. In that event, if the counter evidence is “significant and probative,” then the ALJ has a duty to address and resolve the inconsistency. Wischmann v. Kijakazi , 68 F.4th 498, 505 (9th Cir. 2023) (citation omitted).

This case concerns that very circumstance. Here, we face the question of whether the ALJ erred by failing to address and resolve the inconsistencies between a claimant ’s job-number evidence and that of a VE. We answer in the affirmative. In so holding, we apply the rule articulated in our most recent decision, Wischmann , and clarify the analysis used to determine whether evidence is “probative.” At base, because the claimant’s contrary job -number evidence here is significant and probative, the ALJ erred by failing to sufficiently address it. Accordingly, we reverse with instructions to remand to the agency.

Daniele Powley filed her Title II application for disability insurance benefits in October 2014 and her Title XVI application for supplemental security income in February 2015. In applying for disability benefits, Powley alleged disability due to several conditions, including multiple sclerosis, Sjogren’s syndrome, sleep apnea, osteoarthritis, chronic depression, post-traumatic stress disorder, and anxiety. On August 1, 2017, Powley appeared Page 5for a hearing before an ALJ regarding her claim for benefits, and on September 7, 2017, the ALJ issued a decision finding Powley not disabled. The Appeals Counsel denied review, and Powley appealed to the district court, which remanded for reasons not at issue in this appeal. Following the remand from the district court, Powley appeared for a second hearing before a different ALJ. The ALJ's determination in the second hearing is the subject of this appeal, and we discuss it in some detail below.

ALJ 's Findings: The Social Security Administration ('SSA') has established a five -step sequential evaluation process for determining whether an individual is disabled. See 20 C.F.R. §§ 404.1520(a), 416.920(a). The ALJ applied the five-step sequential framework and made the following findings of fact and conclusions of law. At step one, the ALJ found that Powley has not engaged in substantial gainful activity since the alleged onset date of March 14, 2013. At step two, the ALJ found that Powley has the following severe impairments: multiple sclerosis, Sjogren’s syndrome, asthma, cervical spine degenerative disc disease, and major depressive disorder. And at step three, the ALJ found that Powley does not have an impairment or combination of impairments that meets or medically equals the severity of one of the listed impairments in 20 C.F.R. Part 404, Subpart P, Appendix 1.

Before continuing to step four, the ALJ determined Powley's residual functioning capacity ('RFC'). See 20 C.F.R. §§ 404.1520(e), 416.920(e). The ALJ determined that Powley has the RFC to perform sedentary work as Page 6defined in 20 C.F.R. § 404.1567(a) and § 416.967(a) with the following exceptions:

· “She is limited to no more than occasional balancing, stooping, crouching, crawling, kneeling, or climbing of ramps and stairs.” · “She [is] precluded from climbing ropes, ladders, or scaffolds.' · “She can use bilateral foot controls occasionally.' · 'She is further limited to no more than occasional exposure to vibrations, temperature extremes, unprotected heights, moving machinery, and similar hazards.” · She “would need to avoid concentrated exposure to dust, fumes, gases, poor ventilation, and other noxious odors.” · She is ' limited to simple, repetitive, routine tasks.”

With this RFC established, the ALJ moved to step four and concluded that Powley is unable to perform any past relevant work. Finally, at step five, the ALJ determined that “[b]a sed on the testimony of the vocational expert, . . . considering [Powley] 's age, education, work experience, and residual functional capacity, [Powley] is capable of making a successful adjustment to other work that exists in significant numbers in the national economy.'Page 7 Based on his finding at step five, the ALJ determined that Powley was 'not disabled.'

Vocational Expert 's Testimony: The testimony of the VE was central to the ALJ’s determination at step five of the disability analysis. Therefore, we recount the VE’s testimony here. On March 9, 2021, the ALJ heard testimony from Vernon Arne, a VE with over 30 years of experience. During the hearing, the ALJ posed a hypothetical to Arne. First, the ALJ asked Arne to think of “an individual with the same age, education, and work experience as [Powley],' who has her same RFC and limitations. The ALJ then asked Arne if this person “ would . . . be capable of performing any jobs that exist . . . in significant numbers in the national and regional economy.” Arne responded in the affirmative and identified three jobs: (1) eyeglass assembler, identified in the Dictionary of Titles (“DOT”) as 713.687 -018, with 39,425 positions; (2) jewelry preparer, identified in the DOT as 700.687-062, with 24,700 positions; and (3) fishing equipment assembler, identified in the DOT as 732.684-062, with 14,700 positions.

