Kyle Polich avatar

Dexa/Kyle Polich

Learn more
Published Mar 20, 2020

Visualizing Uncertainty

Dive into the world of statistical data with Dr. Loni Besancon as she unravels the complexities of p-values, the cliff effect, and visualization techniques to promote transparency and enhance data interpretation, paving the way for better research practices and understanding.
Episode Highlights
Data Skeptic logo

Topics covered

Popular Clips

Episode Highlights

  • P-Values

    and engage in a lively discussion about the role of p-values in statistical research. Loni challenges the conventional use of the 0.05 p-value threshold, describing it as arbitrary and emphasizing the need for a more nuanced understanding. She argues that smaller p-values indicate a stronger ability to reject the null hypothesis, but the specific threshold should not be rigidly applied across all fields 1.

    Being on 0.05 doesn't really mean anything for me. It's just a binary threshold that people put there that's completely random.

    ---

    Kyle reflects on the broader implications of relying on p-values, questioning their role in the reproducibility crisis and the potential shift away from traditional statistical methods 2.

       

    Transparency

    The conversation shifts to the importance of transparency in statistical methodologies. Loni argues against banning p-values and confidence intervals, suggesting that the focus should be on educating researchers to use these tools correctly and ensuring data transparency 3. She highlights the risks of a replication crisis in computer science due to a lack of transparency and dichotomous interpretations in research 4.

    I think transparency is the one thing that matters the most and not the tools per se.

    ---

    Kyle agrees, emphasizing the need for openness and the potential benefits of alternative visualizations to improve data interpretation 3.

Related Episodes