Helen Mihailovich v. Gary Laatsch and Law Office of Pavalon, Gifford, Laatsch & Marino

ILANA DIAMOND ROVNER, Circuit Judge. A one-car automobile accident on a curved roadway maintained by Cook County, Illinois (the “County”) caused Helen Mihailovich to suffer a spinal cord injury. Mihailovich filed suit against the County in state court, alleging that it had been negligent in its care and upkeep of the roadway and that the condition of the roadway was a contributing cause of the accident. Attorney Gary Laatsch and the law firm now known as Pavalon, Gifford, Laatsch & Marino (collectively “Laatsch”) filed the suit on Mihailovich’s behalf but withdrew *894 from the case approximately six months prior to the scheduled trial date. The attorney who succeeded Laatsch was unable to convince the court to continue the trial date, and after Mihailovich’s new counsel was unable to retain an expert and to identify other individuals who might testify on her behalf, the court entered successive orders barring Mihailovich from presenting witnesses and then granting summary judgment in favor of the County. Mihailovich, now a citizen of Florida, subsequently filed suit in diversity against Laatsch, charging him with legal malpractice. The malpractice suit proceeded to trial, and the jury held in Laatsch’s favor. Mihailovich appeals, contending that the district judge erred in certain evidentiary rulings that constrained her ability to show that the roadway on which her accident occurred was dangerous and that she would have prevailed in her suit against the County but for Laatsch’s alleged malpractice. Because we conclude that the district court abused its discretion in excluding evidence of other accidents that had occurred at the location of the Mihailo-vich accident, we vacate and remand for a new trial. I. Early in the evening of July 1, 1986, Mihailovich’s husband Mike picked her up after work at the Olympia Way train station in Olympia Fields, Illinois, a suburb southwest of Chicago. On leaving the train station, they proceeded south on Kedzie Avenue with Mr. Mihailovich at the wheel of their 1978 Chevrolet Nomad van. Between 205th and 207th Streets, they encountered what is known as the Kedzie curve, a reverse S-curve that connects Kedzie Avenue with Olympian Way. As the “S-curve” appellation suggests, the Kedzie curve comprises two successive curves: a northern curve that turns to the west, and a southern curve turning back to the east. Mihailovich did not know exactly how fast her husband was driving, but later testified that “[h]e was going with the traffic and the speed was regular.” Tr. 207. She noticed nothing unusual in his manner of driving. It was raining or drizzling as it had been throughout the day, and the pavement was wet. Somewhere along the S-curve (Mihailovich could not recall where), their van slid off the roadway, tumbled into a ditch, and came to rest upside down. Mihailovich could not recall whether or not her husband had braked prior to the accident. The event was “just like a bullet,” she later testified (see Tr. 209, quoting her deposition testimony); it happened “in a spare moment, in a spare second” (Tr. 182). As a result of the accident, Mihailovich suffered a spinal cord injury that required months of hospitalization and rehabilitation and permanently limited her mobility. 1 Police officer Mark Fazzini was summoned to the scene of the accident at approximately 6:40 p.m. Fazzini served on the Olympia Fields police force for twenty-five years and retired in 2000 as its chief. Over the course of his work for the Village, he traveled the Kedzie curve many times each day. When Fazzini reached the scene, he discovered the overturned Mi-hailovich van on the west side of a ditch in front of a residence located at 20540 Ked-zie Avenue, a position near the south end of the northern curve. Fazzini described the Kedzie roadway as a two-lane road that was long and straight on either end of the S-curve, with only a broken yellow line separating the two lanes of traffic, each moving in an opposite direction (north and *895 south). So far as Fazzini knew, no maintenance work had been done on the curve from 1980, by which time the County had assumed responsibility for the curve from the State, through the date of the Mihailo-vich accident in 1986. Fazzini’s written report of the accident indicated that there were “no defects” in the road. By way of background, we pause at this point to note that the Kedzie curve had something of a notorious reputation among local officials as a traffic hazard. Accidents occurred regularly along the S-eurve, particularly when the pavement was wet, and there was a perception among officials that the frequency of accidents had increased since the County had resurfaced and widened the roadway to the immediate north and south of the Kedzie curve (but not the bulk of the S-curve itself) in 1981-1982. Concern about the accident rate had moved the board of trustees of Olympia Fields in 1984 to enact a resolution calling upon the County to address the problem. Partly in response to that resolution, the County in late 1984 and early 1985 had installed additional signage alerting motorists to the winding character of the road, reducing the advised speed through the curve, and warning drivers that the pavement was slippery when wet. These signs were in place at the time of the Mihailovich accident in 1986. However, the County had not resurfaced or otherwise addressed the condition of the roadway itself, and notwithstanding the additional warning signs, accidents continued to occur with regularity until the County reconstructed the S-curve in 1988-1989. At that time, the County increased the radius of the north curve, added a four-foot center rumble strip, added a su-perelevation of four percent, and resurfaced the roadway. In the wake of the reconstruction, the accident rate dropped from an average of approximately twenty per year to two or three per year. The failure to take such steps prior to the Mihailovich accident is what gave rise to Mihailovich’s negligence claim against the County. The jury in this case was aware that the Kedzie curve was perceived as dangerous by local officials, but heard no evidence about the other accidents that had occurred at the S-curve in the years immediately before and after the Mihailovich accident. Because Mihailovich has made the district court’s limitation her lead issue on appeal, we shall summarize both the evidence that the court allowed and the evidence that the court excluded on this point. Following those summaries we shall describe the other evidence presented in support of Mihailovich’s malpractice claim before turning to the merits of her appeal. Admitted Evidence Regarding the Curve’s Potential Hazard Arthur Kaindl worked as a highway engineer for Cook County for thirty-eight years, rising to the post of Chief Engineer in the Bureau of Maintenance, which he held at the time of his retirement in August 1986. In September of 1983, Kaindl was the Chief Engineer of Transportation and Planning. In that capacity, he received a memorandum from Joseph Marsik, Chief Engineer of Maintenance, forwarding a second memorandum that Marsik had received from Frank Reno, the District Engineer of Maintenance for the district that included Olympia Fields and the Kedzie curve. Although Kaindl was not permitted to read either of the two memoranda to the jury, he indicated that Reno’s memorandum expressed a concern that the Kedzie curve was in a dangerous condition. Tr. 555. Upon receipt of these memoranda, Kaindl ordered personnel under his supervision to undertake a review of the curve, including field surveys to assess, among other things, *896 whether the existing signage and posted speed limit at the curve were appropriate to the condition of the curve. The propriety of the speed limit was evaluated using a ball-bank study, which by measuring the degree of sideways sway of a vehicle driving through the curve ascertains the maximum speed at which a motorist will feel comfortable. That study indicated that thirty miles per hour was an appropriate, comfortable speed through the curve and that was the speed recommended on advisory signs later posted at the curve. Kaindl acknowledged that a ball-bank study is conducted on dry pavement, that it does not measure the coefficient of friction, 2 and that it does not determine a safe speed for the roadway when the pavement is wet. After completing its review of the curve (apparently in 1984), the County did not resurface or reconstruct the curve at any time through August of 1986, when Kaindl retired. Fred Unger served as the. police chief of Olympia Fields from 1971 to 1990. He traveled through the Kedzie curve and observed its condition hundreds of times over the course of his tenure. According to Unger, the pavement at the curve tended to become slick when dampened by precipitation. “During inclement weather, accompanied by precipitation, the roadway would take on a slick surface, and the oil from the pores of the macadam surface would be floated to the top by the amount of water that was hitting the pavement.” Tr. 429. Unger indicated that a light rain was sufficient to bring about this condition. Robert E. Field has lived in Olympia Fields since 1974. He served as a trustee of the Village from 1981 through 1989 and as its president from 1991 through 1997. His home is located one block south and west of the Kedzie curve, and he traveled it at least twice daily through the date of the Mihailovich accident, although both Field and his wife avoided the S-curve during inclement weather. Field described the S-curve as turning to the west and then back to the east and going downhill as it did so; according to Field, a driver could not see from one end of the curve to the other as he entered it. The roadway through the curve comprised two lanes without a center divider, the edges of the pavement were crumbling, the shoulders were gravel rather than paved and were a couple of inches lower than the road itself, the surface of the roadway was very smooth, and over time traffic had worn grooves into the pavement. Field testified that his primary reason for seeking a seat on the Village board of trustees in 1980 was to do something about the condition of the Kedzie curve, which he had grown to believe was a hazard. He was particularly concerned that the curved roadway became dangerously slick when wet. Following his election and at his urging, the board first enacted a resolution directing the police and public works departments of the Village to assess the condition of the roadway and prepare a report. Upon receipt of that report, Field and 'another trustee drafted a resolution calling on the County to address the condition of the curve. The Village board adopted that resolution by a unanimous vote on October 10,1984. Anita Healy served on the Olympia Fields board of trustees beginning in 1979 and as its president from 1981 through 1989. Healy read to the jury the following redacted version of the Village’s 1984 resolution regarding the Kedzie curve: WHEREAS, it is a fact that a portion of Kedzie Avenue lying between 207th Street and 206th Street in the Village of *897 Olympia Fields is a county maintained roadway; 3 and WHEREAS, ... it appears to be an eminently dangerous roadway particularly under certain hazardous weather conditions; and WHEREAS, the Village Board of the Village of Olympia Fields as the governing body of the said Village wishes to inform the County Board of Commissioners of the County of Cook, State of Illinois[,] of its concern and alarm over the state of repair, condition, design and road surface material of the aforesaid section of Kedzie Avenue; and WHEREAS, it has been proposed that the Village by official action of this resolution inform the Board of Commissioners of the County of Cook of the State of Illinois officially and request that immediate action and remedies be instituted to correct the condition of the roadway herein set out: NOW, THEREFORE, BE IT RESOLVED by the Board of Trustees of the Village of Olympia Fields, Illinois, that they hereby petition on behalf of the Village of Olympia Fields, Illinois, a municipal corporation, to the Board of Commissioners of the County of Cook of the State of Illinois that they cause to have undertaken immediate remedial actions to correct the dangerous, unsafe and ill-constructed condition of county road commonly known as Kedzie Avenue between its intersection of 207th Street and 206th Street that lies within the Village of Olympia Fields, Illinois. This resolution shall be in full force and effect from and after its passage and approval as provided by law. R. 1 Ex. G; Tr. 235-37. 