Defendants are charged with two counts of criminal mistreatment and one count of hindering prosecution. The indictment charges, in the criminal mistreatment counts, that defendants left two of their daughters, who were aged 13 and 15,1 in their residence under circumstances that were likely to endanger the childrens’ health and welfare. In the hindering prosecution count, the indictment charges that defendants harbored and concealed Jose Juan Gonzales Cruz (Gonzales Cruz), who was allegedly guilty of sex crimes involving the 13-year-old daughter. Before the charges were filed, the police obtained a search warrant based on the affidavit that is the subject of this appeal. Defendants moved to suppress the evidence discovered as the result of the execution of the warrant, and the trial court granted the motion. The state appeals and argues that the trial court erred in holding that the officer’s affidavit in support of the application for the warrant was insufficient to establish probable cause. We reverse.

We begin by considering several legal issues, because the resolution of those issues affects how we review the information in the affidavit. The first issue concerns the effect of the trial court’s rulings on defendants’ motion to controvert the allegations in the affidavit. ORS 133.545(4) requires that an affidavit in support of an application for a search warrant “particularly set[ ] forth the facts and circumstances tending to show that the objects of the search are in the places, or in the possession of the individuals, to be searched.” A party who files a motion to controvert the assertions in an affidavit for a search warrant has the burden of proving by a preponderance of the evidence “that the evidence presented before the issuing authority was not offered in good faith, was not accurate and was not truthful.” ORS 133.693(3). If the court finds inaccuracy, untruthfulness, or lack of good faith, it must then determine whether, under the *599applicable law, what it found requires suppression. ORS 133.693(5).

In this case, the trial court held an extensive hearing on the motion to controvert, after which it found two different kinds of problems with the affidavit. First, it found that the officer’s statements were incomplete or inaccurate in several respects. As an example, the affidavit stated that defendants were uncooperative during an attempt to check on the welfare of their 13-year-old daughter, but it did not mention that the attempted welfare check occurred at 2:30 a.m. Those inaccuracies do not affect the essential statements in the affidavit and have only a minor effect on our review of its sufficiency.2

When the court finds inaccuracies in an affidavit, it does not “correct” the affidavit by adding evidence to it. Rather, the judge’s assessment of an affidavit after allowing a motion to controvert is one of subtraction, not addition. See State v. Harp, 299 Or 1, 9-10, 697 P2d 548 (1985). However, one way of subtracting from an affidavit is to use the additional information developed on the motion to controvert in order to weaken the force of the officer’s statements in the affidavit. That is what the court did in this case, and it is consistent with Harp. See also State v. Sloan, 66 Or App 269, 273, 673 P2d 567 (1983), rev den 296 Or 536 (1984) (when additional facts would affect the weight of facts that are otherwise sufficient to show probable cause, reviewing court determines whether magistrate, as reasonable person, would have found probable cause if omitted facts had been known and correct inferences drawn). We review the sufficiency of the affidavit in light of the trial court’s findings of inaccurate information.

*600Secondly, the trial court excised a number of paragraphs and partial paragraphs from the affidavit, not because defendants had controverted them, but because the court concluded that the information that they contained was either irrelevant or legally insufficient to support the issuance of the warrant. Although parties and courts at times refer to excising material from search warrants on those grounds, see. e.g., State v. Poppe, 131 Or App 14, 19-20, 883 P2d 905, rev den 320 Or 492 (1994), in fact that action exceeds the trial court’s authority under ORS 133.693. The statute does not provide for removing information from an affidavit because it is irrelevant or legally insufficient; that remedy is available only if the information is inaccurate, untruthful, or not offered in good faith. Rather than excising irrelevant or legally insufficient information, the trial court should simply have ignored it, because it does not contribute to a showing of probable cause. The information, however, remains part of the affidavit and on appeal we may conclude, contrary to the trial court, that the information is relevant or legally sufficient. If it is, we are free to consider the information in reviewing the trial court’s decision.

Our purpose in reviewing the sufficiency of an affidavit supporting an application for a search warrant is to determine whether, on the basis of the information in the affidavit, a neutral and detached magistrate could have concluded that there was probable cause to believe that a search would discover the specified things in the specified places. State v. Young, 108 Or App 196, 200, 816 P2d 612 (1991), rev den 314 Or 392 (1992). In reviewing the issuance of a search warrant, we defer to the issuing magistrate’s determination of probable cause and resolve doubtful cases by the preference for searches that are conducted under the authority of warrants rather than without prior judicial authorization. See State v. Prince, 93 Or App 106, 112, 760 P2d 1356, rev den 307 Or 246 (1988).

Because the information in the affidavit upon which we rely is provided by named informants and the police, the test to be applied is a “totality of the circumstances” test. Young, 108 Or App at 202-03.

*601“Our function, when faced with [the question of the sufficiency of such an affidavit] is to determine whether a neutral and detached magistrate could conclude, based on the facts and circumstances shown by the affidavit, that there was probable cause to believe that the search would discover things specified in the affidavit in the places requested to be searched. See ORS 133.555(1), (2). We are to construe the supporting affidavit in a common-sense and realistic fashion.” State v. Villagran, 294 Or 404, 408, 657 P2d 1223 (1983) (citations omitted).

