Mark Scott, who pleaded guilty to possessing child pornography, reserved the right to appeal from the district court's order denying his motion to suppress the evidence that police found at his home when they executed a search warrant. Officials in Wisconsin obtained the warrant, from a state judge, after arresting Scott for attempting to have sexual relations with a boy who Scott (age 58 at the time) believed was 14 years old. Actually the "boy" was an agent of the state's Department of Justice, who had impersonated a teenager in response to an ad that Scott placed on Craigslist. Scott and "Kyle" (the officer's alias) had an extensive electronic exchange. Scott sent Kyle both sexually explicit emails and sexually graphic photos; one was of a man, on a bedspread, with an object in his rectum. Scott asked Kyle to reciprocate with "a pic of you in your compression shorts, with a hardon if you can".
The affidavit in support of the request for a search warrant told the state judge about the ad on Craigslist, the sexual photos and messages sent by computer, Scott's request that Kyle send a picture of himself with an erection, Scott's offer to "host" the sexual encounter, and assurance by an investigating agent that in his "training and experience" pedophiles collect child pornography. The state judge issued a warrant authorizing a search of Scott's home, including his computers, for child pornography. Executing the warrant, officers found plenty. The federal district judge denied the motion to suppress, concluding that the warrant is supported by probable cause. He sentenced Scott to 178 months in prison, to go with an 11-year sentence imposed by a state judge.
Scott assumes that on appeal we will make an independent (de novo ) assessment of probable cause, ignoring the state *844judge's finding. We will not. The decision of the judge who issued the warrant receives "great deference". See Illinois v. Gates , 462 U.S. 213, 236, 103 S.Ct. 2317, 76 L.Ed.2d 527 (1983) ; United States v. McIntire , 516 F.3d 576, 577-79 (7th Cir. 2008). With the benefit of "great deference," this warrant is valid.
Scott observes that the rendezvous was not at his home and asks us to infer that any child pornography also would not be at home-maybe on his cell phone, Scott allows, but not at home. Yet Scott had offered to "host" the encounter, something possible at home. Scott also sent Kyle a picture of a man on a bedspread; it does not stretch credulity to infer that the picture had been taken at Scott's home and that other pictures also could be found there. What's more, many cell phones are set up to transmit the pictures they take to home computers, which with larger screens and access to file-sharing software are better than phones for acquiring and viewing pictures of all kinds.
Scott also contends that there is no reason to think that pedophiles create or acquire child pornography. Yet Scott asked Kyle to send a picture of himself with an erection. Scott insists that such a picture would have been child erotica (because he asked Kyle to wear compression shorts) rather than child pornography. Maybe so; the line between the two is hazy. But we have held that the collection of pictures focused on children's genitalia supplies reason to believe that the collector had child pornography. United States v. Lowe , 516 F.3d 580, 586 (7th Cir. 2008). The state judge did not exceed the bounds of "great deference" in drawing the same inference.
Several courts of appeals have held that particular affidavits did not adequately link pedophilia to the collection of child pornography. United States v. Cordero-Rosario , 786 F.3d 64, 70-71 (1st Cir. 2015) ; United States v. Falso , 544 F.3d 110, 123 (2d Cir. 2008) ; Virgin Islands v. John , 654 F.3d 412, 419 (3d Cir. 2011) ; United States v. Doyle , 650 F.3d 460, 472 (4th Cir. 2011) ; United States v. Hodson , 543 F.3d 286, 289 (6th Cir. 2008) ; Dougherty v. Covina , 654 F.3d 892, 898-99 (9th Cir. 2011). In some of these decisions the affidavit was just ipse dixit , not connecting pedophilia to pornography. The affidavit supporting a search of Scott's home, by contrast, included several connections-the electronic communications, the explicit pictures Scott sent to Kyle, and the request that Kyle send a sexual picture of his own. In other decisions we have cited there was a bit of evidence, but not enough; here there was more. And to be complete we must add that the Eighth Circuit has held that the link between pedophilia and child pornography is so strong that proof of one always supplies probable cause to search for the other. United States v. Colbert , 605 F.3d 573, 578 (8th Cir. 2010). We need not go that far to conclude that, with the benefit of great deference to the issuing judge's decision, this search warrant is valid.