4 Healy sent a copy of the resolution to the County Superintendent of Highways on October 25, 1984. In December of 1984, the County installed three types of signs that were intended to improve the safety of the curve: a sign advising motorists to reduce their speed to thirty miles per hour, a reverse curve sign, and a “Slippery When Wet” sign. 5 On January 7, 1985, the County’s Superintendent of 'Highways, Richard Golterman, wrote a letter to Village president Healy noting that the County had investigated the “dangerous condition” alluded to in Healy’s letter of October 25, 1984, forwarding the Village’s resolution to the County. Golterman’s letter described the signs that the County had installed and concluded, ‘We trust that these corrected measures will serve to alleviate this ‘dangerous condition’, but respectfully suggest that the Village continue to monitor this portion of Kedzie Avenue and keep us advised.” Tr. 240; R. 1 Ex. 1. *898 Excluded Evidence Regarding Other ' Accidents at the Kedzie Curve Plaintiffs retained expert, civil engineer William Berg, reviewed data maintained by Olympia Fields regarding accidents that had taken place at the Kedzie curve from 1983 through July 15, 1987, and prepared a report which, in relevant" part, summarized that data. 6 Berg’s report indicated that a "total of ninety-four accidents occurred at the Kedzie curve during this period of four and one-half years. In 1983, there were twenty-five accidents, seventeen of which, or sixty-eight percent, occurred when the pavement was wet. In 1984, fifteen accidents occurred, fourteen of them, or ninety-three percent, on wet pavement. In the eighteen months immediately preceding.the Mihailovich accident (January of 1985 through June of 1986), a total of twenty-nine accidents took place, twenty-two of them,1 or seventy-six percent, on wet pavement. In the (approximately) one-year period following the Mi-hailovich accident (July 1, 1986 through July .15, 1987), another twenty-five accidents occurred, twenty of them, or eighty percent, when the pavement was wet. Berg also reviewed police reports that were available for twenty-four of the twenty-nine accidents that occurred in the eighteen months prior to the Mihailovich accident. According to Berg, in twenty-one of those twenty-four accidents, a vehicle had been proceeding southbound on Kedzie Avenue into the curve and had either left the roadway altogether inside or outside of the curve or had crossed the center line and collided with a northbound vehicle. R. 62 Ex. 6. 7 Village president Anita Healy would have testified that the occurrence of serious accidents at the curve, particularly during inclement weather, had been a topic of discussion among Village officials for a number of years. She knew of one fatality and multiple injuries that had resulted from such accidents. To Healy’s knowledge, no other curved roadway in the Village had a comparable rate of accidents, injuries, and fatalities.. When Healy sent the Village board of trustees’ 1984 resolution regarding the Kedzie curve to the County Superintendent of Highways on October 25 of that year, she wrote in her transmittal letter that “[w]e have been receiving increased numbers of complaints about the roadway in that area and requests to remedy the situation.” R. 1 Ex. H. Her letter concluded: “It is our hope *899 that the Highway Department will look into this problem and will determine a way to remedy this dangerous situation.” Id. According to Healy, after the curve was reconstructed between 1988 and 1989, the accident rate dropped to an “[ajhnost negligible” rate, and there had been no fatalities since then. Tr. 250-51. Retired Olympia Fields police officer Mark Fazzini would have testified that between 1980 and 1986, “dozens and dozens and dozens of automobile accidents [had occurred] on the Kedzie curve,” far more than at any other curve in the Village. Tr. 252. According to Mihailovich’s offer of proof: [I]t was [Fazzini’s] opinion from investigating dozens of collisions that when it rained, a film developed on the surface of the highway, that it became slick, , it looked like an oil slick, and that the automobilef] collisions that he investigated in the wet inclement conditions always lost traction in the roadway and slid up, either straight ahead, either going into the woods at the far end of the curve, or after sliding would then go off the roadway to the west. Tr. 252. Fazzini also would have testified that the Village police department had made efforts to slow southbound traffic entering the curve in order to reduce the accident rate: [BJecause of the Village’s concern about the cars continuing to pile up and injure and kill citizens out there, ... the police department literally set up speed traps at the highway to try to get drivers to slow down because they knew that cars coming from the north down into this curve had been traveling through a semi-rural area and that, as we know in this case, Kedzie Avenue just north of the curve, was resurfaced in 1981, 82 at Biden and this resurfacing went right up to the beginning of this north curve and the cars would come into the curve at a high rate of speed because of their assurance they were on a good, sound highway. Tr.- 253. According to Fazzini, the Olympia Fields police department had notified the County of the situation with respect to the Kedzie curve “on .more than one occasion,” but nothing was done in response. Id. Once the curve was resurfaced in 1988-1989, accidents were virtually eliminated. Village -trustee Robert Field would have additionally testified, based on his own observations as a driver who traveled the ■curve regularly and as a Village trustee, that a-large number of accidents occurred at' the curve, that he knew of no other location in the Village with such a high accident rate, and that most or all of the accidents occurred in inclement weather. Tr. 678. Recall that following Field’s election to the board of trustees in 1980, he had successfully urged the board to commission a study of the curve. Field would have testified that in the course of that study, the Olympia Fields police department had compiled a multi-year inventory of-traffic accidents ■ that had occurred at the curve, and that -the police chief at that time, Fred Unger, had presented a report including that compilation to the Village board. That -report, ■ according to Field, indicated that “more traffic accidents had occurred on the Kedzie curve than in the rest of the Village combined,” (Tr. 680), and that the resurfacing of the roadway to the north and south of the Kedzie curve in 1981-1982 appeared to have resulted in an increase in the number of accidents. Tr. 681-82. Field also was familiar with the additional warning signs that the County had installed at. the curve in response to the board’s October 1984 resolution. Field recalled that according to the Village’s police department, that signage had not had any discernible impact on the accidents Occurring at the curve. Tr. 681. However,- after the curve, was reconstructed and *900 resurfaced in 1988-1989, the accident rate dropped to near-zero. Tr. 683-84. Frank Reno, who served in the County’s employ for thirty-eight years, was the District Engineer of Maintenance in the early 1980s for the district that included Olympia Fields and the Kedzie curve. In August or September 1983, more than one year before the Olympia Fields board of trustees enacted and transmitted its resolution about the curve to the County, Reno became aware that the rate of accidents with fatalities had increased in the area of the curve in the preceding years. Reno thought that the County should look into the situation at the curve, and for that reason he wrote the September 19, 1983 memorandum to Joseph Marsik, Chief Engineer of Maintenance, that Arthur Kaindl had referred to in his testimony. Reno stated in his memorandum: In the last two years it has been observed that the accident rate in the above subject area has increased considerably, with fatalities involved. There is a change of alignment with a curve in the pavement alignment between Kedzie Ave. and 01ympia[n] Way, and this curve does not seem that sharp and dangerous. The pavement on Olympiad Way is very smooth and could contribute to the cause of these accidents. The accident rate has increased since Kedzie Ave. was improved and repaved. It is recommended that this matter be relayed to our design and traffic division. Tr. 434-35; R. 62 Ex. 10. Neither Marsik nor Reno was permitted to relay the contents of Reno’s memorandum to the jury. Marsik had received Reno’s memorandum and agreed with his recommendation that the curve be investigated. Marsik had forwarded Reno’s memorandum to Kaindl. In his cover memorandum to Kaindl, Marsik had written: Attached herewith is a memorandum from District Engineer Frank Reno pertaining to an alarming increase in traffic accidents at subject location (reversed curve section) between 01ympia[n][Way] and Kedzie Avenue. It is requested that this section be reviewed. It is our belief that the traffic is going too fast on the straight roadway, entering the curve and then has a problem negotiating the curve. Tr. 500; R. 62 Ex. 9. Evidence Relating to Malpractice Following her accident in 1986, Mihailo-vich decided to sue the County. She eventually engaged attorney Eugene Pavalon, who in turn assigned the case to his associate (later his partner) Gary Laatsch. Laatsch filed suit on Mihailovich’s behalf in June 1987, alleging that the curve was dangerous, that the County had negligently permitted the curve to remain in an unsafe condition, and that the County had failed to alert motorists to the danger. R. 1 Ex. C. Laatsch believed there was sufficient merit to Mihailovich’s claim to warrant discovery. Laatsch represented Mi-hailovich from 1987 through early 1998, when he formally withdrew from the case. During his representation of Mihailo-vich, Laatsch took the following steps (among others) in an effort to build a case on her behalf: (1) He met with Mihailovich at her home on multiple occasions and prepared her for her deposition by the County. (2) According to his trial testimony, Laatsch obtained a copy of the Village’s 1984 resolution calling on the County to investigate the Kedzie curve. Tr. 371, 373-74. 8 (3) In 1990, Laatsch ob *901 tained copies of the field studies that the County had completed in 1984 in order to ascertain the appropriate speed limit on the Kedzie curve and appropriate traffic advisory signs. (4) He spoke with structural engineer Herbert Miller, who had inspected the scene of the accident two months after it occurred together with Mi-hailovich’s first attorney. Miller had taken pictures of the accident site and had prepared a report for the prior attorney; Laatsch had copies of those materials. But according to Laatsch, Miller told him that “he could never testify” that the pictures he had taken depicted the area where the Mihailovich vehicle had left the roadway and that Miller, in fact, had not been able to determine where precisely the accident had occurred. Tr. 390-92.(5) Laatsch also issued one set of written interrogatories to the County, and as a result of those interrogatories he obtained a list of potential witnesses from the County. He deposed one of these witnesses — Daniel Waters, a sign technician, whose knowledge was limited to the signage at the Kedzie curve and who had no responsibility for the surface of the roadway. (6) Laatsch contacted William Berg, a civil engineer, for purposes of obtaining an expert opinion about the condition of the roadway at the time of the Mihailovich accident. Laatsch sent Berg, among other things, a copy of Waters’ deposition, along with certain correspondence between Olympia Fields and the County Highway Department. Based on these materials, Berg informed Laatsch that the signs posted at or near the Kedzie curve were sufficient to inform drivers of the condition of the curve. Berg could not otherwise determine whether the roadway was defective at the time of the Mihailovich accident. R. 62- Ex. 5 at 44.