Statistics in Environmental Sciences

Valerie David

• Publisher: Wiley-ISTE
• Year: 2019
• Print ISBN: 9781786305077
• eBook ISBN: 9781119649717
• Edition: 1
• Topic: Sozialwissenschaften
• Subtopic: Wissenschaftliche Forschung & Methodik

Statistical tools are indispensable for the environmental sciences. They have become an integral part of the scientific process, from the development of the sampling plan to the obtainment of results. Statistics in Environmental Sciences provides the foundation for the interpretation of quantitative data (basic vocabulary, main laws of probabilities, etc.) and the thinking behind sampling and experimental methodology. It also introduces the principles of statistical tests such as decision theory and examines the key choices in statistical tests, while keeping the established objectives in mind. The book examines the most used statistics in the field of environmental sciences. Detailed descriptions based on concrete examples are given, as well as descriptions obtained through the use of the free software R (whose usage is also presented).

1
Working with the R Software¹

1.1. Working with the R software

1.1.1. Why and how to work with the R software?

The R software is both a computer programming language and a working environment for statistical data processing. It is used to manipulate data, plot graphs and perform statistical analyses on these data. Among its advantages, R is:

• – a “free” software that can be downloaded and installed on personal computers;
• – a “cross-platform” software that runs on Windows, Linux, Mac OS, etc.

As a result, many scientists use it internationally and do not hesitate to share their statistical knowledge by developing new functions and communicating through forums. Everyone can contribute to its improvement by integrating new functionalities or analysis methods that have not yet been implemented.

This therefore makes it a software that is in rapid and constant evolution. Thus, many statistical analyses that are both simple and complex are available (descriptive and inferential statistics, parametric or non-parametric tests, linear or nonlinear models, multivariate analyses, etc.). No commercial software is able to provide such a choice of methods. It also has many useful and professional-quality graphic functions. Lastly, the tool is very effective because, when we master the R language, we are able to create our own tools (or “scripts”). This makes it possible to perform very sophisticated analyses by writing a script with a series of functions on the data (functions that are already incremented or created by ourselves) and to reproduce them very quickly on other data sets. In this book, we will be using the integrated development environment “Rstudio”, which is more user-friendly and powerful.

It should be noted that not all functions are available in the basic software and that it will often be necessary to use additional libraries (or packages) that are also free. For example, the MASS package. When first used, the installation of the package requires an Internet connection. There are two steps when using it for the first time: installation with the install.package() function, then loading it into the working session with the library() function:

---

  >install.package(‘MASS’) >library(MASS)  

---

Once installed, each new session can just be loaded without an Internet connection:

---

  >library(MASS)  

---

1.1.2. Recommended working method

The environment is frugal: everything is done through command lines (the “functions”). Working with the R software therefore requires a working method. If it is respected, it makes it possible to be more efficient and faster in analyses.

1.1.2.1. Creating a working directory

First of all, it is advisable to create our own working directory (called the current directory). All analyses will be centralized there (or even all libraries could be centralized there by copy-pasting zip files from the website). This will make it easier to search for files (data, scripts, sessions, etc.). To do this, we must create a folder in the session called “R”, for example. The first thing to do each time the R software is opened is to change the current directory:

• – either via File/Change the current directory in the toolbar (at the top);
• – or via setwd() by specifying the address of the current directory (this is a fake address). The getwd() function is used to check that the working s...

Table of contents

• Cover
• Table of Contents
• Preface
• Introduction
• 1 Working with the R Software
• 2 Fundamental Concepts in Statistics
• 3 Developing a Sampling or Experimental Plan
• 4 Principle of a Statistical Test
• 5 Key Choices in Statistical Tests
• 6 Comparison Tests of Unilateral and Bilateral Parameters
• 7 Classical and Generalized Linear Models
• 8 Non-parametric Alternatives to Linear Models
• Conclusion
• References
• Index
• End User License Agreement