“Probable cause” arises from information provided by named informants when a reasonable person, based on the totality of the circumstances, would believe that it is probable (more likely than not) that the objects of the search will be found at the location to be searched. State v. Anspach, 298 Or 375, 380, 692 P2d 602 (1984). In reviewing the sufficiency of an affidavit, we rely on the facts as stated together with any reasonable inferences that can be drawn from them. State v. Poulson, 150 Or App 164, 170, 945 P2d 1084 (1997).

The dissent is critical of our analysis based on the above principles. It asserts that we have failed to evaluate the evidence from named informants “with the care that our case law requires,” 172 Or App at 613, and it apparently believes that the Aguilar/Spinelli rule, embodied in ORS 133.545(4),3 is “the primary framework for evaluating the totality of the circumstances concerning the reliability of information from named informants.” 172 Or App at 614 (emphasis added). It relies on State v. Milk/Sales, 127 Or App 397, 400-01, 872 P2d 988 (1994), and State v. Wilson, 120 Or App 382, 387, 852 P2d 910, rev den 317 Or 584 (1993), for that proposition. If the notion advanced by the dissent is that this court has modified the test for named informants set out in State v. Farrar, 309 Or 132, 144, 786 P2d 161, cert den 498 US 879 (1990), to be synonymous with the statutory test and the Aguilar / Spinelli rule for unnamed informants based on what we said in Milk/Sales and Wilson, its argument proves *602too much. In Farrar, the Supreme Court said, “[W]here the information in the affidavit is provided by a named informant, as in the instant case, the Aguilar / Spinelli standard is not required by statute or by the Oregon Constitution.” Farrar, 309 Or at 144. Citing Farrar, we implicitly adopted that understanding in Milk /Sales when we said that “we look to the totality of the circumstances in the affidavit to determine if the information provided by the [named] informant is sufficiently reliable[.]” Milk/Sales, 127 Or App at 400, quoting Young, 108 Or App at 202. In Wilson, we reiterated the rule expressed in ORS 133.545(4) and then made the statement quoted by the dissent. See 172 Or App at 614. In that context, we said, citing Farrar, that information from a named informant is “subject to essentially the same analysis to determine whether a reasonable magistrate could conclude, under the totality of the circumstances, that there was probable cause to support the issuance of the warrant.” Wilson, 120 Or App at 387. That statement is merely another way of saying that in the overall analysis, the basis of an informant’s knowledge is one germane factor in determining probable cause. It says nothing about what weight must be given to that factor. In this case, we adhere to the rule in Farrar that the analytical framework provided in ORS 133.545(4) is but one consideration in determining the reliability of the information from named informants and that the proper legal standard is the “totality of the circumstances” test. We turn to the affidavit with that understanding.

As the state recognizes, the affidavit is poorly organized and confusing, and it contains a great deal of material that is at best tangentially relevant. We will state the facts described in the affidavit as the officer who sought the warrant stated them, correcting for the omissions and the inaccuracies that the trial court found after hearing the motion to controvert, and we will ignore material irrelevant to our decision. For instance, some statements relate to the officer’s training and experience.4 Most of the other material either *603summarizes the results of other police investigations, using such introductory terms as “it was learned,” or repeats statements from various named and unnamed informants, most of whom were allegedly involved in other sex and drug offenses, either as perpetrators or victims. In some instances, there is nothing in the affidavit to support those conclusory statements. Consequently, we ignore them for purposes of determining whether probable cause to search exists within the four corners of the affidavit.

The search warrant in this case authorizes the search of the premises located at “171 Woodbine, Blue Ridge, located in Astoria, Oregon” for evidence of controlled substances and sexual activity between adults and juveniles. It also authorizes the search of Ricardo Jesus Aguilar Escorcia (Escorcia) and Gonzales Cruz, as “suspects,” and defendants’ daughters as “victims,” for medical evidence of sex crimes. The supporting facts in the affidavit pertinent to our inquiry begin with the recitation in paragraph 13 that Tomas Rodriguez Santos was arrested on December 5,1997, after he was found in bed with a 13-year-old girl, C.G. According to paragraph 24, law enforcement officers “were directed to 171 Woodbine via specific information provided by [C.G.].” She told the officers that defendants’ 14-year-old daughter had recently given birth to a son fathered by Escorcia and that their 13-year-old daughter was “reported to be having an ongoing live-in sexual relationship with an adult male, age 24.”

On December 6, officers made contact with defendants at the residence at 171 Woodbine but were refused entrance. However, Daniel Boyer told the officers that Gonzales Cruz “does stay at [his] apartment at times.” By checking other public record sources, the officers ascertained that both Escorcia and Gonzales Cruz listed the Woodbine address as their residence. In addition, the affidavit recites, “this officer and Detective Reed saw Mr. Gonzales Cruz recently enter the residence at 171 Woodbine in the late afternoon. He did not use a key nor knock, and was unaccompanied by either the Boyer children or Mr. and Mrs. Boyer, the formal renters of the apartment.”

*604On December 30, 1997, the police arrested Martin Valdivia Villarreal, who was charged with the rape of a 14-year-old girl. At the time of the arrest, Villarreal was accompanied by a 17-year-old female, Brandy Huff-Olvera. On January 2,1998, Huff-Olvera was arrested for conspiracy to compel prostitution. In an interview with Detective Reed, Huff-Olvera said that she knew Gonzales Cruz and that he lived at 171 Woodbine “with [defendant’s 13-year-old daughter] and her family.” She told Reed that Gonzales Cruz was the 13-year-old daughter’s “boyfriend” and that he paid rent to the Boyer family. She also asserted that Gonzales Cruz was selling narcotics at the Boyer residence and that she recently purchased drugs from him there.5 She said, “Gonzales Cruz sells the most and best cocaine in the Blue Ridge neighborhood.”

