Week 9

• Non-parametric data


  when two variables do not meet parametric conditions

  • interval data
  • approx equal variance
  • normality
• Kendall's tau


  Better than Spearman's for small samples
• Spearman's rho


  Pearson's correlation on ranked data

  
• Kendall's tau

  
• Bootstrapping


  • another alternative to non-parametric correlations
  • a statistical procedure that re-samples a single dataset to create many simulated samples
• Partial correlation


  • measures relationship between two variables
  • controlling for the effect that a third variable has on them both
• Semi-partial correlation


  • measures the relationship between two variables
  • controlling for the effect that a third variable has on only one of the variables
• Use of Gaussian function (normal)

  Statistics

  Noisy measurements
• Use of Quadratics


  Non-monotonic responses (i.e stress responses or cellular responses to temp changes)

  Trend that can change direction
• Use of sigmoids


  To model a system with two states which can switch between the states

  
• Use of Exponential Growth

  Growth rate under unlimited growth

  
• Exponential decay


  Decay of drugs metabolised by the body, radioactive decay

  
• Use of Logistic growth

  Population growth with limited resources

  
• Early career researchers (ECRs)


  enthusiastic about improving reproducibility
• Limitations of ECRs


  Low statistical power leads to unreliable results, wasting time and resources.

  • ECRs face challenges due to resource constraints in conducting adequately powered studies.
• Solutions for ECRs


  • Pivot: Focus on related, feasible research questions.
  • Collaborate: Work within larger consortia to share resources and data.
  • Use shared data: Leverage openly available datasets to address research questions.
  • Embrace theory and computation: Focus on modeling, analysis, and computational approaches as alternatives to costly data collection.
• Cons of reproducibility


  • Improved reproducibility practices (e.g., pre-registration, larger samples) reduce productivity in terms of publications, which may hurt job prospects in the short term.
  • Advocacy for scientific integrity and prioritizing robust methods over "splashy" findings is necessary.
  • The academic system still emphasizes quantity and high-impact results, which undermines reproducibility efforts.
• Reproducibility


  the ability to repeat an experiment or study and obtain the same or very similar results

  • ensures that findings are reliable and not just the result of random chance or errors.
  • when research is reproducible, other scientists can follow the same methods and check if they get the same outcomes, strengthening the validity of the conclusions.
• Reproducibility vs Replicability


  Reproducibility

  • ability to achieve the same results when using the original data and methods provided by the original researchers.
  • getting the same outcomes when you use the same setup, tools, and data

  Replicability

  • the ability to achieve similar results when a different research team independently conducts the study with new data, possibly following the same methodology.
  • about the consistency of results when the study is repeated in different settings.