On cross examination, Powley’s attorney asked Arne to identify his sources for the job numbers. Arne initially responded that he used 'a number of different sources , ” including standard Bureau of Labor documentation, some commercial services, and “manufacturer’s databases.” Arne then testified that the commercial products used were Data Axle or SkillTRAN , and the manufacturer’s database was a 'business directory , ' like those one 'might find through Chamber of Commerce or through other . . . business publi cations.” 1 Powley's attorney then inquired about Page 8which specific method was used for each of the three identified jobs. For eyeglass assembler, Arne testified that he used the manufacturer database method — noting that it provided him information as to the number of optical labs in the United States. For jewelry preparer and fishing equipment assembler, Arne testified that he relied on SkillTRAN.

Powley’s post -hearing evidence: Following the hearing, Powley submitted objections to Arne’s job numbers and provided contrary evidence using the same or similar methods deemed acceptable by the SSA. In the objection letter, Powley noted that the job numbers Arne provided for jewelry preparer (DOT 700.687-062) and fishing reel assembler (DOT 732.684-062) came “ directly from SkillTRAN 's Job Browser Pro (JBP)' but asserted that “Arne is either not using an updated version of JB P or is not correctly interpret ing the information.' While Arne testified that there are 24,700 positions for jewelry preparer, Powley’s attorney found that JPB reports “ not applicable ' for jewelry preparer. And while Arne testified that there are 14,700 positions for fishing reel assembler, Powley’s attorney found that JPB shows only 19 positions.

Further, Powley acknowledged that for optical or eyeglass assembler (DOT 713.687-018), Arne had “consulted a manufacturer database” to conclude “that there are 39,000 people working as final assembler, optical goods.' Powley, however, highlighted that this number reflects only a single part of the entire optical industry that Page 9Position Eyeglass assembler

would otherwise be composed of skilled jobs and other nonskilled positions. Powley then challenged this number, pointing out that according to the 2014 census, there were only 24,910 positions in the entire optical industry. Powley asserted that 'it is extremely unlikely that the [optical] industry has increased in size so much ” that the number of people working in a single part of that industry today amounts to almost double the number of people who worked in the entire industry in 2014. Powley also provided a JPB report showing only 27 jobs nationally for optical or eyeglass assembler (DOT 713.687-018), in contrast to Arne’s number of 39,425.

The chart below shows the differences between Arne’s job numbers and Powley’s job numbers, and their respective data sources:

VE Arne Powley’s

Powley’s

Job VE Source Job

Source

Numbers Numbers

Manufacturer SkillTRAN

database and U.S.

(DOT 39,425 27

/business Census

713.687- directory Bureau 018)

Jewelry

Preparer

(DOT 24,700 SkillTRAN N/A SkillTRAN 700.687-

062)

Page 1014,700 SkillTRAN 19 SkillTRAN

ALJ Response to Counter Evidence: The ALJ only briefly discussed Powley’s counter evidence. He stated:

Subsequent to the hearing, the claimant's representative submitted objections to the vocational expert’s job numbers. The undersigned has considered these objections, but finds them unpersuasive. Rather, the undersigned is more persuaded by the vocational expert’s testimony, because it was based on his consideration of specific hypotheticals posed at the hearing, in conjunction with his professional experience, knowledge, and expertise, which includes over 30 years as a vocational rehabilitation counselor.