Based on the foregoing information, probable cause exists to believe that Gonzales Cruz was living at defendants’ residence at the time of the issuance of the warrant. Daniel Boyer’s and Huff-Olvera’s statements, the officers’ observations and the public records search confirm that fact. The remaining issue is whether the affidavit establishes, based on the information provided by the police and the named informants, that there is probable cause to believe that evidence could be found that defendants’ 13-year-old daughter was engaging in a sexual relationship with Gonzales Cruz, and that Gonzales Cruz was selling drugs from defendants’ residence.

The “veracity” of an informant refers to the extent to which the informant can be considered trustworthy. State v. Alvarez, 308 Or 143, 147, 776 P2d 1283 (1989). Veracity can be established in several ways including corroboration by the police, corroboration by another informant, the informant exposing himself to liability for filing a false report or by the *605informant’s statements against penal interest. State v. Wheelon, 137 Or App 63, 72, 903 P2d 399 (1995), rev den 327 Or 123 (1998). We turn back to the contents of the affidavit to first examine it on the issue of the veracity of the informants.

The affidavit states, in relevant part

“It was later learned through interviews with [the mother of C.G.], [M.R.], and [A.M.] that [Gonzales Cruz] was indeed present in the residence (171 Woodbine, Blue Ridge) when officers visited.

* :ji * *

“According to interviews with those girls who are friends of [the 13-year-old daughter], and visit the residence, Mr. Gonzales Cruz does housework for the Boyers, watches the girls when the parents are away, runs errands using his and the parents’ vehicles and sleeps in both the living room or [the 13-year-old’s] room.

“[Gonzales Cruz] is identified by [C.G.], [A.M.], [J.M.], [M.R.], [J.McP.], and [Huff-Olvera] as the adult male presently having a sexual relationship with [defendant’s 13-year-old daughter].”

The above informants are named informants, and their information is corroborated in part by the police investigation as well by the fact that each of the informants gave identical information to the police. For the reasons previously mentioned, those factors weigh in favor of probable cause. As to the veracity of Huff-Olvera, her statements regarding her drug purchase are statements against her penal interest, a factor that weighs in favor of probable cause. State v. Carlile, 290 Or 161, 167, 619 P2d 1280 (1980). Additionally, her credibility and the credibility of the other informants is strengthened by the fact that Daniel Boyer admitted to the police that Gonzales Cruz resided at his apartment.

This case is similar in one respect to State v. Payne, 150 Or App 469, 946 P2d 353 (1997), rev den 326 Or 390 (1998), in which the warrant was issued on the basis of information provided by a named informant who revealed to the police that he had stolen marijuana from the defendant’s shed. We noted that police corroboration of details from an *606informant’s statement such as physical characteristics of the property, the vehicles present and the description of the residence’s owner were additional indicia of the reliability of the information. We concluded that “[t]he statement against penal interest made by a named informant, under these facts, is alone sufficient to establish the reliability of [the informant’s] statements.” Payne, 150 Or App at 474 (emphasis added). Here, Huff-Olvera’s statements also were against penal interest and were corroborated by the police and the other named informants.

We next examine the basis of the knowledge of each informant. The information that Huff-Olvera gave to the police about Gonzales Cruz’s sale of drugs at 171 Woodbine is based on her presence and participation in the transaction at the residence. It is thus inferable that her information regarding Gonzales Cruz’s relationship with defendants’ daughter was acquired in the course of her contacts with Gonzales Cruz or a member of defendants’ household. As to the means by which the other named informants’ information was acquired, the affidavit contains no express statement by any of the minor girls as to how they learned of Gonzales Cruz’s sexual relationship with defendants’ daughter. However, there is no statutory or constitutional requirement that the basis of a named informant’s knowledge be set forth expressly in the affidavit.

A comparison of the facts in this case with the facts in State v. Villagran, 294 Or 404, 657 P2d 1223 (1983), is instructive in that regard. In Villagran, the issue was whether information from a disinterested named citizen informant, who was unconnected with the crime being investigated, should be deemed credible when police questioned him about a suspect. At issue was the validity of a search warrant authorizing the search of premises near Umpqua, Oregon. After discovering marijuana growing in a barn located in Lookingglass, Oregon, the police seized items from that location, including an address in Roseburg. By checking public records, they learned that Villagran and Waterbury resided at the Roseburg address. The police also learned from an individual who lived near the Lookingglass property that Madsen had built the barn from which the marijuana was *607seized. A member of the sheriffs office talked to Madsen. Madsen confirmed that he had built the Lookingglass barn for Waterbury and that

“ ‘the guy he built the bam for was building an ‘underground’ house in Umpqua which is described as being on the driveway adjacent to the driveway for the Cal Henry residence which is located at 1054 Cal Henry Road. Going up the driveway to the ‘underground’ house, to the left of the driveway is a trailer house and a barn behind said trailer. The “underground” residence is located further down the driveway.’ ” Villagran, 294 Or at 407 n 3 (quoting affidavit for search warrant in that case).

Based on that information, the trial court issued a search warrant for the Umpqua property, and the police seized evidence from it, that the defendant sought to suppress. She argued in support of her motion that the affidavit failed to show the basis of Madsen’s knowledge and that the warrant therefore had issued without probable cause.

