Before Wiener, Southwick, and Duncan, Circuit Judges. Stuart Kyle Duncan, Circuit Judge: “[W]hile the material inside adult bookstores and movie theaters is speech, the consequent sordidness outside is not.” City of Los Angeles v. Alameda Books, Inc., 535 U.S. 425, 445 (2002) (Kennedy, J., concurring in the judgment). Communities can therefore regulate the so-called “secondary effects” of sexually oriented businesses (or “SOBs”), like crime and blight, without running afoul of the First Amendment. See generally City of Renton v. Playtime Theatres, Inc., 475 U.S. 41 (1986); Young v. Am. Mini Theatres, Inc., 427 U.S. 50 (1976). Acting on that authority, the City of Dallas passed Ordinance No. 32125 in 2022. The Ordinance requires licensed SOBs, such as cabarets, escort agencies, and adult video stores, to close between 2:00 a.m. and 6:00 a.m. The Ordinance was backed by ample data—from the City’s own police task force, other comparable cities, and academic research— supporting a link between SOBs’ late-night operation and increased crime. Plaintiffs, a group of SOBs and their trade association, challenged the Ordinance under the First Amendment. After a hearing, the district court found that the City lacked reliable evidence to justify the Ordinance and that the Ordinance overly restricted Plaintiffs’ speech. It therefore preliminarily enjoined the Ordinance. The district court erred. Under longstanding Supreme Court precedent, the Ordinance is likely constitutional. The City’s evidence reasonably showed a link between SOBs’ late-night operations and an increase in “noxious side effects,” such as crime. Alameda Books, 535 U.S. at 446 (Kennedy, J., concurring in the judgment). The Ordinance also left the SOBs ample opportunity to purvey their speech at other times of the day and night. We therefore VACATE the preliminary injunction and REMAND for further proceedings. I. From late 2020 to early 2021, a rash of shootings in or around Dallas SOBs left multiple people dead.[1] The police responded by forming a task force to patrol near SOBs on busy nights after midnight.[2] Operating for about eight months during 2021, the task force made 123 felony arrests, responded to 134 calls for service, issued over 1,100 citations, and made more than 350 drug and weapon seizures. The police also compiled and analyzed 2019–21 data on crime occurring within a 500-foot radius of licensed SOBs. They broke this data down based on the number of arrests, crimes reported, and 911 calls. The analysis focused on the nighttime hours, comparing the 10:00 p.m.-to-2:00 a.m. and the 2:00 a.m.-to-6:00 a.m. windows. During those timeframes, the data showed over 1,600 custodial arrests. And while most property crime occurred from 10:00 p.m. to 2:00 a.m., the opposite was true for violent crime: roughly 67% of all aggravated assaults, rapes, robberies, and murders occurred from 2:00 a.m. to 6:00 a.m. In 2021, that percentage jumped to 76%. The data told a similar story about 911 calls. The police received over 4,500 calls between 10:00 p.m. and 6:00 a.m., over half of which came between 2:00 a.m. and 6:00 a.m. Over half of the Priority 1 calls—those requiring an immediate emergency response—also came during that window. The same was true with respect to calls to the fire department. After months of heightened patrols, the department began presenting its findings to the city council—twice to committees and once to the entire council. It also provided summaries of three academic studies linking SOBs to increased crime rates. And it noted that two other Texas cities, Beaumont and Amarillo, had issued reports finding a correlation between SOBs’ hours of operation and increased crime. Based on this evidence, the department recommended that the council close SOBs from 2:00 a.m. to 6:00 a.m. The council unanimously passed the Ordinance in January 2022. The Ordinance stated it was restricting SOBs’ hours to “reduce crime and conserve police and fire-rescue resources” because “the operation of [SOBs] between 2:00 a.m. and 6:00 a.m. is detrimental to the public health, safety, and general welfare.” The Ordinance listed the evidence it relied on, including recent “multiple shootings,” the increase in violent crime and 911 calls during those hours, the three academic studies, and the Beaumont and Amarillo reports. Plaintiffs immediately sued to enjoin the Ordinance, arguing it violated the First Amendment. Specifically, they claimed the Ordinance was a content-based restriction on their speech and that the City enacted it “without valid empirical information to support it.” The district court held a hearing and, largely agreeing with the Plaintiffs, granted a preliminary injunction. The court declined to decide whether intermediate or strict scrutiny applied, noting our court’s unsettled caselaw on the continuing validity of the secondary effects doctrine. But it held that the Ordinance likely failed under either standard. The district court then scrutinized the City’s evidence and concluded that it