I don't know what you mean by margin of error. The results in the paper are a little confusing because they report mean and standard deviation while t-tests use

I don't know what you mean by margin of error. The results in the paper are a little confusing because they report mean and standard deviation while t-tests use mean and standard error. In Table 3, they report the 95% confidence interval for the treatment condition (i.e., receiving cookies) as 0.67–11.62, so it seems that the difference between the two conditions is significant at the 5% level. Interestingly, the reported p-value here is "just" 0.02 instead of the <0.001 that they include in the abstract.