The Supreme Court pointed out that Madsen “appears to be a wholly disinterested citizen providing information to the police at their request.” Villagran, 294 Or at 411. Similarly, here the information provided by C.G., A.M., J.M., M.R. and J.McP. comes from citizen informants who had no complicity in the criminal conduct being investigated at defendants’ residence. The Villagran court then turned to the basis of Madsen’s knowledge. It stated:

“Under the circumstances disclosed by the affidavit, it is reasonable to conclude that his knowledge was obtained in the course of working for or with Waterbury on the barn, and that Madsen was credible and that his information was accurate. Moreover, it was obtained at police request rather than volunteered. Unlike the Aguilar / Spinelli anonymous informant cases or those such as [State v. Montique, 288 Or 359, 605 P2d 656, cert den 449 US 846 (1980),] where the informer is fingering someone and there is no independent ground for believing the story, here there is no reason for viewing the informer’s information with suspicion or caution. Madsen’s information, in a sense verifies itself. Reasonable persons routinely rely upon factual information obtained in a manner such as this. One could naturally and not unreasonably conclude from the information that *608Waterbury had arranged for the barn to be built by Madsen and that Waterbury was building a house on Cal Henry Road and might be living there. Accordingly, we hold that the affidavit was sufficient to establish the truth of Madsen’s statements.” Villagran, 294 Or at 412.

The same quality of self-verification that existed in Villagran exists in this case. The information is from named informants, and its particular nature gives rise to a reasonable inference that it was learned by the informants from first-hand contacts with defendants’ residence or members of their household. The affidavit states that the police conducted interviews with C.G., A.M., J.M. and M.R. It also states that, “according to interviews with those girls who are friends of [the daughter] and who visit the residence, [Gonzales Cruz] does housework for the Boyers, watches the girls when the parents are away, runs errands using his and the parents’ vehicles, and sleep[s] in both the living room or [the daughter’s] room.” (Emphasis added.) It also recites that Gonzales Cruz is “identified” by C.G., A.M., J.M., and M.R. “as the adult male having a sexual relationship” with defendants’ daughter, and says that A.M. and M.R. knew about the contact between the police and defendants on December 6 and that they knew that Gonzales Cruz was present in the residence at the time that Daniel Boyer prevented the authorities from entering.

Although the affidavit is ambiguous about who the “friends” are who “visit” the residence, it is subject to the same kind of reasonable inferences as the affidavit was in Villagran. Here, it is reasonably inferable from the affidavit as a whole that the visitors to defendant’s household are the same girls who are listed by name in the preceding paragraphs of the affidavit. In those paragraphs, the named informants give information about activities inside defendants’ household about which outsiders would likely not know, but of which visitors to the household or friends of its occupants would likely be aware. That particularized information given by the named informants followed by the paragraph regarding the visitors’ observations gives rise to a reasonable inference that the named informants had been inside the home, and that they are the girls referred to as the “friends” and the “visitors.”

*609 The dissent disagrees that such an inference can reasonably be drawn. If there is a common thread to its analysis, it is the premise that the only reasonable inference to be drawn from the affidavit is that the named informants acquired their information through community rumor, rather than first-hand knowledge. However, “the fact that information in an affidavit may give rise to other inferences does not mean that the affidavit is insufficient.” Wheelon, 137 Or App at 72. “The information in an affidavit need not point to only one inference to the exclusion of others.” State v. Donahue, 93 Or App 341, 347, 762 P2d 1022 (1988), rev den 307 Or 303 (1989). Again, probable cause to search means that the facts upon which the warrant is premised would lead a reasonable person to believe that seizable things will more likely than not be found in the location to be searched. Anspach, 298 Or at 381. Doubtful or marginal cases are to be determined by the preference accorded to search warrants. Donahue, 93 Or App at 346. The trial court and the dissent err because they fail to apply those principles. Specifically, they refuse to draw an inference that the visitors to defendants’ residence are some of the named informants when such an inference' is available and reasonable from the affidavit.

The reasonableness of the inference that the girls’ knowledge was acquired first-hand arises from a number of facts in the affidavit. First, there is the cross-corroboration among the informants. Second, there is no hint or suggestion in the affidavit that the named informants’ reports of sexual activity in the residence are derived from community rumor. For example, the affidavit does not say that the girls heard at school from their friends that such a relationship was occurring. Instead, the affidavit uses phrases that indicate that the information came from first hand information: phrases such as Gonzales Cruz was “identified by [the named informants],” “[name of informant] said,” or “[name of informant] confirmed knowing.” In addition, the detailed information provided by the girls, such as what chores Gonzales Cruz did inside defendants’ home, would not ordinarily be known by the general community.

Finally, as discussed above, some of the most persuasive evidence leading to a reasonable inference that some *610of the named informants had first-hand knowledge of the private details of defendants’ residence is the fact that A.M. and M.R. reported to officers that Gonzales Cruz “was indeed present in the residence * * * when officers visited.” The fact that the girls possessed that information shows that they knew of the police visit, that they knew of the time of the visit, and that they knew who was in the house at that time. Although it could be inferred from all of the above facts that the girls were reporting only rumors, the dissent does not explain why it disregards the equally permissible inference that the girls knew, in the light of the information that their reports describe, about Gonzales Cruz’s sexual activities because of their contacts with defendants’ household.