failed to support the “stated rationale for the Ordinance.” In particular, the court found that the City’s crime data was unreliable and that, regardless, it did not adequately link SOBs to secondary effects such as crime and increased 911 calls. Finally, the court concluded the Ordinance failed to leave SOBs’ protected speech sufficiently accessible. The City now appeals. II. “We review a preliminary injunction for abuse of discretion, reviewing findings of fact for clear error and conclusions of law de novo.” Tex. All. for Retired Ams. v. Scott, 28 F.4th 669, 671 (5th Cir. 2022) (citation omitted). To obtain a preliminary injunction, the Plaintiffs must show: (1) they are substantially likely to succeed on the merits; (2) a substantial threat of irreparable harm absent the injunction; (3) the threatened injury outweighs any harm caused by granting the injunction; and (4) the injunction is in the public interest. Clarke v. CFTC, 74 F.4th 627, 640–41 (5th Cir. 2023). On appeal, the parties contest only the first factor, whether the Plaintiffs are likely to succeed on the merits of their First Amendment claims. III. As a threshold matter, the parties dispute the First Amendment standard governing a municipality’s regulation of SOBs. For over three decades, the Supreme Court has analyzed such regulations under a two-step test adopted in City of Renton v. Playtime Theatres, 475 U.S. 41 (1986).[3] The first step asks whether the measure “ban[s]” SOBs or regulates only the “ time, place, and manner” of their operation. Id. at 46. If the latter, the second step asks whether the regulation is “designed to combat the undesirable secondary effects” of “businesses that purvey sexually explicit materials” rather than to restrict their “free expression.” Id. at 48–49. A regulation satisfying both steps is “reviewed under the standards applicable to ‘content-neutral’ time, place, and manner regulations,” namely intermediate scrutiny. Id. at 50. Accordingly, the regulation will be upheld if it “is designed to serve a substantial governmental interest and allows for reasonable alternative avenues of communication.” Ibid. Plaintiffs argue that Renton is no longer good law. And even if it is, they contend that the Ordinance is content-based under recent Supreme Court precedent and thus subject to strict scrutiny. We reject both arguments. Plaintiffs’ first argument depends on our now-overruled decision in Reagan National Adverting of Austin, Inc. v. City of Austin (Reagan I), 972 F.3d 696 (5th Cir. 2020), rev’d and remanded sub nom. City of Austin v. Reagan Nat’l Advert. of Austin, LLC, 142 S. Ct. 1464 (2022). There, we applied strict scrutiny to a law distinguishing on-premises from off-premises signs. Ibid.[4] To reach that conclusion, we read Reed v. Town of Gilbert, 576 U.S. 155 (2015), to require strict scrutiny whenever “a regulation of speech on its face draws distinctions based on the message a speaker conveys,” even if the law had a “benign motive” or “content-neutral justification.” Id. at 702 (quoting Reed, 576 U.S. at 163, 165) (internal quotation marks omitted). We further suggested that Reed abrogated many of our precedents—including cases applying Renton that we listed in a footnote See id. at 703 n.3.[5] Plaintiffs’ argument here relies heavily on that footnote. The problem for Plaintiffs—and it is a fatal one—is that the Supreme Court reversed Reagan I and rejected that decision’s understanding of Reed as “too extreme.” City of Austin, 142 S. Ct. at 1470–71. The Court clarified that its precedents “have consistently recognized that restrictions on speech may require some evaluation of the speech and nonetheless remain content neutral.” Id. at 1473; see also id. at 1474. Based on that principle, the Court held that the sign ordinance at issue was content neutral because it drew only location-based distinctions and had no illicit purpose. Id. at 1471. Importantly, the Court emphasized that an overly strict reading of Reed would “contravene numerous precedents” upon which “Reed did not purport to cast doubt.” Id. at 1474. The upshot for our case is obvious. Any shadow cast on the secondary effects doctrine by our Reagan I opinion has been dispelled by City of Austin. Specifically, Plaintiffs are mistaken that the Reagan I footnote somehow survived the decision’s reversal. To the contrary, that footnote depended on a view of Reed that City of Austin repudiated. The footnote, in other words, was not spared in the fall of Reagan I.[6] Alternatively, Plaintiffs argue that the Ordinance should now be analyzed as content-based under City of Austin‘s clarification of Reed. We disagree. Both Reed and City of Austin concerned physical signs and said nothing about SOBs or the secondary effects doctrine. The Court “does not normally overturn . . . earlier authority sub silentio.” Shalala v. Ill. Council on Long Term Care, Inc., 529 U.S. 1, 18 (2000).