Common Table 1 mistakes |
This discusses two problems that often occur in studies' descriptive summaries (usually presented as "Table 1"). |
Common mistakes when describing a study population (the typical Table 1) |
Insufficient and spurious precision |
This explains the concepts of insufficient and excessive precision and provides suggested guidelines for how many decimals to present for P-values, odds ratios, relative risks and hazards, correlation coefficients, fold-effects, and regression coefficients. |
Avoiding Insufficient or Excessive Precision |
Use of "significant" alone |
This explains why it is important to avoid use of "significant" or "significantly" alone in manuscripts. |
Use of \x93Significant\x94 Alone in Publications |
Interval censored data |
This discusses three different approaches for modeling interval-censored survival data, along with how to implement them in R or SAS. It concludes with a recommended general strategy for doing regression modeling with interval-censored survival data. |
Interval-Censored Survival Analysis |
Results interpretation |
This provides general conceptual guidance on how to interpret results of statistical analyses, with emphasis on avoiding the common problem of focusing only on whether or not P-values are <0.05. It also links to an interactive application that provides example text, based on user inputs, giving interpretations that reflect the estimates and confidence intervals, rather than just the P-values. |
Proper Interpretation of Results |
Log transformation |
This explains and illustrates conceptual and practical issues about when to use logarithmic transformation of outcome or predictor variables and how to report results. |
Logarithmic Transformation |
Infinite estimates |
This describes options for how to perform and report regression analyses in cases where standard approaches break down because estimated effects are infinite. It includes specific methods for SAS and some for Stata, describes advantages and drawbacks of several methods, and has links to additional web pages that explain many of the key concepts used. |
Infinite Estimates |
Reporting followup time |
This explains what to report about followup time in time-to-event studies and provides an example. |
Reporting Followup Time |
Reporting Power |
This briefly explains why statistical power is irrelevant for interpreting the results of completed studies, provides references documenting strong support for this fact, and discusses why this calls into question the supposed need to report power calculations in publications of studies\x92 results. |
Reporting Power |
Common Biostatistical Problems |
This discusses in article form the common biostatistical problems and best practices from a lecture developed for Biostat 209, and also provides lecture slides, lecture notes, and source files. Detailed subheadings in the article format permit linking directly to specific issues. |
Common Biostatistical Problems |
Ethics and Sample size |
This summarizes past literature on the issue of whether having too small a sample size makes a study unethical, and then has a side-by-side structure for presenting reasoning on both sides of various issues and points. |
Ethics and Sample Size |