1. Bad Science Flashcards
Dirk Smeesters
Had 7 papers retracted due to p-hacking
Diedrick Stapel
Data fabrication of at least 30 papers
P-Hacking
Manipulating data to get a significant result (p>0.05)
Replication Crisis
It is is important to replicate studies to determine the reliability of results. Replication is rarely done due to lack of interest or funding, or bad practice making replication difficult
File Drawer Problem
Studies with non-significant results are far less likely to be published
Confidence Intervals
This is a range computed from a specific set of data. In this range, we have the estimated population of scores in data can fall under. We can determine the probability of the true value of a given test being within this range of values based on the confidence level. E.g. an interval with the confidence level of 99% has a 99% probability that the true value is somewhere within the range.