[7] So, it would be a mistake to interpret those decisions as silently spelling Renton‘s demise. To the contrary, City of Austin cautioned inferior courts against doing exactly that. See City of Austin, 142 S. Ct. at 1474 (warning that overreading Reed would “contravene numerous precedents” upon which “Reed did not purport to cast doubt”). More to the point, whether to overrule or modify Renton is the High Court’s business, not ours. “Our job, as an inferior court, is to adhere strictly to Supreme Court precedent, whether or not we think a precedent’s best days are behind it.” United States v. Vargas, 74 F.4th 673, 683 (5th Cir. 2023) (en banc) (citing Mallory v. Norfolk S. Ry. Co., 143 S. Ct. 2028, 2038 (2023)). Renton and its longstanding secondary effects doctrine has “direct application in [this]case,” Rodriguez de Quijas v. Shearson/Am. Exp.,Inc., 490 U.S. 477, 484 (1989), and so we are bound to apply it to the challenged Ordinance, Freedom From Religion Found., Inc. v. Mack, 4 F.4th 306, 315 (5th Cir. 2021). To that we now turn. IV. Under Renton, the Ordinance must be upheld if it “is designed to serve a substantial governmental interest and allows for reasonable alternative avenues of communication.” Renton, 475 U.S. at 50.[8] The district court found the Ordinance likely failed both requirements. We address each in turn. A. The district court concluded that the Ordinance failed Renton‘s first requirement because of flaws in the City’s supporting evidence. We disagree. The district court held the City’s evidence to a standard of exactitude not required by the Supreme Court’s precedents. The Supreme Court explicated Renton‘s evidentiary standard in City of Los Angeles v. Alameda Books, Inc., 535 U.S. 425, 433 (2002) (plurality opinion); see also id. at 444–53 (Kennedy, J., concurring in the judgment).[9] An SOB regulation is “designed to serve a substantial government interest” when the municipality can “provid[e] evidence that supports a link” between the regulated business and the targeted secondary effects. Alameda Books, 535 U.S. at 434, 437; see also id. at 449 (Kennedy, J., concurring in the judgment) (agreeing that the plurality “gives the correct answer” to the question of how much evidence is needed to satisfy Renton).[10] A municipality may rely on evidence “reasonably believed to be relevant,” id. at 438 (quoting Renton, 475 U.S. at 51–52), but not on “shoddy data or reasoning” that does not “fairly support” the ordinance’s rationale. Ibid.; see also id. at 451 (Kennedy, J., concurring in the judgment). Plaintiffs may show evidence is “shoddy” either because it “does not support [the ordinance's] rationale,” or because Plaintiffs’ own evidence counters the municipality’s findings. Id. at 438–39. Doing so shifts the burden back to the municipality to provide additional evidence. Id. at 439. The pertinent inquiry here, then, is whether the City could reasonably believe that its evidence linked SOBs’ operation between 2:00 a.m.–6:00a.m. and the secondary effects targeted by the Ordinance. See Baby Dolls Topless Saloons, Inc. v. City of Dallas, 295 F.3d 471, 481 (5th Cir. 2002) (stressing that this is a “reasonable belief standard”) (emphasis omitted). The district court answered that question in the negative after closely scrutinizing the City’s evidence. “[W]e review a district court’s findings as to the existence of a city’s evidence for clear error, but we review de novo whether that evidence” is “shoddy” or unreliable within the meaning of Alameda Books. H and A Land Corp., 480 F.3d at 338. To begin with, the district court found the City’s data flawed in that it “artificially enhance[d]” the association of crime with SOBs. The court cited four main reasons. First, the data included crimes committed at locations that held an SOB license but were not operating as an SOB.[11] Second, the court found the data “inaccurately inflate[d]” the numbers by counting all crime within a 500-foot radius around SOBs, thus bringing in crimes that might have occurred at a nearby restaurant or motel. Third, the court noted that, while the data analyzed crime occurring from 10:00 p.m.– 6:00 a.m., not every SOB was always open during those hours (for instance, some closed before 6:00 a.m. on weekdays). Finally, the court believed the very existence of the police task force distorted the data—for instance, by generating stops that would not have otherwise occurred or that were unrelated to SOBs, like traffic stops. For these reasons, the court concluded that the crime data did not reasonably link SOBs to secondary effects. The court also criticized the three academic studies cited in the Ordinance. Its basic objection was that the studies were not sufficiently similar to the Ordinance to be relevant. The court noted that, while all three studies linked SOBs with increased crime rates, they either did not “show increasing crime rates associated with late-night hours” or did not “address[] any particular time of day.” Thus, in the court’s view, the City could not have reasonably relied upon such studies to curtail nightly hours of operation. The district court applied Renton‘s