This case involves a long-time employee of a railroad who suffered a knee injury as a result of the railroad’s negligence. Because of the injury, the employee was no longer able to work at his job. At the time of the injury, the employee was nearly fifty-nine, and he would have been eligible to retire on full benefits at age sixty. Indeed, he had previously checked the railroad’s website to determine the benefits he would receive if he retired at age sixty.
Following the injury, the employee sued the railroad in the Polk County District Court under the Federal Employers’ Liability Act (FELA). In the litigation, the employee claimed he had planned to work until age sixty-six and, on that basis, sought approximately $755,000 in economic damages. To challenge this asserted retirement date, the railroad attempted to introduce evidence that the plaintiff was *715eligible to retire on full benefits at age sixty and had checked the railroad’s website regarding his retirement benefits, as well as statistical evidence that most railroad employees in the plaintiffs position retire at age sixty.
Based on its reading of the federal collateral source rule applicable to FELA cases, the district court excluded the railroad’s evidence. The jury subsequently awarded $1.25 million to the plaintiff in a general verdict covering both economic and noneconomic damages. The railroad appeals, arguing the district court’s reading of the federal collateral source rule in FELA cases was incorrect.
On our review, we agree with the railroad in part. When a railroad employee makes a claim of lost earning capacity based on a hypothetical retirement age, we do not believe federal law precludes the introduction of statistical evidence as to when railroad workers in the plaintiffs position typically retire. Because this excluded evidence was relevant and important to the railroad’s defense, we reverse and remand for a new trial on damages.
I. Facts and Procedural History.
John Giza was born in 1950. In 1969, after graduating from high school, he went to work in Crestón for the CV & Q Railroad as a brakeman-switchman. Except for a stint in the Navy, Giza continued to work for the railroad, which eventually became part of BNSF, for the next forty years. In 1978, Giza’s seniority enabled him to become a conductor. This meant he still had the physical job duties of a brakeman-switchman but also had paperwork and supervisory responsibilities.
Giza’s everyday work required him to assemble and disassemble trains by gathering up railcars from customers and breaking them down for customers along a stretch of railroad between Crestón and Shenandoah. Giza had to climb ladders, ride on moving railcars, walk railcars, release and connect the “knuckles” between railcars, and walk on ballast.
On October 9, 2009, Giza was riding on the ladder of a railcar in the Red Oak yard as a locomotive was slowly pushing the railcar backwards. Giza was “protecting the point,” that is, he was watching the crossing toward which the car was being pushed while talking on a handheld radio with the locomotive’s engineer. Suddenly, Giza heard a bumping sound, indicating the train had derailed. Giza was thrown off the railcar and landed on his left foot. He instantly heard a pop in his left knee and felt excruciating pain.
Giza suffered a tear of his anterior cruciate ligament, a sprain of his medial collateral ligament, and a medial meniscal tear. Orthopedic surgery was performed on the knee on November 20, 2009. This was followed by physical therapy, manipulation; and injections on the knee. None of these gave Giza the relief from pain or flexibility he needed to go back to his former job. Giza could no longer climb ladders, walk on uneven surfaces, or stand for long periods of time. At the time of trial, Giza was still enduring constant knee pain at some level, and the pain became considerably worse when he tried to walk.
Giza brought suit against BNSF under FELA. See 45 U.S.C. § 51 (2012).1 He *716alleged that the railroad’s negligence caused his injuries. Although BNSF initially disputed liability, by the time of trial, it had admitted fault and was only contesting damages.
Giza was nearly fifty-nine years old when the accident occurred. Giza sought recovery of lost earning capacity and loss of the value of household work services. Giza also requested damages for his pain and suffering.
Before trial, Giza acknowledged in deposition testimony that he was eligible to retire at age sixty on full pension, having worked at least thirty years for the railroad. He had checked the railroad’s website before the accident to determine his estimated retirement benefits. However, he testified that he personally had intended to wait until age sixty-six to retire.