Appeals Counsel Review: Powley appealed the ALJ’s decision to the Appeals Counsel, arguing that the ALJ erred by failing to discuss her counter evidence and by failing to provide any reasonable explanation for rejecting the counter evidence in whole or in part. In her request for review, Powley reasserted that “Counsel has generated re ports from Job Browser Pro on the three identified jobs' and these reports contradict the numbers provided by Arne. Powley argued that her numbers demonstrate that jobs do not exist Page 11in significant numbers in the national economy that someone with Powley’s RFC could perform. The Appeals Counsel denied review, relying on the ALJ’s deference to Arne.

District Court Review: Powley appealed the ALJ's decision to the district court. The district court affirmed, reasoning that even if Powley used the same methodology as Arne, “alternative job numbers are not considered significant or probative if no information about how the job numbers were produced is provided.” The district court concluded that Powley did not explain who generated the alternative numbers, whether that individual had the requisite level of expertise, or what methodology was employed.

We have jurisdiction under 28 U.S.C. § 1291. We “review the district court’ s order affirming the ALJ ’ s denial of social security benefits de novo, and will disturb the denial of benefits only if the decision contains legal error or is not supported by substantial evidence.” Lambert v. Saul , 980 F.3d 1266, 1270 (9th Cir. 2020) (quoting Tommasetti v. Astrue , 533 F.3d 1035, 1038 (9th Cir. 2008)).

We begin by setting forth the legal framework established by our precedent. We then apply that framework to the facts of this case and conclude that the ALJ erred by failing to address and resolve the inconsistencies between the VE’s and Powley’s job -number evidence.

The question here is whether the ALJ erred in failing to address and resolve discrepancies between a claimant's counter evidence and the evidence of a VE. Four cases Page 12govern our analysis: Buck v. Berryhill , 869 F.3d 1040 (9th Cir. 2017), Kilpatrick v. Kijakazi , 35 F.4th 1187 (9th Cir. 2022), White v. Kijakazi , 44 F.4th 828 (9th Cir. 2022), and Wischmann v. Kijakazi , 68 F.4th 498 (9th Cir. 2023).

In Buck , the VE testified that there were 600,000 bottling-line attendant jobs, 8,800 bottle-packer jobs, and 235,000 conveyor-belt maker jobs in the national economy. 689 F.3d at 1047. Using the same software relied on by the VE, Buck's attorney found significantly different job numbers. Id. He found only 231 bottling-line attendant jobs, 2,039 bottle-packer jobs, and 26 conveyor-belt maker jobs nationally. Id. The ALJ curtailed cross-examination of the VE at the hearing but allowed Buck to submit his objections post hearing. Id. The ALJ then denied Buck benefits without addressing his post-hearing objections. Id. We reversed , holding that 'the vast discrepancy between the VE ’ s job numbers and those tendered by Buck, presumably from the same source, is simply too striking to be ignored.” Id. at 1052. In so doing, we concluded that where such a vast inconsistency is presented, the ALJ must address it. Id.

Kilpatrick built on the Buck framework. There, we held that “an ALJ may not ignore significant probative evidence that bears on the disability analysis.” Kilpatrick , 35 F.4th at 1193. If a claimant produces evidence counter to the VE that is significant and probative, the ALJ must address it and explain why that evidence is being rejected. Id. Applying this standard, we rejected Kilpatrick’s arguments. First , we concluded that it was no surprise that Kilpatrick 's attorney arrived at different job numbers because his attorney did not replicate the VE's same methodology but rather used his own unique methods. Id. at 1194. We concluded that “there is no basis to conclude that these results qualified as significant probative evidence that the ALJ was required Page 13specifically to address' because , inter alia , Kilpatrick’s attorney had no identified expertise in calculating the figures and the attorney had conceded that his figures “may not be completely accurate.” Id.

Less than three months later, we again addressed this issue in White . In that case, we determined that White’s evidence “was produced using a data source and methodology frequently relied on by the SSA,' which produced 'job estimates [that] differed substantially from those of the VE.” White , 44 F.4th at 837. Specifically, we recognized that 'White estimatedusing SkillTRAN Job Browser Pro and the same DOT codes the VE had used — that there were 2,957 table worker, 0 assembler, and 1,333 film touch-up inspector jobs in the national economy, ” whereas '[t]he VE had testified that there were 72,000, 65,000, and 32,000 jobs, respectively, in these same occupations.” Id. We found this discrepancy comparable to the one in Buck and remanded “to allow the ALJ to address the evidence and to resolve the inconsistency between the jobnumber estimates provided by White and by the VE.” Id.

