Cleophas T.J., Zwinderman A.H. Clinical Data Analysis on a Pocket Calculator: Understanding the Scientific Methods of Statistical Reasoning and Hypothesis Testing

New York: Springer, 2016. - 328 p.In medical and health care the scientific method is little used, and statistical software programs are experienced as black box programs producing lots of p-values, but little answers to scientific questions. The pocket calculator analyses appears to be, particularly, appreciated, because they enable medical and health professionals and students for the first time to understand the scientific methods of statistical reasoning and hypothesis testing. So much so, that it can start something like a new dimension in their professional world. In addition, a number of statistical methods like power calculations and required sample size calculations can be performed more easily on a pocket calculator, than using a software program. Also, there are some specific advantages of the pocket calculator method. You better understand what you are doing. The pocket calculator works faster, because far less steps have to be taken, averages can be used. The current nonmathematical book is complementary to the nonmathematical "SPSS for Starters and 2nd Levelers" (Springer Heidelberg Germany 2015, from the same authors), and can very well be used as its daily companion.Data Spread, Standard Deviations.
Data Summaries, Histograms, Wide and Narrow Gaussian Curves.
NullHypothesis Testing with Graphs.
NullHypothesis Testing with the TTable.
OneSample Continuous Data (OneSample TTest, OneSample Wilcoxon Test).
Paired Continuous Data (Paired TTest, Wilcoxon Signed Rank Test).
Unpaired Continuous Data (Unpaired TTest, MannWhitney).
Linear Regression (Regression Coefficient, Correlation Coefficient and Their Standard Errors).
KendallTau Regression for Ordinal Data.
Paired Continuous Data, Analysis with Help of Correlation Coefficients.
Power Equations.
Sample Size Calculations.
Confidence Intervals.
Equivalence Testing Instead of NullHypothesis Testing.
Noninferiority Testing Instead of NullHypothesis Testing.
Superiority Testing Instead of NullHypothesis Testing.
Missing Data Imputation.
Bonferroni Adjustments.
Unpaired Analysis of Variance.
Paired Analysis of Variance.Variability Analysis for One or Two Samples.
Variability Analysis for Three or More Samples.
Confounding.
Propensity Scores and Propensity Score Matching for Assessing Multiple Confounders.
Interaction.
Accuracy and Reliability Assessments.
Robust Tests for Imperfect Data.
Nonlinear Modeling on a Pocket Calculator.
Fuzzy Modeling for Imprecise and Incomplete Data.
Bhattacharya Modeling for Unmasking Hidden Gaussian Curves.
Item Response Modeling Instead of Classical Linear Analysis of Questionnaires.
Metaanalysis of Continuous Data.
Goodness of Fit Tests for Identifying Nonnormal Data.
Nonparametric Tests for Three or More Samples (Friedman and KruskalWallis).Data Spread: Standard Deviations, One Sample ZTest, One Sample Binomial Test.
ZTest for CrossTabs.
Phi Tests for Nominal Data.
Chisquare Tests.
Fisher Exact Tests Convenient for Small Samples.
Confounding.Interaction.
ChiSquare Tests for Large CrossTabs.
Logarithmic Transformations, a Great Help to Statistical Analyses.
Odds Ratios, a ShortCut for Analyzing CrossTabs.
Logodds, the Basis of Logistic Regression.
Log Likelihood Ratio Tests for the Best Precision.
Hierarchical Loglinear Models for Higher Order CrossTabs.
McNemar’s Tests for Paired CrossTabs.
McNemar’s Odds Ratios.
Power Equations.
Sample Size Calculations.
Accuracy Assessments.
Reliability Assessments.
Unmasking Fudged Data.
Markov modeling for Predicting Outside the Range of Observations.
Binary Partitioning for CART (Classification and Regression Tree) Methods.
Metaanalysis of Binary Data.
Physicians’ Daily Life and the Scientific Method.
Incident Analysis and the Scientific Method.
Cochran QTest for Large Paired CrossTabs.