Giza’s expert, Dr. John Ward, estimated Giza’s lost earnings using the retirement age of sixty-six. His report initially stated, “Age 66 is the full benefit retirement age of all railroad workers born in 1950 as reported by the Railroad Retirement Board.” Later this was amended to read, “For persons born in 1950, age 66 is the age at which an individual may receive an unreduced benefit at retirement under the Social Security Act.”2 He estimated loss of income at approximately $755,000 for those seven years (fifty-nine to sixty-six) if Giza were not able to secure and maintain alternative employment.
BNSF’s expert, Mark Erwin, filed a report noting that railroaders with thirty or more years of service retire on average at age 60.7, and over sixty-two percent of them retire at age sixty. He also pointed out that railroad retirement benefits are largely exempt from federal taxes. Erwin discussed Giza’s retirement benefits and concluded that based on the relative financial impact of working as opposed to retiring, it was “unlikely” Giza would have worked past age sixty even if he had not been injured.
Before trial, Giza filed a motion in limine seeking to exclude (1) all evidence of the average age of retirement for railroad workers and (2) all evidence regarding potential railroad retirement benefits. The district court ultimately granted the motion. Originally, the court said it would allow BNSF to show that Giza would be eligible to retire at age sixty but would not allow the railroad to go into the retirement benefits available to him. The court explained that its ruling was based upon FELA caselaw.
Both sides took issue orally with this ruling. Upon further consideration of FELA precedents, the court ruled that BNSF could not go into a railroader’s retirement age, even without referring to benefits. The court also ruled BNSF could not introduce evidence that, prior to his injury, Giza had checked his retirement benefits on the BNSF website.
In his trial testimony, Giza reiterated that if he had not been injured, he planned to work until he was sixty-six. Dr. Ward testified he had calculated $755,000 in lost income based on Giza’s statement that he intended to retire at age sixty-six. Dr. Ward acknowledged his statement in his report that sixty-six is the age at which an individual may receive an unreduced Social Security benefit. Under cross-examina*717tion, Dr. Ward added that Giza’s retirement
is his individual decision of what he wants to do. He is simply going to weigh how much he could earn by working against what he would earn not working; what benefits he would receive; his own health and his own enjoyment of the job, and it is his opinion that it is age 66.
Q. Okay. Would you agree with me that a person, in making those types of decisions, are going to weigh the very factors that you outlined, weigh what benefits they would receive, what income they would earn, comparing that versus retirement versus employment to determine when to retire? A. That is correct. He would be weighing, basically, earnings of approximately $106,000 a year, plus benefits of health insurance for he and his wife, versus what he would get by not working.
Q. [Y]ou would agree with me that it would be fair, and certainly economically supported, to look at what other people do in testing that testimony or that information that you were provided by Mr. Giza? A. I have no problem with that, no.
With the jury excused, BNSF asked the court to reconsider its ruling on the motion in limine, arguing:
. [A]s the Court’s ruling stands right now, the railroad has been precluded from doing anything to test Mr. Giza’s testimony that he would have retired at age 66. We have not been allowed to explore the topic as to his understanding as to when he could retire. And we have not been allowed to explore the efforts that he undertook prior to this incident to evaluate and look at what benefits he would have received upon retirement. We have not been permitted to introduce evidence as to the statistical or actuarial studies that would indicate when railroaders actually do retire, nor explore with either Dr. Ward or Mr. Giza as to what benefits he would have received upon — upon retirement.
The court declined to reconsider its ruling but permitted BNSF to make an offer of proof.
In the offer of proof, BNSF introduced statistical and actuarial tables showing when railroaders usually retire.3 It also had Dr. Ward acknowledge that 62.1% of railroaders with thirty years of service retired at age sixty according to these data. Dr. Ward further admitted that railroad retirees generally have health insurance available upon retirement. In addition, BNSF introduced pages from Giza’s deposition and Erwin’s curriculum vitae and report.
Later, at the jury instruction conference, the court declined to give BNSF’s proposed instruction 36, which read:
You are not to award damages for any injury or condition from which the Plaintiff may have suffered or may now be suffering unless it has been established, by a preponderance of the evidence in the case, that such injury or condition was caused by the accident in question.
The court ruled the instruction “at best is cumulative and unnecessary, at worst is confusing.”
