Replication and Reproducibility in Research

The video discusses the importance of replication in scientific research, highlighting the reproducibility crisis where many studies fail to replicate. It explores causes such as statistical variability, misuse of p-values, and publication bias. The video suggests solutions like more replication studies, better understanding of p-values, and sharing data. It uses examples like social priming and power posing to illustrate these points.

Why is replication considered essential in scientific research?

• It helps in confirming the validity of results.

• It ensures that all studies are published in top journals.

• It allows for faster publication of studies.

• It reduces the cost of research.

What percentage of cancer treatment studies were successfully replicated by Amgen scientists?

• 50%

• 75%

• 25%

• 11%

What is one of the less nefarious reasons for the lack of replicability in scientific studies?

• Lack of funding for research

• Variability in data analysis methods

• Over-reliance on technology

• Intentional data falsification

What is a common misunderstanding about p-values in scientific research?

• They indicate the importance of a result.

• All of the above

• They measure the size of an effect.

• They are the only factor in determining statistical significance.

What is the impact of small sample sizes on study results?

• They increase the likelihood of skewed results.

• They ensure all results are significant.

• They reduce the need for replication.

• They lead to more accurate results.

What did the American Statistical Association emphasize about p-values in 2016?

• They should be the sole basis for scientific conclusions.

• They do not measure the size or importance of an effect.

• They are irrelevant in modern research.

• They should always be below 0.01.

What is the false discovery rate in the social priming example provided?

• 60%

• 25%

• 42.9%

• 10%

What is one proposed solution to improve the reproducibility of scientific research?

• Increasing the p-value threshold to 0.1

• Publishing more null results

• Reducing the number of replication studies

• Focusing only on significant results

What is one reason why replication studies are not as popular as new discoveries?

• They are less expensive.

• They attract less media attention.

• They are easier to conduct.

• They always confirm the original findings.

What is one benefit of sharing research data more widely?

• It increases public trust in science.

• It eliminates the need for replication.

• It reduces the need for peer review.

• It allows for faster publication.