statistical analysis

Bilingual

Regatta Starting Stations
2 months ago
- Henley Royal Regatta's historic data shows Berkshire station wins 53.5% of 7555 races, far above the expected 50% for equal stations.
- Statistical analysis confirms the Berkshire station provides a real advantage, as the deviation exceeds 1.645 standard deviations.
- Men's win rate from Berkshire (53.7%) vs women's (52.7%) is not statistically significant, with a chi-squared test yielding a p-value > 0.12.
- A refined chi-squared test comparing heats vs advanced rounds shows significance (p < 0.05), suggesting skill amplifies the Berkshire advantage.
- The Berkshire advantage may be due to wake flow disturbances affecting the Buckinghamshire station more.
- The author warns against p-hacking by regrouping data post-hoc to achieve significance.
Entropic Thoughts
2 months ago
- The claim that the Monaco Grand Prix winner is determined 90% of the time by starting position is false; data from the past 25 years shows first-row winners only 80% of the time.
- Monaco has a higher first-row win rate (80%) than some tracks, but it is not the highest; other tracks like Spain (86%) have higher rates.
- The effect of starting position is confounded by equipage capability (driver and car team performance), which influences both qualifying results and race wins.
- Using a proxy for equipage capability (worst qualifying round results), a logistic regression model for 2022–2025 shows that when holding capability constant, pole position has a 53% win chance on average, while Monaco gives a 70% win chance from pole.
- Monaco's pole position advantage (70%) is significant but not exceptional compared to other tracks like Australia (88%) or Canada (74%).
- The original claim of nine-out-of-ten determination is an overstatement; Monaco is decided by qualifying to some extent, but driver skill and other factors still matter.
Intuition for Distribution Differences
2 months ago
- Cumulative distribution curves show the proportion of the population below a given score.
- A lower cumulative curve indicates higher scores overall, as seen when comparing city-dwellers to the general population.
- The median score is found at the 50% point on the y-axis, with the corresponding x-axis value.
- A treatment can increase score variation, leading to both better scores for most and worse scores for the bottom third.
- Comparing only averages or medians can be misleading; visual inspection of distributions reveals nuanced effects like increased variation.
- Statistical significance testing is the next step to confirm observed differences.
Morphometrics: Introduction to the Analysis of Shape
2 months ago
- Morphometrics is the quantitative study of form, including size and shape, used to analyze distinctions, changes, and taphonomic effects.
- Principal Components Analysis and Principal Coordinates Analysis are key methods for summarizing multiple variables in morphometrics.
- Morphospaces graphically represent quantified shape; theoretical morphospaces use predetermined axes, while empirical ones derive from actual data analysis.
- Function of extinct forms is reconstructed via analogies, phylogenetic inferences, biomechanics, trace fossils, and correlations, though some features may be non-functional 'spandrels'.
- Landmarks are crucial points for comparing specimens, and traditional morphometrics involves measuring distances between these points.
- Phylogenetically independent contrasts correct for phylogenetic biases in statistical analyses, requiring a known phylogeny for meaningful correlations.
- PCA and related techniques reduce multiple variables for analysis, applicable beyond morphometrics to fields like ecology.
- Data acquisition methods include linear and outline methods like Fourier analysis, with adaptations for 3D morphometrics.
- Raup's morphospace exemplifies theoretical morphospaces; advantages include pre-set axes, but some areas may be unoccupied due to functional constraints or taphonomy.
- Empirical morphospaces use PCA-derived axes, better occupying the space but requiring reanalysis for new data.

statistical analysis

Regatta Starting Stations

Entropic Thoughts

Intuition for Distribution Differences

Morphometrics: Introduction to the Analysis of Shape