State v. Harenda Enterprises, Inc. 307 Wis. 2d 604, 746 N.W.2d 25, 2008 WI 16

ANN WALSH BRADLEY, J.

¶ 1. The petitioner, State of Wisconsin, seeks review of a published court of appeals decision reversing a circuit court judgment in favor of the State and imposing civil penalties and surcharges on Harenda Enterprises, Inc. (Harenda).1 The judgment and sanctions imposed were for violations of Wis. Admin. Code Ch. NR 447 in connection *610with Harenda's contract to inspect the Milwaukee Auditorium for possible asbestos-containing material (ACM).

¶ 2. The case centers on the question of the proper method for testing whether material constitutes asbestos-containing material under Wis. Admin. Code § NR 447.022 and 40 C.F.R. Pt. 763, Subpt. E., App. E, § 1.7.2.1. The State asserts that the court of appeals erred in concluding that the rule prescribing the method of testing clearly requires the averaging of the test results. It contends that the rule is ambiguous and that we should give deference to the clarifications of the rule issued by the United States Environmental Protection Agency (EPA), which explain that each layer of a multi-layer sample must be tested. It maintains that under the rule, ACM is present if a single layer of the sample contains greater than one percent asbestos.

¶ 3. Harenda argues that the State's interpretation is inconsistent with the plain language of the rule, which requires averaging of the test results. It maintains that the clarifications issued by the EPA are inconsistent with § 1.7.2.1 and should therefore be accorded no deference. Harenda further argues that the clarifications outlining the single layer test method constitute impermissible rule making. Finally, it argues that the State's enforcement action violates its substantive due process rights.

¶ 4. We determine, first, that the language of § 1.7.2.1 is ambiguous. Giving deference to an agency's interpretation of its own rule, we conclude that the EPA's interpretation is controlling because it is neither *611inconsistent with § 1.7.2.1 nor clearly erroneous. We further determine that the clarifications do not constitute impermissible rule making; rather, they are valid interpretive rules. Finally, we determine that the circuit court's judgment does not violate Harenda's substantive due process rights. We therefore reverse the court of appeals.

I

A

¶ 5. To better understand the issues presented, it is helpful to examine the regulatory framework. We begin with a brief summary of the federal and state regulations at play in this case.

¶ 6. Under the federal Clean Air Act, the United States Environmental Protection Agency (EPA) is authorized to publish a list of hazardous air pollutants and to establish national emission standards (National Emission Standards for Hazardous Air Pollutants, or "NESHAPs") for each pollutant on the list. 42 U.S.C. § 7412; United States v. American National Can Co., 126 F. Supp. 2d 521, 523 (N.D. Ill. 2000). Asbestos was one of the first pollutants designated as hazardous under the Clean Air Act. National Can, 126 F. Supp. 2d at 523.

¶ 7. The original asbestos NESHAP was published in 1973, and included standards governing removal of asbestos prior to building demolition. 38 Fed. Reg. 8,820 (1973). In 1975, the asbestos NESHAP was expanded to address the handling of asbestos during building renovations. 40 Fed. Reg. 48,293 (1975); National Can, 126 F. Supp. 2d at 523. The current asbestos NESHAP which was published in 1990, is found at 40 C.F.R. § 61.140 et seq.

*612¶ 8. The Wisconsin Department of Natural Resources (DNR) is authorized by the Wisconsin Statutes to promulgate rules implementing clean air standards consistent with chapter 285 of the Wisconsin Statutes and the federal Clean Air Act. Wis. Admin. Code § NR 447.01(2);3 Wis. Stat. §§285.11, 285.13, 285.17, 285.27.4 Pursuant to this authority, the DNR promulgated rules concerning asbestos inspection, identification, and abatement. These rules are set forth in chapter NR 447 of the Wisconsin Administrative Code.

¶ 9. Chapter NR 447 is patterned after the federal NESHAP standards. See note to Wis. Admin. Code § NR 447.01. Most importantly with respect to the present matter, chapter 447 has adopted measures requiring owners and operators to inspect facilities prior to demolition or renovation. Wis. Admin. Code § NR 447.06(1). If such an inspection reveals enough asbestos, the notification and abatement requirements of sections NR 447.07 and 447.08 apply to each owner or operator. Wis. Admin. Code § NR 447.06(2). Violations of the regulations are strict liability offenses. United States v. B & W Inv. Properties, 38 F.3d 362, 367 (7th Cir. 1994).

¶ 10. As discussed more fully below in Part III A, the regulations define ACM as material that contains *613"more than 1% asbestos as determined using the method specified in Appendix E to Subpart E, 40 CFR part 763, section 1, Polarized Light Microscopy... Wis. Admin. Code § NR 447.02(l)(a). The regulations describing the proper way to analyze a multi-layered sample include the following instructions: "When discrete strata are identified, each is treated as a separate material so that fibers are first identified and quantified in that layer only, and then the results for each layer are combined to yield an estimate of asbestos content for the whole sample." 40 C.F.R. Pt. 763, Subpt. E, App. E, § 1.7.2.1. Wisconsin adopted both the definition and the instructions describing the method of analysis. Wis. Admin. Code § NR 447.02; Wis. Admin. Code § NR 484.04(28)(July 2007).

¶ 11. The text of the instructions, however, quickly became a source of confusion. On two occasions the EPA issued notices of clarification to address how multi-layered samples are to be analyzed. 59 Fed. Reg. 542 (Jan. 5, 1994); 60 Fed. Reg. 65,243 (Dec. 19, 1995).

¶ 12. The December 1995 clarification explained that § 1.7.2.1 continued EPA's past policy that separate layers in multi-layered systems were to be analyzed separately, such that "no averaging or dilution by combining layers of asbestos-containing material with nonasbestos-containing material was allowed." 60 Fed. Reg. 65,243 (Dec. 19, 1995).

¶ 13. The clarification allowed that a source sending a multi-layered sample to a laboratory for testing may request that certain samples first be composited for analysis in an effort to reduce time and the costs associated with the sample. It advised that when the composite analysis indicates that the average of the sample's layers is greater than one percent, the sample is deemed to be ACM and an individual analysis of the *614layers is not necessary. However, when the composited sample analysis results in less than one percent asbestos, but greater than zero, an "analysis by layers is required to ensure that no layer in the system contains greater than one percent asbestos." Id.

