The A–Z of Error-Free Research

Phillip I. Good

Publisher:

Chapman & Hall/CRC

Publication Date:

2013

Number of Pages:

249

Format:

Paperback

Price:

49.95

ISBN:

9781439897379

Category:

General

MAA Review
Table of Contents

[Reviewed by

Tom Schulte

, on

04/9/2013

]

“A–Z” suggests a completeness probably not possible in roughly 240 pages. The subtitle on the back is both more detailed and more accurate for this ABC (and R) of applied statistics in research: “A Practical Guide with Step-by-Step Explanations, Numerous Worked Examples, and R Code”. R is a free programming language and environment for statistical computing. The R language is widely used among statisticians. R's popularity has increased substantially in recent years. This book follows has a similar trend: R content increases exponentially as the page numbers rise. The material at the end includes an R primer and function index. Being conversant in R, and probably planning to use R in research, is a prerequisite to getting the most out of this book. Understanding and embracing R is implicit and basically taken for granted throughout, although there are pointers.

I have not been involved in any research that was truly error-free. Indeed, errors are milestones in research progress. The author does not make a concerted attempt to define “error-free research” and charting a course toward it, but there is very practical advice on surveying (keep the list of questions short and make an effort to follow-up on non-responders) and in proper design, analysis, and reporting of experiments, clinical trials, and more. I particularly appreciate the author’s encouragement to step back and plan for improving data quality, such as defining a well-defined response (as in polling) in detail prior to collecting data. More of the author’s experience comes out in advice on considering missing variables and such pieces of wisdom as “goodness of fit is not prediction”, etc.

Making the transition from student to professional researcher can be a daunting experience. This book can serve as a valuable refresher on hypothesis testing, coping with variation, data collection, sample size decisions and more, along with cursory explanation of R output largely based on freely available data sets. After all the work is done, research needs to be presented. The author speaks directly to publication, but it surprises me that graphics are left until nearly the final chapter. R is rich in graphics capability and much research disappears into the black hole of missed points when research output is ineffectually presented to non-scientific decision makers.

This is high-level material to aid the reader in becoming a confident researcher and assumes that the reader has already progressed beyond the subtle intricacies of levels of significance and can glean insight from an ANOVA table. For reader that wants to put theory to practice, and do it in R, this work can be a guide to success in analyzing and collection categorical data, detecting confounding, bootstrap approaches, case-control and cohort studies, and more.

Tom Schulte spent nearly a decade developing and overseeing the development of analytical applications for statistical process control in the Plex Online ERP system.

Research Essentials
Prescription
Fundamental Concepts
Precautions
Will the Data Require Statistical Methods?
Summary

PLANNING
Hypotheses and Losses
Prescription
State the Objectives of Your Research
Gather Qualitative Data
Formulating Hypotheses
Specify the Decisions and Associated Costs
Specify the Alternatives
Summary
To Learn More

Coping with Variation
Prescription
Start with Your Reports
List All Outcomes of Interest
List All Sources of Variation
Describe How You Will Cope with Sources of Variation
Establish a Time Line
Should the Study Be Performed?
To Learn More

Experimental Design
Prescription
Define the Study Population
The Purpose of Experimental Design
K.I.S.S?
Summary
To Learn More

DATA COLLECTION
Fundamentals
Prescription
How Will You Make Your Measurements?
Formal Descriptions of Methods and Materials
Put Your Data in a Computer and Keep It There
Forestall Disaster
To Learn More

Quality Control
Prescription
Potential Sources of Error
Preventive Measures
Make Baseline Measurements
Conduct a Pilot Study
Monitor the Data Collection Process
Monitor the Data
To Learn More

ANALYZING YOUR DATA
Describing the Data
Prescription
Box and Whiskers Plot
Which Statistic?
Interval Estimates
Confidence Intervals for the Population Mean
Confidence Intervals for Proportions
Estimated from Randomized Responses
Confidence Intervals for Other Population
Characteristics
An Improved Bootstrap
Summary
To Learn More

Hypothesis Tests
Prescription
Types of Data
Analyzing Data from a Single Population
Comparing Two Populations
Comparing Three or More Populations
Experimental Designs
To Learn More

Multiple Variables and Multiple Tests
Prescription
Multiple Variables
Multiple Tests
To Learn More

Miscellaneous Hypothesis Tests
Prescription
Hypothesis Tests and Confidence Intervals
Testing for Equivalence
When Variables Are Not Identically Distributed
Testing for Trend

Sample Size Determination
Prescription
Prepare a Budget
Final Sample Size
Initial Sample Size
To Learn More

BUILDING A MODEL
Ordinary Least Squares
Prescription
Linear Regression
Improving the Fit
Increasing the Number of Predictors
Analysis of Variance
Summary
To Learn More

Alternate Regression Methods
Prescription
LAD Regression
Quantile Regression
Errors-in-Variables Regression
Generalized Linear Models
Classification
Modeling Survival Data
Principal Component Analysis
Summary
To Learn More

Decision Trees
Prescription
How Trees Are Grown
Incorporating Existing Knowledge
Using the Decision Tree as an Aid to Decision Making
Summary
To Learn More

REPORTING YOUR RESULTS
Reports
Prescription
Choose a Journal
Methods and Materials
Results
Reporting Your Analyses
Discussion
Introduction
Abstract
Bibliography
Responding to Rejection
To Learn More

Oral Presentations
Prescription
Text
Graphs
Tables

Better Graphics
Prescription
Creating Graphs with R
To Learn More

NONRANDOM SAMPLES
Cohort and Case-Control Studies
A Worked-Through Example
Prescription
Examples
To Learn More

R Primer

Bibliography

Author Index

Subject Index

R Function Index

Tags:

Data Analysis