reasonable belief standard too strictly. “[R]equiring proof to this degree of exactitude set the bar too high.” N.W. Enters., 352 F.3d at 181. The district court demanded the City link SOBs to secondary effects with a degree of certainty that outstrips what Renton envisioned. To the contrary, all Renton demands is evidence “reasonably believed to be relevant” to the problem. Renton, 475 U.S. at 51; see also N.W. Enters., 352 F.3d at 180 (under the “deference” demanded by Renton, “legislators cannot act, and cannot be required to act, only on judicial standards of proof”). Indeed, because “a city must have latitude to experiment” in addressing secondary effects, “very little evidence is required.” Alameda Books, 535 U.S. at 451 (Kennedy, J., concurring in the judgment); see also Doe I v. Landry, 909 F.3d 99, 109 (5th Cir. 2018) (“The evidentiary burden to support the governmental interest is light.”). The standard does not require a city to forge an ironclad connection between SOBs and secondary effects or to produce studies examining precisely the same conditions at issue. See Alameda Books, 535 U.S. at 437 (explaining that the evidence must only “support[] a link” between SOBs and the asserted secondary effects); see also Ctr. For Fair Pub. Pol’y, 336 F.3d at 1168 (“The record here is hardly overwhelming, but it does not have to be.”). The City’s evidence here meets the Renton standard. Consider first the context of the Ordinance’s enactment: responding to multiple shootings at Dallas SOBs in the late hours of the night, the City formed a task force to increase police presence around SOBs. The task force operated for the better part of a year and devoted over 1,200 man-hours to patrols. It made over 100 felony arrests, answered over 100 911 calls, and made over 350 weapons and drug seizures. To be sure, as the district court noted, not every arrest or seizure was related to an SOB. But the City was still entitled to rely on this type of boots-on-the-ground experience in crafting the Ordinance. “[C]ourts should not be in the business of second-guessing fact-bound empirical assessments of city planners” because they “know[] the streets” of their cities “better than we do.” Alameda Books, 535 U.S. at 451–52 (Kennedy, J., concurring in the judgment); see also, e.g., N.W. Enters., 352 F.3d at 180 (emphasizing that, under Renton and Alameda Books, courts must “respect[] local legislators’ superior understanding of local problems”). So long as a city’s “inferences appear reasonable, we should not say there is no basis for its conclusion.” Alameda Books, 535 U.S. at 452 (Kennedy, J., concurring in the judgment).[12] The City could reasonably infer from the lengthy experiences of its police department—which was presented three times to the city council—that SOBs were responsible in significant part for the noxious secondary effects targeted by the Ordinance. As one officer testified during the preliminary injunction hearing, SOBs are “powder keg[s]” for violent crime in the late hours of the night, because they attract crowds of young men consuming alcohol and drugs. The City’s other evidence reinforces that conclusion. While considering the Ordinance, the city council had before it five other Texas cities’ hours-of-operation restrictions on SOBs, including those of Fort Worth, San Antonio, and El Paso. The Ordinance itself noted that Amarillo and Beaumont had issued reports showing “a positive correlation between the hours of operation of [SOBs] and higher crime rates.” And all this was in addition to the three studies that, in the district court’s words, “suggest that SOBs are associated with an increase in overall crime.” Thus, the City was hardly pushing the envelope. Both the Supreme Court and this court have found the reasonable belief standard satisfied on records much more tenuous than this one. See, e.g., Alameda Books, 535 U.S. at 451–52 (Kennedy, J., concurring in the judgment) (allowing an ordinance to survive summary judgment although supported only by “a single study and common experience”); H and A Land Corp., 480 F.3d at 339–40 (finding the standard satisfied based on two surveys, conducted in other cities, in which real estate appraisers “predicted that the presence of an adult bookstore would negatively affect real estate value in the surrounding area”). The district court also faulted the City for failing to include crime data associated with non-SOBs. Because crime could also occur at “other late-night establishments,” the court reasoned, a comparison was necessary to “conclude that the secondary effects are linked to the SOBs, as opposed to some other, unrelated factor.” We disagree. The City was entitled to make reasonable inferences from the information it had without needing to rule out other possibilities. Alameda Books has already settled this point. The Supreme Court faulted the lower court for “implicitly requir[ing] the city to prove that its theory [was] the only one that can plausibly explain the data.” 535 U.S. at 437. To the contrary, a