¶ 14. Pursuant to its regulatory authority the DNR initiated an enforcement action against Harenda. In this case we examine whether we owe deference to an agency's interpretation of its own rule concerning the testing of multi-layered samples for asbestos content.

B

¶ 15. The background facts of this case are not in dispute. The circuit court set forth those facts in its Findings of Fact and Conclusions of Law entered along with its order granting judgment in the State's favor.

¶ 16. This case stems from a decision by the Wisconsin Center District to renovate the Milwaukee Auditorium. In 2001, the Wisconsin Center District retained Harenda to conduct an inspection of the Auditorium for possible ACM prior to renovation. Among the areas Harenda inspected were the walls of the Auditorium's second floor bowl area (the "disputed area"). As part of its pre-demolition inspection, Harenda took samples from the disputed area and sent them to a testing laboratory. The test results indicated that none of the samples contained greater than one percent asbestos.

¶ 17. The demolition of the disputed area, which was carried out by a contractor, commenced in March 2002 and proceeded without following chapter NR 447 asbestos abatement regulations. After demolition had started, representatives of the State took samples from the walls of the disputed area for testing. The labora*615tory that tested the samples found a layer in a multi-layered sample that contained greater than one percent asbestos. Demolition ceased, and the State collected three further samples from the disputed area. The tests on these samples showed that the samples each contained greater than one percent asbestos.

¶ 18. Two days later, the State collected ten samples of plaster material from the disputed area that had been demolished. It provided a "split" of each of these samples to Harenda. The laboratory found that four of the State's ten samples were multi-layered samples containing a single layer of material with one percent asbestos or more, but with an overall asbestos content of less than one percent.

¶ 19. Harenda sent its split samples to an independent laboratory for testing. The laboratory found five of the ten samples were multi-layered samples containing a single layer of two percent asbestos and one layer of material that was a non-detect for asbestos. In contrast to the laboratory used by the State, Harenda's laboratory did not obtain an overall asbestos content for the five multi-layered asbestos-containing samples.5

*616¶ 20. The State, upon the request of the DNR, filed a complaint against Harenda. Based on the tests of its samples from the disputed area, the State alleged that material from the disputed area contained a sufficient amount of ACM for it to be considered regulated asbestos-containing material. It alleged that Harenda was therefore liable for violations of several provisions of Wis. Admin. Code Ch. NR 447. These include failure to adequately wet ACM in violation of § NR 447.08(6)(a), failure to carefully lower ACM in violation of § NR 447.08(6)(b), failure to remove ACM prior to demolition activity in violation of § NR 447.08(1), discharge of visible emissions to the outside air during the disturbance of ACM in violation of § NR 447.13(1), and failure to file an accurate notice of intent to renovate a facility containing ACM in violation of § NR 447.07.

¶ 21. The parties entered into a stipulation according to which Harenda would owe $37,138.50 in penalties and surcharges if it were held liable. They further stipulated that Harenda is an "operator" under chapter NR 447 and that the asbestos testing results from the various laboratories are "accurate within the analytical testing method that each respective laboratory utilized."

¶ 22. The circuit court granted a motion for summary judgment by the State. It determined that the State satisfied its burden of demonstrating that there was ACM under § NR 447.01(l)(a) and § NR 447.02(l)(b) and the relevant federal regulation, 40 C.F.R. Pt. 763, Subpt. E., App. E, § 1.7.2.1. The circuit court further determined that the State met its burden in showing that the ACM was stripped or removed without complying with the requirements of chapter NR 447 as alleged in the complaint.

*617¶ 23. Harenda appealed. The court of appeals determined that the State's interpretation of 40 C.F.R. Pt. 763, Subpt. E., App. E, § 1.7.2.1, which was based upon clarifications of the rule issued by the United States Environmental Protection Agency, was inconsistent with the plain language of § 1.7.2.1. State v. Harenda Enterprises, Inc., 2006 WI App 230, ¶ 9, 297 Wis. 2d 571, 724 N.W.2d 434. It therefore reversed the judgment of the circuit court. The State petitioned for review.

I

¶ 24. The central issue in this case concerns the interpretation of the method used to test for asbestos-containing material (ACM) pursuant to chapter NR 447 of the Wisconsin Administrative Code. This case is before us on the circuit court's grant of summary judgment in favor of the State. We review the grant or denial of summary judgment independently, but apply the same methodology as used by the circuit court. Wis. Mall Props., LLC v. Younkers, Inc., 2006 WI 95, ¶ 19, 293 Wis. 2d 573; 717 N.W.2d 703 (citing Green Spring Farms v. Kersten, 136 Wis. 2d 304, 315, 401 N.W.2d 816 (1987)). Summary judgment is appropriate where there are no genuine issues of material fact and the moving party is entitled to judgment as a matter of law. Wis. Stat. § 802.08(2).

¶ 25. Administrative code provisions are interpreted according to principles of statutory construction.6 Orion Flight Servs., Inc. v. Basler Flight Serv., 2006 WI 51, ¶ 18, 290 Wis. 2d 421, 714 N.W.2d 130. *618When an administrative code provision is ambiguous, we turn to extrinsic sources in order to determine agency intent. Id. An administrative agency's interpretation of its own regulations is controlling "unless the interpretation is inconsistent with the language of the regulation or is clearly erroneous." Id. (quoting Bergmann v. McCaughtry, 211 Wis. 2d 1, 7, 564 N.W.2d 712 (1997)).7 Whether an agency's interpretation of a regulation is inconsistent with the regulation or clearly erroneous is a question of law that we review independently of the determinations rendered by the circuit court and court of appeals. Bergmann, 211 Wis. 2d at 8.

¶ 26. In addressing whether the regulation is clearly erroneous we consider the intent or purpose of the regulation. In resolving ambiguities, "[i]t is fundamental that we must favor a construction of a statute or regulation which will fulfill the intent of the statute or regulation over one which defeats its manifest object." Baierl v. McTaggart, 2001 WI 107, ¶ 21, 245 Wis. 2d 632, 629 N.W.2d 277 (citing Shands v. Castrovinci, 115 Wis. 2d 352, 356, 340 N.W.2d 506 (1983).