(7) In March 1987, Laatsch visited the scene of the accident. He walked a half block in each direction from the point where the Mihailovich van had come to rest, and he also drove through the curve in both directions. He observed no significant potholes or other defects in the pavement, nor did he recall the shoulders being gravel and/or lower than the roadway itself. Based on the information available to him (including the accident report), however, he could not determine where the Mihailovich van had begun to slide or where it left the roadway. He also spoke to an eyewitness to the accident (Harold Moore, the police chief of nearby Crestwood), but did not take a formal statement from him. (8) Finally, he obtained records from the County indicating that it inspected the roadway twice per month and periodically'maintained it. However, Laatsch did not undertake the following steps: (1) He never spoke with Mark Fazzini, the Olympia Fields police officer who investigated the scene of the Mihailovich accident, choosing instead to rely on Fazzini’s accident report. (2) He neither contacted nor deposed any Olympia Fields official, including Healy or Field, to discuss the Village’s 1984 resolution regarding the Kedzie curve or the basis for that resolution. (3) Other than Waters, who disclosed in his deposition that he was not responsible for the road surface of the Kedzie curve, Laatsch did not depose any of the witnesses disclosed in the County’s response to his interrogatories or that Waters named in his deposition-including County Highway Superintendent Golterman, Chief Engineer *902 Kaindl, or Village President Healy. In fact, Waters was the sole individual that Laatsch deposed. (4) Laatsch did not attempt to engage an expert until 1992, when he contacted Berg. (5) Laatsch did not obtain data regarding the other accidents that had occurred at the Kedzie curve. (6) He did not attempt to ascertain whether the signs that the County had installed at the curve in 1984-1985, partly in response to Olympia Fields’ resolution, had affected the accident rate and/or addressed the concerns that had prompted the Village to pass the resolution. (7) He failed to respond to the interrogatories that the County had issued, and he did not comply with the state court’s order to identify the witnesses who might testify on Mihailovich’s behalf at trial. Mihailovich’s case against the County proceeded languorously toward a trial date. The case was initially set for trial in September 1992. Having contacted Berg in June of that year and having been unable to obtain from him an opinion regarding the curve that would have been helpful to Mihailovich’s case, Laatsch opted to voluntarily dismiss the suit when it was called for trial. He refiled it the following year. However, Laatsch engaged in no further discovery after the case was restored to the court’s docket. Neither did he inform Mihailovich that he had dismissed and then refiled the suit. On March 11, 1997, Laatsch appeared in court for a ease management conference and informed the judge that “extensive discovery” had been completed and that Mihailovich would probably have an expert witness testify at trial. Tr. 419. When the judge suggested a trial date of January 13, 1998, Laatsch replied, “That will be fine.” Tr. 420. The court entered an order giving Laatsch until June 9, 1997, to disclose his witnesses. Laatsch never filed a list of witnesses, however, and in December 1997, the County moved to bar Mihailovich from calling any witnesses to testify. In response, Laatsch moved to continue the trial date. On December 16, 1997, the court granted the latter motion and struck the trial date. Approximately two weeks after the court cancelled the January 1998 trial date, Laatsch'moved to withdraw from the case. For nearly a year prior to Laatsch’s motion, Mihailovich and her son had been dissatisfied with the progress of her suit and had been searching for someone to replace Laatsch. In February 1997, Mi-hailovich’s son had written Laatsch directing him to send her case file to the Law Offices of Edward Vrdolyak for review. Vrdolyak’s firm had declined to take the case, however. Two other firms — Ross & Hardies and Corboy & Demetrio — had likewise turned the case down. On January 27, 1998, the court granted Laatsch’s motion to withdraw and gave Mihailovich forty-five days in which to retain new counsel. The trial date was re-set to July 22,1998. In April or May of 1998, attorney Marc McKenna agreed to assume representation of Mihailovich provided that the court would agree to postpone the trial date further (beyond July). When McKenna received the Mihailovich case file from Laatsch, it included only the police report describing the accident, a videotape of the roadway, some photographs of the accident scene, some printouts of information from the County (the contents of which he couldn’t recall), and two depositions— those of Mihailovich and a County employee (presumably Daniel Waters). McKenna twice sought a continuance from the court and was twice denied one. The court did grant McKenna additional time to disclose Mihailovich’s witnesses, including any experts who might testify on her behalf. McKenna contacted two experts hoping to obtain an opinion as to the condition of the Kedzie curve and whether it contributed to the Mihailovich accident, but neither indi *903 vidual was able to form an opinion based on the limited materials McKenna could provide from the case file. McKenna thus found himself with no experts to name, and as for other witnesses, he would later testify that he did not know whom to disclose. He conceded on cross-examination, however, that he could at least have named Mihailovich, her physician, and County employee Waters. In any event, when McKenna failed to name anyone, the County renewed its motion to bar Mihailo-vich from presenting any witnesses at trial. The court granted that motion on July 8, 1998. That order in essence left Mihail-ovich with no evidence to present in support of her claim, and, accordingly, on July 24, 1998, the court entered summary judgment against Mihailovich on the County’s motion. The following year, Mihailovich filed the instant suit against Laatsch and the law firm of Pavalon, Gifford, Laatsch & Marino for legal malpractice. Mihailovich contended, in essence, that Laatsch had conducted so little discovery and had so poorly prepared the case for trial that by the time he withdrew, there was little that his successor reasonably could have done to avoid dismissal. At the same time, according to Mihailovich, there was ample evidence available that would have supported her negligence claim against the County. For example, nearly 100 accidents had occurred on the Kedzie curve between 1983 and 1987; and discovery conducted in this case revealed that Olympia Fields officials and' County employees, among others, were aware of the high accident rate at the curve and expressed concern well in advance of the Mihailovich accident that the curve was (or might be) unsafe. Moreover, when supplied with additional information about the Kedzie curve, including data regarding the other accidents that had occurred between 1983 and 1987, William Berg — the same expert who had been unable to tell Laatsch whether the curve was defective — opined that the roadway was hazardous and that the condition of the road had contributed to the Mihailo-vich accident. After , reviewing the evidence that Mihailovich had marshaled in support of her malpractice claim, the district court ■ denied Laatsch’s motion for summary judgment (R. 123), and the case proceeded to trial. At trial, John Lowrey, a personal injury attorney with more than thirty years’ experience, testified on Mihailovich’s behalf as an expert regarding the standard of care for attorneys handling personal injury cases. Lowrey opined that Laatsch had deviated from that standard of care in a number of respects, including the following: (1) He had failed to properly investigate the Mihailovich accident and ascertain the underlying facts. (2) He had failed to depose the County employees who were responsible for the care and maintenance of the Kedzie curve. He had not deposed anyone named in the County’s answer to the interrogatories he had propounded other than Waters, nor had he pursued the leads that Waters himself had supplied in his deposition. (3) He had failed to speak to Olympia Fields officials about the circumstances that led them to pass the 1984 resolution. Village president Healy, for example, “would have been a gold mine,” in Lowrey’s view. Tr. 449.(4) Laatsch had failed to engage a highway expert early on in the discovery phase of the case, not simply to supply him with an opinion as to the state of the Kedzie curve but also to help- him identify the information necessary in order to render such an opinion. Although Laatsch had contacted Berg, he ,had not provided Berg with the information that the engineer needed in order to form an opinion about the Kedzie curve. (5) By informing the trial court in March 1997 that the parties had engaged in extensive discovery, Laatsch had misled the court about the actual amount of discovery *904 taken (which was modest). (6) Also in March 1997, Laatsch had acceded to an earlier trial date (January 13, 1998) than Laatsch’s level of preparation warranted. (7) He had allowed discovery deadlines to pass without disclosing Mihailovich’s witnesses. (8) Laatsch had withdrawn from the case in January 1998, eleven years into the litigation, with the trial date fast approaching and the ease file in a state of disrepair. Lowrey also opined, based on a review of the depositions of Village President Healy, Village trustee Field, Village police chief Unger, County engineer Reno, Harry Abbott, the County Highway Department’s Chief Engineer of Design, and other witnesses, as well as the Olympia Fields resolution regarding the Kedzie curve, the available information regarding the Mihail-ovich accident, and records of the accidents that had occurred at the Kedzie curve, that Mihailovich’s negligence suit against the County was a winnable case. In an effort to show that Mihailovich would not have won her suit against the County even if Laatsch had taken the steps that Mihailovich’s legal expert criticized him for not taking, Laatsch presented the testimony of accident recon-structionist Kenneth Baker. Baker had reviewed the plans for the construction of the Kedzie curve in the early 1970s as well as the specifications for the Chevrolet Nomad that the Mihailoviches were driving at the time of the accident, and he had entered the pertinent data from both into a computer simulation program designed to recreate electronically the conditions at the time of the accident. Based on his study and the results of the, simulation, Baker opined that there were no defects in the design of the Kedzie curve. Baker admitted, however, that his knowledge of the actual condition of the curve at the time of the accident was limited. He had not inspected the scene of the accident, he could not describe the condition of the pavement at the time of the accident, he had not reviewed any of the internal correspondence between County employees regarding the condition of the curve, and he did not know whether the surface of the curve was highly polished and what its properties were with respect to the retention of water. Baker also acknowledged that the coefficients of friction he had assigned to the curve for purposes of his review might have been inconsistent with the true condition of the curve. He further allowed that if the pavement of the curve allowed oil to rise to its surface when wet, the antual coefficient of friction at the time of the accident might well have been lower (and the surface thus more slippery) than he had assumed. Tr. 735. At the conclusion of the trial, the jury returned a general verdict in Laatsch’s favor, precipitating this appeal. II. Under Illinois law, in order to prevail on a claim of attorney malpractice, a plaintiff must succeed in proving four elements: (1) an attorney-client relationship giving rise to a duty on the attorney’s part; (2) a negligent act or omission by the attorney amounting to a breach of that duty; (3) proximate cause establishing that but for the attorney’s negligence, the plaintiff would have prevailed in the underlying action; and (4) actual damages. E.g., Woidtke v. St. Clair County, Ill., 335 F.3d 558, 562 (7th Cir.2003) (Illinois law); Cleveland v. Rotman, 297 F.3d 569, 572 (7th Cir.2002) (Illinois law); Mitchell v. Schain, Fursel & Burney, Ltd., 332 Ill. App.3d 618, 266 Ill.Dec. 122, 773 N.E.2d 1192, 1193-94 (Ill.App.Ct.2002). These elements effectively demand that the malpractice plaintiff present two cases, one showing that her attorney performed neg *905 ligently, and a second or predicate “case within a case” showing that she had a meritorious claim that she lost due to her attorney’s negligence. E.g., Ignarski v. Norbut, 271 Ill.App.3d 522, 207 Ill.Dec. 829, 648 N.E.2d 285, 289 (Ill.App.Ct.1995); Sheppard v. Krol, 218 Ill.App.3d 254, 161 Ill.Dec. 85, 578 N.E.2d 212, 214 (Ill.App.Ct.1991); Claire Assocs. v. Pontikes, 151 Ill.App.3d 116, 104 Ill.Dec. 526, 502 N.E.2d 1186,1190 (Ill.App.Ct.1986). Because legal malpractice claims must be predicated upon an unfavorable result in the underlying suit, no malpractice exists unless counsel’s negligence has resulted in the loss of the