The dissent advances one more reason why the state should not be given the benefit of a reasonable inference from the affidavit that the named informants are the same girls who visited defendants’ residence. The dissent says that the state has “rejected” such an inference. It asserts that the state has “expressly treated the two groups as separate,” 172 Or App at 619, and focuses on phrases in the state’s brief such as “unnamed friends” and unnamed neighbors.” It concludes that “[T]he only way to read the state’s argument is that it asserts that the named girls are not the same as the unnamed friends.” 172 Or App at 620. The state will be surprised to learn that the dissent has understood its brief in that way. The two references in the state’s brief on which the dissent relies are used to describe information that is also furnished by named informants; the same named informants (A.M. and M.R.) who knew that Gonzales Cruz was hiding inside defendant’s residence when the police contacted them. There is no express repudiation or concession in the state’s brief that the girls who visited defendants’ residence are not the same girls named in the affidavit. Unlike the dissent, we are not willing to read into the state’s position a concession that it does not expressly make or to conclude that the state must have implicitly conceded that fact because of the positions that it does take.

Actually, the state raised to the trial court the precise issue regarding named informants upon which our analysis is based. Below, the state expressly argued to the trial court that the information that it should consider came *611from named informants, including the description of the activities in defendants’ household. The prosecutor told the trial court, in part:

“I think its important for the court to — there’s been a lot of talk about informants and unnamed informants. There were no [unnamed] informants in this case, that’s something that the Court needs to consider. There were named people, named girls, that gave information either directly to Officer Walker or to Detective Reed who then supplied that information to Officer Walker [the affiant].” (Emphasis added.)

On appeal, the state reiterates its arguments made below:

“In this case, the trial court used a two-step process to determine that Officer Walker’s affidavit was insufficient. First, the trial court determined that the affidavit relied largely on the statements of named informants, but failed to express explicitly the informants’ basis of knowledge or provide indicia of their reliability. That failure to satisfy the Aguilar-Spinelli test, the court believed, was an aspect of the totality of the circumstances that defeated probable cause for the search. The trial court then ‘excised’ from the affidavit any information that it viewed as irrelevant in establishing probable cause. After reevaluating the affidavit in light of the information added pursuant to defendants’ motions to controvert, it determined that what remained in the affidavit was insufficient to justify the issue of a warrant.

“The state respectfully submits that the court’s analysis was flawed in both respects. Whether an affiant explicitly sets forth an informant’s basis of knowledge and otherwise shows the veracity of the informant certainly [is] part of the totality of circumstances affecting a determination of whether the affidavit provides probable cause. That does not mean, however, that unless an affidavit relying on informants’ statements expressly meets the AguilarSpinelli test, it is legally insufficient.”

The statement from the state’s brief shows that it continues to rely on its assertion below that the girls named in the affidavit are the informants whose basis of knowledge is in dispute. In addition, the state’s brief on appeal takes care to distinguish information gained from “unnamed neighbors” from information supplied by M.R. and C.G., all of *612“who[m] substantiated Huff-Olvera’s claim that Gonzales Cruz and [the 13-year-old daughter] were sleeping together.” Although the dissent infers that the “friends who visited the household” could only be the “unnamed neighbors” mentioned in the affidavit, the context of the state’s brief and the contents of the affidavit suggest otherwise. The unnamed neighbors are referred to in the brief in addition to the named informants. In the affidavit, references to unnamed neighbors appear in paragraph 10, in the context of a general statement about contacts with neighborhood “residents,” and in paragraph 20 in reference to “friends in the neighborhood” and Huff-Olvera’s use and sale of illegal drugs. Beginning with paragraph 21 and through paragraph 27 of the affidavit, every fact in the affidavit is attributed to a named informant or to information learned by the police directly. In those paragraphs, the affidavit consistently refers to the informants as “girls.” Paragraph 26 says, in relevant part, “According to interviews with those girls who are friends of [defendants’ daughter], and visit the residence[.]” (Emphasis added.) It is apparent that the state’s brief should be reasonably understood to distinguish between the named informant “girls” and the “unnamed neighbors.” The dissent’s argument in this regard is again illustrative of its approach to draw only one of the available inferences from the facts before it.6

In summary, we disagree with the trial court and the dissent that, after irrelevant and unsupported material is disregarded, the remaining information in the affidavit does not establish probable cause that evidence of sex crimes and controlled substances would be found at defendants’ residence. It is important not to permit the analysis to isolate any one informant in the affidavit. The test is one of the totality of the circumstances, and no single factor as to a particular *613informant or group of informants is dispostive. The information can be divided into three separate categories of sources: the information from Huff-Olvera, the information from C.G., A.M., J.M., M.R., and J.McP. and the information from the police. The information from each source corroborates the information from the other sources, thereby bolstering the strength of each. In this inquiry of whether it was more likely than not that evidence of criminal activity pertaining to illegal drug sales and sexual activities with minors would be found at defendants’ residence, we hold that the cumulative effect of all of the above information establishes that there was probable cause that Gonzales Cruz was residing at that address, that he had sold drugs from there and that he was engaged in a sexual relationship with defendants’ 13-year-old daughter. For those reasons, the trial court erred in allowing defendants’ motion to suppress.

Reversed and remanded.

dissenting.

