Lecture 1 & 2 - Research and Experimental Design

____ can be defined as the search for knowledge, or as any systematic investigation, with an open mine, to establish novel facts, solve new or existing problems, prove new ideas, or develop new theories

Research

______ is the design of any information-gathering exercises where variation is present, whether under the full control of the experiment or not

Experimental design

trapping

marking

capture and recapture

Field Methods - Trapping and Marking

VES (visual encounter survey) / Transect Survey

Radiotelemetry

PIT tag / Camera Trapping

field methods - monitoring

Satellite and Remote Sensing

field methods - GIS and GPS

focal animal sampling

scan sampling

Field methods - observational methods

Nutrition

Metabolism

Reproduction

Laboratory / Museum methods - physiological measures

Chromosomes

Protein Electrophoresis

DNA Sequences (PCR)

Laboratory / Museum methods - genetics and mol techniques

Systematic Collections

Specimens (Morphometrics)

Phylogenetics (phylogeography)

Laboratory / museum methods - museum specimens

Until recently, wildlife science is mostly based on ____

descriptive research (traditional wildlife science)

Wildlife profession is built on ____ observations in the wild

natural history

Conclusions from associations between wildlife populations and other factors:

Weather / climate

Habitat

Predation

Competition

Behavior

Traditional wildlife science is built upon a ____

weak foundation or unreliable knowledge

____ shined some light on this problem of ‘Laws of Association’ rather than on ‘experimental tests of hypotheses’

Romesburg (1981, JWM)

Last ____, we have moved to wildlife management based on a series of carefully tested hypotheses and conclusions based on sound scientific inquiry

4-5 decades

Rigor in wildlife science: Difficulties in controlling causal factors

Much natural variation confounds results and conclusions:

Habitat, predators, competition, weather, behavior, invasive species, etc

These factors change spatially and temporally with no standard anthropogenic control

Most research prior to _____ was descriptive

1985

Descriptive research is still _____

Example: breeding ecology of wildlife species; measure variables such as nesting chronology, clutch size, hatching success, nesting habitat, etc

very common and very useful

We use results of descriptive research to generate _____

research hypotheses

_______ : a conceptual model or theory describing causal relationships

Hypothesis

Hypotheses - We observe phenomena but need to

explain why those phenomena occur

Hypothetico-deductive method (Popper 1959, 1968):

‘Multiple working hypothesis’

A circular process in which previous information is synthesized into a theory and predictions are deduced and tested through investigations

Reasoning Process:

Theory

Predictions

Hypotheses

Investigation Process

Data

A set of inter-related concepts, definitions, and propositions that present a systematic view of phenomena by specifying general relations among variables, with the purpose of explaining and predicting the phenomena, which with a deduction leads to predictions

Theory:

Tentative propositions state the relations among 2 or more phenomena or variables, which with a deduction leads to hypotheses

Predictions:

Restatement of predictions in the form of testable statements of the relation between 2 or more variables

• What we know, think we know, or conjecture determines relationships that we investigate measurements that we gather, and results that we expect to obtain

Hypotheses:

Information about the world obtained from 1 or more of the following 3 approaches

Model, New Model, Experiment

Data:

Construct a conceptual, analytical, or simulation model

Data: Model

conduct a survey based on probability sampling methods or use sophisticated comparison techniques to minimize bias in ex post facto comparisons or quasi-experiments

Data: New Observations

perform a manipulative experiment in which treatments are assigned at random

• Synthesizing information may confirm our theories or lead us to develop or change them using inductive reasoning and statistical tests

Data: Experiment

Scientific method must identify the _____, applied or basic

research problem

______ is usually related to a management problem

I.e. identifying critical habitat for endangered species

Applied Research

_____ focuses on underpinnings of ecological process controlling behavior reproduction, density, competition, mortality, habitat use, and population dynamics

Basic Research

Identify the research problem

Conduct a literature review

Identify broad research objectives

Collect preliminary data if needed (pilot studies)

Conduct exploratory data analysis

Formulate research hypotheses

Scientific Method: A Checklist

Problem should be important and significant

Ask, should we dedicate resources to solving the problem?

• Often research is conducted on questions we already know the answers to

  • Example: how many deer diet studies do we need

Scientific Method: problem identification

Conduct a literature review because it is an integral part of research

Develop a complete understanding of literature related to the problem

Otherwise, risk reinventing the wheel

Scientific method: Literature review

Preliminary, short-term trial or study

These can be helpful

• Is the project even feasible?

Often disclose hidden costs

Reveal basic logistical problems

Help estimate the needed sample size

Preliminary data analysis on pilot study data and data from other research projects

Pilot Study

A conceptual model or theory describing causal relationships

Focused on ‘why’ questions

• We observe phenomena but need to explain why those phenomena occur

Adult hens have greater nest success than yearling hens

• Why?

Research Hypothesis

Adults out-compete yearlings for best habitat:

• Ho: % yearling nest cover = % adult nest cover

• Ha: % yearling nest cover < % adult nest cover

Adults are more experienced than yearlings:

• Ho: inexperienced adult productivity > yearling productivity

• Ha: inexperienced adult productivity = yearling productivity

Statistical Hypotheses

Understand why you are doing what you are doing

• How will you collect the data?

