Experimental Design

1
Experimental Design
Ecologists- collect data used for testing
hypotheses or describing nature Hypotheses
without data are not very useful Data without
hypotheses are wasted
2
Experimental Design 101
Limit yourself to useful data What is useless
data - unreliable or unrepeatable -
irrelevant to problem - inappropriate spatial
temporal scales - hopeless experimental
design cannot be statistically tested
3
Rules of the Road
1. Not everything that can be measured should
be Do you have an ecological or biological
reason for hypothesizing x, y, or z
4
Rules of the Road
2. Find a problem and state your objectives
clearly The purpose (or objective) of the
present study was to
5
Rules of the Road
3. Collect data that will achieve your
objectives and make your statistician
happy Sound design should include - large
sample size - replicate sample design - random,
independent samples - take statistical test into
account
6
Rules of the Road
4. Some ecological questions are impossible to
answer at the present time Know when to say when
7
Sampling, Statistics and Ecology Oh My!
Samples Measured
Study Population
Population of Interest
8
Designing Field Studies
Descriptive Statistics used to summarize data
explain, describe complex systems mean, st.dev,
cluster Analytical Statistics used to test
hypotheses t-test, ANOVA Experimental Design
for each type of statistics very different
9
Designing Field Studies
Need to know something about what you plan to
examine to correctly design a study Pilot
studies are an important aspect Allows you
determine feasibility of - techniques -
sampling sites - sample size - statistics
10
Spatial and Temporal Scales
11
Scales of Measurement
Correct data scale nominal, ranking,
ratio Discrete or continuous Significant figures
12
Rules of the Road
5. Decide on the number of significant figures
before you begin the experiment
13
Rules of the Road
6. Never report an ecological estimate without
some measure of its error Problems with
ecological studies - biological problems are
not statistical problems - temporal-spatial
variability in ecological systems - stats only
cope with random errors
14
Rules of the Road
7. Be skeptical about stats results Just
because it is statistically significant does not
mean it is important
15
Rules of the Road
8. Never confuse statistical significance with
biological significance Questions must have
theoretical concept to be ecologically significant
16
Rules of the Road
9. Record data in computer speadsheets soon
after data is collected
17
Rules of the Road

• Garbage in Garbage out
• The conclusions of your study are only as good as
  your data
• The more energy you put into the design of the
  study the better the payoff at the end