The majority takes a disorganized and conclusory affidavit and discovers in it facts and inferences that, the majority believes, are sufficient to show that there was probable cause for the magistrate to issue the search warrant. The majority’s apparent success in panning a few nuggets of gold out of vast beds of gravel, however, depends on two fundamental errors: the majority fails to evaluate the evidence from named informants with the care that our case law requires, and it relies on an essential inference that the state, which is the appellant, expressly repudiated on appeal. When we instead analyze the affidavit according to our existing case law and consider only the arguments that the state did not foreclose on appeal, we must affirm the trial court’s decision to invalidate the warrant and suppress the evidence that the police seized under its authority. Because I would follow our usual practices in those respects, I dissent from the majority’s decision.

With a few exceptions, the information in the affidavit did not come from the officer’s observations, in which case there would be no hearsay issue, or from other police sources or public records, where the hearsay issues would not usually be significant. Rather, essential evidence comes from *614individuals whom the police interviewed at various times and under various circumstances and thus consists of hearsay. Oregon cases provide clear rules for evaluating such hearsay, rules that the majority does not fully follow. Those rules differ depending on whether the affidavit identifies the hearsay informant by name. So far as unnamed informants are concerned, the affidavit must comply with ORS 133.545(4), which requires the affiant to “set forth facts bearing on any unnamed informant’s reliability and [to] disclose, as far as possible, the means by which the information was obtained.” That statute codifies the rule of Aguilar v. Texas, 378 US 108, 84 S Ct 1509, 12 L Ed 2d 723 (1964), and Spinelli v. United States, 393 US 410, 89 S Ct 584, 21 L Ed 2d 637 (1969).1

ORS 133.545(4) and the Aguilar /Spinelli rule do not directly apply to named informants. Instead, as the majority correctly notes, we look to the totality of the circumstances to determine whether their information is sufficiently reliable. State v. Farrar, 309 Or 132, 144-45, 786 P2d 161, cert den 498 US 879 (1990). What the majority ignores is that Oregon cases treat the Aguilar / Spinelli rule as the primary framework for evaluating the totality of the circumstances concerning the reliability of information from named informants. See State v. Milks/Sales, 127 Or App 397, 400-01, 872 P2d 988 (1994) (“The Aguilar / Spinelli test, although not directly applicable to named informants, provides a framework for evaluating the ‘totality of the circumstances’ confronting the magistrate.” (emphasis in original)); State v. Wilson, 120 Or App 382, 387, 852 P2d 910, rev den 317 Or 584 (1993) (information from named informant “is subject to essentially the same analysis [as information from an unnamed informant] to determine whether a reasonable magistrate could conclude, under the totality of the circumstances, that there was probable cause to support the issuance of the warrant”). In Farrar itself the Supreme Court analyzed evidence from named informants according to the Aguilar / Spinelli criteria. *615Farrar, 309 Or at 145-46. That does not mean, as the majority seems to think, that I am saying that the test for named and unnamed informants is the same. Rather, the cases require that, in evaluating information from a named informant, we look for indicia both of the informant’s basis of knowledge and of the credibility and the reliability of his or her information. The fact that an affidavit names an informant makes it more likely that it will satisfy those concerns and thus affects how we evaluate the various factors.2 That an informant is named, however, does not by itself permit us to assume that the named informant had a basis for knowing what he or she told the police. There must still be something beyond the informant’s bald statement to allow the magistrate to conclude that the informant had a reason to know what he or she said.

I agree that the affidavit contains probable cause to believe that Gonzales Cruz was living at defendants’ residence, and it may also include probable cause to believe that he was selling drugs from that location. However, in order to justify the warrant that the magistrate issued, and the search that the police conducted, the affidavit must also contain probable cause to believe that a search would disclose evidence of an illegal sexual relationship between Gonzales Cruz and defendants’ 13-year-old daughter. If there is not probable cause to support the warrant in its entirety, the entire search is unauthorized. See State v. Beagles/White, 143 Or App 129, 135, 923 P2d 1244, rev den 324 Or 487 (1996) (warrant is overbroad if it permits seizure of things for which no probable cause exists); see also State v. Reid, 319 Or 65, 71, 872 P2d 416 (1994) (although affidavit contained probable cause to support a warrant to search the defendant, the search was improper because the warrant also permitted the search of persons for whom there was no probable cause).

The majority relies primarily on statements from a number of named informants that an illegal relationship existed. Those informants were either underage girls who *616were sexually involved with adult men other than Gonzales Cruz or, in one case, the mother of such a girl. Nothing in the affidavit provides any indication of how those informants, other than Brandy Huff-Olvera,3 knew what they said. The majority suggests several ways of avoiding that problem, but none of its suggestions works. First, the majority points out that all of the girls said the same thing. That, it argues, was the kind of corroboration that made their statements more reliable. That argument, however, ignores the fact that the affidavit provides no information about how the girls knew what they said. Without some indication of the basis of their knowledge, telling the same story simply means that they repeated the same rumor.

The majority next argues that this case is similar to State v. Villagran, 294 Or 404, 657 P2d 1223 (1983), in which the Supreme Court held that it was possible to infer the basis for a named informant’s information from the information itself. In Villagran the informant, a disinterested citizen who had no apparent involvement in any criminal activity, had built a barn for the defendant’s associate. He told the police that the person for whom he had built the barn was building an underground house in a different location. The Supreme Court concluded that it was reasonable to believe that the informant had learned the information about the house in the course of working on the barn for or with the defendant’s associate. That was the kind of information on which people routinely rely, and it was sufficient to establish the truth of the informant’s statements.