*718During closing argument, Giza’s counsel made the point three times that Giza planned to work until age sixty-six. In his rebuttal argument, Giza’s counsel added:
If someone asked or wants to debate how do we know John was going to retire at age 66?....
Well, first of all, John Giza has always testified, when he talked to Mr. Thrall, when he talked to you, he’s always said that. And, again, 51 percent, but there is a lot more. Again, more than 51 percent. This is a job he worked very hard to get to. Forty years of seniority. This is the job he had chosen to work since they put it on because it was a steady job, five days a week steady. Now, he knew when he was going home, he got his weekends off. He’s making almost $96,000 a year at this job. This was the job that you work all those years for to get there. There is — no evidence has been — none—introduced in this trial there was anything in John Giza’s mind about retirement until age 66. Our burden has been met.
Giza’s counsel also said the following during rebuttal argument:
If someone wants to talk about, hey, the railroad admitted it’s their fault, they should get some credit for that, remind them that’s not the law. You cannot do that. You can only consider the harms and the losses that they have caused to Mr. Giza. There is a concept called the concept of repentance.
At this point, BNSF’s counsel objected, and a conference was held outside the presence of the jury. BNSF’s counsel explained, “I can’t imagine that there would be proper closing arguments on repentance as part of this particular case.” The district court told Giza’s counsel to use other language and overruled BNSF’s motion for mistrial. Giza’s counsel then resumed his rebuttal argument.
The jury returned a verdict of $1,250,000 for Giza. The district court denied BNSF’s motion for new trial. This appeal followed.
On appeal, BNSF raises five arguments. First, it contends it should have been able to introduce evidence that Giza would have been eligible to retire at age sixty with full retirement benefits. Second, it maintains it should have been able to introduce evidence of when railroad workers typically retire, without going into their retirement benefits. Third, it argues the district court should have granted a mistrial based on Giza’s improper “repentance” argument. Fourth, it insists the jury’s verdict should have been overturned as a product of passion and prejudice. Finally, BNSF contends the district court erred in refusing to give its proposed jury instruction 86.
II. Standard of Review.
We review a district court’s evi-dentiary rulings for an abuse of discretion. Hall v. Jennie Edmundson Mem’l Hosp., 812 N.W.2d 681, 685 (Iowa 2012). “A court abuses its discretion when its ruling is based on grounds that are unreasonable or untenable.” In re Trust No. T-1 of Trimble, 826 N.W.2d 474, 482 (Iowa 2013). “The grounds for a ruling are unreasonable or untenable when they are based on an erroneous application of the law.” Id. (internal quotation marks omitted). Therefore, under our abuse-of-discretion standard, “we will correct an erroneous application of the law.” Rowedder v. Anderson, 814 N.W.2d 585, 589 (Iowa 2012).
Denial of a motion for new trial “based on a discretionary ground such as misconduct” is reviewed for an abuse of discretion. Loehr v. Mettille, 806 N.W.2d 270, 277 (Iowa 2011). Similarly, “[w]e review the district court’s denial of a motion *719for new trial based on the claim a jury awarded excessive damages for an abuse of discretion.” WSH Props., L.L.C. v. Daniels, 761 N.W.2d 45, 49 (Iowa 2008) (internal quotation marks omitted). Additionally, we review a claim that a district court should have given a party’s requested instruction for an abuse of discretion. Mulhern v. Catholic Health Initiatives, 799 N.W.2d 104, 110 (Iowa 2011).
III. Legal Analysis.
A. Eichel v. New York Central Railroad. For the first two issues on appeal, both parties agree our starting point should be the United States Supreme Court’s decision in Eichel v. New York Central Railroad, 375 U.S. 253, 84 S.Ct. 316, 11 L.Ed.2d 307 (1963) (per curiam). Eichel, like the present case, was an action for negligence brought by a railroad employee against his employer under FELA.4 Id. at 253, 84 S.Ct. at 316, 11 L.Ed.2d at 308. The railroad in Eichel had sought to introduce evidence that the employee was receiving a specific sum per month in disability pension payments under the Railroad Retirement Act. Id. This was offered “for the purpose of impeaching the testimony of [the employee] as to his motive for not returning to work and as to the permanency of his injuries.” Id. at 254, 84 S.Ct. at 316, 11 L.Ed.2d at 308. The district court excluded the evidence and the Supreme Court upheld this ruling. Id. at 254-56, 84 S.Ct. at 317, 11 L.Ed.2d at 308-09.