A year later, in Wischmann , we articulated the following rule based on our three prior cases: “to determine whether the ALJ had a duty to address a conflict in job-number evidence (and failed to discharge that duty), we consider on a case-by-case basis whether new evidence submitted by a claimant is 'meritless or immaterial' or has 'significant pr obative' value.' 68 F.4th at 506 (citing Kilpatrick , 35 F.4th at 1193). 'If the new evidence is significant and probative, we must remand to the ALJ to address the inconsistency in the record evidence.” Id. (first citing White , 44 F.4th at 837; and then citing Buck , 869 F.3d at 1052).

Page 14Applying this rule, we held that Wischmann’s “six pages of printouts' were insufficient to meet the significant probative evidence standard. Id. We reasoned that the letter submitted in conjunction with this evidence did not show that the proffered data was produced using the same methodology as that of the VE and provided no information about how the job numbers were produced, other than the software name. Id. at 506-07. Additionally, we noted that the letter provided no information about what queries were entered into the program, what variables were changed, or what filters were applied to the data, and “[n]either the letter nor the pages themselves state[d] that the printout data was produced with [JBP]. ” Id. The court concluded that since the letter and attachments were not “ probative evidence, ' they did “ not give rise to the sort of inconsistency in the evidence that an ALJ is required to resolve.' Id. (citing Kilpatrick , 35 F.4th at 1193-94).

Pursuant to Wischmann , our inquiry here is twofold. First, we must determine if Powley's counter evidence is “significant and probative” requiring the ALJ to address it. Id. at 506. If her counter evidence is significant and probative, then we must determine whether the ALJ sufficiently addressed and resolved the inconsistencies created by the counter evidence. Id. ; see also White , 44 F.4th at 837. Our prior cases begin by addressing whether counter evidence is probative before looking to whether the counter evidence is significant. See, e.g. , Wischmann , 68 F.4th at 506 (“We begin by considering whether the evidence . . . is probative.”). Our analysis follows suit, reviewing first the issue of whether Powley's evidence is probative, before turning to the question of significance.

Powley argues that her counter evidence is probative because it “ is more like the rebuttal evidence submitted in Buck and White , and less like the evidence submitted in Wischmann . ” She asserts that '[u]nlike the evidence submitted in Wischmann , the reports in this case show that they are generated by SkillTRAN LLC, and explain the method used by SkillTRAN LLC to produce DOTspecific . . . data, ” and “ [t]he numbers in the reports match the numbers described in the explanatory memorandum.' Powley also argues that the evidence is significant given the large discrepancy between Arne’s job numbers and hers. We agree. Powley’s counter evidence is both significant and probative.

Probative Analysis: Arne testified that he used SkillTRAN as the source for his job numbers for jewelry preparer (DOT 700.687-062) and fishing equipment assembler (DOT 732.684-062). Powley mimicked Arne’s method by using SkillTRAN, specifically SkillTRAN 's JBP, to develop counter evidence for both jewelry preparer (DOT 700.687-062) and fishing equipment assembler (DOT 732.684-062). For eyeglass assembler (DOT 713.687-018), Arne testified that he used a manufacturer database, which is “basically a business directory , ” to calculate the job numbers. Powley produced counter evidence by pulling data from SkillTRAN and the U.S. Census Bureau. Both SkillTRAN and the U.S. Census Bureau are sources relied upon by the SSA. See White , 44 F.4th at 837 (citing Purdy v. Berryhill , 887 F.3d 7, 14 (1st Cir. 2018) (“SkillTRAN’ s software has been recognized . . . to be widely relied upon by vocational experts in estimating the number of relevant jobs in the national economy.” (omission in original))); see also 20 C.F.R. §§ 404.1566(d)(2), 416.966(d)(2).

