Week 10

Question 1

replicability vs reproducibility

Answer 1

reproducibility (of results) requires replicability (methodology)

Question 2

replicability depends on assumptions

Answer 2

normality- parametric vs nonparametric
the correct model (linear vs non linear) for research q and data

Question 3

replicability depends on transparent methods

Answer 3

being able to repeat the tests in another lab
justifiable decisions and hypotheses that are not made after results are known
correcting for multiple tests
researcher degrees of freedom

Question 4

Simpson’s paradox

Answer 4

across-subject correlations are reversed within individuals present in a sample

Question 5

Generalisability

Answer 5

depends on lack of bias
bias when data is missing and not included
can use imputation to avoid this bias

Question 6

imputation

Answer 6

method for modelling what missing data is likely to have been given its associations with other variables that are not missing
can avoid bias

Question 7

Solutions for false positives

Answer 7

full methodology reported
ethical data and code sharing
effect sizes instead of only p values
pre-registration and registered reports
specification curve analyses
meta analysis
more journals accepting null results
stop incentivising producing most number of publications over most rigorous methodology
keep doing replication work