¶ 1 Although at the beginning of trial, the superior court admonished the jury to “not consult any source, such as ... the [Ijnternet for information,” and then reminded the jury to observe the admonition throughout the trial, two jurors conducted Internet research on the legal definitions of terms in the court’s final instructions, communicated their research to other jurors, and three additional jurors considered the research before joining the other jurors in unanimously convicting Jesus Valdez Aguilar and eodefendant Francisco Ibarra Norzagaray (collectively, “Appel*300lants”) of attempted first degree murder.1 Because the State failed to prove beyond a reasonable doubt the jurors’ misconduct in this case did not taint those verdicts, we reverse the superior court’s denial of Appellants’ motions for a new trial and remand for further proceedings.2
FACTS AND PROCEDURAL BACKGROUND
¶ 2 Shortly after the court discharged the jury following the conclusion of Appellants’ trial, the bailiff, in his normal duties, discovered “extraneous documents” in the notebook of the jury foreman. These documents consisted of printouts of one definition of first degree murder and three definitions of second degree murder, as obtained from three different Internet sources (the “Internet definitions”). See infra ¶¶ 21-23 and notes 5-7. The court informed the parties of the bailiffs discovery, and Appellants subsequently moved for new trial, arguing the jurors’ use of this material deprived them of a fair trial.
¶ 3 The superior court held a series of evidentiary hearings in which counsel and the court questioned each juror as to his or her knowledge of and reliance on the Internet definitions during jury deliberations. Juror eight, the jury foreman, testified that after the first day of deliberations, he did a “Google” search at home on “first degree murder Arizona” (emphasis added), spending about one-half hour researching the issue. He printed the Internet definitions, brought them into the jury room, and discussed his research with other members of the jury. The foreman was not the only person who accessed the Internet to obtain definitions; so too did juror number nine, who acknowledged he had researched “premeditation” (unless otherwise noted, included in the “Internet definitions”). Jurors discussed and considered these Internet definitions during deliberations. See infra ¶¶ 26-28.
¶ 4 The superior court found the State had “defeated the presumption of prejudice by proving beyond a reasonable doubt that the [Internet definitions] considered by the jury ... did not taint the jury’s verdicts,” and denied Appellants’ motions. Appellants timely appealed. We have jurisdiction pursuant to Article 6, Section 9, of the Arizona Constitution and Arizona Revised Statutes (“AR.S.”) sections 12-120.21 (2003), 13-4031 and -4033(A)(1) (2010).
¶ 5 On appeal, Appellants contend the State failed to prove beyond a reasonable doubt the introduction of the Internet definitions into jury deliberations did not contribute to the verdicts on attempted first degree murder, and thus the superior court should have granted their motions for a new trial. We agree.3
¶ 6 We will not reverse the superior court’s decision to grant or deny a new trial based on alleged jury misconduct absent an abuse of discretion. State v. Hall, 204 Ariz. 442, 447, ¶ 16, 65 P.3d 90, 95 (2003). The superior court abuses its discretion when it misapplies the law or bases its decision on incorrect legal principles. State v. Jackson, 208 Ariz. 56, 59, ¶ 12, 90 P.3d 793, 796 (App.2004). Similarly, an abuse of discretion also occurs when a discretionary finding of fact is “not justified by, and clearly against, reason and evidence.” State v. Chapple, 135 Ariz. 281, 297 n. 18, 660 P.2d 1208, 1224 n. 18 *301(1983). A defendant is entitled to a new trial if it cannot be concluded beyond a reasonable doubt the extraneous information did not contribute to the verdict.4 Cf. Hall, 204 Ariz. at 447, ¶ 16, 65 P.3d at 95 (extraneous evidence rather than extraneous legal definitions). Further, once a defendant shows the jury received and consulted extraneous information, prejudice must be presumed and a new trial must be granted unless the State proves beyond a reasonable doubt the information did not taint the verdict. Id.
¶ 7 In State v. Cornell, 173 Ariz. 599, 601, 845 P.2d 1094, 1096 (App.1992), we stated, “reference to outside sources, including dictionaries, usually has been found to be harmless error.” Nevertheless, in that case, we held a juror’s use of a dictionary to review definitions of “aggravate” and “assault” contributed to the verdict and was, therefore, not harmless. Id. at 602, 845 P.2d at 1097. The juror in Cornell testified that reading the definitions was “like a light bulb going off in [my] head,” and “made my decision for me,” and changed his position from “holdout” to siding with the remaining members of the jury. Id. at 600-01, 845 P.2d at 1095-96.
