Transcription of GraphPad Statistics Guide
1 1995-2014 GraphPad Software, is one of three companion guides to GraphPad Prism are available as web pages on Statistics GuideGraphPad Software Statistics Guide2 1995-2014 GraphPad Software, of ContentsForeword0 Part IPRINCIPLES OF 91 The big 9 When do you need statistical calculations? .. 10 The essential concepts of Statistics .. 13 Extrapolating from 'sam ple' to 'population' .. 13 Why Statistics can be hard to learn .. 14 Ordinal, interval and ratio variables .. 16 The need for independent sam ples .. 17 Intuitive Biostatistics (the book) .. 182 The Gaussian 18Im portance of the Gaussian distribution.
2 19 Origin of the Gaussian distribution .. 21 The Central Lim it Theorem of Statistics .. 21 The lognorm al distribution .. 233 Standard Deviation and Standard Error of the 23 Key concepts: SD .. 25 Com puting the SD .. 26 How accurately does a SD quantify scatter? .. 28 Key concepts: SEM .. 28 Com puting the SEM .. 29 The SD and SEM are not the sam e .. 30 Advice: When to plot SD vs. SEM .. 31 Alternatives to show ing the SD or SEM .. 314 Confidence 32 Key concepts: Confidence interval of a m ean .. 34 Interpreting a confidence interval of a m ean .. 36 Other confidence intervals .. 36 Advice: Em phasize confidence intervals over P values.
3 37 One sided confidence intervals .. 38 Com pare confidence intervals, prediction intervals, and tolerance intervals .. 39 Confidence interval of a standard deviation .. 415P 41 What is a P value? .. 41 The m ost com m on m isinterpretation of a P value .. 42 More m isunderstandings of P values .. 43 One-tail vs. tw o-tail P values .. 45 Advice: Use tw o-tailed P values .. 46 Advice: How to interpret a sm all P value .. 47 Advice: How to interpret a large P value .. 48 How Prism com putes exact P values .. 496 Hypothesis testing and statistical 49 Statistical hypothesis testing .. 50 Extrem ely significant?
4 50 Advice: Avoid the concept of 'statistical significance' w hen possible .. 51A Bayesian perspective on interpreting statistical significance 3 Contents3 1995-2014 GraphPad Software, 53A legal analogy: Guilty or not guilty? .. 53 Advice: Don't keep adding subjects until you hit 'significance'.. 557 Statistical 56 Key concepts: Statistical Pow er .. 57An analogy to understand statistical pow er .. 58 Type I, II (and III) errors .. 59 Using pow er to evaluate 'not significant' results .. 61 Why doesn't Prism com pute the pow er of tests .. 63 Advice: How to get m ore pow er .. 648 Choosing sample 64 Overview of sam ple size determ ination.
5 66 Why choose sam ple size in advance? .. 68 Choosing alpha and beta for sam ple size calculations .. 69 What's w rong w ith standard values for effect size? .. 71 Sam ple size for nonparam etric tests .. 729 The problem of multiple 72 The m ultiple com parisons problem .. 74 Approach 1: Don't correct for m ultiple com parisons .. 75 Approach 2: Correct for m ultiple com parisons .. 76 Approach 3: False Discovery Rate (FDR) .. 77 Lingo: Multiple com parisons .. 78 Multiple com parisons traps .. 82 Planned com parisons .. 84 Exam ple: Planned com parisons .. 87 The Bonferroni m ethod .. 8810 Testing for 88 Key concepts: Equivalence.
6 89 Testing for equivalence w ith confidence intervals or P values .. 9211 Nonparametric 92 Key concepts: Nonparam etric tests .. 92 Advice: Don't autom ate the decision to use a nonparam etric test .. 93 The pow er of nonparam etric tests .. 94 Nonparam etric tests w ith sm all and large sam ples .. 95 Advice: When to choose a nonparam etric test .. 96 Lingo: The term "nonparam etric" .. 98An overview of outliers .. 99 Advice: Bew are of identifying outliers m anually .. 99 Advice: Bew are of lognorm al distributions .. 101 How it w orks: Grubb's test .. 102 How it w orks: ROUT m ethod .. 104 The problem of m asking.
7 105 Sim ulations to com pare the Grubbs' and ROUT m ethods .. 10913 Analysis 109 Unpaired t test .. 111 Paired t test .. 113 Ratio t test .. 114 Mann-Whitney test .. 115 Wilcoxon m atched pairs test .. 116 One-w ay ANOVA .. 118 Repeated m easures one-w ay ANOVA .. 120 Kruskal-Wallis test GraphPad Statistics Guide4 1995-2014 GraphPad Software, 121 Friedm an's test .. 122Tw o-w ay ANOVA .. 124 Repeated m easures tw o-w ay ANOVA .. 125 Contingency tables .. 126 Survival analysis .. 128 Outliers Part IISTATISTICS WITH PRISM 1301 Getting started with Statistics with 130 What's new in Prism 6 ( Statistics )?.
8 131 Statistical analyses w ith Prism .. 132 Guided exam ples: Statistical analyses .. 1332 Descriptive Statistics and frequency 134 Colum n Statistics .. 134 How to: Column 136 Analysis checklist: Column 138 Interpreting results: Mean, geometric mean and 139 Interpreting results: Quartiles and the interquartile 141 Interpreting results: SD, SEM, variance and coef f icient of variation (CV).. 142 Interpreting results: Skew ness and 143 Interpreting results: One-sample t 144 Interpreting results: Wilcoxon signed rank 147 Interpreting results: Normality 148 Frequency Distributions .. 148 Visualizing scatter and testing f or normality w ithout a f 149 How to: Frequency 153 Graphing tips: Frequency 154 Fitting a Gaussian distribution to a f requency 156 Describing curves.
9 156 Smoothing, dif f erentiating and integrating 159 Area under the 162 Row Statistics .. 162 Overview : Side-by-side 163 Row means and 1633 Normality 164 How to: Norm ality test .. 164 How norm ality tests w ork .. 165 Interpreting results: Norm ality tests .. 166Q&A: Norm ality tests .. 1694 Identifying 169 How to: Identify outliers .. 172 Analysis checklist: Outliers .. 1735 One sample t test and Wilcoxon signed rank 174 How to: One-sam ple t test and Wilcoxon signed rank test .. 174 Interpreting results: One-sam ple t test .. 175 Interpreting results: Wilcoxon signed rank test .. 1786t tests, Mann-Whitney and Wilcoxon matched pairs 178 Paired or unpaired?
10 Param etric or nonparam etric? .. 178 Entering data f or a t 179 Choosing a test to compare tw o 181 Options f or comparing tw o groups5 Contents5 1995-2014 GraphPad Software, 182 What to do w hen the groups have dif f erent standard deviations?.. 185Q&A: Choosing a test to compare tw o 186 The advantage of 188 Unpaired t test .. 188 How to: Unpaired t test f rom raw 189 How to: Unpaired t test f rom averaged 191 Interpreting results: Unpaired 193 The unequal variance Welch t 195 Graphing tips: Unpaired 196 Advice: Don't pay much attention to w hether error bars 198 Analysis checklist: Unpaired t 200 Paired or ratio t test.