underlying action. Plaintiff is required to establish that but for the negligence of counsel, he would have successfully prosecuted or defended against the claim in the underlying suit. Damages will not be presumed, and the client bears'the burden of proving he suffered a loss as a result of the attorney’s alleged negligence. Ignarski, 207 Ill.Dec. 829, 648 N.E.2d at 288 (citations omitted). Laatsch, of course, was no longer representing Mihailovich when the Illinois court entered orders barring her from presenting any witnesses and then granting-summary judgment in favor of the County; by that time McKenna had stepped into the case. However, in cases of successive representation, Illinois law recognizes that a prior attorney’s negligence may be the proximate cause of a plaintiffs damages where the plaintiffs underlying claim is no longer viable when his representation ends. See Mitchell, 266 Ill.Dec. 122, 773 N.E.2d at 1194-95. As noted, this is Mihailovich’s theory here. By the time Laatsch withdrew from the case, she asserts, her claim against the County was no longer viable given a trial date six months’ hence and the minimal degree to which Laatsch had prepared the case for trial. As our summary of the testimony above reveals, much of the trial was devoted to the merits of Mihailovich’s negligence claim against the County — the “case within a ease” whose meritorious nature Mihailo-vich had to establish in order to prevail on her malpractice claim against Laatsch. Indeed, all of the issues that Mihailovich has raised on appeal relate to her efforts to prove that aspect of her malpractice claim. In briefing those issues, both parties have assumed that Mihailovich presented sufficient evidence as to the other elements of the claim, including proof that Laatsch breached the duty of care that he owed Mihailovich as her attorney. Having thoroughly reviewed the record, we agree that Mihailovich presented evidence adequate to support a jury finding on that point and the other elements of her malpractice claim. Because the jury rendered a general verdict in Laatsch’s favor, we must therefore assume that it might have found against Mihailovich based solely on her failure to show convincingly that she had a meritorious claim against the County and had lost that claim due to Laatsch’s alleged negligence. Any error that preju-dicially interfered with Mihailovich’s ability to prove that aspect of her malpractice case would therefore require a new trial. We thus turn our attention to the first and primary issue she has raised on appeal, which concerns- the district court’s decision to exclude all evidence of other accidents that had occurred at the Kedzie curve. Because we conclude that the district court erred in excluding that evidence and that a new trial is warranted on that basis,’ we need not resolve the other issues Mihailo-vich has presented: See Fed.R.Civ.P. 61. We shall, however, devote some attention at the conclusion of our opinion to the exclusion of Mihailovich’s proffered expert testimony, a matter that we believe warrants reconsideration on remand. *906 A. Exclusion of Other Accidents As we have noted, the district court granted Laatsch’s motion in limine to exclude all evidence regarding other accidents at the Kedzie curve. See Tr. 247-48, 685-88. Mihailovich’s purpose in offering that evidence was to establish that the curve was dangerous when wet. Tr. 685. But in the court’s view, there was already ample evidence in the record establishing that the surface of the curve became slick when wet, including testimony that the County itself had installed a “Slippery When Wet” sign at the curve in 1984-85 to warn motorists of that condition. Tr. 686. Indeed, this was a common-sense proposition that the jury was likely to appreciate 'without such evidence. Tr. 685-86. Therefore, the court viewed the probative worth of the other-accidents evidence as “minuscule to say the least.” Tr. 685. At the same time, looking to this court’s decision in Kelsay v. Consolidated Rail Corp., 749 F.2d 437, 442-46 (7th Cir.1984), the court saw a significant potential for undue prejudice in this evidence. By her own admission, Mihailovich was not contending that the condition of the roadway was the sole cause of all other accidents at the curve; and the variables that might have contributed to those other accidents (including the direction of travel, weather, speed, vehicle weight, seasonal conditions, and other traffic) were “almost infinite.” Tr. 687. The potential for a collateral trial to occur as to the causes of those accidents thus struck the district court as substantial. Tr. 687-88. Finally, in the absence of evidence that any of the motorists involved in the other incidents had filed suit against the County, the court found no reason to believe that those accidents bespoke negligence on the County’s part. Tr. 688. Although relevant evidence is presumptively admissible, see Fed.R.Evid. 402, a court has the authority to exclude it if the risks posed by the introduction of the evidence significantly outweigh its probative worth. Although relevant, evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury, or by considerations of undue delay, waste of time, or needless presentation of cumulative evidence. Fed.R.Evid. 403. Rule 403 thus calls upon the district court to weigh the need for and probative value of the evidence against potential harm that its admission might cause. Id., Advisory Committee Note (1972); see also, e.g., Manuel v. City of Chicago, 335 F.3d 592, 596 (7th Cir.2003). “ ‘The balancing of probative value and prejudice is a highly discretionary assessment, and we accord the district court’s decision great deference, only disturbing it if no reasonable person could agree with the ruling.’ ” Id., quoting United States v. Thomas, 321 F.3d 627, 630 (7th Cir.2003). In order to show that she would have won her suit against the County but for Laatsch’s alleged malpractice, Mihailovich had to prove that the County, which was responsible for maintaining the Kedzie curve in a reasonably safe condition for its normal and intended uses, negligently discharged that responsibility. See generally DiBenedetto v. Flora Tp., 153 Ill.2d 66, 178 Ill.Dec. 777, 605 N.E.2d 571, 574 (Ill.1992); First Nat’l Bank in DeKalb v. City of Aurora, 71 Ill.2d 1, 15 Ill.Dec. 642, 373 N.E.2d 1326, 1331 (Ill.1978); 745 Ill. Comp. Stat. 10/3-102, 10/3-105; see also, e.g., Storen v. City of Chicago, 373 Ill. 530, 27 N.E.2d 53, 55 (Ill.1940); Santelli v. City of Chicago, 222 Ill.App.3d 862, 165 Ill.Dec. 277, 584 N.E.2d 456, 459 (Ill.App.Ct.1991). Her theory in relevant part was that years of wear had polished the surface of the Kedzie curve, so that when precipitation dampened the pavement and oil rose to its surface, the pavement became extremely *907 slick; thus, by failing to make appropriate repairs or changes to the road, notwithstanding requests that it do so, the County had permitted a hazard to arise and persist that caused motorists to lose control of their vehicles when the pavement was wet. See, e.g., Tr. 771, 813. In this context, there is no doubting that the other accidents proffered were relevant to establish that the worn pavement of the curve constituted a hazard- for motorists. If, as Mihailovich theorized, the Kedzie curve was dangerous and had been for some time, then one would naturally expect there to have been other accidents besides hers over the years. In fact, as the record in this case reveals, the number of accidents that occur at a particular location is one of the first indicators that officials concerned with roadway safety look to for an indication that the location is dangerous. We know, of course, from the proffers of excluded testimony that the reason why the Olympia Fields board of trustees passed the 1984 resolution calling on the County to address the curve was the high number of accidents that had occurred at that curve; employees of the County’s Highway Department themselves had cited the high accident rate in calling for studies of the curve in 1983, more than a year before the Olympia Fields board passed its resolution. More than this, the frequency with which accidents occur would appear to be an accepted reference point in civil engineering for the purpose of identifying potentially hazardous roadway locations. The February 21, 2001 report of civil engineer Paul Box, which Mihailovich submitted in opposition to Laatsch’s motion to exclude the evidence of prior accidents, noted that although -there is no national standard as to the number of accidents necessary in order to label a particular location as “hazardous,” experts in the field had conducted research and developed a graph indicating that the occurrence of more than two accidents over a period of three years signaled that the location was “hazardous,” while the occurrence, of more than ten- accidents in three years indicated that it was “very hazardous.” R. 105, Box Report at 6 ¶ 18, citing J.L.Taylor and H.T. Thompson, Determining Hazardousness of Spot Locations, Transportation Research Record 630 (1977); see also J.I. Taylor and H.T. Thompson, Federal Highway Administration, Identification of Hazardous Locations, Report Nos. FHWA-RD-77-81 through 83 (1977) (identifying accident rate as among factors relevant to hazard-ousness of particular location). Assuming that Box has correctly interpreted and applied this criterion, the total of ninety-four accidents that occurred at the Kedzie curve in the four and one-half years surrounding the Mihailovich accident obviously would exceed the Taylor & Thompson threshold for a “very hazardous” location by many multiples. If the frequency of accidents at a particular location constitutes a category of data that both experts in civil engineering and officials charged with the maintenance of roadways consider important, then that information would be highly relevant to the, jury’s own assessment of whether or not the curve posed a hazard. 9 *908 Indeed, our own precedents recognize that other accidents are generally deemed admissible both to prove the existence of a defect or danger in a location or a product and to show that the defendant had notice of the defect or danger, so long as the other accidents are “substantially similar” to the accident at issue in the litigation. Weir v. Crown Equip. Corp., 217 F.3d 453, 457 (7th Cir.2000); Ross v. Black & Decker, Inc., 977 F.2d 1178, 1185 (7th Cir.1992); Estate of Carey by Carey v. Hy-Temp Mfg., Inc., 929 F.2d 1229, 1235 n. 2 (7th Cir.1991); Nachtsheim v. Beech Aircraft Corp., 847 F.2d 1261, 1268 (7th Cir.1988); see also Walker v. Trico Mfg. Co., 487 F.2d 595, 599 (7th Cir.1973); Lever Bros. Co. v. Atlas Assur. Co., 131 F.2d 770, 777 (7th Cir.1942). Where, as here, the other accidents are offered as proof of a dangerous condition, 10 courts insist on a greater degree of similarity between those accidents and the one at issue than they do when the other incidents are offered solely to establish notice of a potential danger to the defendant. Nachtsheim, 847 F.2d at 1268 n. 9. This is not to say, however, that the accidents must correspond with one another in every detail. “In applying this standard, we have emphasized that ‘substantially similar’ does not mean ‘identical,’ and that the range between similar and identical is a matter to be addressed on cross-examination.” Buscaglia v. United States, 25 F.3d 530, 533 (7th Cir.1994), citing Carey, 929 F.2d at 1235 n. 2. The particular defect or danger alleged by the plaintiff will serve to define the degree of commonality that there must be among the accidents in order for them to be considered substantially similar. Jackson v. Firestone Tire & Rubber Co., 788 F.2d 1070, 1083 (5th Cir.1986). Looking to the established facts underlying both the plaintiffs accident and the other accidents she has proffered, the court must consider whether those facts reasonably support an inference that all of the accidents share a common cause — i.e., the danger that the plaintiff has alleged. See Nachtsheim, 847 F.2d at 1269. If the facts do support such an inference, then the other accidents are admissible barring other factors suggesting that their admission will result in undue prejudice. See id. Laatsch suggests at the outset that Mi-hailovich did not make enough of a record regarding the other accidents at the Ked-zie curve to show substantial similarity between those accidents and her own, but we disagree. The record before us includes a year-by-year tabulation, prepared by Unger, then the chief of police of Olympia Fields, of the accidents that had occurred at the Kedzie curve from January I, 1983 through July 15, 1987; 11 police reports for some (but by no means all) of the accidents