¶ 8 Despite Cornell and its discussion of Arizona case law, neither our supreme court nor this court has addressed in detail the factors a court should consider to determine whether the State has met its burden of proving, beyond a reasonable doubt, the extraneous legal definitions received and considered by the jury — here, the Internet definitions — did not taint the verdict. See State v. Holden, 88 Ariz. 43, 50, 352 P.2d 705, 710-11 (1960) (juror’s reading of portions of “California Jury Instructions in Criminal Cases” to other jurors did not result in prejudice because it made her “more considerate and more fair to the defendant”); Lane v. Mathews, 74 Ariz. 201, 206, 245 P.2d 1025, 1028 (1952), rev’d on other grounds, 75 Ariz. 1, 251 P.2d 303 (1953) (without quoting words in question or explaining context, finding harmless error when jurors consulted dictionary in civil action); Cornell, 173 Ariz. 599, 845 P.2d 1094 (discussed supra ¶ 7).
¶ 9 Although the Cornell court focused on the divergence between the dictionary definitions and the jury instructions, 173 Ariz. at 602, 845 P.2d at 1097, other jurisdictions have considered additional factors in deciding this issue. See Sassounian v. Roe, 230 F.3d 1097, 1108 (9th Cir.2000) (multifaetor analysis to assess whether extrinsic evidence caused juror prejudice); Mayhue v. St. Francis Hosp. of Wichita, 969 F.2d 919, 924 (10th Cir.1992) (multifactor analysis to assess whether extrinsic legal definitions caused juror prejudice); see also McNeill v. Polk, 476 F.3d 206, 226-27 (4th Cir.2007) (no prejudice because extrajudicial definition of “mitigate” not incompatible with court’s instruction, citing Mayhue); TIG Ins. Co. v. Liberty Mut. Ins. Co., 250 F.Supp.2d 1197, 1200 (D.Ariz. 2003) (no persuasive evidence dictionary definition prejudiced jury, citing Mayhue); Steele v. State, 216 Ga.App. 276, 454 S.E.2d 590, 592-93 (1995), disapproved on other grounds, Kennebrew v. State, 267 Ga. 400, 480 S.E.2d 1 (1996) (extrajudicial definition of law held prejudicial because information not in accord with state law and referred to sentencing, a matter outside purview of jury); George L. Blum, Prejudicial Effect of Juror Misconduct Arising from Internet Usage, 48 A.L.R.6th 135, 165-74, §§ 11-13 (2009) (citing eases); Jean E. Maess, Prejudicial effect of jury’s prvcurement or use of book during deliberations in criminal cases, 35 AL.R.4th 626, 645-52, § 5 (1985 & Supp. 2007) (citing cases).
¶ 10 Although Mayhue is a civil case, we find instructive its analysis of the factors a court should consider when jurors consult extraneous definitions. In Mayhue, following the verdict in a race discrimination ease, the court’s staff found a handwritten note in the jury room containing, inter alia, definitions of the words “discriminate” and “p[re]judiee.” 969 F.2d at 921. The court affirmed the district court’s finding of prejudice, after considering the following factors:
(1) The importance of the word or phrase being defined to the resolution of the ease.
*302(2) The extent to which the dictionary definition differs from the jury instructions or from the proper legal definition.
(3) The extent to which the jury discussed and emphasized the definition.
(4) The strength of the evidence and whether the jury had difficulty reaching a verdict prior to introduction of the dictionary definition.
(5) Any other factors that relate to a determination of prejudice.
Id. at 924.
¶ 11 The factors identified in Mayhue are consistent with certain factors identified by our supreme court in determining whether extrinsic evidence contributed to a jury’s verdict. Hall, 204 Ariz. at 448, ¶ 19, 65 P.3d at 96. In Hall, the jury convicted the defendant of felony murder, armed robbery, kidnapping, and theft. Id. at 445, ¶ 1, 65 P.3d at 93. To connect the defendant to the victim’s disappearance and death, the State relied in part on “grainy” convenience store surveillance videos it argued showed the defendant attempting to use the victim’s credit card. Id. at 446, ¶ 9, 65 P.3d at 94. After the bailiff told at least one juror the defendant had tattoos on his wrist, some jurors “methodically” looked for tattoos on the person in the videos during deliberations. Id. at 447, ¶ 13, 65 P.3d at 95. The supreme court reversed the defendant’s conviction and remanded for a new trial because it could not say beyond a reasonable doubt the improperly introduced evidence of the defendant’s tattoos did not contribute to the verdict. Id. at 449, ¶ 25, 65 P.3d at 97. In so holding, the supreme court identified several factors to assist courts in determining whether extrinr sic evidence has contributed to a verdict:
1. whether the prejudicial statement was ambiguously phrased;
2. whether the extraneous information was otherwise admissible or merely cumulative of other evidence adduced at trial;
3. whether a curative instruction was given or some other step taken to ameliorate the prejudice;
4. the trial context; and
5. whether the statement was insufficiently prejudicial given the issues and evidence in the case.
