PHOENIX NEWSPAPERS, INC., an Arizona corporation, Plaintiff-Appellee, Cross Appellant, v. Lisa Graham KEEGAN, Superintendent of Public Instruction of the State of Arizona; The State of Arizona Board of Education, a public body; and The State of Arizona Department of Education, a public body, Defendants-Appellants, Cross Appellees.
¶ 1 This case requires us to review a superior court's decision that some of the questions in a statewide academic test are open to public inspection under Arizona's public records law. The superior court, exercising special action jurisdiction, granted partial relief to the plaintiff, a newspaper publisher. The court ordered that the defendants, a State officer and State agencies, allow the newspaper to inspect and copy most questions from the first administration of the test known as “AIMS” (Arizona Instrument to Measure Standards).
¶ 2 Dissatisfied with the superior court's decision, both sides appealed. The State defendants contend that they should not be required to disclose any test questions. The newspaper's cross-appeal argues that all questions should be open to public inspection. The newspaper also challenges the superior court's decision not to award attorneys' fees incurred by the newspaper in pursuing disclosure.
¶ 3 The document in dispute is the first version, known as “Form A,” of AIMS. AIMS is the result of State legislation requiring the State Board of Education to “[d]evelop and adopt competency tests for the graduation of pupils from high school in at least the areas of reading, writing and mathematics and shall establish passing scores for each such test.” Ariz.Rev.Stat. (“A.R.S.”) § 15-701.01(A)(3) (Supp.2000). Form A was administered to about 50,000 Arizona high school sophomores in preparation for additional testing. Subsequent tests, consisting of six total versions, ultimately would be administered to all Arizona high school students from their sophomore to senior years. A passing AIMS score will be needed to graduate from high school. See A.R.S. § 15-701.01(A)(3) (Supp.2000).1
¶ 4 Form A was administered in the spring of 1999. In November, the State released the results. About 92 percent of the students failed one or more of the three basic parts of the examination: mathematics, writing and reading. The failure rate on the mathematics portion alone was about 88 percent. This administration of AIMS was not used to grant or withhold high school diplomas, but the same form could be used later for this purpose.
¶ 5 Sample questions representative of examination items were released in advance of the test administration. These included answering multiple choice items, writing essays, narratives, letters and short answers, interpreting data and graphs, diagraming, and solving of equations.2
THE REQUEST FOR INSPECTION
¶ 6 Immediately following the State's release of the results, the newspaper asked to inspect and copy Form A. Public documents are generally subject to inspection and copying upon request. A.R.S. § 39-121.01(D)(1) (Supp.2000). The State declined to permit copying, but offered limited viewing. The State's offer allowed the newspaper to view the document for a limited time, but note-taking or copying was forbidden. The State also required a non-disclosure agreement.3 The newspaper was permitted additional viewing time if it appeared justified to the State.
THE SPECIAL ACTION
¶ 7 The disclosure battle was joined when the newspaper filed a special action seeking a superior court order that the State permit inspection and copying of Form A. Not every conceivable issue was disputed in the special action, however. For example, the State conceded that Form A is a public record subject to Arizona's public records laws. For its part, the newspaper did not seek to expand its quest for disclosure to include later forms of the test, including at least one prepared before the superior court decided this matter. Thus, this case involves what the parties agree are public records, and it involves only the disclosure of Form A and not other versions of AIMS.
¶ 8 The dispute in the superior court focused upon this principal question: Must the public's right to inspect this public record yield to some greater State interest? The State defendants vigorously argued that disclosure of Form A would render worthless their efforts to devise the several forms of AIMS. The State argued that it would require millions of dollars to replicate the testing regime. The newspaper responded that disclosure of Form A would not prevent valid test results even if Form A were used again. Both sides relied on expert opinions.4
¶ 9 The superior court found a middle ground between the parties' positions. It determined that Form A contained two types of questions: “anchor” questions that would be repeated in future forms and all other questions. The anchor questions need not be disclosed, the court ruled. All other questions must be made available for public inspection. More than one of every five test items is an anchor.
