error and reproducibility

Definition of error in science

mistakes in measurements or procedures that have an impact on research results.

Two types of error

systematic and random

Systematic errors are

consistent and repeatable errors (miscalibrated instruments)

Random errors are

unpredictable and varied errors (human error)

Sampling errors

• small sample sizes are bad - unrepresentative samples.

• Important statistically.

• Eg. Biased selection of subjects.

Why does design matter w/ experimental designs?

control groups and variables.

Randomisation and blinding.

Role of stats in research… and why?

basic concepts (mean mode median).

So we can understand data.

When interpreting data make sure…

• to avoid pitfalls.

• Not to misinterpret results.

• Thorough analysis

Reproducibility definit.

• ability to replicate study findings using the same methods.

• repeating experiments to verify results.

• Important for scientific validity.

Why reproducibility is important

scientific progress.

Building on reliable research.

Key to success of drug development.

Drug efficiency and safety.

Consequences of irreproducibility

waste resources and delay drug development.

Factors influencing reproducibility

methodology and reporting.

Clear protocols.

Detailed reporting.

Improving reproducibility

standardise methods.

Open sharing of data.

What is peer review

when someone looks at someone else’s work and reviews it - checking for errors

Ethical considerations

honesty & transparency.

Consequences of unethical behaviour

Roles of funding

influences research - transparency in funding sources & impact on research priorities.

Reproducibility in Pharma

important for drug development.

Industry standards and practices.

Key issues leading to errors (3):

1. Complex biological systems - small change affects outcome.

2. Experimental design errors - poor controls / biases / selective reporting.

3. Preclinical-to-clinical transition issues - differences in human and animal biology.

Case example: failed cancer drug development

• cancer drugs show 80% tumor reduction preclinically.

• Follow up trials couldn’t reproduce the results.

• Contaminated reagent = the cause of error.

• Result = millions in losses and project delays.

3 consequences of poor reproducibility

1. Financial costs.

2. Delays in drug approval.

3. Public health impact

Improving reproducibility in drugs development

1. Better experimental design - standardise protocols.

2. Open science & data sharing - transparency to verify.

3. Collaborative validation - independent replication projects.

4. AI integration - to flag inconsistencies & predict reproducibility.