city need not “rule[] out every theory for the link between [SOBs and secondary effects] that is inconsistent with its own.” Ibid. It can instead reasonably interpret the available information without courts “replac[ing] the city’s theory . . . with [their] own.” Id. at 437–38; Fantasy Ranch Inc. v. City of Arlington, 459 F.3d 546, 561 (5th Cir. 2006) (requiring deference to the “legislative process” even if the evidence allows a “different and equally reasonable conclusion” (citation omitted)). So, here, the City could plausibly infer that the best explanation for violent crime and 911 calls near SOBs was the SOBs themselves rather than some other factor. One final word. At times, the district court appeared concerned with whether the Ordinance would be successful in reducing secondary effects. For instance, the court noted that, without data about 911 calls from non-SOBs, it was “impossible to know” whether closing SOBs from 2:00 a.m. to 6:00 a.m. would really conserve City resources. This was the wrong focus, however. Cities’ latitude to experiment means, by definition, that they need not show that their “ordinance[s] will successfully lower crime,” at least “not without actual and convincing evidence from plaintiffs to the contrary.” Alameda Books, 535 U.S. at 439 (emphasis added); see also Doe I, 909 F.3d at 110 (“The State need not demonstrate through empirical data, though, that its regulation will reduce [secondary effects].”); Baby Dolls Topless Saloons, 295 F.3d at 481 (rejecting the argument that there must be “specific evidence linking” the ordinance “to reducing secondary effects”) (emphasis removed). Once again, the district court demanded evidentiary precision from the City that Renton does not require. In sum, the City is substantially likely to show that the Ordinance was “designed to further a substantial government interest” under Renton.[13] B. The Ordinance must also “allow[] for reasonable alternative avenues of communication.” Renton, 475 U.S. at 50. It is mostly here that Justice Kennedy’s controlling opinion in Alameda Books differs from the plurality. See World Wide Video of Wash., Inc. v. City of Spokane, 368 F.3d 1186, 1195 (9th Cir. 2004), as amended, (July 12, 2004) (noting that Justice Kennedy’s concurrence “dovetails with the requirement that an ordinance must leave open adequate alternative avenues of communication”). As he cautioned, restrictions on SOBs must “leave the quantity and accessibility of the speech substantially undiminished.” Alameda Books, 535 U.S. at 445 (Kennedy, J., concurring in the judgment). A city may not reduce secondary effects simply by reducing speech in the same proportion. Rather, “the necessary rationale for applying intermediate scrutiny is the promise that zoning ordinances . . . may reduce the costs of secondary effects without substantially reducing speech.” Id. at 450. The district court found that the Ordinance failed this requirement because closure would cost the SOBs significant revenue while depriving many patrons and dancers of access to protected speech during those hours. We disagree. A regulation need not be costless to be valid. See Lakeland Lounge of Jackson, Inc. v. City of Jackson, 973 F.2d 1255, 1260 (5th Cir. 1992) (upholding a location-based regulation even though it required SOBs to relocate to places that “d[id] not seem particularly desirable for economic reasons”). And Plaintiffs do not argue that the Ordinance will be so costly as to drive them out of business. See Ent. Prods., Inc. v. Shelby County, 721 F.3d 729, 741 (6th Cir. 2013) (rejecting a profitability argument where plaintiffs did “not allege that [the ordinance] . . . makes their businesses unprofitable”). Thus, Plaintiffs still have a “reasonable opportunity to open and operate” their businesses. See Renton, 475 U.S. at 54; see also N.W. Enters., 352 F.3d at 181 (interpreting Justice Kennedy’s concurrence to mean that “the City may not use its regulation to eliminate businesses as a means to reduce their secondary effects” (emphasis added)). On this record, we cannot say that the Ordinance substantially or disproportionately restricts speech. It leaves SOBs free to open for twenty hours a day, seven days a week, while also, in the City’s reasonable view, curtailing the violent crime and 911 calls with which the City was concerned. Other circuits have found similar restrictions valid in the wake of Alameda Books. See, e.g., Ctr. For Fair Pub. Pol’y, 336 F.3d at 1162–63; Deja Vu of Cincinnati, L.L.C. v. Union Twp. Bd. of Trs., 411 F.3d 777, 791 (6th Cir. 2005). We see no reason to conclude otherwise.[14] *** To sum up, we hold that Renton remains good law and thus apply intermediate scrutiny to the Ordinance. We further conclude that, under Alameda Books, Plaintiffs have not shown a likelihood of success on the merits of their First Amendment claims. A preliminary injunction was therefore unwarranted. V. We VACATE the preliminary injunction and REMAND for further proceedings consistent with this opinion.