¶ 10 Because the superior court stayed its order pending appeal, the State was not obliged to immediately release the non-anchor questions. Nevertheless, the State released many of the Form A questions, publishing them on the Arizona Department of Education's website.5 The State does not suggest that its release satisfied the superior court order by revealing all of the non-anchor questions that the court directed be open to inspection. The release also does not include the anchor questions that the newspaper still wishes to inspect. The public release of these questions therefore did not resolve this dispute.
STANDARD OF REVIEW
¶ 11 Our review does not simply re-decide all questions already resolved by the superior court. We defer to any findings of fact the court made, such as the difference between anchor questions and other questions. See Scottsdale Unified Sch. Dist. No. 48 v. KPNX Broad., 191 Ariz. 297, 302, 955 P.2d 534, 539 (1998) (applying “clearly erroneous” standard of review). However, the matter of the public's right of inspection is a question of law that we decide de novo. Id. at 302, 955 P.2d at 539.
¶ 12 The parties present many questions to us, but the central question is this: Did the superior court properly determine that the balance of public interest favored disclosure of all but the anchor test questions? 6
¶ 13 We affirm the superior court in all respects. We agree that, under the facts, the law requires disclosure of some but not all of the document. We also see no abuse of discretion in the court's denial of attorneys' fees to the newspaper.
¶ 14 We first address a procedural question. The State defendants attack the superior court's decision as “premature.” The court upheld the public's right of inspection after considering whether some greater State interest in non-disclosure outweighed the presumptive right of inspection. See Carlson v. Pima County, 141 Ariz. 487, 491, 687 P.2d 1242, 1246 (1984). The parties agreed that the court should perform this balancing task, but apparently the State did not anticipate an outcome in which some but not all of the Form A questions would be disclosed.
¶ 15 The State complains that the court needed more evidence and therefore that its decision was “premature.” The State's argument is no more than an admission that it did not anticipate the outcome and an assertion that it would have adopted different tactics had it known.
¶ 16 The superior court acted only after both sides had full opportunity to produce evidence. Some of the evidence distinguished anchor questions from other test questions. One of the possible outcomes was to require partial disclosure. In fact, the State recognizes that precedent supports redaction of documents to allow disclosure when possible. See KPNX-TV v. Super. Ct., 183 Ariz. 589, 594, 905 P.2d 598, 603 (App.1995). Thus, the court's order was fully consistent with the evidence presented and well within its authority. The parties were not entitled to new evidentiary hearings merely because they did not foresee this particular outcome.7
BALANCING OF INTERESTS
¶ 17 We now turn to the pivotal issue: Did the court correctly balance the competing interests? The general statutory right of inspection of public records is unqualified. A.R.S. § 39-121.01(D). However, consistent with certain statutory exceptions from disclosure, a common law limitation on disclosure exists. Carlson, 141 Ariz. at 490, 687 P.2d at 1245. Although the statute creates a presumption of disclosure, id. at 491, 687 P.2d at 1246, this presumption can be outweighed by other public interests.
¶ 18 A public official can withhold inspection by showing that non-disclosure serves “confidentiality, privacy or the best interests of the state․” Id. This “best interests of the state” standard is not confined to the narrow interest of either the official who holds the records or the agency he or she serves. It includes the overall interests of the government and the people. Of course, the government's administrative interests are not excluded from the weighing process: The public interest includes consideration of how disclosure would adversely affect the agency's mission. But it also may include other ways in which the public would be affected by disclosure or non-disclosure.
¶ 19 To justify withholding public documents, the State's interest in non-disclosure must “outweigh the general policy of open access.” Id. The official who wishes to withhold public documents must prove specifically how the public interest outweighs the right of disclosure. See Cox of Ariz. Publ'ns v. Collins, 175 Ariz. 11, 14, 852 P.2d 1194, 1197 (1993); KPNX Broad., 191 Ariz. at 300, 955 P.2d at 537 (“The State has the burden of overcoming ‘the legal presumption favoring disclosure.’ ” (quoting Collins, 175 Ariz. at 14, 852 P.2d at 1198)).
