Filter Company News
  • Revisiting DOAR’s “Making a Murderer” Coercion Research
    Revisiting MAM

    Last week, lawyers for Brendan Dassey, whose conviction was documented in the Netflix series “Making a Murderer,” asked the U.S. Supreme Court to review a federal appeals court decision that ruled a young Dassey’s confession was voluntary. He was 16 years old at the time he confessed to helping his uncle, Steven Avery, rape and murder photographer Teresa Halbach in 2005, and according to court filings, he has suffered from intellectual disabilities most of his life. Dassey’s attorneys claim investigators took advantage of his youth and mental deficiencies to coerce a false confession, and now, the case that became a cultural phenomenon is potentially poised to have a widespread impact on our legal system should SCOTUS throw that confession out.

    “Making a Murderer” and the popular podcast “Serial,” in which “This American Life” Producer Sarah Koenig investigated the suspicious circumstances surrounding the conviction of Adnan Syed, both came out in 2015. At the time, DOAR wanted to measure how these kinds of documentaries might be shaping potential jurors’ perceptions, including their willingness to accept that false confessions do occur. Research conducted prior to the release of “Making a Murderer” found that jurors who hear a confession are more likely to convict than those who don’t and even when they could identify an interrogation as coercive, most still believed false confessions to be unlikely .

    DOAR’s 2016 study, Making a Murderer May Make Better Defense Jurors, indicated a shift toward potential jurors’ willingness to accept the idea that false confessions do occur, and that the change in perceptions may have been linked, at least in part, to these types of documentaries. It is interesting to revisit this research as another chapter of the case may play out in our country’s highest court.

Share This: LinkedIn Twitter Google+ Facebook Email Reddit Digg

Leave a Reply or Subscribe

Your email address will not be published.