Id. at 448, ¶ 19, 65 P.3d at 96 (citing Jeffries v. Wood, 114 F.3d 1484, 1491-92 (9th Cir.1997)). Relevant to our analysis here, and sharing similarities with the Mayhue factors, is factor four, the trial context, which the supreme court described as including:
[Wjhether the material was actually received, and if so, how; the length of time it was available to the jury; the extent to which the jurors discussed and considered it; whether the material was introduced before a verdict was reached, and if so at what point in the deliberations; and any other matters which may bear on the issue of the reasonable possibility of whether the extrinsic material affected the verdict.
¶ 12 Combining the principles of and factors identified in Hall, Cornell, and Mayhue, we agree with Appellants the superior court should have granted them a new trial on the attempted first degree murder charge. See Hall, 204 Ariz. at 446-49, ¶¶ 10-25, 65 P.3d at 94-97; Cornell, 173 Ariz. at 601-02, 845 P.2d at 1096-97; Mayhue, 969 F.2d at 923-26.
I. Importance of the Words
¶ 13 To assess the importance of the Internet definitions, we must view them against the evidence presented in this case. The victim, J., testified Norzagaray asked him to meet in the parking lot of a shopping mall under the guise of obtaining directions to the wedding of J.’s sister later that day. Norzagaray told J. to get in a white four-door car so he could “get directions in writing.” J. entered the ear and sat in the back seat next to Norzagaray who was sitting behind the driver; Aguilar was in the front passenger seat. Once inside the car, Aguilar confronted J. concerning a $900 drug debt J. owed him. J. testified Aguilar “had a gun pointed to my face,” and as he tried to escape from the car, Aguilar, Norzagaray, and the driver “wrestl[edj” with him to keep him in the car. The car turned to the left, J.’s shii't ripped, and as he fell out of the car, he “heard a gunshot.” J. later realized he had *303been shot. J. did not see who shot him; he “guess[ed]” it was Aguilar because “he had the gun.”
¶ 14 Aguilar, on the other hand, denied having the gun and any knowledge of the gun until after he heal’d the shot. Aguilar testified he looked over his right shoulder when he heard J.’s door open, saw J. “hanging out of the door,” and heard a shot. Aguilar then saw the driver hide a gun. According to Aguilar, J. owed the drug debt to the driver, not to him (Aguilar), Norzagaray was not involved in the disputed debt between J. and the driver, and at no time did he see Norza-garay “grabbing” J.
¶ 15 Although this conflicting evidence supported an inference Appellants acted with the required premeditation for the attempted first degree murder charge, it also supported a contrary inference — that the shooting occurred impulsively, during the struggle with J. to keep him in the ear. As we explain below, given these conflicting inferences that could be drawn from the evidence, the definitions of “first degree murder” and “premeditation,” and the jury’s understanding of those terms were critical to the determination of Appellants’ guilt for attempted first degree murder. Cf. Mayhue, 969 F.2d at 924-25 (“discriminate” and “prejudice” are critical to a 42 U.S.C. § 1981 action).
II. Extent to Which Dictionary and legal Definitions Differ
¶ 16 The superior court instructed the jury:
The crime of attempted first degree murder requires proof that the defendant:
1. Intentionally engaged in conduct that would have been a crime if the circumstances relating to the crime were as the defendant believed them to be; or
2. Intentionally committed any act that was a step in a course of conduct that the defendant planned would end in the commission of a crime; or
3. Engaged in conduct intended to aid another person to commit a crime, in a manner that would make the defendant an accomplice, had the crime been committed or attempted by the other person.
FIRST DEGREE MURDER
The crime of first degree murder requires proof that the defendant:
1. Caused the death of another person; and
2. Intended or knew that he would cause the death of another person; and
3. Acted with premeditation.
“Premeditation” means that the defendant intended to kill another human being or knew he would kill another human being and that after forming that intent or knowledge, reflected on the decision before killing. It is this reflection, regardless of the length of time in which it occurs, that distinguishes first degree murder from second degree murder. An act is not done with premeditation if it is the instant effect of a sudden quarrel or heat of passion. The time needed for reflection is not necessarily prolonged, and the space of time between the intent or knowledge to kill and the act of killing may be very short.