As the Court explained,
In our view the likelihood of misuse by the jury clearly outweighs the value of this evidence. Insofar as the evidence bears on the issue of malingering, there will generally be other evidence having more probative value and involving less likelihood of prejudice than the receipt of a disability pension. Moreover, it would violate the spirit of the federal statutes if the receipt of disability benefits under the Railroad Retirement Act of 1937, 50 Stat. 309, as amended, 45 U.S.C.s 228b(a)(4), were considered as evidence of malingering by an employee asserting a claim under the Federal Employers’ Liability Act. We have recently had occasion to be reminded that evidence of collateral benefits is readily subject to misuse by a jury. It has long been recognized that evidence showing that the defendant is insured creates a substantial likelihood of misuse. Similarly, we must recognize that the petitioner’s receipt of collateral social insurance benefits involves a substantial likelihood of prejudicial impact. We hold therefore that the District Court properly excluded the evidence of disability payments.
Id. at 255, 84 S.Ct. at 317, 11 L.Ed.2d at 309 (footnotes and internal citations omitted).
In Snipes v. Chicago, Central & Pacific Railroad, we applied Eichel. See 484 N.W.2d 162, 166-67 (Iowa 1992). There we affirmed a district court’s refusal to admit evidence that an injured employee had received a monthly annuity under the Railroad Retirement Act (RRA). Id. As we put it, “The federal law is well settled that, under the FELA, the collateral source rule operates to prevent consideration of RRA disability pension payments in mitigation of damages suffered by an injured employee.” Id. at 166. We declined to consider the railroad’s policy ar*720guments, noting that “our decision must be guided by federal case law.” Id. at 167.5
B. Applying Eichel to This Case. Strictly speaking, the retirement benefits involved here are not collateral source payments. They are not paid on account of an injury, nor are they compensation for an injury. See Pexa v. Auto Owners Ins. Co., 686 N.W.2d 150, 156 (Iowa 2004) (“The collateral source rule is a common law rule of evidence that bars evidence of compensation received by an injured party from a collateral source.”); see also Schonberger v. Roberts, 456 N.W.2d 201, 202 (Iowa 1990) (“Under the collateral source rule a tortfeasor’s obligation to make restitution for an injury he or she caused is undiminished by any compensation received by the injured party from a collateral source.”). The railroad’s aim here is not to show that Giza is receiving other compensation, but to call into question his claim that he would have worked until he was sixty-six if he had not been injured.
Having said that, some intermediate state appellate courts following Eichel have refused to allow evidence of the availability of railroad retirement benefits even for this purpose. In Griesser v. National Railroad Passenger Corp., the Pennsylvania Superior Court held Amtrak could not introduce evidence that its employee would be eligible to retire at age sixty with full pension benefits to counter an expert calculation of lost earning capacity based on an expected retirement age of sixty-five or seventy. See 761 A.2d 606, 612-13 (Pa.Super.Ct.2000). The court acknowledged: “The instant case presents a more attenuated link between the injury and the benefits at issue. The benefits at issue are future retirement pension benefits and not current disability benefits. Thus, the benefits are not related to the injury.” Id. at 610. Nonetheless, the Pennsylvania court warned of “the danger that the jury would use this evidence for the improper purpose of mitigating Appellant’s damages or reducing Amtrak’s liability.” Id. at 613.