¶ 20 The superior court's balancing of these interests is assailed from both sides. The State says that the court should not have released even a single question from Form A; the newspaper says that not a single question should be withheld from public view.
¶ 21 We can decide whether the superior court correctly balanced the interests only after considering what it found as facts. Some of the facts were established by the parties' joint statement of facts; others were contested. By long-established rule, we imply any findings of fact supported by the evidence that are necessary to uphold the court's judgment. Coronado Co. v. Jacome's Dep't Store, Inc., 129 Ariz. 137, 139, 629 P.2d 553, 555 (App.1981).
¶ 22 The AIMS test differs from more familiar academic tests. AIMS is a “high-stakes,” “criterion-referenced” test. It is high-stakes because graduation from high school will depend on passing the test. It is criterion-referenced because it measures how well students know information or perform tasks that they are expected to know and perform as a result of high school study. In contrast, norm-referenced tests measure how well students perform in relation to other students.8 The Scholastic Aptitude Test (SAT) is a familiar example of a norm-referenced test, a result of which is that students are informed of how their scores compare with those of other test-takers.9
¶ 23 Because the stakes-a high school diploma-are so great with AIMS, the State is understandably concerned with the fairness of the examination. To enhance fairness, students will be given multiple opportunities to pass the examination during their sophomore, junior and senior years of high school. The State plans to devise a total of six forms of the examination and use them in rotation.
¶ 24 It is the use of several forms of the examination that implicates “anchor questions.” Anchor questions are repeated from one form to another, although not every anchor question need appear on each form. The results on the anchor questions from different forms are compared. The results are then compared to results from non-anchor questions on both forms. The resulting comparison informs the testers whether one form is significantly more difficult than another.10 Thus, the stated purpose of this process is to promote fairness by ensuring the equivalency of difficulty among the various AIMS forms. The expert testimony referred to this process as “equating.”
¶ 25 The State's evidence of the importance of equating and of its need to use anchor questions to equate tests was essentially unchallenged. The newspaper did not dispute the importance of equivalency. It did not contend that administering six different forms of the test minimized the need for test-equating. Nor did the newspaper suggest alternatives to anchor questions as a means of achieving equivalence.11
¶ 26 The superior court decided that anchor questions from Form A could not be disclosed because they must be used again in subsequent forms. Disclosing these questions in advance would “limit the ability of the test results to be compared to the results of other similar tests taken in Arizona in years to come,” the court said. Disclosure of all Form A questions would also require substantial cost to devise new questions. Although the State emphasized the allegedly costly consequences of disclosure of the entire Form A, the evidence suggested that rewriting anchor questions would require much greater effort and cost than remedying disclosure of other questions.
¶ 27 The newspaper argued that disclosure of all test questions would not render future results based on the same questions “invalid.” 12 The superior court found the impact of disclosure on the testing process to be a disputed fact and decided to the contrary. The evidence supports that decision. The evidence was that if anchor questions were disclosed, they could not be used again. Public release of all questions, without replacing the anchors, would significantly increase test scores. Releasing the anchor items would threaten the equating function that enables testers to evaluate the relative difficulty of test forms and thus the comparability of scores from one form to another.
¶ 28 The court directed that non-anchor Form A questions be open to public inspection. It thus found that a lesser public interest weighed against disclosing the non-anchor questions. The record supports the court's decision that disclosure of the remainder of the examination would cause much less difficulty and expense than would disclosure of the anchor questions. Among other things, the testing contractor may be able to use “off the rack” questions it previously devised for use in other states, just as it did in writing Form A. Cf. State ex rel. Rea v. Ohio Dep't of Ed., 81 Ohio St.3d 527, 692 N.E.2d 596, 599 (1998) (Ohio statewide academic test questions drawn from a bank of about 14,000 questions). In addition, some test items may not be vulnerable to pretest preparation, such as questions that call for essay answers. Other questions may be altered in form for use in a retest so that a test subject could not provide a memorized answer. It may also be possible to rotate questions among forms. Rotating questions means that a test subject would find it difficult to predict, and thus prepare for, which 135 questions of the total of at least 810 questions from the six forms will be asked. Of course, the total number of possible questions could be far greater than 810 if any additional questions were substituted from a question bank. The duration of the test-an average of nine hours-also somewhat limits the effectiveness of memorizing answers in advance.