Although the Supreme Court did not expressly say so, Villagran is in large part an application of Spinelli, in which the United States Supreme Court held that information can be sufficiently detailed to permit the magistrate to infer that the informant had a basis for knowing what he or she said. The Court referred to an earlier case in which the informant stated that a drug courier had left Denver by train on a certain date, would return by train on one of two specified days with three ounces of heroin, and would be wearing

*617clothes that the informant described with minute particularity. Spinelli, 393 US at 416-17. When the police in that earlier case went to the train station on the appropriate morning, they saw the defendant dressed precisely as the informant had described. That confirmation of the informant’s information was the final step that established probable cause for the arrest. Id. at 417-18. Villagran did not require that level of detail because the informant was a named disinterested citizen, but it nevertheless required that the information be sufficiently detailed to permit an inference that the informant had a basis of knowledge.

In this case, in contrast, there is nothing to indicate any connection between the named girls, on the one hand, and defendants or their daughters on the other, beside the fact that they live in the same section of town.4 The girls do not claim to be friends of the family or the daughter, nor do they claim ever to have been inside defendants’ residence. The details that they give of the alleged relationship between Gonzales Cruz and the 13-year-old daughter do not come close to the details in Villagran. Thus, there is no basis to infer that they gained their knowledge of that relationship in the course of other contacts with either of them.5 The only relevance of Villagran to this case is to emphasize again that an affidavit must provide some basis for a named informant’s knowledge.6

The state’s argument on this point is different from the majority’s. According to the state:

*618“[The informants] must have known that Gonzales-Cruz and [the 13-year-old daughter] were sleeping together, based on their observation of behavior permitting that inference, on an admission by one of the two, or on community knowledge. That is apparent from the affidavit, which shows that [C.G.] and [M.R.] were both victims of sexual abuse; that [C.G.] had a sexual relationship with Santos, and that Santos was friends with either defendants or Gonzales-Cruz; and by [C.G.’s] apparent volunteering of the information to police on the night that Santos was arrested for abusing her.”

(Emphasis added.) As the use of “must have known” indicates, that argument is pure speculation. Nothing in the affidavit shows that the named informants had observed behavior permitting the inference that the state describes or that either the 13-year-old daughter or Gonzales Cruz had admitted anything. The state’s speculation about their possible bases of knowledge is not a substitute for evidence on that point. Finally, it is difficult to determine what the state means by “community knowledge” other than rumor or hearsay; in any case, nothing in the affidavit refers to such knowledge as a basis for the informants’ statements.

Apparently recognizing that the affidavit provides no basis for the informants’ knowledge, the state also suggests that we should ignore their basis of knowledge in determining the reliability of their statements. That, however, is inconsistent with our previous cases, under which the Aguilar / Spinelli criteria remain important considerations in evaluating statements from named informants. Under that analysis, an informant’s basis of knowledge is one requirement for showing probable cause, while the other requirement is to show either that the informant is credible or that his or her information is reliable. The affidavit must satisfy both requirements for unnamed informants. See 2 Walter R. LaFave, Search and Seizure § 3.3(a), at 90-94 (1996). For named informants, the ability to satisfy both requirements, while not as absolutely controlling as with unnamed informants, remains an important aspect of determining whether the totality of the circumstances supports a finding of probable cause. Indeed, in Farrar the court expressly pointed out *619that the affidavit described the named informants’ basis of knowledge. Farrar, 309 Or at 145.

Ignoring a named informant’s basis of knowledge is also inconsistent with the requirement of probable cause, because without some basis for an informant’s knowledge of a hearsay statement there would be no reason for the issuing magistrate to believe that the statement is sufficiently reliable to justify the warrant. The total lack of information in the affidavit concerning the factual foundation for the informants’ statements so undercuts the reliability of those statements that, under the totality of these circumstances, the statements are of no value in finding probable cause.

If there is probable cause for this affidavit, it must come from the two other sources on which the majority relies: first, its inference that the named girls were the same people as the friends who, according to the affidavit, had visited the 13-year-old daughter at her home, and, second, the testimony of Huff-Olvera. I will deal first with the inference that the named girls were also the unnamed friends.7 The majority states that, “[although the affidavit is ambiguous about who the ‘friends’ are who ‘visit’ the residence,” it is reasonable to infer that they are the named underage girls. Based on that essential inference, the majority attributes all of the information from the “friends” to those named girls, thereby providing a basis for them to know about a relationship between Gonzales Cruz and the 13-year-old daughter.

The fundamental problem with the majority’s position is that the state expressly treated the two groups as separate. Thus, the majority is basing its reversal of the trial court’s order on an argument that the appellant expressly failed to make to this court. The relevant statements appear in two paragraphs in the affidavit. In the first paragraph, the officer described evidence that indicated that Gonzales Cruz lived at defendants’ residence. Part of that evidence was the statement that:

*620“According to interviews with those girls who are friends of [the 13-year-old daughter] and visit the residence, Mr. Gonzales Cruz does housework for the Boyers, watches the girls when the parents are away, runs errands using his and the parents’ vehicles, and sleeps in both the living room or [the daughter’s] room. He also pays the Boyers rent for his housing arrangement.”

The paragraph then ends with a reference to an automobile that Gonzales Cruz owned and that was routinely parked in the area used by those living at defendants’ residence. The next paragraph begins:

“Mr. Jose Juan Gonzalez Cruz is identified by [C.G.], [A.G.], [J.M.], [M.R.], [J.McP.], and Ms. Brandy Huff-Olvéra as the adult male presently having a sexual relationship with [the 13-year-old daughter], DOB 11-05-84.”