¶ 27. It is often difficult to discern the difference between an interpretive rule and a legislative rule, as *619they lie upon a "hazy continuum." American Hospital Ass'n v. Bowen, 834 F.2d 1037, 1045 (D.C. Cir. 1987). "Determining whether a given agency action is interpretive or legislative is an extraordinarily case-specific endeavor." Id.; see Harry T. Edwards and Linda A. Elliott, Federal Standards of Review, 134-35. In making such a determination courts "consider the agency's own characterization of the particular action and will generally give deference to the agency's views." Beverly Health & Rehab. Servs. v. Thompson, 223 F. Supp. 2d 73, 103 (D.D.C. 2002)(quoting American Hosp. Ass'n, 834 F.2d at 1056 and British Caledonian Airways Ltd. v. Civil Aeronautics Bd., 584 F.2d 982, 991 (D.C. Cir. 1978))(internal punctuation and citations omitted).

¶ 28. In this case we also address the issue of whether the circuit court's judgment violated Harenda's constitutional right to substantive due process. Such an inquiry presents a question of law which we review independently. Kenosha County Dep't of Human Servs. v. Jodie W., 2006 WI 93, ¶ 22, 293 Wis. 2d 530, 716 N.W.2d 845.

I

A

¶ 29. Under Wis. Admin. Code § NR 447.02(l)(b), ACM is defined as material "containing more than 1% asbestos as determined using the method specified in Appendix E to Subpart E, 40 CFR part 763, section 1... ."8 The parties agree that the provision of Appen*620dix E, section 1 that is relevant in this case is § 1.7.2.1, "Gross Examination." The parties disagree, however, as to whether the language is clear or ambiguous. Harenda contends that § 1.7.2.1 is clear, and that the State's interpretation conflicts with the plain meaning of that section. To assess Harenda's argument, we begin by examining the language of the rule. Section 1.7.2.1 provides as follows:

Bulk samples of building materials taken for the identification and quantitation of asbestos are first examined for homogeneity at low magnification with the aid of a stereomicroseope. The core sample may be examined in its container or carefully removed from the container onto a glassine transfer paper or clean glass plate. If possible, note is made of the top and bottom orientation. When discrete strata are identified, each is treated as a separate material so that fibers are first identified and quantified in that layer only, and then the results for each layer are combined to yield an estimate of asbestos content for the whole sample.

40 C.F.R. Pt. 763, Subpt. E, App. E, § 1.7.2.1 (emphasis added).

¶ 30. The focus of our inquiry is on the last sentence of the above rule. It describes two different types of testing measurements: (1) a quantification of *621asbestos fibers in each discrete layer, and (2) an "estimate of asbestos content for the whole sample" achieved by combining the results from discrete layers. However, it is unclear which of these measurements is relevant in determining whether material is ACM.

¶ 31. The circuit court aptly describes the ambiguity in the two main clauses of the sentence, noting that they say different things:

[I]f they can discern discrete strata, they're supposed to come up with a number for each layer. It says that in the first clause of the final sentence that I read. Arid then it's got a conjunction — and—which supports [Harenda's] argument, and the second clause, which says, And then you composite or come up with a total for all the layers. And the question what does it mean? .... I'm going to borrow from my early life as an English teacher. There's two clauses in one sentence, and they say two different things. They don't rule each other out, either. They say two different things, and they're joined by a linking conjunction. ...

[I]t doesn't say whether the violation occurs in both parts of the sentence, just one, or the other. It is not clear. However, I think it's significant that each half of the sentence doesn't rule the other out. And so then the question becomes, is this interim method rule ... clear? No, it's not clear. And that's pretty obvious.9

¶ 32. The second clause of the sentence reveals further ambiguity with respect to how to arrive at an "estimate of asbestos content for the whole sample." It directs that "and then the results of each layer are combined." However, it is unclear what is meant here by *622the word "combined." The State asserts that it is ambiguous, but posits that a reasonable interpretation is that the results are added together. Harenda maintains that "combined" means that the results are averaged.

¶ 33. The court of appeals, without explanation, concluded that "combining" the results from each layer requires that the percentage of asbestos for each layer be averaged. Harenda Enterprises, 297 Wis. 2d 571, ¶ 4. An obvious problem with the court of appeals' and Harenda's "plain meaning" interpretation is that the word "average" cannot be found in the text of the rule.

¶ 34. Moreover, "combine" means to "bring into a state of unity; merge." American Heritage Dictionary of the English Language, 377 (3rd ed. 1992). "Average" is defined as the "arithmetic mean." Id. at 127. However, "arithmetic mean" refers to the "value obtained by dividing the sum of a set of quantities by the number of quantities in the set." Id. at 99. Thus, averaging requires combining quantities and then dividing. The language of the rule, however, states only that quantities are to be combined.

¶ 35. Further, the language of the section leaves unclear which layers are combined to yield an estimate of the asbestos content of the whole sample. There are two plausible interpretations. The first is that only the results from each layer in which there are asbestos fibers are combined to yield an estimate of the asbestos content of the whole sample. The second interpretation is that the results from every layer identified, including those in which no asbestos fibers are identified and quantified, are combined to yield the estimate of asbestos content for the whole sample. This appears to be the interpretation embraced by the court of appeals.

¶ 36. In light of these interpretations, we reject Harenda's contention that the text of the rule is unam*623biguous. It is unclear how the first clause of the relevant sentence relates to the second clause. Likewise, the meaning of the word "combine" is uncertain, and there are competing interpretations as to what layers are to be combined to yield an estimate of asbestos content. We therefore determine that the language of § 1.7.2.1 is ambiguous.

B

¶ 37. The determination that § 1.7.2.1 is ambiguous, however, does not end our inquiry. We must next examine whether the State's interpretation is inconsistent with the regulation or clearly erroneous. As we have noted, an administrative agency's interpretation of its own regulations is controlling unless the agency's interpretation is "inconsistent with the language of the regulation or is clearly erroneous." Orion Flight Servs., 290 Wis. 2d 421, ¶ 18.

¶ 38. The State maintains that under § 1.7.2.1, material is ACM if a single layer from a multi-layered sample contains greater than one percent asbestos. It bases its position on the interpretations advanced in two clarifications issued by the EPA in response to questions regarding the testing of multi-layered samples.

