NHSTs Shortcomings + Meta Analysis - Week 9 Flashcards
1
Q
Two main issues of NHST
A
- Distracts from deeper question: H0 rarely captures hypothesis / challenges it
- Focus on effects distracts attention from better questions
2
Q
What should we focus on in NHST to overcome issues
A
- P values ≠ direct truth
- Non significant results that are missing literature, effect sizes are over estimated
3
Q
What three things should we be focussing on to conduct a good analysis of a study?
A
Effect sizes (regression slope)
Precision (CI)
Better / Simpler explainations
4
Q
Why are effect sizes often overestimated?
A
Non-significant results not published
Researchers run study over and over until they find a significant result by chance
5
Q
Define meta analysis
A
Combining effect sizes across multiple studies to get a better estimate of population effect size
Larger N = More accurate estimate