Page 16As in White , Powley’s evidence was produced using data sources and methodologies frequently relied on by the SSA. See 44 F.4th at 836-37 (deciding White's conflicting job numbers were probative because the evidence “was produced using a data source and methodology frequently relied on by the SSA'). Accordingly, b ecause Powley’s evidence was produced “using . . . data source[s] and [a] methodology frequently relied on by the SSA, ” and because we see no other factors which call the accuracy of this evidence into question, we conclude that it is probative. Id. at 837.

The government’s arguments to the contrary are unavailing. Citing Kilpatrick , 35 F.4th at 1194, the government argues that because some of Powley’s evidence is based on different sources than Arne’s evidence, it cannot be probative. Specifically, the government notes that Arne extrapolated information “ from national optical laboratories ” and used “ commercial services like Data Axle that provided a listing of employers by SOC codes ” to determine the number of eyeglass assembler jobs. This, the government contends, required 'Mr. Arne to 'draw some assumptions’ from his review of manufacturer databases.” In contrast, Powley’s eyeglass assemble r evidence is from the 2014 U.S. Census Bureau County Business Patterns and SkillTRAN.

The government overreads Kilpatrick to require that a claimant's counter evidence mirror that of a VE 's in source and methodology. In Kilpatrick , we expressed skepticism about the probative nature of the counter evidence because 'Kilpatrick's attorney did not replicate the VE's same methodology ' or identify his expertise in calculating alternative job numbers. 35 F.4th at 1194. Importantly, however, these were but two factors, among several, that led Page 17us to find the counter evidence not probative. Other major issues tainted Kilpatrick 's evidence. For instance, Kilpatrick’s attorney conceded that his figures “may not be completely accurate” and that he had used a seemingly selfcreated equation to calculate his job numbers with no evidence that the SSA frequently relied on such a method. Id. These combined issues led us to conclude that the evidence was not probative. Id. at 1194-95. Thus, whether a claimant replicates a VE’s methodology or pulls from the same sources are relevant factors we may consider in our probative analysis, but they are not dispositive.

Here, unlike in Kilpatrick , Powley’s attorney made no concession about the accuracy of his counter evidence and used only data sources “frequently relied on by the SSA.” See White , 44 F.4th at 837. Moreover, though Powley’s attorney did not exactly replicate Arne's methodology or pull from all the same sources, he did largely replicate it by relying on either the same source — SkillTRAN — or other sources commonly utilized by the SSA, like data from the U.S. Census Bureau. Therefore, we hold that Powley’s counter evidence, unlike Kilpatrick's evidence, clears the probative bar.

The government then turns to Wischmann to support its position that Powley's evidence is neither significant nor probative. As to Powley’s eyeglass assembler evidence, the government argues that Powley’s attorney failed to identify if he had “ expertise in calculating jobs figures in the national economy ” and inappropriately used data that was over seven years old. This argument is not persuasive. Pursuant to both Kilpatrick and White, whether a claimant’s counter evidence is produced by someone with expertise, though relevant, is not dispositive. Rather, the data creator’s expertise is one of several factors that we look to in determining whether such Page 18evidence is probative. See Kilpatrick , 35 F.4th at 1194 (listing at least four issues with the counter evidence, only one of which related to the data creator’s expertise); White , 44 F.4th at 837 (requiring only that the evidence be “produced using a data source and methodology frequently relied on by the SSA”). Accordingly, although we agree that there is limited evidence in the record regarding Powley's attorney’s expertise in calculating jobs figures, that deficiency is not sufficient to doom Powley's counter evidence for eyeglass assemblers.

As to Powley’s jewelry preparer and fishing equipment assembler evidence, the government argues that “[s]imilar to Wischmann , Powley’s objection to Mr. Arne’s job numbers was based on data that an unspecified person copied from [JBP] based on DOT codes' and , just as in Wischmann , Powley did not state “‘ who produced the [JBP] outputs[, ]’ or whether that person had any ‘ identified expertise in calculating job figures in the national economy. ’” We disagree. In Wischmann , the counter evidence was contained in “six pages of printouts” with no indication that the data was produced with JBP. 68 F.4th at 506-07. The letter explaining the six pages of data provided “no information about how the job numbers were produced, other than the name of the software progra m,” gave “no information about what queries were entered into the computer program, what variables were changed, or what filters were applied to the data,” and failed to state “ which version of the program was used. ” Id. at 507.