¶ 39. In the first, issued in January 1994, the EPA stated that the clarification was a response to members of the regulated community who had "frequently asked" questions regarding the analysis of multi-layered samples.

The Agency has learned that some of the regulated community have questions concerning the analysis of samples which may contain multiple layers, any or all of which may be asbestos containing materials *624(ACM).... Because these questions are frequently asked, EPA is making this clarification.

59 Fed. Reg. 542 (Jan. 5, 1994).

¶ 40. The clarification provided that when multi-layered samples are tested, the results from each layer should be reported. It stated that "[i]n general, when a sample consists of two or more distinct layers or materials, each layer should be treated separately and the results reported by layer (discrete stratum)." Id.

¶ 41. Thus, the EPA's interpretation of § 1.7.2.1 is that the results of each layer are relevant in determining whether material is ACM. This addresses the first ambiguity in § 1.7.2.1 that we describe above. It sheds light on the circuit court's question of "whether the violation occurs in both parts of the sentence, just one, or the other." The January 1994 statement clarifies that a violation occurs in the first part of the sentence.

¶ 42. The second clarification in which the EPA interprets its rule was issued in December 1995 in response to continued questions regarding the testing of multi-layered samples. The EPA explains in the second clarification that prior to adopting the procedures set forth in 40 C.F.R. Pt. 763, Subpt. E, App. E, § 1.7.2.1, on November 20, 1990, the EPA had informal policies of treating each layer in multi-layer systems separately and against diluting asbestos-containing layers by combining them with layers not containing asbestos:

EPA's unwritten policy based on the definition of "friable asbestos material" was that each layer in a multi-layered system was to be analyzed as a separate material (no averaging or dilution by combining layers of asbestos-containing material with nonasbestos-containing material was allowed).

*62560 Fed. Reg. 65,243 (Dec. 19, 1995).

¶ 43. Further, the December 1995 statement explained that the January 1994 clarification interpreted § 1.7.2.1 as precluding averaging. It states that in multi-layered systems, "results were not allowed to be combined to determine average asbestos content (continuing the policy that dilution of an asbestos-containing material is not allowed)." Id.

¶ 44. The EPA's December 1995 clarification therefore provides guidance on a second ambiguity in § 1.7.2.1. "Combining" the results from discrete layers does not mean averaging the content of those layers where doing so dilutes the results.

¶ 45. The document also describes a potentially cost-saving method for testing multi-layered samples. It explains that in testing multi-layered samples, labs may composite layers first in order to determine whether it is necessary to perform potentially more costly and time-consuming testing of discrete layers:

Any source sending multi-layered bulk samples to a lab may request that certain sample(s) or portions of sample(s) be composited for analysis first (to potentially reduce time and cost of sampling).

(Note: A composite sample does not mean that multiple samples may be composited into one sample. It means that multiple layers of one core sample may be compos-ited for analysis.)

Id. 10

¶ 46. The clarification then explains that if the analysis of composited layers shows that the average content of asbestos for the whole sample is greater than *626one percent, the system must be treated as ACM. However, if the analysis shows the presence of asbestos, but in a concentration of less than one percent, each discrete strata must be treated separately. If a single layer is found to contain greater than one percent asbestos, then that layer must be treated as ACM.

If the result of the composite analysis shows that the average content for the multi-layered system (across the layers) is greater than one percent, then the multi-layered system must be treated as asbestos-containing and analysis by layers is not necessary. If the result of the composite sample analysis indicates that the multi-layered system as a whole contains asbestos in the amount of one percent or less, but greater than none detected, then analysis by layers is required to ensure that no layer in the system contains greater than one percent asbestos. If any layer contains greater than one percent asbestos, that layer must be treated as asbestos-containing. This will have the effect of requiring all layers in a multi-layered system to be treated as asbestos-containing if the layers can not be separated without disturbing the asbestos-containing layer. Once any one layer is shown to have greater than one percent asbestos, further analysis of the other layers is not necessary if all the layers will be treated as asbestos-containing.

Id.

¶ 47. The December 1995 statement therefore reiterates the position of the January 1994 statement that the measure of asbestos content from a single layer may render multi-layer material ACM. It also provides an opportunity to perform first a composite test in order to determine more quickly and efficiently that a sample is ACM. Notably, it does not imply that the "estimate of asbestos content for the whole sample" described in § 1.7.2.1 is an average of all layers. Rather, *627it clarifies that averaging may not be used to dilute the measure of asbestos content in multi-layer samples. The composite test allows averaging, but it is a separate test used to forestall the expense of analyzing individual layers and not as a means of dilution.

¶ 48. The State's argument regarding the testing method is therefore supported by the EPA's clarifications, which are not inconsistent with § 1.7.2.1. Moreover, the EPA's interpretation is not clearly erroneous, as its interpretation is supported by the purpose of regulation and basic principles of statutory construction.

¶ 49. The manifest purpose of chapter NR 447 of the Wisconsin Administrative Code, which incorporates 40 C.ER. Pt. 763, Subpt. E, App. E, § 1.7.2.1, is to protect workers and the public from air pollution from asbestos. Prohibiting the dilution of ACM by averaging the asbestos content of multi-layer systems serves that purpose.11

*628¶ 50. The interpretation suggested by Harenda, and adopted by the court of appeals, runs contrary to the purpose of the rule. At oral argument, the State explained that averaging layers would allow that a layer of wall containing (for example) 1.9 percent asbestos would constitute ACM if it were standing alone, but not constitute ACM if attached to a layer of wall with no asbestos. The State explained, however, that whether the asbestos-containing layer is attached to a non-asbestos-containing layer does not diminish the amount of asbestos that disperses into the air when a worker strikes the wall with a hammer or a wrecking ball.12

¶ 51. The State's and EPA's interpretation is also required in order to harmonize the language of § NR 447.02(l)(b) and § 1.7.2.1. It is a basic principle of statutory and administrative rule construction that "provisions relating to the same subject matter should be read together and harmonized when possible." State v. Morford, 2004 WI 5, ¶ 21, 268 Wis. 2d 300, 674 N.W.2d 349.

