# Data Analysis

Resource Center

## "Exact" P values for multiple comparisons tests after ANOVA

The #1 suggestion we've heard from scientists to improve Prism 5.

The phrase "exact P value" can be interpreted in several ways, and Prism 6 offers two approaches.

Following Bonferroni, Tukey or Dunnet multiple comparisons testing, Prism 6 can compute a multiplicity adjusted P value. For each comparison, this is the smallest significance level (applied to the entire family of comparisons) where this particular comparison would just barely be declared to be "statistically significant". An adjusted P value is an “exact P value” reported for each comparison. But the value of the adjusted P value depends on the number of comparisons you are making, and on the data for all the comparisons.

The second approach Prism 6 uses to report "exact P values" is the unprotected Fishers Least Significant Difference (LSD).

## Statistics Guide

The GraphPad Statistics Guide is both a guide to doing statistical analyses with GraphPad Prism 6, and a concise review of statistical principles, emphasizing clear explanations with very little math.

**Selected topics:**

- The essential concepts of statistics
- Ordinal, interval and ratio variables (surprisingly, our most popular page)
- When to plot the SD vs SEM
- The most common misunderstanding of P values, and more misunderstandings.
- How to interpret a small P value or a large P value
- An analogy to understand statistical power
- Why Prism doesn't compute the power of tests
- Multiple comparisons traps
- The False Discovery Rate (FDR)
- The lognormal distribution
- A Bayesian perspective on interpreting statistical significance
- The confidence interval of a standard deviaiton
- Fisher's Least Significant Difherence (LSD) test
- Holm-Sidak multiple comparisons
- Multiple comparisons after repeated measures ANOVA

## Curve Fitting Guide

The GraphPad Curve Fitting Guide is both a guide to fitting curves (and lines) with GraphPad Prism 6, and a concise nonmathematical explanation of the principles of linear and nonlinear regression.

**Selected topics:**

- The many uses of global nonlinear regression
- Comparing fits of nonlinear models
- Identifying outliers
- Comparing linear and nonlinear regression
- Comparing linear regression and correlation
- Advice: Avoid Scatchard, Lineweaver-Burke plots, etc.
- Why you shouldn't fit a model to smoothed data
- Interpolating from a standard curve
- Fitting a curve without a model using spline, lowess, and polynomial fits
- Understanding dependency and the covariance matrix
- Plotting a function
- 50% of what? Defining the EC50.
- EC80, EC90, ECanything
- Competitive, noncompetitive and uncompetitive enzyme inhibition
- Finding the Kcat of an enzyme