¶ 17 The first degree murder jury instruction used by the superior court was drawn from our supreme court’s decision in State v. Thompson, 204 Ariz. 471, 65 P.3d 420 (2003). In that case, the supreme court interpreted the legislature’s definition of premeditation. Id. at 475, ¶ 12, 477, ¶¶ 23-27, 65 P.3d at 424, 426-27. As background for its analysis, the court explained that for most of Arizona’s history, first degree murder explicitly required proof of “premeditation,” i.e., “a plan to murder was formed after the matter had been made a subject of deliberation and reflection.” Id. at 476, ¶ 16, 65 P.3d at 425 (quoting State v. Magby, 113 Ariz. 345, 352, 554 P.2d 1272, 1279 (1976)).
¶ 18 Because premeditation involves a defendant’s thought processes, over the years, the courts attempted to describe what type of evidence could show premeditation. Id. at 476, ¶ 17, 65 P.3d at 425. In 1978, the legislature redefined premeditation to mean “the defendant acts with either the intention or the knowledge” he will kill another person “when such intention or knowledge precedes the killing by a length of time to permit *304reflection.” Id. at 476, ¶ 18, 65 P.3d at 425 (quoting A.R.S. § 13-1101(1) (1978)). The legislature also instructed, however, premeditation did not include an act “if it is the instant effect of a sudden quarrel or heat of passion.” Id. Despite the legislature’s efforts to redefine premeditation, litigants “injected confusion” into the analysis through “inappropriate emphasis” of the time element in eases in which there was evidence, whether direct or circumstantial, of actual reflection. Id. at 477, ¶ 21, 65 P.3d at 426.
¶ 19 To resolve this conflict and clarify the distinction between first and second degree murder, our legislature amended the definition of premeditation in 1998 to include the clause “[pjroof of actual reflection is not required.” Id. at 477, ¶ 23, 65 P.3d at 426 (quoting A.R.S. § 13-1101(1) (Supp.1998)). The Thompson court recognized by using this phrase the legislature “sought to relieve the state of'the often impossible burden of proving premeditation through direct evidence” but had not intended to eliminate “the requirement of reflection altogether or to allow the state to substitute the mere passing of time for the element of premeditation.” Id. at 478, ¶ 27, 65 P.3d at 427.
¶ 20 The court found support for its interpretation of premeditation in the statutory admonishment “an act is not done with premeditation if it is the instant effect of a sudden quarrel or heat of passion.” Id. at 478, ¶ 28, 65 P.3d at 427 (quoting A.R.S. § 13-1101(1) (Supp.1998)). The court explained this language “distinguishes impulsive killings from planned or deliberated killings and confirms the legislature’s intent that premeditation be more than just a snap decision made in the heat of passion.” Id. The Thompson court concluded by noting evidence of reflection may include, inter alia, threats made by the defendant to the victim, a pattern of escalating violence between the defendant and the victim, defendant’s acquisition of a weapon before the killing, and even the passage of time as long as the passage of time does not just serve as a “proxy” for premeditation. Id. at 478-80, ¶¶ 29, 31, 33, 65 P.3d at 427-29.
¶ 21 In contrast with our supreme court’s carefully crafted instruction concerning premeditation and description of reflection, the foreman’s Internet definition of first degree murder did not speak of reflection and did not acknowledge any distinction between a planned or deliberated killing and a killing caused by a “snap decision made in the heat of passion.” The printed Internet definition for first degree murder, from “legal-dictionary.thefreedictionary.com,” read:
[Although it varies from state to state, it is generally a killing which is deliberate and premeditated (planned, after lying in wait, by poison or as part of a scheme), in conjunction with felonies such as rape, burglary, arson, involving multiple deaths, the killing of certain types of people (such as a child, a police officer, a prison guard, a fellow prisoner), or with certain weapons, particularly a gun. The specific criteria for first degree murder are established by statute in each state.... It is distinguished from second degree murder in which premeditation is usually absent, and from manslaughter which lacks premeditation and suggests that at most there was intent to harm rather than to kill.
¶ 22 Not only did the Internet definition of first degree murder focus only on whether the killing was “deliberate” and “planned” with no mention of the word “reflection” or its derivatives and no reference to “heat of passion,” but it also emphasized the nature of the crime, how it was carried out, the victim, and whether certain weapons were used, “particularly a gun,” factors that, under Arizona law, are not necessarily indicative of a killing that is premeditated. For example, under the Internet definition, the mere.use of a gun-in contrast to the prior acquisition of a gun — would constitute evidence of premeditation. See id. at 479, ¶ 31, 65 P.3d at 428.