*629¶ 52. Section 447.02(l)(b) states that" 'Category II nonfriable ACM' means any material" that contains greater than one percent asbestos under § 1.7.2.1. (Emphasis added.) Section 1.7.2.1 is explicit that where a sample contains discrete strata, "each is treated as a separate material." (Emphasis added.) Reading these two provisions together, discrete strata must be treated as separate material under § 1.7.2.1, and any material containing greater than one percent asbestos is ACM under § NR 447.02(l)(b). It therefore follows that a discrete strata containing greater than one percent asbestos is ACM.

¶ 53. Interpreting the regulations such that a discrete strata in a multi-layer sample cannot be ACM would require that "material" have different meanings in § 1.7.2.1 and § NR 447.02(l)(b). However, it seems implausible that § NR 447.02(l)(b) uses the word "material" to mean one thing, but employs and directly refers to a test for ACM that uses "material" to mean something altogether different.

¶ 54. Harenda argues, however, that the EPA's interpretation of § 1.7.2.1 contradicts the section because it renders the second clause of the sentence (estimating the asbestos content of the entire sample) superfluous. Construction of statutes and administrative rules should avoid whenever possible interpretations that render language superfluous. Hutson v. State Pers. Comm'n, 2003 WI 97, ¶ 49, 263 Wis. 2d 612, 655 N.W.2d 212. Harenda's argument misses the mark for two reasons.

¶ 55. First, its argument assumes that "combining" the results from discrete layers to "yield an estimate of asbestos content for the whole sample" means averaging the results. The EPA was clear in the January 1994 and December 1995 documents that layers *630could not be averaged as a means to dilute layers containing greater than one percent asbestos. 59 Fed. Reg. 542 (Jan. 5, 1994); 60 Fed. Reg. 65,243 (Dec. 19, 1995). The discussion of averaging the asbestos contents for multiple layers occurs in the context of providing testers the opportunity to do a preliminary cost- and time-saving test in order to establish whether analyzing discrete layers is necessary. Id.

¶ 56. Second, it is Harenda's interpretation of § 1.7.2.1 that renders language superfluous. If the asbestos content of an entire multi-layer sample is the only measure relevant in determining whether material is ACM, there would be no need to treat each layer "as a separate material so that fibers are first identified and quantified in that layer only." Material could be determined ACM without analyzing discrete strata. Harenda's argument is unpersuasive.13

*631¶ 57. We therefore conclude that the EPA's January 1994 and December 1995 clarifications Eire not inconsistent with the language of § 1.7.2.1. We also conclude that the EPA's interpretation is not clearly erroneous. The EPA's interpretation fulfills the purpose of the asbestos regulations, whereas the alternative interpretation Harenda proposes defeats that purpose. Moreover, the EPA's view is supported by basic principles of statutory construction. Giving deference to an agency's interpretation of its own rule, we conclude that the interpretation of § 1.7.2.1 explained in the EPA's January 1994 and December 1995 clarifications is controlling.14

*632IV

¶ 58. Harenda also argues that the clarifications impose new obligations, and absent the clarifications, there would be no basis for the State's enforcement action. It contends that the clarifications are therefore legislative rules, and that they are unenforceable because they were not enacted through the proper administrative rule making procedures pursuant to 5 U.S.C. §553.

¶ 59. Under the federal Administrative Procedures Act (APA), an agency may issue a legislative rule only if it uses the note and comment procedure described in 5 U.S.C. § 553(b) or establishes an exception under 5 U.S.C. § 553(b)(3)(B). Hemp Indus. Ass'n v. DEA, 333 F.3d 1082, 1087 (9th Cir. 2003). However, an agency need not follow the note and comment procedure in order to issue an interpretive rule. Id. A legislative rule promulgated without complying with the procedures of the APA is invalid. Id. It is not disputed that the clarifications were issued without the procedures required to issue a legislative rule. Rather, the question is whether the clarifications are interpretive or legislative rules.

¶ 60. The difference between interpretive rules and legislative rules is, roughly speaking, that interpretive rules merely explain substantive law, and legislative rules create rights, impose obligations, or effect a change *633in existing law. Yesler Terrace Community Council v. Cisneros, 37 F.3d 442, 449 (9th Cir. 1994). A rule is legislative if, "in the absence of the rule, there would not be an adequate legislative basis for enforcement action." Hemp Indus. Ass'n, 333 F.3d at 1087. However, as noted above, interpretive rules and legislative rules lie upon a "hazy continuum" and distinguishing them is a case-specific endeavor. American Hospital Ass'n, 834 F.2d at 1045.

¶ 61. Harenda cites to a recent, unpublished case from Southern California in support of its claim that the clarifications are legislative rules. In U.S. v. San Diego Gas & Electric Co., 2006 LEXIS 84856 (S.D. Cal. Nov. 21, 2006) (SDG&E), San Diego County sued San Diego Gas and Electric for violating asbestos work practice standards. Id. at 4. The county's indictment stated that the alleged ACM contained multiple layers and was tested for asbestos according to the method outlined in the clarifications. That is, the county's test would find that the material is ACM if "any of the layers, standing alone ... [contains] over 1% asbestos and [is] friable." Id. at 9.

¶ 62. The court determined that the "single-layer" method described in the clarifications conflicts with the test set forth in § 1.7.2.1. It interpreted the § 1.7.2.1 test to mean that "material is not regulated unless the combined result yields more than 1 percent asbestos." Id. at 22. The court therefore held that the government's failure to allege that multi-layered material contained greater than one percent asbestos as determined by averaging the asbestos content of the layers was fatal to its indictment. Id. at 27-28.

¶ 63. The court's conclusion in SDG&E, however, turns on an interpretation of § 1.7.2.1 that we have rejected. As explained above, it is our determination that *634a reasonable interpretation of § 1.7.2.1 is that multilayer material is ACM if a single layer has an asbestos content of greater than one percent. That interpretation is supported by EPA's clarifications, the manifest purpose of the rule, and principles of statutoiy construction.15 Further, the SDG&E court uncritically adopts the view, which we reject, that in § 1.7.2.1 "combine" means average.

