Distribution of Midterm Grades

1
Distribution of Midterm Grades
2
Approach Which tool in the Toolbox

• How do you decide which to use?
• Consider the strengths and weaknesses of each
• Experiment when you can (gives you best shot at
  addressing your hypothesis)
• Dont experiment if its impossible to have a
  well-scaled experiment
• Comparative Analysis
• Long Term Assessment
• Other questions might require other approaches
• Estimation
• Modeling / Prediction

3
If you experiment

• Dont confuse experimental design with
  experimental method
• Experimental design concerns the allocation of
  treatments and controls across time and space
• One way experimental design
• Two way experimental design
• BACI Design

4
Experimental Methods

• These are the types of experiments you might run
• Lab, microcosm, mesocosm
• Enclosure / Exclosure
• Whole-ecosystem
• They concern the actual physical structure or
  scale of the experiment
• What kind of experimental design is used in the
  goby exclosure experiment?

5
Sampling Designs vs. Sampling Methods

• Sampling designs are the allocation of sample
  points in time and space
• Random design, fixed distance, stratified random,
  adaptive clustered design
• Sample methods are the gears used to collect
  samples

6
Hypothetico-Deductive Method

• HD method is not analogous to experimentation
• You can use HD to test hypotheses using
  historical or paleontologic data
• The key to HD reasoning is that you have a
  hypothesis, then have a prediction based on that
  hypothesis. You evaluate the prediction, and if
  it is false the hypothesis must be false (by
  deduction)
• You can do this on ALL SORTS of data, not just
  experiments
• The important point is that there are other ways
  of gaining understanding

7
Asides on Comparative Approaches

• We often use comparative approaches in lieu of
  experiments
• We often try to follow similar designs, though
  allocation of treatment is NOT randomized (by
  definition)
• This means you still need replication to make
  inferences (you cant compare two locations and
  imply any cause-effect relationship)

8
Comparative Analyses Question What limits
food chain length? Why do we care? Highly
variable among ecosystems Influences community
structure and ecosystem function Affects
contaminant concentrations in top predators
9
Alternative Hypotheses Productivity Hypothesis
More productive ecosystems have longer food
chains (energy constraint) Ecosystem Size
Hypothesis Larger ecosystems have longer food
chains (greater habitat diversity, species
diversity) Productive Space Hypothesis A
combination of ecosystem size and productivity
lead to longer food chains (energy constraint)
10
Results Ecosystem size hypothesis best explains
variation in food chain length
Post et al. 2000
11
Long Answer Grades
default score
Figure 2. Frequency histogram of scores on the
long answer question, Fish 210 midterm 2008.
12
How to tackle Q 19A

• Part 1 figure out which approach to use
• Can you use experimentation?
• Why or why not.
• If not, what alternatives would be most
  appropriate
• Appropriate tractable and liable to produce
  data that test the hypothesis

13
How to Tackle Q. 19 B

• How do you design your experiment
• One way, two way, BACI design?
• What size experiment?
• Randomize treatments and replicate
• What are your response variables?
• How to you design a comparative analysis
• Select MULTIPLE ecosystems for comparison
• What are your response variables?
• What is your sample design? (not sampling
  method!)
• In time, space