¶ 29 The State emphasizes expense and difficulty of administration as the costs of disclosure. As noted above, the superior court did not err in distinguishing between the anchor questions and the remainder of Form A. The evidence suggests that rewriting the anchor questions is impracticable but that remedying the disclosure of other questions is not.
¶ 30 The State's argument is deficient in another respect: The State fails to recognize the significant interests arrayed against its asserted interests of expense and convenience. The public's right to know any public document is weighty in itself. In this case, however, the public documents are of broad and intense interest. The specific public interest in disclosure must be weighed in balancing that interest against the State's asserted harm. “[T]he court must evaluate carefully the public interest that plaintiff seeks to vindicate in requesting those documents․” S. New Jersey Newspapers, Inc. v. Township of Mt. Laurel, 141 N.J. 56, 660 A.2d 1173, 1184 (1995) (applying balancing test and holding that firearm permit disclosure related to “a matter of fundamental public concern, ․ the proper issuance of firearm licenses only to persons qualified to receive them ․”).
¶ 31 Consideration of the public interest resolved a request for disclosure of standardized achievement test results in Citizens for Better Ed. v. Bd. of Ed. of the City of Camden, 124 N.J.Super. 523, 308 A.2d 35 (1973). The court balanced the competing interests in favor of disclosure: “Petitioners are endeavoring to ascertain what deficiencies, if any, exist in any grade or school. This is a matter of legitimate public concern and clearly outweighs any conceivable interest there may be in maintaining the confidentiality of the information in the form sought.” Id. at 38.
¶ 32 In Arizona, AIMS has been the subject of significant public debate. The administration of a test that might deny a diploma to nearly all Arizona high school students, is a matter of clear public interest. The Ohio Supreme Court ordered disclosure of academic test results after considering the public significance of the information.
The tremendous implications for students who take such tests or assessments cannot be underestimated. The instruments that evaluate students and determine their capabilities should not be enshrouded in a cloak of secrecy, isolated from the scrutiny and oversight of the general public, concerned parents, and students themselves.
Ohio Dep't of Ed., 692 N.E.2d at 602. Moreover, the high failure of AIMS test subjects implies either that the test instrument is inapt or that Arizona schools have failed their mission. According to expert testimony, review of the test questions would enable an evaluation of the test.
¶ 33 The superior court did not err in weighing the public interest in disclosure more heavily than the State's cost and inconvenience in remedying that disclosure. “The core purpose of the public records law is to allow the public access to official records and other government information so that the public may monitor the performance of government officials and their employees.” I91-004 Ariz. Op. Att'y Gen. 418, 420 (1991). “[E]ducation of its citizenry is one of the most important functions of the state․” Ohio Dep't of Ed., 692 N.E.2d at 601. In this light, we see no error in the order requiring disclosure.
¶ 34 The State also contends that no questions should be revealed because they are “trade secrets.” If they were trade secrets, they would be protected by the confidentiality exception to disclosure. See Carlson, 141 Ariz. at 491, 687 P.2d at 1246.
¶ 35 The State did not present this contention to the superior court. The trade secrets argument was raised only by the intervenor, the test developer. The developer did not challenge the superior court's judgment and is not a party to this appeal. The State may not advance an argument for the first time on appeal. Dugan v. Fujitsu Bus. Communications Sys., 188 Ariz. 516, 521, 937 P.2d 706, 711 (App.1997).
