Halkides: Expectation Bias and the Patricia Stallings Case

Ed. Note: Chris Halkides has been kind enough to try to make us lawyers smarter by dumbing down science enough that we have a small chance of understanding how it’s being used to wrongfully convict and, in some cases, execute defendants. Chris graduated from the University of Wisconsin-Madison with a Ph.D. in biochemistry, and teaches biochemistry, organic chemistry, and forensic chemistry at the University of North Carolina, Wilmington.

Patricia Stallings took her infant son Ryan to the hospital because of lethargy, vomiting and other symptoms. The hospital tested Ryan’s blood and found what it believed to be evidence of ethylene glycol and metabolic acidosis (his blood was too acidic). The key ingredient of antifreeze, ethylene glycol, produces low pH and can be fatal when it is ingested and metabolized into oxalate, which forms crystals with calcium ion in the brain and kidneys.

Ryan was released into foster care, and Patricia and her husband were allowed weekly visits.  Ryan became ill after one of these visits and, about four days later, he died. His blood pH was even lower than after the first episode and the level of ethylene glycol was even higher, about 900 milligrams per liter.

Patricia Stallings was arrested and placed into custody. The main evidence against her was that two labs had claimed to find evidence of ethylene glycol (some antifreeze was found at their home).  She was subsequently convicted of murder and sentenced to life without parole.

Patricia Stallings was already pregnant with another son, David, at the time of her arrest.  After David was born, he was cared for by others, yet he began to show similar symptoms. Eventually he was diagnosed with methylmalonic acidemia (MMA).  MMA is caused by a failure of an enzyme to work properly, leading to a build-up of compounds including methylmalonic acid and propionic acid in the blood and urine.  MMA is a rare genetic disease, but the fact that David suffered from it indicated that Ryan had a one-in-four chance of having it as well.

Although this diagnosis was known by the start of Patricia Stallings’ trial, the judge prevented her attorney from presenting this theory to the jury. The defense attorney later said that he might not have understood exactly what he needed to do in order to get the judge to allow him to make this argument.

To understand what went wrong in this case, one must have some knowledge of clinical chemistry. Gas chromatography (GC) is a technique in which substances pass through a column of a waxy liquid based upon their chemical properties. The period it takes to emerge is the retention time. Different substances ordinarily have different retention times, but any two substances may have the same retention time within experimental error. GC is the technique that the two labs used to identify ethylene glycol. The first lab claimed it had identified ethylene glycol despite an apparent discrepancy in the retention times between the unknown substance in Ryan’s blood and standard of ethylene glycol, and the second lab did not run a standard at all.

Although it can be used to find the quantity of a substance, GC gives only one piece of information that bears on the identity of that substance, the retention time.  Mass spectrometry (MS) gives the mass of molecule and usually provides additional information, based upon the mass of fragments of the molecule. Gas chromatography and mass spectrometry (GC/MS) are particularly powerful when they are coupled; gas chromatography separates mixtures, often into pure molecules, and the combination of retention time and the mass of the molecule and its fragments identifies it.

Dr. James Shoemaker’s initial finding of methylmalonic acid in a sample of Ryan’s urine suggested that he also suffered from MMA, and this information was given to the District Attorney and through that office to Patricia Stallings’ lawyer. But the lawyer did not fully appreciate the information.

A biochemist, William Sly, and the man who later became Patricia Stallings’ lawyer, Robert Ritter, independently saw an episode of the television program Unsolved Mysteries about the Stallings case. Mr. Ritter visited Mrs. Stallings in custody and became convinced of her innocence. Professor Sly and a geneticist Piero Rinaldo continued to examine the case, along with Dr. Shoemaker.

Their tests, performed under essentially identical conditions to that employed by one of the labs, showed that ethylene glycol and propionic acid standards emerged from the column with similar but distinct retention times. When a sample of Ryan’s blood was spiked with an internal standard of ethylene glycol, it was unequivocally a distinct substance from the unknown compound, which was shown to be propionic acid by GC/MS. Clearly Ryan also suffered from methylmalonic acidemia. The prosecutor acknowledged that Patricia Stallings’ first lawyer had been ineffective, and she was granted a new trial.  Several months later, the prosecutor’s office dismissed all charges.

One of the most important factors in this wrongful conviction was expectation bias. This caused the first lab to brush off the mismatch between standard and control and caused the second lab to assert a conclusion not based on evidence. Moreover, MMA is a rare disease; Bob Mead implied that if the doctors treating Ryan were more familiar with ethylene glycol poisoning than genetic diseases, it might have colored their interpretation. Indeed, when samples of serum spiked with propionic was sent to three labs, two of the three misidentified the substance as ethylene glycol.

Another key factor was uncritical reliance on a chemical technique that provided limited information. Bob Mead also highlighted the potential contrary information that was ignored: Ryan would have had to consume 300 liters of ethylene glycol to account for the amount of unknown substance found after the second episode (ethylene glycol is rapidly excreted from the body, and about 100 hours elapsed between Patricia’s visit and the test). The innocent presence of antifreeze in the Stallings’ garage was interpreted as evidence of guilt.

With respect to this case, Judge Rick Teitelman said, “I don’t think there is another lawyer in America who could have done what Bob Ritter did.” Clearly Drs. Shoemaker, Sly, and Rinaldo (who was refreshingly blunt about the poor quality of the initial laboratory work) deserve considerable credit for engaging with the legal system. Prosecutor George McElroy III behaved in a reasonable fashion. But there were also elements of chance: David Stallings’ being born with the same metabolic error and that two of the principals happened to independently watch an episode of a television show.


