Stat 5870 section 2 - Copies of handouts and computer-drawn graphs and pictures shown during lecture and lab

Lecture on 30 Aug

• Histograms for creativity data
• Four data sets used to illustrate dot and box plots. (rotated version)
• Graphs to illustrate spread

Lab on 28 Aug

• Quick comparison of JMP, R, and SAS software

Lecture on 30 Aug

• Results from lab 1
• Results from last year's lab 1
• Tables from Svedberg 1922
• Tables from a 2010's paper

Lecture on 4 Sept

• Randomization distribution for tomato yield data
• Randomization dist. with p-value computation
• Samples from bank salary population
• Standard error of the mean, bank salary population
• Kelley 1924, Science paper on reporting statistics

Lecture on 9 Sept

• Randomization distributions for 4 data sets
• Randomization distributions of diff/se for 4 data sets
• T distributions for 4 different degrees of freedom
• Comparison of randomization distribution and T distributions for 4 data sets
• Guinness brewery, Dublin
• First page of "Student" 1908

Lecture on 11 Sept

• Bumpus's sparrows, weight dot plot
• Bumpus's sparrows, weight box plot
• Schizophrenia study (case 2.2), data as 2 samples
• Schizophrenia study (case 2.2), data as paired

Lecture on 13 Sept

• Comparison of 2 sample and paired data
• Variability of differences: effective and not effective pairing

Lecture on 16 Sept

• 95% confidence intervals for 100 data sets
• 99% confidence intervals for same 100 data sets

Lecture on 18 Sept

• Cognitive Load study (case 4.2)
• Hamburger E.coli data
• Hamburger study info and Wilcoxon test
• Cognitive Load: Randomization distribution of rank sum statistic

Lecture on 20 Sept:

• NP analyses of Schizophrenia data

Lecture on 23 Sept:

• Residual diagnostic plots for creativity and hamburger data
• Normal QQ plots for creativity and hamburger datasets.
• Normal QQ and residual vs predicted plots for tomato data
• Normal QQ plots for simulated data. All normal. 4 plots with 20 observations; 4 plots with 200 observations.
• Residual vs predicted value plots for simulated data, n = 10 per group
• Residual vs predicted value plots for simulated data, n = 100 per group
• Display 3.4 and 3.5 from book - robustness to non-normality and unequal variances
• Display 3.6 from book Strategy for outliers
• Flowchart showing reasons why subjects excluded. One example of something required by the CONSORT document.
• Link to CONSORT statement web site Reporting standards for for human trials, including nutrition
• Link to Meridian web site Collects reporting standards for for research on animals. Includes links to Arrive, Reflect and others.

Lecture on 27 Sept

• Dot plot of bacterial content of apple cider
• Residual plot of bacterial content of apple cider
• Dot plots of log transformed bacterial content of apple cider

Lecture on 2 Oct

• Standard error of estimated probability

Lecture on 4 Oct

• Log odds ratio for pairs of probabilities

Lecture on 11 Oct

• Treatments in the diet and longevity study
• Plot illustrating equal means (red) and separate means (blue) models for the mouse diet and longevity study
• Plot illustrating equal means (red) and separate means (blue) models for 6 sets of randomly generated data. In all 6 cases, the true model is equal means.

Lecture on 14 Oct

• Boxplots for CCK enzyme activity data
• Residual Plots for Diet study and CCK enzyme activity data
• Residual Plots for CCK enzyme activity data Top panel is analysis of raw data; bottom is analysis of log transformed data.
• Boxplot of data from handicap study. Case study 6.1

Lecture on 16 Oct

• Bioreactor product data

Lecture on 21 Oct

• Consequences of adjusting p-values
• Plot of meat pH over time

Lecture on 25 Oct

• Plot of aflatoxin concentration vs % clean peanuts

Lecture on 28 Oct

• Biorenewables 2nd study: Plot of product against incubation time
• German 10 DM note, with CF Gauss
• Galton 1886 title page
• Galton's data
• Galton's prediction machine

Lecture on 30 Oct

• Peanut data plots plots of y vs x, y vs shifted x, confidence interval for y given x, prediction interval for y given x

Lecture on 1 Nov

• Comparison of y vs x and x vs y
• Overlay of the two fitted lines
• Galton's regression lines

Lecture on 4 Nov:

• Lack of fit plot and ANOVA for meat data
• ANOVA tables for components of the LOF test
• Annotated handout on lack of fit model comparison
• R and SAS output (JMP is similar) for ANOVA lack of fit test
• Plots of data and residuals for four situations
• Plot of Anscombe data set #1, r = 0.81, r2 = 0.67
• Plots of Anscombe data sets
• Residual plots for Anscombe data sets
• How choosing values of X changes the correlation
• Earwings
• Monograph on the order
• Correlation of Earwing weight and ear length
• Earwing correlation by island

Lecture on 6 Nov:

• Donner party data
• Donner party logit P[survive] vs age
• Donner party P[survive] vs age
• Donner party P[survive] vs extended age

Lecture on 11 Nov:

• Brain weight scatterplot matrix Untransformed variables.
• Brain weight scatterplot matrix Log transformed variables.
• Log Brain weight: simple linear regression on log litter
• Log Brain weight: multiple linear regression on log body and log litter
• Contour plot multiple linear regression on log body and log litter
• Log Body weight: simple linear regression on log litter

Lecture on 18 Nov:

• Plot of historical corn yield vs rainfall
• Residual plots for linear and quadratic models
• Spline fit to corn yield vs rainfall data
• Spline fit to corn yield vs rainfall data, with extrapolation
• Field of meadowfoam
• White meadowfoam flower
• meadowfoam data
• meadowfoam data with additive model fits
• meadowfoam data with interaction fits
• meadowfoam data with quadratic fits

Lecture on 21 Nov:

• but not in either component
• Outlying X in 2d
• Plots of Anscombe data sets
• Cooks Distance plots for Anscombe data sets
• Residual plots for Anscombe data sets
• Std errors of predicted values and residuals
• Residuals and standardized residuals

Lecture on 2 Dec:

• Illustration of overfitting data when model has too many parameters
• AIC model selection for SAT data in SAS and R

Lecture on 9 Dec:

• Binomial and Poisson probabilities
• Elephant data
• Elephant data
• Elephant data characteristics
• Elephant data + Fitted Poisson regression
• Elephant data + Linear + Quadratic
• Spline fit version of a linear regression
• Spline fit version of a quadratic regression
• Linear spline, 1 knot
• Linear spline, 1 knot, possible fits
• Linear spline, 4 knots
• Linear spline, 4 knots, possible fits
• Data + spline, very wiggly
• Data + spline, no wiggles
• Data + spline, optimal wiggles
• Thin plate spline

Lecture on 9 Dec:

• Regression tree (CART) predictor
• Over- and under-fitting, 1 sample
• Over- and under-fitting, 20 samples and average
• Collard study, as 4 treatments
• Collard study, as 2 way factorial, version 1
• Collard study, as 2 way factorial, version 2