Experts and the Rule 703 Exception

Experts and the Rule 703 Exception

Every rule in this chapter so far has pointed the same way: a witness may speak only to what she perceived. That demand fits an eyewitness, but it fits poorly an arson investigator who reached the scene after the fire, a toxicologist who never met the patient, or an economist who has never visited the market she describes. Rule 602 anticipates the problem in its last sentence, which says the rule does not apply to a witness's expert testimony under Rule 703.¹

The reason for the carve-out lies in what experts are for. A jury calls an expert to supply knowledge it lacks, and in practice experts form opinions by relying on tests run by others, reports written by colleagues, and the accumulated findings of their field. If every expert had to have personally observed every underlying fact, a physician could not rely on another doctor's X-rays, and expertise would become unusable in most real disputes. The law therefore trusts a field's own standards of reliability and leaves the jury to test the expert's basis through cross-examination. That trust has two limits: it belongs to experts only, and it cannot be used to smuggle another person's accusation into a criminal trial. The section takes the expert's permitted sources first, then the limits on telling the jury about them, and then the constitutional boundary that arises when the expert stands in for someone who did the work.

Expert Bases Under Rule 703

Rule 703 identifies three routes by which an expert may acquire the facts on which an opinion rests. She may observe them firsthand before trial, learn them at the trial or hearing itself, or draw on outside facts and data gathered from other sources, provided those sources are of a kind experts in her field would reasonably rely on.² Only the first route resembles what Rule 602 demands of a lay witness. A treating physician who personally examined the patient may describe what she found and then give her opinion, and that foundation is much the same one a lay witness would lay by placing herself at the scene.³

The second route, facts learned at trial, covers two techniques. An expert who is allowed to sit in the courtroom may hear the other witnesses testify, and counsel may also put a hypothetical question, which asks the expert to assume certain facts are true and then to give an opinion based on them.⁴ The question itself is not evidence, only the answer is, so the opinion is worth no more than the assumed facts are worth. A nineteenth-century instruction quoted in Williams v. Illinois makes the point: if the facts assumed in the questions are not supported by the proof, the answers deserve no weight.⁵ Rule 705 removed the old requirement that an expert lacking personal knowledge be examined through a hypothetical question, and it lets her state her opinion and reasons without first testifying to the underlying facts, subject to the court's contrary order and to cross-examination on those facts.⁶ Hypothetical questions had grown so long and complex that reciting one could take fifteen minutes, and Rule 703's willingness to let experts rely on trustworthy outside data lowers the risk of allowing the opinion to come out first.⁷

The third route is the broadest and the one that most departs from Rule 602. An expert may rely on facts or data that experts in the field would reasonably rely on, and those facts or data need not themselves be admissible.⁸ Imagine a psychiatrist who evaluates a defendant's mental state partly by interviewing the defendant's friends and relatives. Offered for their truth at trial, those accounts would be hearsay. As the basis of her opinion they raise a different question, namely whether other psychiatrists would rely on such accounts when assessing a patient's condition, and the answer is yes.⁹ Each profession has its standard sources, such as the notes and X-rays of other doctors, and an expert who draws on them is doing what her field does.¹⁰

The standard is reasonable reliance, so a source's being customary is not enough by itself. In an illustrative arson case, an investigator who arrived after the fire relied on an on-scene firefighter's oral account, given under a duty to report, and on laboratory tests run in the regular course of business, and both were reasonable bases. He also heard that a bystander had calmly reported seeing a man run out of the building, and that a firefighter had heard a rumor that the defendant was a professional arsonist. Neither of those statements carried any assurance of trustworthiness beyond that of hearsay in general: the bystander owed no duty to report, and the rumor was not based on anyone's knowledge of the facts.¹¹ Quantity matters as well as quality, since Rule 702 requires sufficient facts or data, and an expert on safety practices who never inspected the facilities at issue was properly excluded.¹² If the bases turn out to be inadequate, the opinion should be stricken as conjecture or speculation.¹³

The exception stops at the expert. A bystander at the same fire who announces that the department's chemists found an accelerant cannot invoke Rule 703, because the carve-out covers expert testimony only, and her statement is an attempt to repeat what someone else knows.¹⁴

Disclosing Basis Data to the Jury

Relying on an inadmissible fact is a different act from repeating it to the jury, and Rule 703 treats the two differently. When the facts or data would otherwise be inadmissible, the proponent of the opinion may disclose them to the jury only if their probative value in helping the jury evaluate the opinion substantially outweighs their prejudicial effect.¹⁶ This is the mirror image of the Rule 403 balance: Rule 403 excludes evidence only when its dangers substantially outweigh its value, whereas here the burden sits on the proponent and the balance tips toward exclusion.¹⁷ If the court does permit disclosure, it must instruct the jury to consider the material solely as a basis for the expert's opinion and not as substantive evidence.¹⁸

The heightened burden exists because of how basis evidence actually works on a listener. Jurors cannot use a basis statement to judge an expert's reasoning without at least tempting themselves to treat it as true, so the limiting instruction asks them to do something difficult. The presumption shifts when the factfinder is a judge. In a bench trial, the Federal and Illinois rules place no restriction on revealing such information, because a judge is presumed to understand the limited reason for the disclosure and not to rely on it improperly, while in jury trials the rules generally bar an expert from disclosing inadmissible basis evidence.¹⁹ That difference between jury and bench trials becomes important in the next sub-section.

