Posts

In the previous post, we walked through the basic theory of maximum likelihood and coded up some examples. In these models, the relation between the predictors and the response was linear and the error around the expected value was normally distributed.

In this post we will lay out the fundamentals of the theory of likelihood estimation. First, we will sketch out the problem of inference in general. The likelihood theory is one solution to this problem, based on assumptions that we will make explicit.

We return once again to the study that investigated the effect of a grit intervention on math scores. Even though the study was an experiment with thousands of participants, it couldn’t shake off the threat of lurking confounders because the treatment was randomly assigned at the school level.

In a previous post we discussed the design of a field experiment by Doug Rohrer and colleagues, in which math questions were blocked for some classes and mixed for other classes.

According to a number of psychologists, we’ve been teaching math the wrong way. Instead of hammering away at the same topic for weeks, learners of math should mix up, or interleave, the concepts they practice.

In previous posts on the Grit study we saw that it was difficult to estimate the effect of the grit intervention on math scores, even though we have data from a large experiment involving thousands of children.

In the previous post we have explored the data visually and dealt with missing data. Here we turn to inference. Frequentist inference We first attempt to replicate the results of Sule and colleagues by building a classical regression model.

Students are often told that perseverance pays off in the end. But does it? And if it does, can it be instilled in students? According to the authors of a large experiment in Turkey, the answer to these questions is yes.

In recent years, the idea that political elites do not respond to the poor has gained ground. It is held by a diverse set of people, ranging from populist leaders, yellow vest protesters to leftist academics.

Students and teachers are united in their dislike of tests. Students get stressed out by cramming and teachers by grading. According to a growing movement among researchers of education, this sentiment is unfortunate, because they think that taking a test after learning improves the score on a subsequent test.