FW552
Applied Sampling

Class Stuff

  • Instructor: Brian Gerber

  • Classroom: Engineering E102

    • Lecture/Discussion/Problem Solving
  • When: Mo/We 3 - 4:15pm

  • My Office: 202A Wagar, Cooperative Research Unit

    • Office hours: We 1pm - 2pm and by appointment

    • brian.gerber@colostate.edu

  • Computers: Assignments will often require a computer

My Background

What is this course?

Primarily:

  • connecting things we care about (fish/wildlife/habitat) with sampling principles

What is this course?

A mix of…

  • Sampling principles
  • Design-based inference
  • Math / notation
  • R coding

What is this course?

Important Concepts and Terminology

For example,

  • Inference

  • Sampling distribution

  • Sampling variation

  • Sampling Bias

  • Sample size

  • Parametric bias and variance

Questions for you…

  • Name and short background

  • Motivation for taking this course?

  • Experience with the R programming language

  • One vision of an ideal job, research project, study animal, or study location.

Assessment

Assessment Components Percentage of Grade
Course Engagement 10%
Assignments 25%
Quizzes 25%
Tests 20%
Project 20%

Project

????

Course Learning Objectives

Upon successful completion of this course students will be able to:

  • Identify different types of sampling designs and understand when to apply them

  • Understand statistical estimators and their properties

  • Frame a fish and wildlife sampling design problem and apply appropriate statistical tools to estimate parameters of interest in accordance with the selected design.

  • Be able to use fundamental code practices in the R programming language.

Why is this class useful?

  • Fundamental to scientific learning

  • Have options in designing an empirical study

  • Interpret and frame learning when reading the literature

  • Collaborate with colleagues/statisticians

Software

Why learn to code?

  • efficiency
  • transparency
  • flexibility in application
  • shareable
  • marketable skill
  • needed for publications

Why use R?

  • open-source and free
  • small total user base / large in ecology and statistics
  • find help online, e.g., stackoverflow
  • data management
  • statistics
  • plotting / graphics

Why use RStudio?

Statistics in the Modern Age

"The theory and practice of computer-age statistics are, for the most part, a case of new wine in old bottles: The fundamental tenets of good statistical thinking have not changed, but their implementation has."
- Cox and Efron, Sci. Adv. 2017;3: e1700768.

Coding in the Modern Age

  • Software changes all the time

  • Code will become obsolete

  • Base R functions change slower than packages

  • Document/Annotate code and publish it online

  • File management is important!

Teaching Philosophy

  • Learning is a choice (in every moment)

  • An inclusive environment is paramount for learning

  • Communication is everything

  • Everyone has something to teach and something to learn

  • Struggle is good. Solving problems leads to learning. But….






Words Matter

Website

Most Material

https://bgerber123.github.io/FW552/


Quizzes and Assignments

https://canvas.colostate.edu/

Questions

A ‘Sample’

Brook trout habitat in rivers of Massachusetts

A finite part of a target population whose properties are studied to gain information about the whole (i.e., to infer or gain inference).

A ‘Sample’

Brook trout habitat in rivers of Massachusetts

A selection of rivers (sample) to measure habitat characteristics (e.g. gravel beds) to be used to infer about habitat in all of the state.

Not Sampling

Piping plovers nesting at Watch Hill, RI

Census: a complete enumeration of a target population

Sampling

The process of selecting a representative part of the population for the purpose of determining characteristics of the target population.

Why do we need to sample?

  • logistically/physically prohibitive
  • fiscally prohibitive / waste of time and money
  • timeliness
  • sampling may be destructive

Central paradox of sampling

  • Impossible to know whether a sample is ‘good’ (i.e., representative of the target)

  • The sample itself cannot tell us - only the process

  • Given exact same sample, we treat it differently depending on how it was selected.

  • A sample from a reliable process has credentials

How have you used sampling?