that had occurred at the curve in 1981 through 1986; and the compilations and analyses of the accident data reflected in the reports of both Box and *909 Berg. Mihailovich cited and presented this information to the district court prior to trial, both in opposing (successfully) Laatseh’s motion for summary judgment and in opposing (unsuccessfully) the County’s motion in limine to exclude- evidence concerning the other accidents. See, e.g., R. 62 (Plaintiffs Local Rule 56.1 Statement of Undisputed Material Facts) Exs. -6 (Berg Report), 24 (Kedzie curve accident data); R. 105 (Plaintiffs Response to Defendant’s Motion In Limine No. 2) Attachment (Box Report). The police reports for some of the accidents that occurred at the curve in 1981 through 1986 were attached to Mihailovich’s malpractice complaint. R. 1 Ex. J. Collectively, this data, coupled with the other evidence of record, tends to show that: (1) the other accidents occurred along the Kedzie curve; (2) they occurred relatively close in time to the Mihailovich accident; (3) the condition of the curve was no different at the time of the Mihailovich accident than it was at the time of the other accidents (the curve itself had not been resurfaced since the 1970s, and would not be resurfaced until 1988 or 1989); (4) a majority of the accidents (roughly three-quarters of the total from 1983 through mid-July 1987) occurred on wet pavement, just as the Mihailovich accident did; and (5) although the police reports (and Berg’s analysis of the reports for twenty-four of the twenty-nine accidents that occurred in the eighteen months immediately prior to the Mihailovich accident) reveal variations in precisely where each vehicle left the roadway and what happened as a result (for example, some vehicles collided with oncoming traffic, while others came to rest on either side of the roadway), the reports reveal a. pattern of drivers losing control of their vehicles while on the curve, particularly when the roadway was wet. Keeping in mind the particular danger that Mihailovich has alleged was present at the curve — excessive slipperiness when the pavement was wet — we are satisfied that the other accidents are substantially similar to the Mihailovich accident and therefore are admissible as proof that the curve posed a hazard to motorists. We readily grant Laatsch’s point that there is a great deal that the record does not tell us about these accidents. The gross accident data compiled by Unger, for example, reveals only the date and time that each accident occurred, the number of vehicles involved, whether the accident resulted in a fatality, injury, or property damage, the weather conditions at the time of the accident, and the condition of the road surface. See R: 62 Ex. 24. Although Unger’s compilation was apparently derived from police reports regarding these accidents (the compilation lists a report- number for each incident), the record before us does not include all of these reports. The thirty-five or so police reports attached to the complaint reveal a good deal more information about those accidents, including in particular the basic facts of each accident, and also the identities of the motorists involved, the types of vehicles they were driving, where the accident occurred, whether alcohol was involved, the apparent physical condition of the motorists, and so forth. But the reports do not recount precisely what each driver was doing immediately before the accident occurred (how fast he was driving, for example, whether and how he applied his brakes, and so forth), they do not establish with precision where along the S-curve each incident took place, and they do not purport to assign fault for the accident or to identify each of the factors that may have contributed. Even so, the data do reveal a distinct pattern to the other accidents: they occurred at an average rate" of twenty per year in the four and one-half years surrounding the Mihailovich accident (January 1983 through mid-July 1987); a substantial majority (at least sev *910 enty-five percent) occurred when the pavement was wet; and (per the police reports) they almost uniformly involved a loss of control by a driver with the result that the vehicle slid, spun, or skidded into a vehicle proceeding in the opposite direction, into a vehicle stopped directly ahead of him, or off the road altogether. Berg’s analysis also indicates that in twenty-one of the twenty-four accidents occurring the eighteen months prior to the Mi-hailovich accident for which records were available, the subject vehicle had been proceeding southbound into the S-eurve. Given the frequency and regularity with which motorists, and in particular southbound motorists, were losing control of their vehicles at the curve when the pavement was-wet, it is a fair inference that both the condition of the curve’s pavement when wet and the nature of the southbound approach to the curve, played some role in these other accidents. The same inference can be drawn from what we know about the Mihailovich accident: it involved a southbound motorist, it occurred on the curve, when the pavement was wet, and it involved a loss of control that resulted in the van leaving the road. In these respects, the other accidents may be deemed substantially similar to the Mi-hailovich accident. Evidence concerning the other accidents was therefore admissible, barring a reason which would have justified its exclusion under Rule 403. As our summary of the district court’s rationale for excluding the other accidents indicates, the court had two basic concerns: it saw the probative worth of evidence concerning the other accidents as minimal, while the potential for prejudice in such evidence to be great. As we explain below, the court was off-base in its first concern, and in articulating its second concern, the court appears to have been led astray by our fact-specific holding in Kelsay. In weighing the probative value of the evidence, the district court described the relevance of the prior accidents as “minuscule” (Tr. 685), reasoning that other evidence amply established that the surface of the Kedzie curve was slippery when wet and that jurors in any event would have appreciated that fact as a matter of common sense. In that observation, the court was mistaken. Mihailovich’s theory as to the hazard posed by the curve was not simply that it was slippery when wet, but that it was unusually and dangerously so, i.e., to a degree that motorists would not ordinarily anticipate. That so many other accidents occurred on the curve, and that approximately three-quarters of them occurred when the pavement was wet, lends support to the notion that the curve was unreasonably slick when wet. Indeed, the fact that the accidents continued unabated even after the County installed a “Slippery When Wet” advisory sign at the curve in 1984 might be construed as an indication that the degree of slipperiness exceeded even the expectations of motorists who were placed on notice of the potential hazard. The reaction of local officials to the history of accidents occurring at the curve likewise could be construed as a sign that the degree of danger was unusual. Indeed, we must point out that, in view of the factual backdrop of the case, proof of the other accidents that occurred on the curve was one of the few ways in which Mihailovich could have- established that the roadway was in an unreasonably dangerous condition. A more direct way of proving the supposed hazard would have been to measure objectively the slipperiness of the pavement, as by ascertaining the coefficient of friction on the surface of the curve when it was wet. But that was no longer possible by the time Laatsch withdrew from representing Mihailovich — the curve had been reconstructed and resurfaced between 1988 and 1989, and Laatsch *911 had not engaged an expert to inspect the curve and conduct appropriate testing to determine the coefficient of friction prior to that time. So Mihailovich had to pursue more indirect ways of establishing the alleged hazard. The jury of course knew that the Olympia Fields trustees believed the curve to be hazardous and that they had conveyed that concern to the County by way of the 1984 resolution. But the district court permitted the jury to consider that resolution solely as proof that the County had notice of a potential hazard, not as proof that the roadway was, in fact, dangerous. Tr. 288. Beyond this, the jury heard only abbreviated testimony that the surface of the roadway had a tendency to become oily and slick when wet. Consequently, there was little, if any, evidence indicating whether or not the surface of the Kedzie curve in fact became unusually or dangerously slippery when wet and what effect the wet surface had upon motorists who negotiated the curve. Not surprisingly, this was a point that Laatsch’s counsel, seized upon in his closing argument to the jury. E.g., Tr. 795. So, when it engaged in the Rule 403 balancing process, the court did not give the other-accidents evidence the weight it deserved in terms of its probative value. At the same time, on the other side of the scale, the court in assessing the potential reasons for excluding the evidence did not properly focus on the key consideration, which is substantial similarity between the other accidents and the one at issue in this case. Instead, referencing our opinion in Kelsay as its primary guidepost, the court found that the justifications that Kelsay cited for the exclusion of other-accident evidence in that case also weighed in favor of exclusion here. But our decision in Kelsay does not mandate the exclusion of the other-accident evidence tendered here. Kelsay, which involved the collision of a train with an automobile at a railroad crossing, affirmed (over a dissent) the district court's decision to exclude evidence of two fatal accidents that had occurred at the same crossing twelve and thirty years earlier. We- cited four factors that in our view supported the exclusion of the prior-accident evidence: (1) key differences in the conditions and circumstances of the prior accidents, (2) the absence of evidence that lawsuits had been filed on behalf of the deceased motorists in the prior accidents, which suggested that the drivers themselves may have been at fault, (3) the danger that the jury might infer from the prior accidents alone that the railroad crossing at issue was hazardous, and (4) the plaintiffs ability to show through other evidence that the conditions at the crossing had remained unchanged for a number of years, so that the potential danger would have been evident to the railroad. 749 F.2d at 443. Nothing in our opinion suggested that this was an exhaustive list of pertinent considerations or that the factors we identified would apply with the same force in any case involving other-accident evidence. It bears mentioning in that regard that we had first identified these factors in Gardner v. Southern Ry. Sys., 675 F.2d 949, 952 (7th Cir.1982), as being among the “many factors” that the district judge “undoubtedly” had considered in excluding other-accident evidence; Gardner itself did not cite any precedents that focused on these factors. The governing standard, however, as cases both before and after Gardner and Kelsay make clear, was and is whether the other accidents are substantially similar to the accident at issue in the instant suit. See, e.g., Buscaglia, 25 F.3d at 533; Carey, 929 F.2d at 1235 n. 2; Walker, 487 F.2d at 599; Lever Bros., 131 F.2d at 777. Moreover, Kelsay involved a materially distinct set of circumstances. First, the prior accidents involved in that case were, *912 as the court noted, quite remote in time from the incident under litigation, and in each case there was at least one apparent circumstance suggesting that something-other than the hazardous nature of the railroad crossing was at fault. In one of the instances, the windows of the car had been heavily soaped, which obviously would have interfered with the driver’s view of the crossing; and in the other instance, the motorist had arrived at the crossing after the train did, suggesting that the driver’s view of the crossing and the oncoming train was not a cause of the collision. By contrast, the accidents that Mihailovich sought to introduce occurred with regularity in the years just before and just after her accident, with no grossly apparent distinctions that would explain away the condition of the roadway as a cause of those accidents. Second, the prior accidents involved in Kelsay had both resulted in fatalities, so the lack of evidence that lawsuits had been filed in those cases was more telling than it would be here, where the vast majority of the other accidents did not result in deaths. As litigious as Americans may be, one cannot reasonably expect every motorist who loses control of her car on a slippery road to file a lawsuit. Moreover, whether or not another accident resulted in litigation really is of very limited use in weighing the probative value of that accident. As Judge Posner pointed out in his dissent in Kelsay, the question with respect to other accidents is not whose fault they were in a legal sense but what they can tell the jury about the dangerousness of the location where they occurred. 