Because the foundation for an expert's facts can be assembled in several ways, a short checklist helps keep the analysis in order.

The fourth step does the most work in everyday practice, because it is where the field's standards are measured against the particular source offered.

Surrogate Witnesses and Confrontation

Rule 703 is a rule of evidence, and the Confrontation Clause of the Sixth Amendment is a constitutional guarantee that an evidence rule cannot override. The Clause promises the accused the right to be confronted with the witnesses against him, and under Crawford v. Washington a testimonial statement by a witness who does not appear at trial is inadmissible unless the witness is unavailable and the defendant had a prior opportunity to cross-examine.²¹ A statement is testimonial when, for example, it is a formal declaration made for the purpose of establishing a fact in a later prosecution. In Melendez-Diaz v. Massachusetts, sworn certificates reporting that a seized substance was cocaine fit that description, so the analysts who wrote them were witnesses against the accused and could not be replaced by the certificates alone.²² The Court also said that this does not require the prosecution to call everyone who handled the evidence. Gaps in the chain of custody normally go to weight rather than admissibility, and the prosecution decides which links are important enough to prove, but whatever testimony it does introduce must be live if the defendant objects.²³

Bullcoming v. New Mexico applied that principle to a substitute witness. A laboratory analyst named Caylor had certified a blood-alcohol result, and the State, which never claimed he was unavailable and showed only that he was on unpaid leave, called a different analyst, Razatos, to testify in his place.²⁴ Razatos knew the laboratory's procedures and the gas chromatograph machine, but he had not performed or observed the test on this sample. The New Mexico court saw Caylor as a mere scrivener transcribing a machine's number. The Supreme Court answered that Caylor's certification represented more than a number: that the sample arrived with its seal unbroken, that the report and sample numbers matched, that he followed a precise protocol, and that nothing affected the sample's integrity. Those are statements about past events and human actions, and they are suited to cross-examination.²⁵ A surrogate cannot say what Caylor knew or saw, and cannot reveal any lapse or dishonesty on his part, which includes why he was placed on leave.²⁶ The comparative reliability of machine-produced data does not change the result, because the Clause requires that reliability be tested through cross-examination and not assumed.²⁷

The connection to personal knowledge is direct. Razatos had no firsthand knowledge of what happened to Bullcoming's sample, so as a lay witness he could not have testified to the test at all. Rule 703 might seem to rescue him by letting an expert rely on others' work, but the Confrontation Clause shows the limit. An expert may form her own opinion from data compiled by others. She may not serve as a conduit through which the compiler's testimonial statement reaches the factfinder for its truth. The Court noted that the State did not assert that Razatos had any independent opinion about the defendant's blood-alcohol level.²⁸

Williams v. Illinois tested how far that line extends, and it produced no single rationale that commanded a majority, so it must be read with care.²⁹ Lambatos, a state police analyst, testified at a bench trial that a DNA profile from an outside laboratory, Cellmark, matched a profile derived from the defendant's blood. The Cellmark report itself was never admitted. The plurality reasoned that Lambatos testified to the truth only of matters she knew personally, such as the shipping records showing that the swabs went to Cellmark and came back, and the match between the two profiles. The premise that Cellmark's profile came from the victim's swabs was just an assumption built into the prosecutor's question, and the judge was presumed to understand that it was not offered for its truth.³⁰ The plurality added that with a jury the result might differ, because jurors could take her testimony as proof of where the profile came from.³¹

The dissent rejected the not-for-its-truth rationale itself. When a witness repeats an out-of-court statement as the basis for a conclusion, the conclusion is only as good as the statement, so the factfinder cannot evaluate the opinion without judging whether the statement is true.³² The same objection was voiced against the idea that statements introduced to explain an expert's basis serve any legitimate nonhearsay purpose.³³ The dissent's illustration was an officer who testifies that he arrested a man because a reliable eyewitness described a star-shaped birthmark, and that the man has one. Framing the eyewitness's accusation as the basis for the officer's conclusion changes nothing constitutionally, the dissent argued, and a contrary rule would let the prosecution evade the Clause by a wink and a nod.³⁴

Because Williams was fractured, the dependable rules for the exam come from Melendez-Diaz and Bullcoming: a testimonial report offered for its truth cannot be introduced through a witness who did not make it and observed nothing of the work. Williams establishes only that the expert's reliance on outside data is not automatically a violation, particularly in a bench trial where the basis statement is not offered for its truth. Two contrasting witnesses illustrate the boundary. A forensic analyst who personally ran the test may testify to it, and a different analyst may give her own independent opinion about data compiled by others without conveying a testimonial report's assertions. A surrogate who merely reads or summarizes a colleague's testimonial certificate for its truth cannot, unless the colleague is unavailable and was previously cross-examined.