Conger Beasley, Jr., “The Right Man for the Job” Missouri & Kansas Super Lawyers, November 2007.

Bob Mead, “Accused and Convicted—a case study” Proceedings of the Royal Society of Western Australia July 2007.

James Shoemaker, Robert Lynch, Joseph Hoffman, and William Sly, “Misidentification of propionic acid as ethylene glycol in a patient with methylmalonic acidemia,” Journal of Pediatrics 1992, 120:417-421.

Bill Smith, “Not Guilty: How the system failed Patricia Stallings” St. Louis Post-Dispatch


15 thoughts on “Halkides: Expectation Bias and the Patricia Stallings Case

  1. Matt Grawitch

    Expectation bias can be problematic when it comes to the law. Part of the problem is that prior probability estimates (priors) are important for navigating the world, and they are often based on prior experiences and stored information. But the double-edged sword is that they are based on our prior experiences and stored information. That means that as long as the probable explanation comes from that prior experience/stored knowledge, it is a benefit. But when we encounter novel situations or where a novel explanation is the better explanation, it can be quite difficult for us to see and/or determine that. Because much of our information seeking is based on looking for evidence to confirm our priors (as opposed to refuting them), if the evidence lines up with our priors, we are prone to accept the predicted explanation. This is especially the case when we have strong priors. Thus, the more we think some explanation is likely, the more we are subject to filtering in information that confirms that explanation and filtering out or ignoring information that might be misaligned with that expectation.

  2. Hunting Guy

    “and the second lab did not run a standard at all.“


    To me, that statement throws doubt on any work done by that lab.

    You _always_ run standards and blanks in analytical lab work.

    I don’t do biochemistry in my job but we use multiple labs and run the standards against each other and against different labs in our elemental analysis for mining operations. If the results aren’t within a given margin of error we find out why.

    If you don’t use standards there is no guarantee that your instruments are calibrated or your process is done correctly.

    1. Chris Halkides

      Faking a negative control has happened in more than one DNA-based case. The other two problems I have encountered are that the control sample and questioned sample are not treated in an identical manner and that the results of the control are sometimes ignored.

      1. alanlaird

        I’m on the clinical lab side. It does happen that a tech will put out results without good qc in the appropriate time period, but that results in immediate patient reruns and other follow-up.

        Simply ignoring bad qc or, worse, bad calibrations or standards gives me the shivering willies and keeps me up at night.

        1. Grum

          Chris, you are being very diplomatic here. It’s been over 30 years since I was near chemical analysis but the entire process sounds arse-backwards; starting from a conclusion then looking for “evidence” to support it. In fact, I’m cynical enough to suspect that labs who come back with the “wrong” results don’t get too much future business given this level of utter dismissal of any scientific method I’m aware of. The whole point of the bloody thing is to eliminate expectation (or any other) bias.
          Reading between the lines and given the stakes, this is fierce condemnation (yeah, have done GC, MS & NMR in anger, and had to learn the mathematics that make them credible).
          SHG, please, more like this? I already look forward to Seaton’s Friday excursions.

          1. Grum

            And, assuming on the off-chance that you might be unaware of it (addressed to the general readership, not Chris), Richard Feynman’s address on the subject of Cargo Cult Science covers the bases much better than I can and does a job of work demolishing this shit. Respecting SHG’s ban on links, it is nonetheless easily accessible via Google or your search engine of choice.

          2. Chris Halkides

            Dr. Rinaldo was less diplomatic. He said that the quality of the tests were “totally unacceptable,” and he also said that they were, “unbelievable; out of this world.” It is too often the case that the forensic worker receives what is called “domain irrelevant information” that can and sometimes does bias his or her interpretation.

  3. losingtrader

    Can I get a law degree from TV law school?
    Maybe one in a diploma folder that plays the “Forensic Files” theme music when opened.

    The best thing about that show is everything gets resolved in 30 minutes.
    Oh, wait, “Unsolved Mysteries ” is an hour. That’s beyond my attention span , which means you should always select me as a juror as I’m likely to fall asleep and won’t hold it against your client.

    (BTW, no more market advice since I’ve been horribly wrong. My paying client–me–actually did well)

  4. DaveL

    I know very little about analytical chemistry, but I’m pretty floored that the presence of antifreeze at the home was considered in any way probative of anything.

    1. KP

      Yes, its completely stupid to think that is evidence of a person poisoning another. The most basic criminal intelligence would have made sure there was no antifreeze in the house if they’d poisoned someone with it!

      “Ryan would have had to consume 300 liters of ethylene glycol to account for the amount of unknown substance found after the second episode” This should have stopped the trial dead and the science argued about until it was shown the results could be from drinking a feasible amount.

      If you would like to read the law side of novel scientific evidence you can search for ‘The Vicky Calder Case: A Review – Annex Publishers’ where Vicky Calder was tried twice for poisoning her boyfriend with acrylamide. I do science, definitely not law, but the article covers how to get science into the law system & the problems of lay juries inability to handle either.

  5. Mark M.

    This is very, very useful stuff. I should apply for CLE due to the bounty of information and practice resources provided. Thank you to all who participated.

Comments are closed.