¶23 Underscoring the differences in the superior court’s first degree murder instruction with the foreman’s Internet first degree murder definition were three definitions of second degree murder obtained by the foreman that, in attempting to distinguish that crime from first degree murder, described first degree murder in terms either contrary to Arizona law or in a manner that muddied the meaning of premeditation under Arizona law. The foreman’s first definition of second *305degree murder differentiated that crime from first degree murder based solely on whether the killing was “premeditated, intentional” or “vicious.”5 The second definition was worded so poorly it could only have confused the jury over the difference between a premeditated killing and a killing caused by “a snap decision made in the heat of passion.”6 And, the third definition described the crimes as being the same except for the penalties, a factor the superior court instructed the jury it was not to consider.7
¶ 24 Also conflicting with the court’s instruction on premeditation was the Internet research by juror nine. Although he could not recall the exact content of his research on premeditation, he acknowledged “[o]ne was a definition that wasn’t the same as what the Court instructions were or the explanation that the Court gave. The other one was.”
¶25 To summarize: the Internet definitions obtained by the foreman and juror nine were significantly different from the jury instructions given by the superior court on first degree murder and its elements. Although the record fails to show how different juror nine’s Internet definition of premeditation was from that used by the superior court, the record reflects it “wasn’t the same.” And, the record also reflects, indeed, demonstrates, the foreman’s Internet definitions failed to track our supreme court’s first degree murder instruction and muddied the differences between a premeditated murder and a murder committed through a “snap decision made in the heat of passion” or as the “instant effect of a sudden quarrel.” Given the conflicting inferences from the evidence regarding the circumstances of J.’s shooting and whether the shooting was impulsive or premeditated, the significant differences between the Internet definitions and the court’s were of critical importance in this case.
III. Extent to Which Jury Discussed and Emphasized the Internet Definitions
¶ 26 On the final day of deliberations, the jury foreman shared his research, and seven of the other eleven jurors remembered discussing it. One of those seven jurors also remembered juror nine sharing his research. Five jurors remembered discussing second degree murder, even though it was not defined in the jury instructions. One juror testified:
Definitions were read outside from the Court document. People had questions, which I would assume is part of deliberating, as to what the words meant. There was [sic] personal clarifications. There was [sic] people that thought they had the structure definition from the dictionary. There were people that felt they had a law background. There were several discussions within the room.
¶ 27 Not only did jurors discuss the Internet definitions during their deliberations, but *306certain of the jurors used the Internet definitions in considering Appellants’ guilt. Although, as the State points out, the jurors all testified they “relied” on the instructions given to them by the superior court, for several jurors their reliance came only after they had considered the Internet definitions. We therefore disagree with the State’s argument the Internet definitions did not affect some jurors’ interpretations of the court’s instructions, and thus could not have contributed to the verdicts.
¶ 28 Specifically, the foreman “considered” his own research; juror nine testified he was confused about the definition of premeditation after the first day of deliberations, but after he looked up the definition of premeditation on the Internet it “solidified my thinking”; juror eleven testified “I felt that [the foreman’s Internet definitions] helped me understand”; juror two testified the jury “considered” the foreman’s Internet definitions and the information about second degree murder was “important”; and yet another juror, number seven, acknowledged the importance of the foreman’s Internet definitions to her deliberation:
Q: ... You would agree that the information that [the foreman] provided from the Internet was important in your deliberation process? You agree with that?
¶29 The superior court found each juror “did not rely on the Internet pages to reach [his or her] verdicts.” As the preceding excerpts from the record show, however, the record contains clear evidence certain jurors relied on the Internet definitions to develop and shape their interpretations of critical legal terms. The Internet definitions diverged significantly from the jury instructions; juror nine performed research of unknown substance that, he acknowledged, “solidified my thinking”; and the jury discussed, and certain jurors used and considered, the Internet definitions during deliberations. Applying our standard of review, see supra ¶ 6, and the principles espoused in Hall, Cornell, and Mayhue, the record demonstrates the superior court abused its discretion in determining the State had overcome the presumption of prejudice that arose when the jury considered the Internet definitions. See Hall, 204 Ariz. at 449, ¶ 25, 65 P.3d at 97. Therefore, the superior court should have granted Appellants’ motions for a new trial on the attempted first degree murder charge.
¶30 For the foregoing reasons, although we affirm Appellants’ kidnapping convictions and Norzagaray’s forgery conviction, we reverse Appellants’ convictions on attempted first degree murder and remand for a new trial on that charge.
CONCURRING: DANIEL A BARKER, and PETER B. SWANN, Judges.