¶ 64. Moreover, § 1.7.2.1 provides an adequate basis for the enforcement action even without the clarifications. As we discuss above, § 1.7.2.1 is ambiguous, and the EPA's interpretation is consistent with the section. Specifically, § 1.7.2.1 supports the interpretation that if a single layer of a multi-layer sample contains greater than one percent asbestos, the material is ACM. Also as outlined above, that interpretation is supported by the purpose of the regulations and basic principles of construction.

¶ 65. Finally, we are mindful that the EPA characterized the 1994 and 1995 statements as "clarifications to the final rule" rather than rules. When courts consider an agency's characterization of its actions they "generally give deference to the agency's views." Beverly Health & Rehab., 223 F. Supp. 2d at 103. The EPA's view supports our conclusion that the statements are interpretive rules rather than legislative rules. We therefore determine that, contrary to SDG&E, § 1.7.2.1 provides an adequate legislative basis for the State's *635enforcement action.16 The clarifications are not legislative rules. Rather, they are valid interpretive rules.

V

¶ 66. We turn finally to Harenda's due process argument. Harenda contends that its pre-demolition sampling and testing satisfied the applicable regulations and indicated that the material in the disputed area was not ACM. It further maintains that the differences between the results of its tests on the split samples and the State's tests on the split samples create an issue of fact as to whether the material from the disputed area was ACM. Thus, it argues that the circuit court granted the State's motion for summary judgment on the basis of "conflicting test results." It maintains that the decision was therefore arbitrary, and it violated Harenda's right to due process.

¶ 67. Harenda's contention that the circuit court's decision was arbitrary is unpersuasive. Its due process argument is thus without foundation.

¶ 68. The circuit court based its decision on a four-element test as set forth in National Can, 126 F. Supp. 2d 521:

In order to establish liability under the asbestos NES-HAR the government must prove that 1) [American *636Can] was an owner or operator of a facility, 2) a renovation occurred, 3) asbestos was removed or stripped without complying with the requirements and practices delineated in the asbestos NESHAR and 4) a jurisdictional amount of asbestos was disturbed.

Id. at 525, note 1 (citing 40 C.F.R. § 61.140 et seq.)

¶ 69. Under the above test, owners and operators are strictly liable for asbestos abatement regulation violations occurring at their regulated facilities. The State does not have to prove any negligence or intent to violate the regulation. As stated in United States v. B & W Inv. Properties, 38 F.3d 362 (7th Cir. 1994):

Having been deemed an owner or operator, [the defendant] has no valid challenge against application of the Act, regardless of how minimal the company's responsibilities or knowledge may actually have been. The Act imposes strict liability on all owners and operators of properties in violation of the Act.

Id. at 367.

¶ 70. The parties have stipulated that Harenda is an operator of a facility. There is also no dispute that a renovation occurred and that removal of material from the disputed area occurred without following the procedures outlined in chapter 447. Further, Harenda does not contest that if the material removed from the disputed area is ACM, a jurisdictional amount of asbestos was disturbed.17 The issue is whether the material removed from the disputed area is ACM.

*637¶ 71. Harenda is correct that the State has offered no argument that Harenda failed to conduct its pre-demolition sampling and testing according to the regulations. In fact, the State offers no opinion on the issue. Instead, the State asserts that liability does not depend on whether an owner or operator conducts pre-demolition sampling testing in accord with the regulations.

¶ 72. The State is correct. The elements set forth above do not require that the owner or operator sample or test improperly, and Harenda offers no argument that sampling and testing according to the regulations is a defense. Thus, in the present case liability depends on whether the material disturbed was ACM, not on the quality of Harenda's pre-demolition sampling and testing.

¶ 73. Harenda's argument that the circuit court made an arbitrary choice between "dueling test results" is similarly unpersuasive. It makes much of the fact that the tests of the split samples conducted for the State and for Harenda had different results. Harenda contends that such "analytic variability" in testing results undermines the basis of the circuit court's decision and renders it arbitrary.

¶ 74. It is correct that Harenda's tests indicated that one more sample contained ACM than the State's test, and it is also correct that the tests differed with *638respect to the precise amount of asbestos in some samples. However, the samples are not "dueling" with respect to whether the material was ACM under the regulations. The tests on the split samples conducted for Harenda and the tests on the split samples conducted for the State both came back positive for ACM. Regardless of "analytic variability," the two sets of tests were consistent in the relevant respect, namely, whether the material was ACM.

¶ 75. Harenda's pre-demolition sampling and testing was the only set of tests that did not indicate that material from the disputed area was ACM. The State's test during demolition indicated that the material was ACM, and the State's test after demolition was halted indicated that the material was ACM. Further, both the State's and Harenda's tests on the split samples indicated that the material was ACM. The parties have stipulated that the tests were accurate within their testing method. Thus, the array of positive tests is sufficient to show the presence of ACM.

¶ 76. Accordingly, we determine that there was nothing arbitrary in the circuit court's conclusion that the tests showed that the material from the disputed area was ACM. Harenda therefore has no basis for a claim that the decision violated its right to due process.

VI

¶ 77. In sum, we determine, first, that the language of § 1.7.2.1 is ambiguous. Giving deference to an agency's interpretation of its own rule, we conclude that the EPA's interpretation is controlling because it is neither inconsistent with § 1.7.2.1 nor clearly erroneous. We further determine that the clarifications do not *639constitute impermissible rule making; rather, they are valid interpretive rules. Finally, we determine that the circuit court's judgment does not violate Harenda's substantive due process rights. We therefore reverse the court of appeals.

By the Court. — The decision of the court of appeals is reversed.

ANNETTE KINGSLAND ZIEGLER, J.

¶ 78. (dissenting). In an area as crucial as asbestos removal, it is important to have clear, understandable, enforceable testing methods. In general, clarifications may be a useful guide, but the "clarifications" in this case are inconsistent with the plain language of the rule and more than interpretive guidelines. In fact, the "clarifications" effect a substantive change rather than clarify the alleged ambiguity in the 1990 rule. Despite the fact that the "clarifications" do not clarify the alleged ambiguity, are contrary to the promulgated rule, and were not adopted in this state, the majority still gives them the force of law. As a result, I must respectfully dissent.