Expert Bases Under Rule 703

A lay witness may describe only what she perceived herself, and a lay opinion must be rationally grounded in that perception. Expert testimony follows a different rule on this point. An expert may testify about matters she never personally observed, which creates an obvious difficulty: the material she builds her opinion on may itself be inadmissible, either because no witness with knowledge has vouched for it or because it is hearsay.³⁶ Rule 703 answers that difficulty by saying where an expert may get her facts and what quality those facts must have, and it is the reason Rule 602 carves experts out of its own requirement.

The rule recognizes three routes, and it helps to keep them apart because each carries a different kind of support:

1. Firsthand observation before trial. The expert personally examined, tested, or inspected the person, place, or thing she will talk about.
2. Facts learned at the trial or hearing. The expert heard other witnesses testify or answered a hypothetical question that summarized evidence already admitted.
3. Outside facts or data. The expert gathered information before trial from other sources, provided experts in her field would reasonably rely on that kind of information.³⁷

The first route is the familiar one. A physician who examines an injured claimant and later testifies to the diagnosis and the prognosis for recovery is doing what Rule 602 asks of any witness, which is to speak from her own perception.³⁸ The foundation is the same as for a lay witness: preliminary questions show that she actually examined the patient, and the open question is whether she gathered enough data to support the opinion.³⁹ The second route belongs to experts alone. An expert who sits in the courtroom listening to earlier witnesses has learned the facts at trial, and an expert asked to assume certain facts can reason from them without ever having seen them. Suppose a lawyer asks a medical expert to assume that the plaintiff breathed a stated concentration of asbestos for a stated number of years, and then asks for the likely medical consequences. The question is not evidence, only the answer is, so the assumed facts must be supported by admissible evidence elsewhere in the record.⁴⁰

The third route is the broadest, and the reason for it is practical. Every profession has standard sources that its members use in forming judgments: doctors rely on notes and X-rays prepared by other doctors, accountants on returns prepared by others, and economists on valuations prepared by others.⁴¹ Much of that material would be inadmissible if offered directly, and requiring each underlying record to be proved separately would force a fresh dispute over every item. Borrowing the field's own standard avoids that, on the theory that a profession's habits of reliance reflect what has proved trustworthy in its work. The rule also formalized what had long been true in practice. Before the 1975 Rules, courts acknowledged only the first two categories, even though trained experts have always drawn on books, articles, conversations, and experiments that could never be put in evidence.⁴²

The test for outside data is reasonable reliance, and it has a qualitative and a quantitative side. Qualitatively, the source must be of a kind that experts in the field would reasonably use. A physician who reads another doctor's X-rays satisfies it. Imagine instead that a medical examiner's opinion on a cause of death rested on a neighbor's rumor about what the decedent had been doing, a rumor that came from no one's knowledge of the facts. No physician would treat that as a basis for a diagnosis, so it falls outside the rule even though the examiner relied on it in fact.⁴³ Quantitatively, Rule 702 requires sufficient facts or data. In a First Circuit decision, Pelletier v. Main Street Textiles, the court affirmed the exclusion of a safety-practices expert who never inspected the facilities and equipment at issue, because he lacked enough information to support his opinion.⁴⁴ A source can therefore pass the quality test and the opinion can still fail for want of quantity.

Two features keep this generosity within bounds. First, the ordinary result is that the underlying information stays out of evidence. The expert may rely on it, and she may state her opinion, but the basis itself ordinarily is not admitted.⁴⁵ Second, there is a narrow path by which otherwise inadmissible basis facts may reach the factfinder, because they can help it understand the expert's reasoning and decide what weight her opinion deserves. Because such evidence would normally be excluded, Rule 703 puts a heavy burden on the proponent: the probative value of the material in helping evaluate the opinion must substantially outweigh its prejudicial effect. That is the mirror image of the Rule 403 balance, and it tips toward exclusion.⁴⁶

Surrogate Witnesses and Confrontation

A witness who did not see or perform a test cannot testify to what happened during it, and in a criminal case the Sixth Amendment adds a second barrier. The Confrontation Clause gives the accused the right to be confronted with the witnesses against him. Under Crawford, a testimonial statement by a person who does not appear at trial is inadmissible unless that person is unavailable and the defendant had an earlier opportunity to cross-examine.⁴⁸ Rule 703 lets an expert build an opinion on facts she did not personally perceive, but a rule of evidence cannot waive a constitutional guarantee. That gap is where personal knowledge and confrontation meet, and the forensic-report cases show how it is policed.