Likewise, in Norfolk Southern Railway v. Tiller, the Maryland Court of Special Appeals held a trial court correctly refused to let a railroad show an injured employee would have been eligible to retire at age sixty, notwithstanding the employee’s testimony that he planned to retire at age sixty-five. See 179 Md.App. 318, 944 A.2d 1272, 1274-75, 1286 (2008). The court reasoned the situation was covered by Eichel:
The use the defendant railroad sought to make of the disability pension benefits in Eichel was closely analogous to the use Norfolk Southern sought to make of the future pension benefits in this case. The New York Central was trying to show that the injured employee in that case had a motive not to go back to work because he was then collecting disability pension benefits. Norfolk Southern was trying to show that Tiller *721would have had a motive for not continuing to work past age 60 because he could then have been collecting retirement pension benefits. In each case, the motive not to work because of benefits as an alternative to work was the same. We do not see the difference in the tenses as compelling a different result.
Id. at 1281.6
One can debate this point. The defendant’s goal in introducing the disability payments in Eichel was to show the plaintiff was not injured as badly as he claimed and was not motivated to go back to work. Yet as the Supreme Court noted, “[0]n the issue of malingering, there will generally be other evidence having more probative value and involving less likelihood of prejudice than the receipt of a disability pension.” Eichel, 375 U.S. at 255, 84 S.Ct. at 317, 11 L.Ed.2d at 309. Moreover, there is a significant danger of prejudice when the jury learns the plaintiff is receiving other compensation for the same injuries for which he is seeking compensation. By contrast, in Tiller, the purpose of the evidence was not to show the plaintiff was a malingerer or that he was not injured as severely as he claimed to be. See 944 A.2d at 1286. Rather, the evidence was intended to show that he would have retired at age sixty even if he had not been injured. Id.
Nonetheless, in Griesser, the Pennsylvania Superior Court remained concerned “the jury could conclude that [the railroad] was liable for lost wages to age 65 or 70, but then decline to award such damages because of the fortuitous existence of equivalent retirement benefits.” 761 A.2d at 612. Otherwise stated, the concern appears to be that the jury, notwithstanding any limiting instruction, would: (1) believe the plaintiffs testimony he intended to re- ’ tire at age sixty-five; (2) assume the plaintiff now can retire at age sixty; and (3) therefore award the plaintiff less than the full lost wages through age sixty-five by offsetting some amount for the retirement benefits available at age sixty.
Of course, the jury would have no reason to conclude the retirement benefits were “equivalent” (and the record in this case indicates they are not) without evidence as to their amount.7 Also, the same potential prejudice noted in Griesser can arise in any case where a plaintiff seeks damages for loss of future earning capacity beyond a possible retirement date; it is not limited to railroad cases.
Some courts have therefore held that Eichel allows railroads to introduce evidence of when their employees are eligible to retire. See, e.g., Cowden v. BNSF Ry., — F.Supp.2d -, 2013 WL 5838718 (E.D.Mo.2013). In Cowden, a railroad employee brought an action against the railroad under FELA to recover damages for his injuries sustained in a rail accident. Id. at -, 2013 WL 5838718 at * 1. Relying on Griesser, the plaintiff sought to “exclude any suggestion that, had he not been injured, he could have retired with benefits at the age of 60.” Id. at -, 2013 WL 5838718 at *16. While the court agreed the defendant could not “introduce evidence regarding the availability or *722amount of retirement benefits,” it also stated the “Plaintiff cannot expect to testify he unequivocally intended to retire at the age of 67, thereby increasing any potential damages, without allowing Defendant an opportunity to challenge his assertions.” Id. at -, 2013 WL 5838718 at *17. Citing a previous federal district court case, it therefore held that the defendant could “offer evidence that Plaintiff was ‘eligible to retire’ at the age of 60.” Id. (citing Stevenson v. Union Pac. R.R., No. 4:07CV00522BSM, 2009 WL 652932, at *3 (E.D.Ark. Mar. 12, 2009)). Likewise, a Missouri appellate court held in a FELA case that the admission of evidence that the plaintiff was eligible to retire at age sixty did not amount to plain error. See Payton v. Union Pac. R.R., 405 S.W.3d 1, 7 (Mo.Ct.App.2013). The court explained, “The collateral source rule applies to evidence of collateral compensation for a plaintiffs injury. Here, neither [of the witnesses who testified about eligibility to retire at age sixty] testified about any collateral compensation.” Id. (citation omitted).
