Examining Wrongful Convictions Allison D Redlich

What We Know About Examining Wrongful Convictions

Allison D. Redlich is a forensic psychologist whose research has become one of the more cited bodies of work on why innocent people end up behind bars. Her approach isn't the typical courtroom drama version of wrongful convictions — it's empirical, data-driven, and occasionally uncomfortable for people who like to believe the system self-corrects. I've spent years reading through her papers and the ones that build on her methodology, so here's a straightforward breakdown of how her work functions and how you can actually use it. Redlich's most influential contribution comes from a 2015 study published in Law and Human Behavior, where she and her colleagues conducted an experiment that tested the false confession pipeline in real time. The setup was relatively simple: participants completed a computer task where they were explicitly warned not to click on a specific area of the screen. Then the experimenter "accidentally" wiped the data. Some participants were offered a chance to restore their data by confessing to pressing the forbidden key — even though there was no technical way to prove they did it. The false confession rate in that study was striking: around 70% of participants who were told a witness had seen them press the button admitted to doing it. A smaller but still significant number confessed even without any witness claim. That finding alone changes how you think about interrogation psychology. Most people assume false confessions happen to vulnerable populations — low IQ, juveniles, people with psychiatric conditions. Redlich's work showed something more troubling: under the right pressure, a large proportion of otherwise normal, functioning adults will confess to things they didn't do. The mechanism isn't compliance alone. It's what she calls internalization, where after prolonged questioning, the person starts genuinely doubting their own memory.

How to Access and Apply Her Work

If you're looking to engage with her research directly, her publications are available through academic databases like PsycINFO, Google Scholar, and SSRN. The full-text versions typically go behind paywalls unless your institution has a subscription. I usually pull her papers through my university's proxy or use the open-access copies on ResearchGate when available. It's not glamorous but it saves about twenty minutes compared to waiting for interlibrary loan. One paper that's particularly dense but worth the effort is her 2008 study on contributing factors to wrongful convictions, which she conducted using case analysis of exonerations documented by the National Registry of Exonerations. What makes her methodology different from earlier reviews is that she coded for specific causal pathways rather than just listing error types. She tracked whether a single factor caused the conviction or whether multiple factors interacted in a compounding way. The distinction matters because policy recommendations based on single-factor models tend to fail when applied to real cases.

Practical Issues You'll Hit

When I first tried to replicate parts of Redlich's coding framework for a project I was working on, I ran into a problem with the interaction-variable classification. Her original study identified scenarios where three or more errors combined to produce a conviction — things like eyewitness misidentification paired with forensic error and official deception. The coding scheme treated these as additive, but in practice I found that some factor combinations appear together because they share an underlying cause, not because they're independent. A rushed investigator might produce both a flawed eyewitness procedure and coerced testimony simply because they entered the interview process with a fixed theory of guilt. Redlich's framework doesn't fully account for that kind of systemic contamination. My workaround was to add a meta-factor code for "investigative tunnel vision" and recode cases where multiple errors seemed driven by a single cognitive bias. This slightly changed the distribution of results — tunnel vision turned out to be a stronger predictor than any individual error type. It's a small adjustment but it made the data actually useful for what I was trying to do, which was building a risk-assessment model for parole hearings in cases with questionable evidence chains.

What Beginners Miss

There's a common misconception about Redlich's work that I see repeated in undergraduate papers and even in some legal scholarship. People treat her research as if it proves the system is broken. It doesn't prove that. What it does is quantify the probability that specific procedural failures lead to wrongful convictions under specific conditions. That's a more useful distinction than most writers acknowledge. The false confession rate from her lab study tells you about susceptibility under experimental conditions — it doesn't tell you the rate of false confessions in actual criminal investigations, which would require different methods and likely produce different numbers. Another thing people overlook: Redlich has done important work on eyewitness identification procedures that predates her false confession research. Her 2006 experiments on line-up administration showed that double-blind procedures reduce false identifications by roughly 30 to 40 percent compared to non-blind procedures where the administrator knows the suspect's identity. This isn't a subtle effect. It's large enough that most jurisdictions that have adopted blind line-up procedures since 2010 should be seeing measurable improvements in conviction accuracy. The data from states that implemented these reforms supports that, though implementation quality varies enormously.

Limitations and Where the Work Doesn't Reach

No researcher is going to tell you this, so I will: Redlich's framework has real limitations when applied to cases involving new forensic evidence or investigative techniques that existed after her major studies were published. She's primarily working with exoneration data from the 1980s through early 2000s. Some of the error patterns in newer cases — false evidence ploys that incorporate DNA database searching, behavioral analysis used as pseudo-scientific interrogation tactics, or the specific ways surveillance data gets misinterpreted by juries — aren't well-covered by her existing coding scheme. Also, her work focuses heavily on procedural errors within the U.S. criminal justice system. If you're applying this to other jurisdictions or to civil commitment proceedings, the error rates and causal pathways shift. I've seen people try to apply her false confession risk factors to immigration detention interviews and get results that don't hold up. The interrogation dynamics in those settings are different enough that you'd need to recalibrate. For people who want a broader baseline before diving into her specific methodologies, the National Registry of Exonerations database at the University of Michigan is the most comprehensive public resource available. It's updated regularly and includes case-level detail that lets you cross-reference Redlich's findings against the actual exoneration records. The raw data downloads are free. The interface is clunky but functional.