749 F.2d at 451 (Posner, J., dissenting). Third, although it is quite possible that a jury might infer from the prior accidents themselves that the Kedzie curve was in fact dangerous, that inference would not be as inappropriate as it might have been in Kelsay. The other accidents proffered here, which are neither isolated nor remote in time from the Mihailovich accident, fall into a pattern suggesting that the Kedzie curve indeed did become dangerously slippery when wet. That is an inference that the Village and some County employees themselves drew from the pattern, and it is also an inference that at least two experts in the field, Box and Berg, drew from a review of the evidence. In Box’s view, for example, the fact that a distinct majority of the reported accidents occurred on wet pavement “represents an extremely unusual circumstance” that pointed to the need for the pavement to be addressed by the County. R. 105, Box Report at 7 ¶ 5. Moreover, it appears that when the curve was reconstructed in 1988-1989, the accident rate dropped dramatically to at most a few accidents per year. Although the district court excluded evidence concerning the reconstruction pursuant to Fed.R.Evid. 407 — and we assume that it did so correctly — the impact that the reconstruction had on the occurrence of accidents nonetheless is something that we may consider in determining whether or not the court abused its discretion in excluding evidence of other accidents. The noteworthy reduction in the number of accidents following reconstruction tends to confirm that the prior condition of the curve in fact did have something to do with the accidents that occurred there. All of this indicates that evidence concerning the other accidents would not have been unduly prejudicial in the sense that it might have led the jury to draw unwarranted inferences from those accidents about the condition of the curve. Fourth, although exclusion of the other-accidents evidence did not prevent Mihail-ovich from proving that the condition of the Kedzie curve had been evident for some time prior to her accident — such that the County was on notice of the potential danger — that is really beside the point. *913 Mihailovich, as the district court recognized, was not seeking to introduce the evidence for purposes of notice, but to prove that the pavement of the curve was dangerously slippery when wet. Almost none of the evidence that was admitted by the court in this case went to that important point. Exclusion of the other-accidents evidence deprived Mihailovich of highly probative proof of the condition of the curve and its impact on motorists. The district court found it noteworthy that Mihailovich, by her own admission, was not asserting that the condition of the roadway was necessarily the sole cause of the other accidents and that the number of other variables that might have contributed to the other accidents — including the direction of the vehicle, the weather, the speed, vehicle weight, seasonal conditions, and other traffic — was “almost infinite.” Tr. 687. That the condition of the curve was not necessarily the sole cause of the other accidents was an entirely reasonable and prudent concession for Mihailovich to make, given that the available data concerning those accidents (including the police reports) do not undertake to assign and/or apportion fault or causation. But so long as the evidence regarding those accidents reasonably supports the inference that the condition of the roadway was at least a cause of those accidents — and we believe it does — then they remain probative of whether the road was unreasonably hazardous and whether the condition of the road was a cause of the Mihailovich accident as well. Likewise, the number of variables that might have contributed to the other accidents does not demand their exclusion. The standard for admission, as we have pointed out, is not identity or near identity but substantial similarity. We are satisfied that the record reveals the other accidents to have been substantially similar to the Mihailovich accident. Laatsch would certainly be free to highlight for the jury the potential differences in the types of variables that the district court cited, but those differences go to the weight of the other-accidents evidence rather than its admissibility. Finally, although the district court was (rightly) concerned about the potential for testimony regarding the other accidents to devolve into a collateral trial over the circumstances in those accidents, there are alternatives to exclusion of any and all evidence about those accidents that would suffice to avoid this peril. The occurrence of other accidents at the curve, and the range of circumstances that might have caused them, is not so complex a subject as to require an overly expansive eviden-tiary record. Mihailovich could establish the pertinent facts regarding these accidents without delving deeply into the circumstances of each occurrence, and Laatsch similarly, could identify the potential distinctions between those accidents and the Mihailovich accident as well as the types of conditions besides the road surface that might have contributed to the other accidents without having to litigate who was at fault in each instance. As always, the court remains free to establish reasonable limits on this line of Inquiry in order to prevent it from becoming a sideshow. Although, for all of these reasons, we conclude that the district court abused its discretion in excluding all evidence regarding other accidents at the Kedzie curve, we must also consider whether the error was prejudicial. Pursuant to Federal Rule of Civil Procedure 61, an error in the admission or exclusion of evidence does not supply a basis for setting aside a verdict unless the error was “inconsistent with substantial justice.” “Our cases hold that ‘evidentiary errors satisfy this standard only if a significant chance exists that they affected the outcome of the trial.’ ” Shick v. Ill. Dep’t of Human Servs., 307 *914 F.3d 605, 611 (7th Cir.2002), quoting Hasham v. Cal. State Bd. of Equalization, 200 F.3d 1035, 1048 (7th Cir.2000); see also Harris v. City of Chicago, 266 F.3d 750, 755 (7th Cir.2001). In this case, we believe that there is a significant chance that the exclusion of the other-accident evidence affected the outcome of the trial. In order to prove her “case within a case” — i.e., that she would have prevailed on her claim against the County but for Laatsch’s alleged negligence — Mihailovich was required to present evidence that the Kedzie curve was dangerous. As we have explained, the most direct method of establishing the hazard — having an expert inspect the curve and conduct tests to determine the condition of its pavement when wet — was foreclosed to her. The Kedzie curve as it was at the time of the Mihailovich accident in 1986 no longer exists; it was reconstructed in 1988-1989, ten years before she filed the instant suit. During the time that Laatsch represented Mihailovich prior to the reconstruction, he did not cause any testing or inspections to be done that might have preserved data about the coefficient of friction on the curve’s pavement when wet. So Mihailovich was confined to indirect means of proving that the curve was dangerous. Primary among these methods was proof that other accidents occurred regularly at the curve when the pavement was wet. Barred from introducing any evidence regarding such accidents, Mihail-ovich was left solely with proof that Village officials and County employees perceived the curve to be a potential hazard and no explanation as to why they held that perception. Had the jury known that both the Village and some employees of the County perceived the curve to be dangerous because motorists were regularly losing control of their vehicles when the pavement was wet, it might well have assessed her prospects of prevailing against the County differently. We therefore conclude that Mihailovich is entitled to a new trial at which she may introduce proof of the other accidents that occurred at the Kedzie curve. The district court, of course, remains free to establish reasonable limits on the type and extent of other-accident evidence that Mihailovich may introduce. For example, to the extent that police reports reveal at least the rudimentary details about each accident, and in that way permit the parties to establish points of comparison with the Mihailovich accident, the court might decide to permit evidence solely regarding those accidents for which such reports are available. 12 We also wish to reiterate that the admission of this line of evidence need not portend a mini-trial as to each and every one of the other accidents. Mihailo-vich may present data regarding the number and basic nature of these accidents, and Laatsch may identify the limits of the facts known about those accidents, the distinctions between those accidents and the Mihailovich accident, and the range of circumstances other than the condition of the curve that might account for those accidents, without inquiry into this area becoming unduly prolonged. B. Exclusion of Expert Testimony by Berg In advance of trial, Laatsch moved to bar William Berg from testifying as an expert pursuant to Fed.R.Evid. 702. Mi-hailovich had identified Berg as a civil engineer from whom she intended to elicit testimony as to, among other things, the condition of the Kedzie curve at the time of the accident, the County’s asserted negligence in maintaining the curve, and the *915 likely cause of the Mihailovich accident. Having reviewed a report summarizing Berg’s opinions on these subjects and. having deposed him on multiple occasions, Laatsch contended that Berg should not be permitted to testify because his analytical methodology did not satisfy the requirements outlined in Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993), and Kumho Tire Co. v. Carmichael, 526 U.S. 137, 119 S.Ct. 1167,143 L.Ed.2d 238 (1999). Berg expressed three basic opinions with respect to the condition of the Kedzie curve at the time of the Mihailovich accident and the cause of that accident. In sum, Berg opined: (1) The County was not maintaining the Kedzie curve in a reasonably safe condition as of July 1, 1986.(2) The hazardous nature of the curve resulted in part from the widening and resurfacing of Kedzie Avenue to the north of the S-curve in 1981. As a result of that project, southbound motorists were approaching the curve at speeds greater than the advisory speed of thirty miles per hour for the curve itself and were engaging in frequent applications of their brakes after they entered the curve. The higher speeds and more frequent brake applications resulted in a higher frictional demand in the curve. When coupled with low or reduced skid resistance in the curve when the pavement was wet, this posed a safety hazard to motorists. (3) The configuration of both the S-curve itself as well as the southbound approach to the curve were substantial causal factors of the Mihailovich accident. The prevailing conditions at the time of the accident and the sequence of events were substantially similar to the continuing pattern of wet-pavement accidents involving southbound vehicles. See R. 109 Ex. B; Tr. 620-21. Berg, a civil engineer, employed a “human factors” model in reaching these conclusions. Broadly speaking, a human factors analysis focuses on the interaction between human behavior and the design of a machine, product, or in this case, a particular roadway. An expert engaging in such an analysis will consider whether, in light of predictable human behavior, the design or condition of the subject item poses a potential hazard. See, e.g., Richard J. Kholmn, Using the Human Factors Expert in Civil Litigation, 40 AM. JUR. TRIALS 629 §§ 1-2. Berg’s opinions reflect this approach to the extent they incorporate certain notions about how motorists likely would have been behaving as they entered the Kedzie curve as it existed at the time of the Mihailovich accident. One of the key assumptions underlying Berg’s analysis is that southbound motorists typically would have entered the curve at speeds of between thirty-five and forty-five