BNSF argues that, at a minimum, it should have been able to introduce statistical data showing railroad employees with thirty years of service tend to retire at age sixty. The highest court in Maryland has agreed with this position. In CSX Transportation, Inc. v. Pitts, decided after the trial in this case, the Maryland Court of Appeals distinguished Tiller and held that “although retirement eligibility information in a FELA case is barred by the collateral source rule, statistics about average retirement age for railroad workers is not.” 430 Md. 431, 61 A.3d 767, 791 (2013). The court elaborated:
Use of industry statistics about average retirement age in this context is not evidence of other compensation the plaintiff would receive for the same damage, but rather, evidence that shows that the full amount of lost wages claimed by the plaintiff may not exist. In other words, the tables may cast doubt on a plaintiffs statement that he would work until a certain age, and thus suggest to the fact-finder that the lost wage claim was exaggerated....
Although the collateral source rule bars evidence of disability and retirement benefits, a defendant railroad should not be defenseless against the plaintiffs “1-2 combo” — self-serving testimony about his retirement plans and expert projections about damages based on that testimony. Moreover, it would be unfair to allow the plaintiff to clothe his own prediction about his retirement date with the protective folds of the economist’s projections about damages, while denying the defendant the right to use cross-examination to cast legitimate doubt on the assumption made by that economist that the claimant would retire at age 68.
Id. at 792. The court added that “statistics discussing an individual’s projected date of retirement, or worklife expectancy, have been widely held to be relevant when future wage loss is at issue.” Id. at 791 (citing cases). The court concluded that the trial court did not have the discretion to exclude evidence relating to railroad work-life expectancy tables, although it affirmed the verdict because the railroad did not ask the right questions. Id. at 794.
Griesser also appears to leave the door open for this kind of evidence. While holding that Amtrak was not entitled to show an employee could have retired on full benefits at age sixty, the court did indicate that Amtrak could have cross-examined plaintiffs expert on the fact that “railroad workers commonly retire at age 60 if they have 30 years of service.” Griesser, 761 A.2d at 613.
*723These kinds of statistical data also have been found admissible in the analogous Jones Act context.8 See, e.g., Madore v. Ingram Tank Ships, Inc., 732 F.2d 475, 478 (5th Cir.1984). In Madore, the court found fault with the district court’s conclusion that a disabled seaman was going to retire at age sixty-five when the parties’ experts both estimated he would retire approximately five years before that based on Department of Labor work-life expectancy rates. Id. The court noted evidence may show “a particular person, by virtue of his health or occupation or other factors, is likely to live and work a longer, or shorter, period than the average.” Id. However, when such evidence is absent, as it was in Madore, “computations should be based on the statistical average.” Id.; see also Earl v. Bouchard Transp. Co., 735 F.Supp. 1167, 1175 (E.D.N.Y.1990) (observing in a Jones Act case that “[sjtatistical charts, such as the mortality tables and work-life expectancy tables prepared by the United States Department of Labor, compile averages and are often deemed authoritative [in determining work-life expectancy], particularly in the absence of contradictory particularized evidence.”), aff'd in part, rev’d in part, and remanded, 917 F.2d 1320 (2d Cir.1990).