¶ 36 Even if the argument were not waived, it lacks merit. The superior court found that disclosure of the test questions would not destroy their economic value. A trade secret is information that “[d]erives independent economic value, actual or potential, from not being generally known to, and not being readily ascertainable by proper means by, other persons who can obtain economic value from its disclosure or use.” A.R.S. § 44-401(4)(a).
¶ 37 The correctness of the court's decision is illustrated by the brief of amicus curiae, The Association of Test Developers. In its brief, the amicus states:
A competitor would have to see the technical manual, setting forth all of the research associated with the test, before it would gain any meaningful information for competitive test-building purposes; thus, merely seeing the specific items would not disclose the trade secret, i.e., the methodology for engineering the test, what items are correlated to what subject matter content and for what level of difficulty. Absent full disclosure of the full research information, the kind of limited disclosure that was provided through the Arizona inspection process would not have resulted in any serious concern by the test developer.
¶ 38 The State's confidentiality argument is also belied by its voluntary release and publication of many of the Form A questions during the pendency of its appeal. The items released can no longer be trade secrets, see A.R.S. § 44-401(4)(b) (trade secret must be “the subject of efforts that are reasonable under the circumstances to maintain its secrecy”), and the release is clearly inconsistent with the contention that the entire test instrument is a trade secret.
¶ 39 The newspaper challenges the superior court's denial of its request for attorneys' fees. The superior court declined to award fees after finding that the State did not act in bad faith, arbitrarily or capriciously in denying public access.
¶ 40 Fees are recoverable in actions to require public access pursuant to A.R.S. § 39-121.02 (1996). That statute requires as a condition to a fee award that the court find that “the custodian of such public record acted in bad faith, or in an arbitrary or capricious manner․” Id.
¶ 41 The superior court did not err in declining to find that the State had acted so improperly that fees were warranted. The State had colorable reasons for its limited access policy. Applying the same statute and the same reasoning, we deny the newspaper's application for fees incurred in this appeal.
¶ 42 The superior court correctly ordered the release of some but not all of the Form A questions. Accordingly, we affirm its order.
¶ 43 I agree with the majority's result but not its reasoning. I would affirm the trial court's Solomonic ruling on the ground that the State upended its non-disclosure argument by disclosing the very questions it now seeks to protect. The State's release of questions to the newspapers, the web and other media undermines its present contention that its interests in confidentiality outweigh the public interest in disclosure. If the State had not so acted, I would hold that the State need not disclose any test questions at all because doing so destroys the very idea of a “test.” We have good authority to keep test questions confidential. See Carlson v. Pima County, 141 Ariz. 487, 687 P.2d 1242 (1984).
¶ 44 The State education department has also been inconsistent regarding anchor versus non-anchor questions. At first, in the joint proposed statement of uncontested facts, it admitted that approximately 45 scored anchor questions appeared on Form A. Later, it stated that there were no set anchor questions redactable from Form A and that all or most of Form A's original questions could eventually evolve into anchor questions. Later, it released to the media Form A questions that it said would not be used. Once the State disavows any lasting distinction between anchor and non-anchor questions, release of any test questions seemingly impairs the utility of the remaining questions. Because there is a basis for a distinction, however fuzzy, between anchor and non-anchor questions, there is support in the record for the trial court's split decision. For that reason I defer to his judgment though, given the State's disclosures, I would get there by a different route.
¶ 45 If not for these behaviors by the State, I would avoid this Alice-in-Wonderland approach to public education and hold, instead, that all the test questions simply be kept private. “While access and disclosure is the strong policy of the law, the law also recognizes that an unlimited right of inspection may lead to substantial and irreparable private or public harm․” Carlson, 141 Ariz. at 491, 687 P.2d at 1246. A student's performance on these disclosed questions could well reflect only the ability to memorize the publicized questions.
¶ 46 Equally troubling is the precedential effect of our ruling on some 33 other government tests such as driver's license and law enforcement exams and college admission tests. Our opinion opens a Pandora's box for all public tests, intruding into an arena best left to the legislature and forcing the courts to use a newly-honed Ockham's razor to sever complex from less complex test questions, a task better left to educators than judges.