I

¶ 79. Asbestos is a group of minerals that naturally occur as fibers. It has been widely used in many industrial products such as flooring, textiles, and insulation.1 Exposure to asbestos can lead to mesothelioma, which is a rare form of cancer that affects the protective sac of most internal organs such as the lungs. "Most people who develop mesothelioma have worked on jobs *640where they inhaled asbestos particles."2 Symptoms, however, may not appear until 30 to 50 years after exposure. The incidence of this rare cancer has increased over the past 20 years but may be stabilizing now. Once symptoms appear and the cancer is diagnosed, the disease is often very advanced.3 Mesothe-lioma is often associated with a horrible death.

II

¶ 80. The Clean Air Act regulates hazardous pollutants, such as asbestos. 42 U.S.C.A. § 7412 (West 1999). The Environmental Protection Agency (EPA) enacted the National Emission Standards for Hazardous Air Pollutants (NESHAP) pursuant to its authority under § 7412. In 1973, the EPA promulgated the first asbestos NESHAP regulation. See 38 Fed. Reg. 8,820, 8,826-30 (Apr. 6, 1973). The asbestos NESHAP specifies procedures for removing certain asbestos materials. 40 C.F.R. Pt. 61, Subpt. M, § 61.145. The 1973 regulation underwent significant changes in 1990 in order to "enhance enforcement and promote compliance with the current standard without altering the stringency of existing controls." 55 Fed. Reg. 48,406, 48,406 (Nov. 20, 1990).

¶ 81. Following the proper rule-making process, the EPA established the "Interim Method" for determining asbestos content in bulk samples.4 See id. at 48,415; 40 C.F.R. Pt. 763, Subpt. E, App. E, § 1.7.2.1, Gross *641Examination. The DNR incorporated by reference the 1990 testing method, but it did not so incorporate the 1994 and 1995 clarifications.5 The 1990 method seems to require the following: (1) examine the core sample for homogeneity; (2) if the sample is homogeneous, identify fibers and quantify asbestos content for the sample; (3) when discrete strata are identified, the fibers in each layer are identified and quantified; "and then" (4) after the layers are quantified, the results of each layer are combined to yield an estimate of asbestos content for the whole sample. See, e.g., State v. Harenda Enters., Inc., 2006 WI App 230, ¶ 4, 297 Wis. 2d 571, 724 N.W.2d 434.

¶ 82. The plain language of the 1990 rule is instructive. It does not stop at the analysis of each layer. It reads: "and then the results for each layer are combined to yield an estimate of asbestos content for the whole sample." (Emphasis added.) If the rule was intended to require removal when any one layer was *642over one percent, the rule would not need the last clause, which requires that layers be combined to yield a result for the whole sample.

¶ 83. The majority, however, concludes that the language in § 1.7.2.1 is ambiguous because of the word "combined." See majority op., ¶¶ 29-36. As a result, the majority relies on the 1994 and 1995 EPA clarifications, which undermine the plain language of the properly promulgated 1990 rule.6 The majority accepts the two EPA "clarifications" even though they do not have the force of law, are contrary to the plain language of the rule, and read the "combined" requirement out of existence. See majority op., ¶¶ 38-49. Because the clarifications (1) do not clarify the ambiguity; (2) were not properly promulgated and are not the law of this state; and (3) fail to fully protect the worker, I must dissent.

A

¶ 84. Even if we look to the 1994 and 1995 clarifications and ignore the fact that the DNR did not adopt the clarifications but adopted only the 1990 rule, the clarifications still do not clarify the alleged ambiguity of *643what "combined" means. In 1994 and 1995, the EPA produced "clarifications" for the 1990 rule. The January 5, 1994, clarification briefly explained, "when a sample consists of two or more distinct layers or materials, each layer should be treated separately and the results reported by layer (discrete stratum)." 59 Fed. Reg. 542 (Jan. 5,1994). Even if this is an explanation, it still fails to explain what actions should be taken to combine layers to yield an estimate for the whole sample.

¶ 85. To add to the confusion, the second clarification, issued in 1995, states that combining layers was never allowed. That contention, however, is clearly contrary to the plain language of the rule. The 1995 clarification states that the "unwritten policy" of the EPA dating back to the 1970s "was that each layer in a multi-layered system was to be analyzed as a separate material (no averaging or dilution by combining layers of asbestos-containing material with nonasbestos-containing material was allowed)." 60 Fed. Reg. 65,243 (Dec. 19, 1995). While this new test — the single layer approach — may be a better approach, it is contrary to the 1990 rule's plain language. Rather than clarify the alleged ambiguity, the clarifications create a new and distinct rule.

B

¶ 86. Under our system of government, any enforceable rule, however, must be vetted through the proper rule-making process. The rule-making process is an important part of our democracy. It gives interested parties the opportunity to be heard. The Administrative Procedures Act governs the promulgation of new federal agency rules, 5 U.S.C.A. § 553, and chapter 227 of the Wisconsin Statutes governs the promulgation of Wisconsin agency rules. Rules must be subjected to a *644notice and comment period before they may take effect. 5 U.S.C.A. § 553(b) and (c); Wis. Stat. §§ 227.16-227.19. The notice and comment period occurred at the state and federal level for the 1990 rule, but it never occurred with respect to the 1994 and 1995 clarifications. Because the clarifications effect substantive change and create a new rule, the rule-making process was required.

¶ 87. When a rule is interpretive rather than substantive, the rule is excepted from the notice and comment period requirements. 5 U.S.C.A. § 553(c). Whether a rule is substantive rather than interpretive, however, depends upon whether it creates rights, assigns duties, or imposes new obligations. See Hemp Indus. Ass'n v. Drug Enforcement Admin., 333 F.3d 1082, 1087 (9th Cir. 2003) (concluding (1) that a legislative or substantive rule promulgated without following the rule-making requirements is invalid; and (2) that a rule is legislative or substantive if it imposes new obligations or changes existing law).7

¶ 88. Here, there can be no question that the 1994 and 1995 clarifications effect a substantive change to the law, impose new obligations on companies such as Harenda, and create significant civil or criminal liability. The "clarifications" change the analysis from a multi-layered approach to a single-layer approach. This changes the rule — without following proper rule-making procedures — because instead of a quantity of asbestos for the whole sample giving rise to an obliga*645tion for special removal procedures, a company, such as Harenda, must follow special removal procedures when only one layer out of many contains greater than one percent asbestos. Because the clarifications impose new obligations upon companies such as Harenda, the clarifications change the rules governing asbestos testing.8 In fact, the State seems to concede that the clarifications change the rule; the State admits that it cannot meet its burden of proof if the clarifications are not the law.9 Thus, the clarifications substantively change the rules governing asbestos testing.