When considering lost earning capacity claims in other contexts, courts have found average retirement ages to be relevant and admissible. See, e.g., Boucher v. U.S. Suzuki Motor Corp., 73 F.3d 18, 20, 23 (2d Cir.1996) (holding the trial court did not abuse its discretion when it allowed the defendant’s expert in an employee’s suit against his employer for damages related to a work injury to present pre and post-injury work-life expectancy testimony “based on widely accepted work-life tables published by the Department of Labor and his expertise in vocational rehabilitation”); Weil v. Seltzer, 873 F.2d 1453, 1465 (D.C.Cir.1989) (noting a defendant, in response to a plaintiffs “self-serving testimony ... concerning the anticipated work-life expectancy of the decedent,” may “produce his own expert to offer a contrary opinion on [the decedent’s] work-life expectancy or he may offer the Department of Labor statistics into evidence and request the expert to base his opinion on the work-life expectancy contained in the Department of Labor’s table”); Finch v. Hercules, Inc., 941 F.Supp. 1395, 1416 (D.Del.1996) (holding in a wrongful termination case that the defendant could present statistical evidence as to when its average employee retires); but see Burrows v. Union Pac. R.R., 218 S.W.3d 527, 540 (Mo.Ct.App.2007) (holding that the trial court did not abuse its discretion in disallowing testimony about the average retirement age of Union Pacific machine operators because it “would not prove or disprove when Plaintiff himself planned to retire”).9
On our review, we agree with the line drawn by the Maryland Court of Appeals in Pitts. Eichel does not extend so far as to bar the railroad from introducing evidence as to when railroad workers with certain levels of experience typically retire. These data are several steps removed from the disability benefits that the Supreme Court ruled inadmissible in Eichel. Furthermore, the entire point of Eichel is to prevent unfairness. Yet it is basically unfair for the railroad to be “defenseless,” Pitts, 61 A.3d at 792, in the face of an *724employee’s claim as to when he or she would have retired, particularly when the employee appears to be relying on jurors’ familiarity with a different retirement system than the one in which the employee actually participates. Dr. Ward’s report, which incorporated Giza’s claimed retirement age of sixty-six, indicated that this age is when an individual can retire on full Social Security benefits. But Giza does not participate in Social Security.10
Giza argues with some force that juries know how to connect the dots, so a jury reading BNSF’s exhibits could reach the conclusion that railroad employees like Giza are able to retire and start receiving pensions at age sixty. Nonetheless, we agree with the Maryland Court of Appeals that the alternative could leave the railroad without a realistic way to challenge the testimony of the plaintiff and the plaintiffs damages expert. See id. Again, most jurors participate in the Social Security system, where full benefits come later in life than age sixty, and in the absence of other evidence would likely assume that Giza’s assertion he planned to retire at age sixty-six was entirely typical and unexceptional — even though it isn’t.
No one disputes that when Giza would have retired if he hadn’t been injured is highly relevant to his claim for lost earning capacity. And as Dr. Ward himself conceded in his testimony, to determine when someone is likely to retire, we would want to look at when other people retire.11
For the foregoing reasons, we hold Eichel does not bar the introduction of evidence as to when railroad employees with thirty years of service typically retire so long as the evidence does not directly or indirectly refer to retirement benefits. We doubt a jury will be improperly influenced by learning of the typical retirement age, when details concerning the pension are not disclosed. On the other hand, keeping this information from jurors could create a false impression while leaving no practical way for the railroad to challenge a plaintiffs claimed anticipated retirement date.
C. Deciding the Appeal. Giza argues that even if the district court’s reading of Eichel was incorrect, we should not reverse. Thus, Giza urges that the district court’s ruling excluding the statistical evidence was based upon Iowa Rule of Evidence 5.403, not Eichel, and that BNSF failed to address that rule in its opening brief, thereby waiving the argument. We disagree. The district court viewed the statistical evidence as covered by Eichel. During trial, the court said “the case law doesn’t allow” receipt of the statistical evidence.12 This does not pre*725vent us from upholding the district court’s exclusion of the evidence on an alternative ground. See DeVoss v. State, 648 N.W.2d 56, 62 (Iowa 2002). Nonetheless, when we perform the rule 5.403 balancing ourselves, we do not believe the danger of unfair prejudice substantially outweighs the probative value of the statistics. See Iowa R. Evid. 5.403 (stating that “relevant!] evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice”); State v. Werts, 677 N.W.2d 734, 737-38 (Iowa 2004) (indicating the appellate court should weigh prejudicial effect against probative value where the district court did not do so and rule 5.403 is raised as an alternative ground for sustaining the district court’s ruling).
The statistics give a norm as a frame of reference and allow the plaintiff to argue to the jury why he or she would deviate from that norm. Just as we would allow statistical data to show the duration of a typical professional football player’s or the typical judge’s career, data that show the duration of a typical railroader’s career are also relevant. The numbers take on added significance given that railroad employees with thirty years’ experience usually retire at an earlier age than the benchmarks that would be familiar to jurors from their common experience — i.e., the Medicare eligibility age or the age when a retiree is eligible to draw full Social Security benefits.
