Sessions are held 10:00-11:00am US Eastern Time on most Fridays. Links labeled “Offline discussion” for each section’s outline below will take you to a dedicated datamethods.org topic for in-depth or post-live session discussions. You can go directlly to the discussion topic for session n by going to `bit.ly/datamethods-bbrn`

e.g. `...-bbr1`

. Short questions and comments made during the live broadcast should be entered in YouTube’s live chat. Here is the link to the YouTube channel, and after each video premieres you can go directly to the video for session by going to `bit.ly/yt-bbrn`

.

**NOTE**: Beginning with Session 15 on 2020-04-03 and lasting at least until the coronavirus crisis subsides, the course is shifting to a “study at your own pace” mode with the scheduled 10am-11am EDT time slot set aside for questions, answers, and discussions. Discussions will take place on the `datamethods.org`

topic listed for each session, supplemented by live video/audio/text chat on Zoom. This mode uses the new HTML version of the course notes at hbiostat.org/bbr/md which will automatically size to any viewing device. When studying the notes before the online meeting, listen to audio narration and watch videos that are linked from the notes in the right margin. For any questions or discussion on `datamethods`

add any mnemonic shown in blue boxes in the right margin of course notes (e.g. `reg:ancova`

) somewhere in the `datamethods`

topic so that the course notes will automatically link to that discussion. Contact Frank Harrell if you would like to have a new mnemonic added to the course notes for linking with a discussion you add in `datamethods`

.

- What is biostatistics?
- Statistical scientific method
- Types of data analysis
- Types of measurements
- Optimum response variables
- Offline discussion

Moderators: Donald Szlosek (IDEXX Inc), Matt Shotwell (Vanderbilt)

- Random variables
- Probability
- Distributions
- Descriptive statistics
**Watch in advance**: 2-3 of the videos linked from the camera icons in Section 3.9- Offline discussion

Moderators: Donald Szlosek (IDEXX Inc), Tom Stewart and Dandan Liu (Vanderbilt)

- Statistical graphics
- Tables | separate video
**Watch in advance**: John Rauser’s video- Offline discussion

Moderators: Donald Szlosek (IDEXX), Matt Shotwell and Tom Stewart (Vanderbilt)

This session will be pre-recorded; for questions/discussion use only the offline discussion link below.

- Statistical inference
- Hypothesis testing vs. asking questions
- Branches of statistics
- Errors in hypothesis testing
- Misinterpretations of p-values
- Interval estimation
- Offline discussion

This session will likely be held live through a YouTube premier by which a recorded video is played but the instructor engages in live chat with participants. If this works the video will be viewed on the BBRcourse YouTube Channel at 10am US Eastern Time.

- One sample test for means (Frequentist + Bayesian)
- One sample method for a probability (Frequentist + Bayesian)
- Paired data and one-sample tests
- Offline discussion

This session will be broadcast on the BBRcourse YouTube Channel at 10am US Eastern Time Friday 2019-12-06.

- Two-sample test for means (Frequentist + Bayesian)
- How not to present results involving P-values
- Offline discussion

- Study design considerations
- Sizing a pilot study using precision of SD
- Problems with standardized effect sizes
- Choice of effect size
- Multiple hypotheses and reasons not to multiplicity-adjust
- More about ordinal outcomes
- Study design big picture

- Crossover experiment example
- Offline discussion

- Comparing two proportions
- Normal approximation and Pearson \(\chi^2\) test
- Why not to use Fisher’s “exact” test
- Sample size
- Relative effect measures, especially odds ratio

- Introduction to logistic regression, for comparing proportions
- Bayesian logistic regression for comparing two proportions

- Nonparametric statistical tests
- When to use them
- One-sample Wilcoxon test
- Two-sample Wilcoxon test
- Confidence intervals for medians and differences in medians

- Generalization: proportional odds ordinal logistic model
- Power and sample size for Wilcoxon test and PO model
- Converting odds ratios in PO model to differences in means/medians
- Advantages of PO model over Wilcoxon-Kruskal-Wallis tests

- Pearson and Spearman correlation coefficients
- Correlation vs. agreement; Bland-Altman plots
- Sample size calculation
- Why not to compare two correlations
- Avoiding overinterpretation when correlating many variables

- Watch any of the 12 videos you haven’t seen yet
- Ask questions about material in session
*n*at http://bit.ly/datamethods-bbrn

- Stratification vs. matching vs. regression
- Purposes and advantages of statistical modeling
- Overview of nonparametric regression (smoothers)
- Simple linear regresion

- Proper transformations vs. percentiling
- Multiple linear regression
- Two-sample t-test vs. linear regression
- Using regression for ANOVA
- ANCOVA

- Course notes: hbiostat.org/bbr/md with audio narration and a few videos
- Zoom recording
- datamethods.org discussion
- Topics
- Analysis of continuous lead exposures with multiple linear regression (Chapter 9 Sections 9.4-9.9)
- Two-way ANOVA (Chapter 10 Sections 10.10.5-)
- Interaction
- Internal vs. external validation

- Course notes: Chapter 14 at hbiostat.org/bbr/md with audio narration
- datamethods.org discussion
- Transformations
- Analysis of paired observations
- What’s wrong with change in general
- What’s wrong with percent change
- Objective method for choosing effect measure
- Example analysis
- Regression to the mean

- Video
- Open discussion about all topics covered so far in the course

- Course notes with audio narration
- datamethods.org discussion
- Topics
- Observational treatment comparison modeling strategy
- Problems and misunderstandings about propensity scores
- Role of interactions

- Course notes with audio narration
- datamethods.org discussion
- Topics
- Analysis of serial data
- Formal modeling approaches
- Summary statistic approach
- Case study

- Course notes
- datamethods.org discussion
- Resources
- Topics
- Reproducible research
- Importance of sound methodology and adequate sample size on reproducibility
- Pre-specification
- System forces working against reproducibility
- Literate programming and reproducible statistical reports

- Course notes with audio narration
- datamethods.org discussion
- Resources
- Topics
- What is a false positive
- Why is Type I assertion probability not an “error”
- What is Type I probability α?
- α vs. decision error probabilities
- Why α is not relevant to sequential Bayesian designs

- Course notes with audio narration
- datamethods.org discussion
- Topics
- Analysis of covariance in randomized clinical trials
- Fundamental clinical question / estimand
- Why unadjusted risk differences, risk ratios, odds ratios, and hazard ratios were intended only for homogeneous patients
- Difference between linear and nonlinear models
- Why are adjusted estimates right?
- How many covariables to use?

- Course notes with audio narration
- datamethods.org discussion
- Topics
- Analysis of covariance in randomized clinical trials
- Differential and absolute treatment effects
- Cost-effectiveness ratios
- Statistical plan for RCTs