Especially in light of the generally poor organization of the affidavit, there is no necessary connection between these references in separate paragraphs to the named girls and to the unnamed friends. Thus the affidavit does not compel the inference that the two groups are the same. The state, however, goes further and reads the affidavit as excluding the inference. In its brief, the state says, in arguing that the affidavit provides probable cause that a search would produce evidence of sex crimes, that it “relies primarily on the statements of Huff-Olvera, [C.G.], [A.M.], [J.M.], and [M.H.], and several unnamed friends of [the 13-year-old daughter].” (Emphasis added.) Later in its brief, the state asserts that numerous individuals substantiated Huff-Olvera’s claim that Gonzales Cruz and the daughter were sleeping together. According to the state, those individuals “included [M.H.], and [C.G.], both victims of sexual abuse by adult males; several unnamed neighbors who told police that they were friends of [the daughter’s], that they had been inside defendants’ residence, and that they knew that [the daughter] and Gonzales Cruz shared a bedroom; and by [J.M.] and [A.M.]” The only way to read the state’s argument is that it asserts that the named girls are not the same as the unnamed friends.8

*621The state did not take this position by accident. As the majority shows, it may have suggested to the trial court that the unnamed friends and the named girls were the same. Thereafter, in its letter opinion the trial court stated, concerning the girls,

“there are no facts establishing the credibility of these informants or setting forth their bases of knowledge. They are named informants, but it does not appear that they initiated the reports and they were not making statements against their penal interests. It’s possible they are the same girls mentioned in paragraph 26 who are friends of [the 13-year-old daughter] and visit her residence, but that’s not clear.”

That statement, in conjunction with the state’s argument below, gave it a perfect opportunity to attack the trial court’s decision on appeal by arguing that the magistrate could infer that the named girls and the unnamed friends were the same people. Instead, the state rejected that position and treated the two groups as separate.9 I would not take a position that the appellant declined to take but, rather, would hold the state to its choice.

When I evaluate the evidence from the named girls and the unnamed friends separately, neither has significant value. The evidence of the named girls is inadequate under the totality of the circumstances test because nothing in the affidavit shows that they had any basis for knowing the information that they gave. This is not a case of a weak basis of knowledge that other factors can rectify; there is no basis of knowledge. The evidence of the unnamed friends fails the *622Aguilar/Spinelli test codified in ORS 135.545(4) because nothing shows that they were truthful or that their information was reliable.

That leaves Huff-Olvera’s evidence as the only possible foundation for the decision to issue the warrant. Her statements must provide the necessary evidence without significant assistance from any other source. Because she is a named informant I evaluate her statements under the totality of the circumstances test. Huff-Olvera is in a different category from the other minor girls. At the time of the affidavit she was 17 years old and married, while the other girls were ■ 13 or 14 and single. Her husband had recently been deported; since his deportation she had lived with at least two other men. The affidavit describes Huff-Olvera’s involvement in a way that makes her a perpetrator rather than a victim. According to three of the minor girls, in return for compensation Huff-Olvera provided underage girls to adult men for sexual purposes. As an apparent result of those or similar accusations, at the time of the affidavit she was under arrest on a charge of compelling prostitution. In addition, HuffOlvera stated that, two weeks before the date of the affidavit, she had purchased cocaine from Gonzales Cruz at defendants’ residence.

Huff-Olvera is the only named informant who claimed any direct contact with either Gonzales Cruz or defendants’ residence. Some of her information could contribute to a finding of probable cause. Her contact with Gonzales Cruz might give her a basis for knowing the relationship between him and the 13-year-old daughter. It clearly gave her a basis for knowing that Gonzales Cruz lived with defendants and that he sold drugs from their residence. Her statements also subjected her to possible criminal liability for making a false police report and for possession of cocaine; at least the statement concerning purchasing drugs was thus against her penal interest. Those factors tend to support the warrant.

Even recognizing those things, Huff-Olvera’s statements do not in themselves provide probable cause for the affidavit that the magistrate issued. First, the crucial statements are general rather than specific. She gives none of the *623details concerning Gonzales Cruz’s role in the household or his relationship to the 13-year-old daughter that the unnamed informants gave. She did not explain how she knew of the relationship. She did not say that one of the participants told her, that she learned it from others with knowledge, that she observed actions that led her to that conclusion, or that, as her arrest for promoting prostitution might suggest, she actually arranged it. Although, as Villagran indicates, we should not reject her evidence solely for this reason, the limited information on her basis of knowledge weakens the force of her conclusory statements. Second, aside from giving her name, the affidavit does not provide a basis for finding Huff-Olvera to be truthful. Rather, it raises questions about her veracity. Her admitted criminal involvement weighs against her credibility.10 See Milks/Sales, 127 Or App at 401. The corroboration of a relatively innocent fact — that Gonzales Cruz lived with defendants despite their denial — is not sufficient to make her information reliable without regard to her credibility.

I agree that Huff-Olvera’s statements could contribute several tesserae toward creating a mosaic of probable cause. When those statements are all that there is, however, the design is only begun. By themselves, they are not sufficient to support a search for evidence of an improper sexual relationship between Gonzales Cruz and the 13-year-old daughter, and the rest of the affidavit does not make up for that insufficiency. I would therefore affirm the trial court’s order and dissent from the majority’s contrary conclusion.