miles per hour. R. 109 Ex. B; R. 62 Ex. 6 at 3 ¶ 6; Tr. 620. That assumption was based on two circumstances. First, the posted speed limit on Kedzie Avenue to the north of the reverse curve was forty-five miles per hour. That limit dropped to thirty-five miles per hour at 203rd Street, or approximately 1700 feet north of the S-curve. Id. Second, the 1981-82 widening and resurfacing of Ked-zie Avenue to the north of the curve had created a roadway that would cause drivers to feel safer and to drive faster. In short, due to the higher posted speed limit and more commodious roadway immediately to the north of the curve, southbound motorists were likely to be driving at a rate somewhere between thirty-five miles per hour and forty-five miles per hour as they entered the Kedzie curve, through which the advisory speed was thirty miles per hour. Id. ¶¶ 6-7; Tr. 626-27. The curve itself required motorists to adjust both the path and speed of their vehicles. In Berg’s view, the location, configuration, and other characteristics of the *916 Kedzie curve gave rise to whát he described as an “expectancy violation” and a “detection/recognition” problem for motorists. Id.; R. 109- Ex. B ¶ 3; Tr. 620. Berg defined an expectancy violation as an unusual situation that catches the motorist by surprise. Tr. 575-76. He believed that the Kedzie curve posed an expectancy violation for southbound motorists in the sense that it required motorists to suddenly both slow and turn their vehicles after having traveled along a stretch of wide, straight roadway. R. 62 Ex. 6 at 3 ¶ 7. A detection/recognition problem is one that a motorist will have difficulty recognizing. Tr. 576. Berg thought that the Kedzie curve also posed this type of problem, because roadside vegetation coupled with a downward grade' of 2.5 percent beginning near the middle of the north curve made it difficult for motorists to visualize the location and tightness of the curve. R. 62 Ex. 6 at 3 ¶ 7. Under these circumstances, Berg explained, motorists required warnings to alert them to the presence and location of the reverse curve, the need to reduce their speed to thirty-five miles per hour or less, and also to the fact that the surface of the curve was slippery when wet. Id. at 3 ¶ 8. The need for the latter warning was demonstrated by the pattern of accidents that had occurred on the curve when the pavement was wet. Id. at 3 ¶ 4. As we know, following the field studies that the County conducted in 1984 in response to the concerns expressed by its own employees and by the Village of Olympia Fields about the accident rate at the curve, additional signage had been installed to alert motorists to each of these circumstances. In Berg’s opinion, those signs were sufficient to provide motorists with the information they needed about those aspects of the curve. Id. ¶ 8. However, accident records for the eighteen-month period beginning in January 1985 (after the additional warning signs were installed) and ending in June 1986 revealed that the wet pavement safety problem continued unabated. Of the twenty-nine accidents reported to have occurred at the curve during that period, twenty-two had taken place on wet pavement and the remaining seven had occurred when the pavement was covered with ice or snow. Based on the police reports available for twenty-four of these accidents, Berg had prepared a collision diagram indicating what had happened to the vehicles involved in these accidents. According to Berg, the diagram demonstrated a clear pattern: twenty-one of the twenty-four vehicles involved had been proceeding southbound on Kedzie Avenue and, after entering the curve, had either left the roadway on the inside or outside of the curve or had crossed the center line and collided with a northbound vehicle. Id. at 3-4 ¶ 9; see also Tr. 621-22. Because Berg was relying on the police reports for data about these accidents, he lacked information as to what the driver in each case had done (in particular, whether he or she had applied the vehicle’s brakes) at a particular moment. Tr. 576-77. But what the available data made clear to Berg was that in each case the driver had lost control of the vehicle after entering the curve. R. 62 Ex. 6 at 3^4 ¶ 9. Berg acknowledged that the available information about the Mihailovich accident did not permit him to determine where, precisely, the Mihailovich van had left the road. Tr. 630-31. But Berg indicated that what was known about the Mihailo-vich accident placed it “dead center” within the pattern that he discerned in the history of accidents at the curve — that is, the vehicle was proceeding southbound on Kedzie Avenue, the pavement was wet, and the driver lost control of the vehicle after entering the curve. R. 109 Ex. B ¶ 3; Tr. 578, 620, 630. *917 As we have noted, Berg opined that two factors in combination were the probable cause of this pattern of accidents: (1) low or reduced pavement skid resistance on the curve, and (2) high frictional demand for vehicles traversing the curve. R. 109 Ex. B ¶ 2. Berg surmised that the surface of the curve had low or reduced skid resistance based on the fact that a distinct majority of the accidents occurred when the pavement was wet. Tr. 574-75. He thought that the reduction in skid resistance was likely due either to a polishing of the pavement resulting from a high volume of cornering traffic and/or to the poor drainage of surface water resulting from the low superelevation rate of the curve and the descending 2.5 percent grade. Tr. 584-85. 13 Berg attributed the high frictional demand to the relatively high speeds (thirty-five to forty-five miles per hour) of vehicles approaching the curve, frequent brake applications by motorists after they entered the curve, and the reduced effectiveness of the advisory signs present at the curve. Again, Berg assumed that southbound motorists were likely traveling at a rate of at least thirty-five to forty-five miles per hour. on approach to the curve based on the higher speed limit and the wider and straighter roadway on Kedzie Avenue north of the curve. Tr. 626-27. He assumed that frequent brake applications occurred on the curve given that motorists were, by his estimate, approaching the curve at speeds in excess of the advisory speed for the curve itself (thirty miles per hour) and given the sheer number of cars passing through the curve on a daily basis (6,200 cars per day). Tr. 573-74. Finally, he assumed that the warning signs were not as effective as they otherwise might have been because the curve did not appear to pose a problem when dry and because the recommended speed of thirty miles per hour through the curve was not dramatically lower than the speed of thirty-five to forty-five miles per hour at which he believed motorists likely were approaching the curve. See R. 62 Ex. 6 at 3-4 ¶¶ 9-10; Tr. 621-22: Consequently, motorists who were used to driving through the curve when it was dry might not have paid as much attention as they should have to the warning signs posted at the curve. Appropriate countermeasures to address the ongoing wet-pavement safety problem, in Berg’s view, would have included applying a skid-resistant pavement overlay within the curve and/or to increase the superelevation of the curve. In addition, the County might have conducted a speed study of southbound traffic to determine whether there was a need for either additional enforcement of the regulatory speed limit or for measures, such as a flashing beacon or rumble strips, to reinforce the warnings that were already in place. Id. at 4 ¶ 11; see also Tr. 628. Berg noted that after the Mihailovich accident occurred, the advisory speed through the curve was reduced to twenty-five miles per hour. Yet, accident records for the year following the accident again showed that the wet-pavement problem persisted. R. 62 Ex. 6 at 4 ¶ 12. What ultimately did solve the problem, Berg noted, was the reconstruction of the curve in 1989. The key aspects of the reconstruction which, to Berg’s mind, improved the safety of the curve were a new road surface coupled with a larger curve radius and superelevation at the north curve which reduced frictional demand on the pavement. Berg noted that as a result of these changes, drivers could safely navi *918 gate the curve at speeds of up to forty-five miles per hour. Id. ¶ 13; Tr. 624-25. Finally, Berg opined that the County had been negligent in failing to identify and address the root causes of the problem at the Kedzie curve sooner than it did. He indicated that since the 1960s, it has been a standard civil engineering practice to prepare collision diagrams for roadway segments that reflect unusually high frequency of accidents. In his view, the County’s Highway Department should have had, but did not have, appropriate procedures in place prior to the Mihailo-vich accident in order to ensure that collision diagrams were prepared when the circumstances warranted. R. 62 Ex. 6 at 4 ¶ 14. Berg believed that had the County prepared such a diagram for the Kedzie curve in 1983, when the high accident rate for that location was first noticed by County employees, the timetable for identifying the cause of the accidents and for implementing improvements sufficient to rectify the problem could have been advanced so that those improvements would have been made prior to. the Mihailovich accident in 1986. Had the curve been reconstructed prior to July 1,1986 as it later was in 1989, for example, the Mihailovich accident likely would not have occurred, in Berg’s' opinion. Id. at 4-5 ¶ 15. Having reviewed Berg’s three principal opinions, his deposition testimony, and his testimony on voir dire, the district court barred him from testifying as an expert but gave little explanation of its reasons for doing so. After noting the factors identified' as pertinent in Daubert, the court simply stated that “it’s 'clear that none of these factors are even remotely present to support [Berg’s] purported analysis and methodology in reaching these three conclusions or opinions .... ” Tr. 590-91. When it will assist the trier of fact, Rule 702 permits a witness whose knowledge, skill, experience, training, or education has granted him expertise in a particular field to testify in the form of an opinion regarding one or more issues in the case, provided that “(1) the testimony is based upon sufficient facts or data, (2) the testimony is the product of reliable principles and methods, and' (3) the witness has applied the principles and methods reliably to the facts of the case.” Fed.R.Evid. 702. The district court functions as a gatekeeper with respect to testimony proffered under Rule 702 to ensure that the testimony is sufficiently reliable to qualify for admission. See Kumho Tire Co., 526 U.S. at 147, 119 S.Ct. at 1174. The Supreme Court in Daubert identified five criteria that bear on that assessment, including whether the theoretical framework -or technique underlying the witness’s testimony (1) is subject to verification through testing, (2) whether it has been subjected to peer review and publication, (3) what its known or potential rate of error is, (4) whether there are standards controlling its application, and (5) whether it is generally accepted within the relevant expert community. 