PS 101 Unit 2 research methods explanation example

What Waiting for a Second Marshmallow Can and Cannot Tell Us: A Research Methods Explanation of a Longitudinal Correlational Design and Its 2018 Replication

Student Name

Psychology, Herzing University

PS 101: Psychology

Instructor Name

January 21, 2026

What Waiting for a Second Marshmallow Can and Cannot Tell Us: A Research Methods Explanation of a Longitudinal Correlational Design and Its 2018 Replication

The Research Question

Does a preschool child's ability to wait for a larger reward predict how that child does years later? That is the question behind one of the best-known studies in psychology, often called the marshmallow test. This paper walks through how the original research answered it, what kind of conclusion its design allows, and what happened when a later team asked the same question with a different sample.

The Design, Step by Step

Who took part

The original work began in the late 1960s and early 1970s with children at a preschool on the Stanford University campus. Mischel et al. (1989) summarized this research program. The children were roughly four years old, and most were children of university faculty, staff and graduate students. That detail matters later.

What was measured first

Each child was shown two rewards, such as one marshmallow and two marshmallows, and told that the experimenter would leave the room. If the child waited until the experimenter returned, the child would get the larger reward; the child could ring a bell at any time to bring the experimenter back and receive the smaller one. The measure was simply how many seconds the child waited. Different versions of the procedure changed details, such as whether the treats were visible, which is itself an important part of the research.

What was measured later

Years afterward, the researchers contacted the families again. Shoda et al. (1990) reported that children who had waited longer as preschoolers were, as adolescents, rated by their parents as more competent and had higher scores on the SAT college entrance exam. The relationships appeared most clearly in particular versions of the original task.

What was compared

The comparison is between two measurements of the same children taken years apart: waiting time at about age four and outcomes in adolescence. No child was assigned to wait longer or shorter. The researchers measured a naturally occurring difference and then looked at whether it went along with later differences.

What the Design Can Establish

This is a longitudinal correlational design. Longitudinal means the same people were measured at different times. Correlational means the researchers measured variables as they occurred rather than manipulating them. A design like this can establish that two things go together and that one came first in time. It can show that children who waited longer tended, years later, to have better adolescent outcomes.

It cannot establish that waiting caused those outcomes. Because nothing was manipulated across children, any third factor that affects both waiting at age four and success at age fifteen could produce the same pattern. The tidy summary often repeated in popular writing, that self-control in early childhood causes later success, goes beyond what the design can support.

Control, and What Was Left Free to Vary

Within each session the procedure was standardized: the same instructions, the same room, the same rewards. That control makes the waiting measure comparable across children. Across children, however, many things were left free to vary, including family income, parents' education, the child's general cognitive ability and the home environment. Those uncontrolled differences are the alternative explanations the design cannot rule out.

Why the Sample Matters

A sample is the group of people a study actually measures, and a population is the larger group the researchers hope to describe. The original children were a convenience sample, recruited because they attended a university preschool, and their families were unusually well educated. That does not invalidate the findings; it narrows how far they can be generalized. If most families in a sample can be relied on to deliver a promised reward, waiting is a sensible choice for nearly every child, and differences in waiting may reflect something about the child. In a broader sample that includes families facing uncertainty about money or food, waiting may reflect what a child has learned about whether promises are kept. The same behavior can mean different things in different populations, which is exactly why replication with a different sample is informative rather than redundant.

Sample size matters too. Follow-up studies lose participants over the years, and the adolescents who could be found and who agreed to take part may not resemble those who could not. A smaller and less representative follow-up group makes any correlation less stable.

What Happened in a Replication

Watts et al. (2018) asked the original question with a different and much larger sample: about 900 children from a national study of early child care, more diverse in family background than the Stanford preschool. They measured delay at age four and achievement at age fifteen. They found an association between waiting and later achievement, but it was about half the size of the earlier findings. When they statistically accounted for family background, early cognitive ability and the home environment, the association shrank by about two thirds.

This is an important lesson about methods. The replication did not show that the original researchers were careless. It showed that a correlational design with a narrow sample can overstate a relationship, and that much of the link between waiting and later success may be explained by factors that affect both.

The Inference the Design Supports

Stated carefully, the inference is this: among the children studied, the ability to delay gratification at age four was associated with later outcomes, more strongly in a narrow sample and more weakly in a broad one, and much of the association overlaps with family background and early ability. Three alternative explanations remain open. Family resources may help children both wait and succeed. General cognitive ability may support both. And children may wait less when they have learned from experience that promised rewards do not always arrive, which would make waiting partly a measure of the child's environment rather than of a stable trait.

To show that self-control causes later success, a researcher would need an experiment: randomly assigning children to a program that improves self-control and then comparing their later outcomes with a control group. Until then, this research tells us something real about how childhood waiting and later outcomes go together, and much less about why.

References

Mischel, W., Shoda, Y., & Rodriguez, M. L. (1989). Delay of gratification in children. Science, 244(4907), 933-938. https://doi.org/10.1126/science.2658056

Shoda, Y., Mischel, W., & Peake, P. K. (1990). Predicting adolescent cognitive and self-regulatory competencies from preschool delay of gratification: Identifying diagnostic conditions. Developmental Psychology, 26(6), 978-986. https://doi.org/10.1037/0012-1649.26.6.978

Watts, T. W., Duncan, G. J., & Quan, H. (2018). Revisiting the marshmallow test: A conceptual replication investigating links between early delay of gratification and later outcomes. Psychological Science, 29(7), 1159-1177. https://doi.org/10.1177/0956797618761661