Paper - Revisiting the Marshmallow Test: A Conceptual Replication Investigating Links Between Early Delay of Gratification and Later Outcomes

  • Metadata:
    • author: Tyler W. Watts, Greg J. Duncan, Haonan Quan
    • title: Revisiting the Marshmallow Test: A Conceptual Replication Investigating Links Between Early Delay of Gratification and Later Outcomes
    • year: 20182018

    • url: Weblink
  • Background:
    • Shoda, Mischel, and Peake‚Äôs (1990) showed in their famous "Marshmallow Test" study that the ability to delay gratification strongly correlated with later success in life.
    • The original sample was lacking in terms of size and diversity (high ses)
  • Research Question:
    • Can the results of the Marshmallow Test be replicated for a larger and more diverse sample? (family background, cognitive ability, environment)
  • Method:
    • Sample size of 10x the originals size
    • 3 measuring points
      • 54 months: Classic Marshmallow test (reward delay task -> measure time (7 minutes max)
      • First grade: WJ-R test for academic success, CBCL (child behavior checklist)
      • 15 years old: WJ-R test, CBCL
  • Results:
    • No statistically significant correlation between the ability to delay gratifiction at 54 months old and either academic success or child behavior at both testing times
    • No effect even if only positive SES is considered, like in the original
  • Interpretation:
    • The results of the Marshmallow experiment could not be replicated for a bigger and more diverse sample. Even a sample that resembles the original one does not yield significant effects.
    • The ability to delay gratification does not correlate with later success in life