509 U.S. at 593-94, 113 S.Ct. at 2796-97. “[T]his list is neither definitive nor exhaustive, but rather flexible to account for the various types of potentially appropriate expert testimony.” Deputy v. Lehman Bros., Inc., 345 F.3d 494, 505 (7th Cir.2003), citing Kumho Tire Co., 526 U.S. at 141, 119 S.Ct. at 1171. We review de novo whether the district court correctly applied the Daubert framework, but assuming that it did so, we review the court’s decision to admit or exclude testimony under Rule 702 for abuse of discretion. E.g., Deputy, 345 F.3d at 505. Review of the district court’s decision to exclude Berg’s testimony is hampered by the fact that the court simply stated without any explanation that Berg’s methodology satisfied none of the criteria identified in Daubert. We do not doubt that the *919 court was aware of the relevant considerations, in view of the fact that the court correctly recited each of the Daubert factors. But given that the Daubert framework is a flexible one that must be adapted to the particular circumstances of the case and the type of testimony being proffered, a court excluding expert testimony would do well to articulate with reasonable specificity the reasons why it believes the testimony is insufficiently reliable to qualify for admission under Rule 702. The lack of such explication makes it difficult to meaningfully review the lower court’s decision. It is not immediately apparent here, for example, in what respects Berg’s methodology departed from those of other civil engineers and why the analysis underlying his opinions is not reliable. Berg appears to have answered each of the questions posed to him on voir dire about his methods and assumptions and identified the data that he relied upon. The admissibility of Berg’s testimony therefore should be freshly addressed on remand. Although we express no opinion as to the appropriate outcome of that analysis, we do make the following observations, which the district court should consider in evaluating the reliability of Berg’s methodology. First, we note that the human factors analysis that colored Berg’s opinions vis-a-vis the Kedzie curve is not novel. It is a recognized analytical approach that is applied in a variety of contexts and may yield legitimate insights as to the hazards that particular products and situations (including roadways) may pose in light of predictable human behavioral patterns. See Kohlman, supra, Using the Human Factors Expert in Civil Litigation, 40 Am. Jus. TRIALS 629. Second, a key assumption underlying Berg’s evaluation of the curve is that drivers approaching the Kedzie curve from the north frequently would have done so at speeds in excess of thirty or thirty-five miles per hour and thus would have applied their brakes as they entered the curve in order to slow down. The district court appears to have had some doubt as to the legitimacy of this assumption. See Tr. 569-74. However, given that the stretch of Kedzie Avenue immediately to the north of the S-curve was more straight, wide, and recently paved than the curve itself, not to mention subject to a higher posted speed limit of forty-five miles per hour, common sense arid experience suggests that southbound motorists frequently might have been traveling at speeds in excess of thirty-five miles per hour as they entered the curve. More to the point, the assumption that they were is consistent with the contemporaneous observations of both County engineers and Olympia Fields police. Recall that when Joseph Marsik, the County Highway Department’s Chief Engineer of Maintenance, forwarded Frank Reno’s memorandum about the possible problem with the curve to Chief Engineer of Transportation and Planning Arthur Kaindl, he wrote, inter alia, “It is our belief that the traffic is going too fast on the straight roadway, entering the curve, and then has a problem negotiating the curve.” Tr. 500. The proffer of Olympia Fields police officer Mark Fazzini likewise reveals a perception that motorists were approaching the curve at speeds substantially in excess of the advisory speed of thirty miles per hour. According to Mihailovich’s counsel, Fazzini would have testified: [T]he police department literally set up speed traps at the highway to try to get drivers to slow down because they knew that cars coming from the north down into this curve had been traveling through a semirural area and that, as we know in this case, Kedzie Avenue just north of the curve, was resurfaced in 1981, 82 at Biden and this resurfacing *920 went right up to the beginning of this north curve and the cars would come into the curve at a high rate of speed because of their assurance they were on a good, sound highway. Tr. 253. Thus, the notion that southbound drivers often would be applying their brakes as they entered the curve, and among other things thereby creating a higher frictional demand on the roadway surface, does not, on its face, appear to be implausible. Whether frequent brake applications would have resulted in a- polishing of the roadway surface, as Berg further hypothesized, is not a matter to which we can speak. Yet, there is ample evidence in the record that the surface of-the curve was worn and smooth, with a tendency to become quite slick when wet. E.g., Tr. 529, 661, 663; see also Tr. 252, 434-35. Third, Berg’s hypothesis that the slipperiness of the pavement caught motorists by surprise notwithstanding the posted warning signs appears, at the least, to be consistent with the facts. The data in the record do reflect an' ongoing' pattern of wet-pavement accidents that continued unabated notwithstanding the warning signs that were added or modified in 1984 (calling motorists’ attention to the curve, advising a speed of thirty miles per hour through the curve, and warning that the pavement was slippery when wet) and again in 1986 (reducing the advisory speed to twenty-five miles per hour). That pattern suggests that the signs, although sufficient in the abstract to alert motorists to a potential hazard (as Berg himself agreed), proved inadequate in practice to avert accidents. Fourth and finally, there appears to be at least some evidentiary support, although weak, for Berg’s opinion that the defects he identified in the roadway likely were at least partially responsible for the Mihailovich accident. That accident does fit the pattern revealed by the data in that it involved a southbound vehicle on wet pavement leaving the roadway (suggesting a loss of control) shortly after entering the curve. Moreover, although Mihailovich could not speak to what her husband was or was not doing as the driver of the vehicle in the crucial seconds before the accident, she did testify that he was driving at a “regular” rate of speed, keeping pace with other traffic, and that there was nothing unusual in the manner of his driving. It is a reasonable inference from her testimony and the pattern of ongoing accidents occurring at the curve under similar conditions that Mr. Mihailovich was not driving in a manner out of character with other traffic, and that he lost control of the vehicle due to the combination of the higher frictional demand his vehicle imposed as it entered the curve and the reduced skid resistance posed by the wet pavement. See Jahn v. Equine Servs., PSC, 233 F.3d 382, 390-92 (6th Cir.2000) (Cudahy, J.) (in order to qualify for admission, expert’s opinion as to causation need not eliminate all other potential causes; expert’s opinion as to probable cause admissible so long as it is based on facts and sound methodology); see also R. 123 at 9-10 (denying Laatsch’s motion for summary judgment as to causation of Mihailovich accident). We leave it up to the district court on remand to determine whether Berg’s methodology reflects sufficient intellectual and professional rigor to render his opinions trustworthy enough to reach the jury. The parties should be fully prepared. to demonstrate to the court the respects in which Berg’s methodology and conclusions do or do not conform to the evidence or to civil engineering standards. III. For the reasons set forth above, the judgment in favor of the defendants-appel- *921 lees is VaCated, and the case is RemaNded for a new trial. Footnotes [1] Mihailovich's husband was killed in a head-on automobile collision that occurred after this accident. The record is therefore confined to Mihailovich’s own limited recollection of the accident. [2] Simply pul, the coefficient of friction is the degree of slip resistance. The higher the coefficient of friction, the less slippery the pavement would be. [3] [Footnote by the court] The evidence indicates that as one moves from north to south, the Kedzie curve actually begins prior to 206th Street, but the record is silent as to why the resolution referred only to the roadway between 207th Street and 206th Street. Field testified that it was his belief that the resolution covered the entire S-curve; he also indicated that the road begins to curve in earnest near 206th Street. [4] The resolution that Healy read to the jury was redacted by agreement between the parties so as to comply with the court's order excluding evidence of other accidents that had occurred at the Kedzie curve. Healy was not permitted to read the following italicized clause from the resolution: Whereas, ... accident records reflect that due to the curvature of this road and the construction thereof, it appears to be an eminently dangerous roadway particularly under certain hazardous weather conditions[.] R. 1 Ex. G. (emphasis ours). [5] The County’s decision to install these signs presumably was based on the studies that it had commenced in response to the concerns raised by the Reno memorandum in 1983 as well as the subsequent Olympia Fields resolution in October 1984. [6] It is unclear from the briefs whether Mihail-ovich, if permitted to introduce evidence regarding the other accidents, would have introduced that evidence primarily through Berg or through another witness. As the ensuing summary of the excluded evidence indicates, a number of witnesses had varying degrees of knowledge about the accident rate at the Ked-zie curve. Berg's written report appears to contain the most complete analysis of those accidents. But engineer Paul Box also looked at the accident data and, like Berg, analyzed and summarized that information. See R. 105 (attachment). Former Olympia Fields police chief Fred Unger also had compiled a report about the curve for the Village trustees which included accident data. R. 62 Ex. 24. We assume that any of these witnesses, who appear to have relied on the same basic data, could have testified about the other accidents that occurred at the Kedzie curve in the years immediately preceding and following the Mi-hailovich accident. We also note that at least some of the underlying accident data itself is in the record. See R. 62 Ex. 24 (accident data from Olympia Fields); R. 1 Ex. J (accident reports). As the parties’ appellate arguments are focused solely upon the admissibility of other-accident evidence, rather than upon how that evidence might have been introduced, we need not explore the latter issue. [7] Berg’s report notes that in the aftermath of the Mihailovich accident, the County reduced the advisory speed through the curve to twenty-five miles per hour. R. 62 Ex. 6 at 4 ¶ 12. Based on the accident data, Berg observed that “the wet pavement safety problem ... continued unabated.” Id. [8] We note that other evidence in the record indicates that Laatsch may not, in fact, have obtained a copy of the resolution. Laatsch himself testified at his pre-trial deposition that he did not believe he had ever received a copy of the resolution. See Tr. 373, 703. Two letters from Laatsch to the Assistant State’s *901 Attorney handling the litigation for the County arguably suggest that Laatsch had not received the resolution. Tr. 393-94. Moreover, the case file that Laatsch later sent to the attorney who replaced him did not include a copy of the resolution. See Tr. 256, 373-74, 702-03. Nonetheless, because we are obligated at this juncture to construe the evidence in the light most favorable to the defendants, we credit Laatsch's trial testimony. [9] It appears that the data regarding other accidents at the Kedzie curve played a role in changing Box's mind as to whether or not the curve was hazardous at the time of the Mi-hailovich accident. Box was one of the two experts that attorney McKenna had contacted in 1998 when he provisionally took over the representation of Mihailovich in her suit against the County. Based on the information that McKenna provided, which did not include any information about the other accidents that had occurred at the curve, Box told McKenna that Mihailovich had "no case.” Xr. 296-97, Later, in the course of, the in *908 stant malpractice suit, Box was supplied with much more information about the history of the Kedzie curve, including the other accidents that had occurred there. Based on the more expansive information with which he was provided, Box concluded that the curve was dangerous at the time of the Mihailovich accident. R. 105, Box Report at 6-7. [10] Evidence of other accidents arguably could have served as proof that the County was on notice of the alleged hazard. However, Mi-hailovich did not offer the evidence for that purpose. In any case, there was other evidence in the record establishing that the County was on notice, including the Village’s 1984 resolution and the County Highway Department's own internal investigation into the safety of the curve. [11] So long as they satisfy the criterion of substantial similarity, accidents that occurred after as well as before the plaintiff's accident are admissible to prove the existence of a danger. Ross, 977 F.2d at 1185; Nachtsheim, 847 F.2d at 1268 n. 8. [12] As we have noted, the appellate record contains only an incomplete set of police records. But we assume that additional records may be available to the parties. [13] Berg admitted that the actual coefficient of friction on the surface of the curve as of the date of the Mihailovich accident could not be ascertained, because the roadway had since been realigned and repaved. Tr. 624.