The first step is to ask whether the out-of-court statement is testimonial. In Melendez-Diaz v. Massachusetts, the Court held that certificates of analysis, prepared by state laboratory analysts to report the results of tests on seized substances, were testimonial because an objective witness would expect them to be used at a later trial. The analysts who wrote them were therefore witnesses against the accused.⁴⁹ The decision has a limit that keeps it workable. The prosecution need not call everyone who touched the evidence, because gaps in the chain of custody ordinarily affect the weight of the evidence rather than its admissibility. The prosecution chooses which links it must prove, but whatever testimony it offers must be live if the defendant objects.⁵⁰

Bullcoming v. New Mexico asked whether a different laboratory employee could stand in for the analyst who wrote a testimonial report. The analyst who signed the blood-alcohol certification, Caylor, did not testify. The State never claimed he was unavailable, and the record showed only that he was on unpaid leave for an undisclosed reason. Instead the State called Razatos, another analyst from the same laboratory, who had neither performed nor observed the test on this sample. Bullcoming had never had a chance to cross-examine Caylor.⁵¹ The state court had accepted the substitution on the theory that Caylor merely transcribed a number produced by a machine, so the machine was the true accuser. The Supreme Court rejected that characterization, because the certification reported more than a number. Caylor represented that the sample arrived with its seal unbroken, that the report and sample numbers matched, that he followed a precise protocol, and that nothing affected the integrity of the sample. Those are statements about past events and human actions, and they are the kind of assertion cross-examination exists to test.⁵²

The reason a substitute will not do follows from what cross-examination is for. The questions that matter concern this analyst and this sample: whether he checked the seal, whether he followed the protocol, whether anything went wrong at the bench. Razatos could speak about the laboratory's general procedures and equipment, but he could not say what Caylor actually did, and he could not be pressed on any lapse or dishonesty on Caylor's part. The Court drew the same line for ordinary observations. An officer who did not see the number above a front door or on a radar gun cannot report it for the officer who did, even if the second officer understands the equipment and the department's procedures.⁵³ Nor does the apparent dependability of machine-generated data change the result. The Clause requires that reliability be tested through cross-examination, so the court does not get to assume it.⁵⁴

The connection to Rule 602 is now visible. Razatos had no firsthand knowledge of what was done to Bullcoming's sample, so a lay witness in his position could not have described the testing at all. Rule 703 might seem to rescue him, since an expert may rely on facts she did not perceive. But an expert who relies on another person's work forms her own opinion from it. She does not become a channel through which that person's testimonial assertions reach the factfinder for their truth, and state evidence rules that permit experts to disclose their bases cannot override the Clause.⁵⁵

Williams v. Illinois tested how far that principle extends, and it must be handled with care because no rationale commanded a majority. A state laboratory analyst, Lambatos, testified at a bench trial that a DNA profile produced by an outside laboratory, Cellmark, matched a profile derived from the defendant. The Cellmark report was never admitted, and she said nothing about what was done at Cellmark or the quality of its work. The plurality held that the words suggesting the Cellmark profile came from the victim's swabs were a premise of the prosecutor's question, not testimony offered for its truth, and that a judge sitting without a jury is presumed to understand the difference. The plurality conceded that with a jury the result might differ, because jurors could take the answer as proof of where the profile came from.⁵⁶

The dissenters rejected the not-for-its-truth rationale itself. When a witness repeats an out-of-court statement as the basis for a conclusion, the conclusion is only as good as the statement, so the factfinder cannot assess the opinion without deciding whether the statement is true.⁵⁷ Their illustration was an officer who testifies that he arrested a man because a reliable eyewitness described a star-shaped birthmark, and the man has one. Labeling the eyewitness's statement the basis of the arrest changes nothing, they reasoned, and a contrary rule would let the prosecution evade the Clause by a formula of words.⁵⁸ Because the Court splintered, the dependable rule for an exam is the one from Melendez-Diaz and Bullcoming, and Williams marks only the point at which the question becomes uncertain.

Two witnesses at the same trial show where the line falls. A scientist who herself performed the retest may testify to her own results, because she is the person who did the work and can be cross-examined about it. A colleague who knows the laboratory's protocols cannot present another analyst's blood-alcohol certificate, because he knows nothing about that particular test and the certificate is a testimonial statement offered for its truth. The facts differ in one respect, which is who stands in the witness chair relative to the work.