Meanwhile, for reasons we have already outlined, we believe the danger of unfair prejudice is relatively low. These statistics do not reveal why railroaders with considerable work experience most often retire at age sixty. Thus, for unfair prejudice to occur, the jury would first have to guess that railroad employees are eligible to receive a retirement pension at age sixty. But even then, for there to be prejudice, a jury would have to believe the plaintiffs testimony as to when he or she plans to retire and disbelieve the railroad while at the same time being willing to penalize the plaintiff by making an unauthorized deduction for retirement benefits without having any idea of the amount of those benefits. This chain of events, while possible, does not appear to be a significant threat to the fairness of the trial.13
Giza does not claim that any error in excluding the statistical evidence would have been harmless. Giza was earning approximately $100,000 per year at the time of the accident. Dr. Ward calculated economic damages of about $755,000 based on over seven years of lost wages. As Giza’s counsel said more than once in closing argument, Giza lost out on the last seven or eight years of his work career. *726Dr. Ward agreed that if sixty rather than sixty-six were the correct retirement age for Giza, the $755,000 would need to be reduced by several hundred thousand dollars. In the end, the jury awarded $1,250,000, including pain and suffering. Because BNSF was improperly precluded from presenting evidence regarding when railroad employees actually do retire, we reverse and remand for a new trial.
We add a caveat. While our ruling concerns the statistical evidence, and only the statistical evidence, a plaintiff may open the door to further exploration of the subject of retirement by the position he or she takes at trial. For example, if a plaintiff testifies on direct examination or the plaintiffs counsel argues that the plaintiff would have kept working until a particular age because of the money she was making, then it may be appropriate for the defendant to show that the plaintiff could make money by not working. See, e.g., Gladden v. P. Henderson & Co., 385 F.2d 480, 483-84 (3d Cir.1967) (holding that notwithstanding Eichel, a defendant may bring up disability payments when the plaintiff claims on direct examination that he only went back to work due to financial distress).
Because we are not convinced that most of BNSF’s remaining appellate issues will arise on remand, we will not address them. However, we do not believe the district court committed reversible error in refusing to give BNSF’s proposed instruction 36. Adhering to the Eighth Circuit’s model jury instructions for FELA cases, the district court instructed the jury as follows:
You must award the plaintiff such sum as you find will fairly and justly compensate the plaintiff for any damages you find the plaintiff sustained and is reasonably certain to sustain in the future as a direct result of the occurrence mentioned in the evidence.
See 8th Cir. Civil Jury Instr. § 7.06A (2011) (“F.E.L.A. Damages — Injury to Employee”). The district court also supplemented that instruction with the following: “[TJhroughout your deliberations you must not engage in any speculation, guess, or conjecture and you must not award any damages by way of punishment or through sympathy.” The Eighth Circuit’s manual says this language “may also be added.” Id. n. 7.
BNSF argues these instructions never told the jury that Giza bore the burden of proof on damages. However, we believe this concept was adequately conveyed by the instructions taken as a whole. See Kiesau v. Bantz, 686 N.W.2d 164, 175 (Iowa 2004) (“The jury must consider the instructions as a whole, and if the instructions do not mislead the jury, there is no reversible error.”). The district court gave a general instruction that “[w]henever a party must prove something they must do so by the preponderance of the evidence.” The court also gave a mitigation of damages instruction that made it clear the defendant bore the burden of proof on that issue — the implication being that the plaintiff bore the burden of proof on other matters. Additionally, Giza’s counsel repeatedly advised the jury during closing argument that Giza had the burden of proof on the remaining damage issues (while asserting Giza had met that burden). “If the concept behind the requested instruction is embodied in other instructions, the district court may properly reject the proposed instruction.” Crawford v. Yotty, 828 N.W.2d 295, 298 (Iowa 2013) (internal quotation marks omitted). We find no error.
IV. Conclusion.
For the foregoing reasons, we reverse the judgment of the district court and *727remand for a new trial. The new trial, like the first trial, should be limited to the question of damages.
REVERSED AND REMANDED.
All justices concur except HECHT, J., who takes no part.