• How long should the study last?

• How many samples do you need?

• What kind of sampling design will you use?

Design Methodology

4 elements of good experimental design

controls

randomization

independence

replication

_____: to verify an experiment by comparison with a standard or by other experiments

Controls must be the same as experimental units except they are not treated

Used to eliminate effects of confounding factors that could potentially influence conclusions or results

Controls distinguish experimental research from descriptive research

Control:

All samples have an equal probability of selection

Sample will therefore be representative or unbiased

Randomization

Freedom from the influence, control, or determination of another or other samples

Independence

_______ is the repetition of an experimental condition so that the variability associated with the phenomenon can be estimated

Replication – to repeat, duplicate, copy, or reproduce in a similar fashion

Helps decrease error and increase sample size

Replication serves to ensure against making a decision based on a single, possibly unusual outcome of a treatment or measurement of a unit

Many replicates (large sample sizes) reduces chances of biased conclusions or results

Replication

The use of inferential statistics to test for treatment effects with data from experiments where either treatments are not replicated or replicates are not statistically independent

Replication vs Pseudoreplication

Should avoid pseudoreplication

But wildlife scientists are often guilty of pseudoreplication

Typically arises from repeating measurements on experimental units and treating measurements as if they were independent measurements

Treating subsampling as sampling

Pseudoreplication

_______ is concerned with the selection of a subset of individuals from within a population to estimate characteristics of the whole population

Sampling the population rather than trying to measure the entire population is more logistically feasible

Designs:

• simple random

• Systematic

• Stratified random

Methodology:

• Plots

• Points

• Transects

• Road-based

Sampling

Requires that every sample unit in the population has an equal chance of being drawn in the sample

Procedure for selecting sample units is truly random

Example:

• Random number generator

• Sampling vegetation plots to measure habitat availability

Simple Random Sampling

Sample taken by selecting units at regular intervals as they are encountered (easier and less subject to investigator error)

Random placement of first plot followed by systematic placement of remaining plots

May be biased with periodic populations

Systematic Sampling

Requires simple random sampling in each strata

Strata are representative of subpopulations

Example:

• Evaluation of white- tailed deer densities in different habitat types

Stratified Random Sampling

_____ are used widely to sample habitat characteristics and count animal numbers and sign

Plot represent small geographic areas that are elements of the geographically defined population

Subset of plots often sampled with assumption that it is representative of the area

Any of the sampling designs are applicable with plots

Plots

_____: A set of points are established throughout the population and measurements are taken from each sample point

Call counts for birds – distance may be measured

Selection of points usually follows a systemic sampling design but others apply

Point Sampling

____: A straight line or series of straight-line segments placed in an area to be sampled

Often used to obtain systematic samples

Plots or points along transect or transects themselves can serve as experimental units

Transect Sampling

Sampling from roads is a widely used technique

But difficult to obtain unbiased samples from roads

Why?

• Roads are placed along ridges or valleys

• Roads often avoid steep or wet areas

• Roads modify habitat and may attract or repel some species

Bias is often ignored due to its cost effective and easy nature

Road-Based Sampling

_____: powerful tool for detecting differences

Study of nesting habitat – measure habitat characteristics at the nest and at random sites

Paired Sampling

_____: a way to control for nuisance variables

Block design

_____: closeness to the true population value

Accuracy

_____: closeness to each other of measurements of the same quantity

Precision

_____: amount of inaccuracy in estimates

Bias

____: refers to the number of independent, random sample units drawn from the research population

In experiments, sample size is the number of replicates to which a treatment is assigned

Increases in sample size can increase precision of estimates

Sample Size

Needed sample size depends on:

Magnitude of the effect to be detected

Variation in the populations

Type of relationship that is hypothesized

Desired power for the test

• what is power? probability of rejecting Ho when Ho is false

Field studies (treatments not random)

• Ex post facto comparisons only

Natural experiments (hurricane, fire)

• Uncontrolled treatments

Field experiments

• Randomly assigned treatments but some factors not under control

  • More realistic scenario

Before-after/control-treatment design

• Control and treatments both on same sample unit but at different times

• Predator removal studies and prescribed fire research

• Natural variation confounds

Laboratory experiments

• Most control but limited scale

• Unrealistic

Integrated research process

• The iterative scientific method

Study Designs

1. Identify the research problem.

2. Conduct a literature review.

3. Identify broad research objectives.

4. Collect preliminary data if needed (pilot study).

5. Conduct exploratory data analysis.

6. Formulate research hypotheses.

7. Formulate testable (statistical) hypotheses.

8. Design methodology for each hypothesis.

9. Write a research proposal.

10. Obtain peer-review and revise proposals.

11. Perform experiments, collect data.

12. Analyze data.

13. Evaluate, interpret, and draw conclusions.

14. Formulate new hypotheses.

15. Submit manuscript to peer-reviewed journal.

16. Repeat the process beginning at step 6 or 7.

Scientific Method