¶ 47 Given the State's prior disclosure of the questions it now insists must be non-disclosed, I concur with the majority's result, having learned in the process a personally useful academic lesson: The next time I face a difficult test, I can sue to discover the test questions in advance.
1. The high school achievement test is but one part of Arizona's pupil assessment plan. Other AIMS tests are to be administered in at least four grade levels. A.R.S. § 15-741(A)(2) (Supp.2000). In addition, standardized norm-referenced tests are to be administered for grades three through twelve. A.R.S. § 15-741(A)(3) (Supp.2000).
2. Student Guide for Arizona's Instrument to Measure Standards (Ariz. Dep't Ed.1999).
3. These restrictions slightly liberalized the inspection policy in effect just prior to the newspaper's request, which had imposed the same restrictions but allowed viewing only up to one hour.
4. The parties chose to submit their experts' opinions by affidavit and deposition rather than by live testimony. The record thus created is not insubstantial but, by comparison to the complex and technical subject of academic achievement testing, is relatively thin and simple.
6. The State also asserts that the public records law does not apply to AIMS because the Legislature gave to the State Board of Education the authority to “adopt and implement” the test, and “determine the manner of implementation.” A.R.S. § 15-741(A)(2). That statute should trump the public disclosure statute because it is later and more specific, the State argues.The State's argument presupposes that the two statutes conflict, and therefore that one must override the other. Its approach does not apply if the statutes can be read harmoniously, giving effect to both. See Lemons v. Super. Ct., 141 Ariz. 502, 505, 687 P.2d 1257, 1260 (1984). These statutes can be so read: The Board has authority to implement the test, but test-related documents that qualify as public records are subject to the public records law. Indeed, the State concedes that the documents involved here are public records.We also note that the parties' joint statement regarding legislative history indicates that several legislative attempts to exempt AIMS from the public records law failed.
7. Although the State did not formally request the superior court to reopen to hear additional evidence, it did state that the court should take further testimony regarding the anchor question distinction. Even a formal request to reopen the evidence would have been within the court's discretion to grant or deny. State v. Doody, 187 Ariz. 363, 378, 930 P.2d 440, 455 (App.1996). No party is entitled to a second chance merely because hindsight suggests that other or additional evidence could have led to its success.
8. The superior court referred to AIMS as “a test that is ‘nationally standardized [and] norm-referenced.’ ” (quoting A.R.S. § 15-741(A)(3), (A)(6) (Supp.2000)). This is a statement not of fact, but a description of what the statutes require. However, it is not a correct description of the statutes. The annual standardized, norm-referenced tests are given to grades three through twelve. See A.R.S. § 15-741(A)(3). In contrast, the AIMS is administered to four selected grades and is not required to be standardized or norm-referenced. The high school AIMS is to be administered not once but twice each year from the sophomore through senior years, and passage is required for high school graduation.
9. It is possible to have a criterion-referenced test with elements that permit some norm-referencing. Conversely, norm-referenced tests have some element of criterion testing in that the questions are selected to measure some particular knowledge or skill. But it is possible to have a criterion-referenced test without regard to norms. An analogy makes this clear. A dog owner wants to build a fence that will keep his dog inside his yard. The owner tests the dog to see how high he can jump, a criterion-referenced test. It does not matter to the owner how high other dogs can jump, so the test is not norm-referenced.
10. Differences in difficulty from one examination to another can be adjusted to achieve equivalent results by scaled scoring. That process converts the raw score-the number of correct answers-to a scaled score.
11. The newspaper's expert stated that non-scored anchors could be used. That is, answers to anchor questions could be used to ensure equivalency but not used in a student's test score. However, Form A uses scored anchors. Moreover, the newspaper could seek to inspect non-scored questions as well as any other.
12. “Validity” is a technical concept in testing, and consists of several types. It appears that the newspaper used the term in the layman's sense rather than as a technical term.
CONCURRING: REBECCA WHITE BERCH, Presiding Judge.