In contrast , Powley’s attorney appears to have used the most up-to-date version of JBP and stated that he used the 2014 U.S. Census Bureau County Business Patterns data, of which the SSA has taken administrative notice. Powley provided 91 pages of supporting evidence, far more than the Page 19six pages in Wischmann , and each JBP printout provided by Powley contains the name of the position and corresponding DOT, which matches the DOT testified to by Arne. Additionally, the JBP printouts discuss the method used to produce the data and list the sources from which the data was pulled. And the census data bears the heading of the U.S. Census Bureau and provides details on the geography covered, the industry it concerns, and other information regarding the data contained therein. Moreover, unlike in Wischmann where the data was produced by an unknown person, the data here was produced by Powley’s attorney as stated in his letter to the appeal counsel, which is “part of the administrative record ' that ' the district court [and this court] must consider.' Id. at 503-04 (quoting Brewes v. Comm ’ r of Soc. Sec. Admin. , 682 F.3d 1157, 1163 (9th Cir. 2012)).

Significant Analysis: Because Powley’s counter evidence is probative, we next evaluate whether it is significant . Significant evidence 'in the context of job -number estimates is a measure of the discrepancy between the VE ’ s estimates (upon which the ALJ relied to render the step-five finding) and the claimant ’s estimates.” Id. at 505. As noted above, the job numbers submitted by Powley are significantly different than those testified to by Arne. For eyeglass assembler, the difference between the respective job numbers is 39,425 versus 27. For fishing equipment assembler, the difference between the respective job numbers is 14,700 versus 19. And for jewelry preparer, the difference is most stark as Powley's counter evidence indicates there are no applicable positions in this DOT, while Arne testified that there are 24,700 positions. The discrepancies between Arne's and Powley's numbers are comparable to those in White and Buck and are therefore significant. See White , 44 F.4th at 837 (finding the counter Page 20evidence to be significant because “ [t]he discrepancy between the[] estimates is comparable to the discrepancy in Buck ” (citing Buck , 869 F.3d at 1047)).

2. The ALPage 21J did not sufficiently address Powley’s Finally, there is no question that the ALJ did not resolve the inconsistency between Arne’s and Powley’s job numbers. As noted above, the ALJ made only a conclusory determination that Arne's testimony was more persuasive because it was based on his experience, knowledge, and expertise. The conflict between the data, however, remains unresolved.

evidence.

Because Powley’s evidence is significant and probative , we next consider whether the ALJ sufficiently addressed it. The ALJ did not. The government argues that the ALJ “need only explain his rejection' of a claimant's competing job numbers and the ALJ here did just that. Not so.

The ALJ is required to both “ address ” the discrepancy and 'resolve the inconsistency between the job -number estimates provided by [Powley] and by the VE.” Id. Here, the ALJ failed to properly 'address' the inconsistencies between the data. The ALJ stated that Arne’s testimony was more persuasive than the counter evidence produced by Powley because Arne’s testimony was based on “hypotheticals posed at the hearing, in conjunction with his professional experience, knowledge, and expertise, which includes over 30 years as a vocational rehabilitation counselor.” Powley’s counter evidence, however, does not attack the hypothetical posed at the hearing, nor does it question Arne’s experience, knowledge, or expertise. Rather, the counter evidence demonstrates a purported inconsistency between job-numbers data from the same database for the same jobs, as well as a conflict between jobnumbers data relied upon by Arne and job-numbers data from alternative reliable sources. Simply noting that Arne’s testimony was based on his experience and his answer to a hypothetical did nothing to address the conflicts between the data. Therefore, the ALJ erred by failing to address Powley's counter evidence.

Powley presented significant and probative evidence to counter that of VE Arne. The ALJ was required to address the discrepancies and inconsistencies between the evidence and resolve them. Because the ALJ did not do so, we reverse the decision of the district court with instructions to remand to the agency for further proceedings consistent with this opinion.