*646¶ 89. The EPA even acknowledged its responsibility to promulgate a new rule so that it could properly enforce the new test. The 1995 clarification itself states that with regard to the composite analysis method, the "EPA intends to amend the asbestos NESHAP in the near future to refer specifically to these procedures." 60 Fed. Reg. 65,243 (Dec. 19, 1995). However, the EPA never took that action.

¶ 90. The 1995 clarification devised an alternative cost-saving test method, which allows one to have the composite analyzed first. Id. If the "composite analysis shows that the average content for the multi-layered" sample is over one percent, then the material must be treated as asbestos containing (ACM). Id. (Emphasis added.)10 However, if the composite analysis yields asbestos below one percent but greater than zero, then analysis by layers is required to ensure that no layer is greater than one percent asbestos content, which would deem that layer an asbestos containing layer giving rise to precautionary measures for the entire sample without combining. Id. Thus, under this clarification, if one layer is greater than one percent, the entire sample is considered ACM. This, however, is contrary to the language of the 1990 promulgated rule, and the DNR never adopted the clarifications. As ac*647knowledged by the EPA, new action is required to promulgate the 1995 testing method.

¶ 91. As we see in this case, the clarifications have been enforced by the DNR as if they have the force of a properly promulgated rule, despite the fact that our state did not incorporate by reference the clarifications11 and they effect a substantive change to the rule. While an agency may interpret rules, they cannot effect a new rule that changes the substantive rule. In effect, new rules governing asbestos testing, which could result in civil or criminal penalties, have been imposed without following proper rule-making procedures. See United States v. San Diego Gas & Elec. Co., No. 06-CR-65-DMS, 2006 WL 3913457, at *6-8 (S.D. Cal. Nov. 21, 2006) (concluding that the clarifications are legislative and thus impose new obligations on companies even though the proper rule-making process was not followed; therefore, a company cannot be held liable based on testing methods articulated by the clarifications).

¶ 92. Proper rule making is important because matters are openly debated and people have the opportunity to be heard. It is fundamentally unfair to subject *648anyone to significant liability — especially criminal liability — when proper rule making did not occur. Here, Harenda was exposed to $4 million in liability.12 If the State intends to hold people or companies liable— potentially criminally liable — the rules must be properly enacted.13 Furthermore, it runs counter to our *649system of justice to hold anyone civilly liable, and especially criminally liable, when the proper testing method is ambiguous as the majority claims.

¶ 93. The rule of lenity must be considered. While the case at hand is a civil action, our interpretation of the rule will affect those who are subject to criminal prosecution for "asbestos violations." The rule of lenity provides "that ambiguous penal statutes should be interpreted in favor of the defendant." State v. Cole, 2003 WI 59, ¶ 67, 262 Wis. 2d 167, 663 N.W.2d 700. "More specifically, the rule of lenity comes into play after two conditions are met: (1) the penal statute is ambiguous; and (2) we are unable to clarify the intent of the legislature by resort to legislative history." Id. The majority concludes that the rule is ambiguous and thus the first prong is satisfied. Perhaps the clarifications could be utilized under the second prong to ascertain the intent of the EPA, but when the legislative history or "clarifications" change the substantive rule's plain language, we cannot rely on it to uphold liability.

*650¶ 94. Here, the clarifications do not merely explain substantive law. Rather, they impose new obligations by changing existing law. Thus, the clarifications introduce substantive changes in the law without following the proper rule-making procedures.14

C

¶ 95. The current testing methods and procedures are dangerously ineffective. Harenda and the DNR both had testing completed under the clarifications method, yet they still acquired different results. Not only were the results different pre- and post-demolition, but the tests conducted post-demolition produced competing results. Harenda's pre-demolition samples did not give rise to special precautions. Post-demolition samples, however, which were divided and given to both Harenda and the State, were not consistent with each other although both results would have required ACM precautions. When the same test yields different results, even on the same sample area, how does a court resolve these competing facts at summary judgment?

¶ 96. While the DNR argues that the clarifications result in safer working conditions, they still do not actually protect the worker. Although this court cannot *651require it, our federal and state government should consider implementing guidelines that actually protect the worker, such as requiring state authorities to conduct testing prior to demolition. If the clarifications are "adopted" by this court, then the DNR should act to properly adopt them. For example, the state of New Jersey enacted 'emergency rules' for the sole purpose of adopting the clarifications. See Robert M. Howard, Patricia Guerrero, David B. McGrath, Drew R. Van Orden, The EPA's Prosecution of Clean Air Act Asbestos NESHAP Cases Based Upon Non-binding Bulk Material Test Methods, 44 San Diego L. Rev. 173, 207-08 (2007). "New Jersey explained that the 1993 test [i.e., the 1994 clarification] method is preferable to the 1990 test method because the enumerated NESHAP 1990 test method underreports asbestos content...." Id. at 208. Our state has not so acted.

III

¶ 97. The "clarifications" do not clarify the 1990 rule. The 1994 clarification promotes a single-layer test, which is not the test articulated in the properly enacted rule. The 1995 clarification confuses the issue even more. The clarifications do not clarify what "combined" means. Moreover, the clarifications are inconsistent with the plain language of the rule and are more than interpretive guidelines. In fact, they effect a substantive change. The clarifications and the majority's interpretation eliminate the last clause of the rule even though the clarifications have not been subjected to the proper rule-making process. The government should be required to clearly articulate and enact the lawful test in order to ensure that citizens are not exposed to asbestos *652and put companies on notice about potentially, significant liability for a violation of the "properly enacted rule."

¶ 98. Here, neither the EPA nor the DNR properly promulgated the clarifications. The clarifications effected a substantive change to the "Interim Method," and they are not the law of this state. In the case at issue, the State's motion for summary judgment should not have been granted.

¶ 99. For the foregoing reasons, I respectfully dissent.

¶ 100. I am authorized to state that Justices DAVID T. PROSSER and PATIENCE DRAKE ROGGENSACK join this dissent.