In finding that the affidavit permitted the state judge to find probable cause, we have not relied on one agent's statement that his "training and experience" demonstrate a link between pedophilia and child pornography. Dougherty holds that such a statement does not supply probable cause, even with the benefit of great deference to the issuing judge, because it is fact free. What training? What experience? Is the training based on data or just intuition? Does the experience show that nine of ten arrested pedophiles possessed child porn? Five of ten? Three of ten? One of ten? Details matter. An officer who testifies on the stand to training and experience-for example, "my training and experience enable me to decode drug jargon"-can be cross-examined to unearth the statement's foundation, but a detail-free assertion of "training and experience" in an ex parte *845presentation does not illuminate the subject.
When an affidavit relies on an unidentified informant's experience, the judiciary demands details. See, e.g., Florida v. J.L. , 529 U.S. 266, 120 S.Ct. 1375, 146 L.Ed.2d 254 (2000). When an affidavit asserts that a dog's training and experience shows the reliability of a drug detection, the judiciary demands details. See, e.g., Florida v. Harris , 568 U.S. 237, 133 S.Ct. 1050, 185 L.Ed.2d 61 (2013). Details likewise are vital when an officer proposes his own training and experience as the basis of a warrant.
Humans can explain, while dogs cannot talk; and humans, who can reason, also are more susceptible to the fallacies of reasoning. Example: An officer may believe that, if child porn was found on the computers of 10 (or 100) pedophiles, then pedophilia always implies possession of child pornography. That's a fallacy. We need to know how many pedophiles do not possess child pornography. Starting with a set of people who have child pornography excludes all of the false positives and is not a reliable means to determine the likelihood that a given person, arrested for attempting to have sexual relations with a youngster, has child porn at home.
The litigants in this court profess certainty (at least, display certitude) about the relation between pedophilia and child pornography. Scott is confident that the two are unrelated. The United States is confident that one is a sign of the other. In support of these opposing views, each side offers-nothing but its own confidence. There is an empirical literature on the relation among different sex crimes, a literature that any expert witness would be obliged to consult and discuss, see Fed. R. Evid. 702, but the litigants have not based their positions on its findings. Indeed, at oral argument counsel for each side professed ignorance of this literature.
A substantial body of work concludes that people who collect child pornography are much more likely than the general population to be pedophiles. See, e.g., Michael C. Seto, James M. Cantor & Ray Blanchard, Child Pornography Offenses Are a Valid Diagnostic Indicator of Pedophilia , 115 J. Abnormal Psych. 610 (2006); Kelly M. Babchishin, R. Karl Hanson & Heather VanZuylen, Online Child Pornography Offenders are Different: A Meta-analysis of the Characteristics of Online and Offline Sex Offenders Against Children , 44 Archives of Sexual Behavior 45 (2015) (discussing other studies); United States Sentencing Commission, Federal Child Pornography Offenses 181 (2012) (reporting that 19% of child-pornography offenders in the study-that is, 310 of 1,654 persons-also had a conviction for a contact sex offense against a child).
Does correlation also run the other way? That is, if you start from a sample of pedophiles, how likely is it that they possess child pornography? The studies we cited above do not address that question. One that does, Janina Neutze, Michael C. Seto, Gerard A. Schaefer, Ingrid A. Mundt & Klaus M. Beier, Predictors of Child Pornography Offenses and Child Sexual Abuse in a Community Sample of Pedophiles and Hebephiles , 23 Sexual Abuse 212, 227 (2011), finds that a little more than half (50 of 95) of a sample of men who conceded sexually abusing children also reported owning and viewing child pornography. Whether self reports are reliable is open to question, and since the sample was drawn from Germany the definitions may not fit categories in the United States.
We think it likely that other empirical findings are available. And we may well have misunderstood those we located on our own. We do not cite these studies to show the validity of the warrant to search Scott's home; indeed, we did not rely on *846the agent's "training and experience" either. Our point, rather, is that inferences from the commission of one crime to the commission of another (e.g., from attempted child molestation to possessing child pornography) ought to be based on data, not on intuitions of the sort that led the lawyers to assert contradictory factual positions with no factual support. Data are available. Police and prosecutors would do well to consult them before making searches and arrests.
AFFIRMED