Hypothesisdriven and Investigative Experimental Design

1
Dr. Matthew Juniper, CUED
Hypothesis-driven and Investigative Experimental
Design
2
What do I mean?
Hypothesis-driven and Investigative Experimental
Design
Investigative
Experiments
Hypothesis-driven
3
Examples
Millikan's oil drop experiment to determine the
charge on the electron (1909)
Investigative
Experiments
Hypothesis-driven
4
Millikan's oil drop experiment (1909)
5
Millikan's oil drop experiment (1909)
Charge on the electron 1.592 x 10-19 Coulombs
6
Millikan's oil drop experiment (1909)
Charge
Charge on the electron 1.592 x 10-19 Coulombs
1909
Year
7
Millikan's oil drop experiment (1909)
Charge
1.602 x 10-19
Charge on the electron 1.592 x 10-19 Coulombs
1909
Year
8
Examples
Millikan's oil drop experiment to determine the
charge on the electron (1909)
Investigative
Michelson-Morley experiment to determine the
speed of the earth through the Aether (1887)
Experiments
Hypothesis-driven
9
Michelson-Morley experiment (1887)
10
Michelson-Morley experiment (1887)
Speed of the earth through the Aether ....
very small !
11
Examples
Millikan's oil drop experiment to determine the
charge on the electron (1909)
Investigative
Michelson-Morley experiment to determine the
speed of the earth through the Aether (1887)
Experiments
Hypothesis-driven
12
Examples
Millikan's oil drop experiment to determine the
charge on the electron (1909)
Investigative
Michelson-Morley experiment to determine the
speed of light in different reference frames
Experiments
Light travels at the same speed in any frame of
reference
Hypothesis-driven
13
Contents
Hypothesis-driven and Investigative Experimental
Design
14
Issue trees
Issue Tree
What ? or How ?
Sub-issue 1a
Issue 1
Sub-issue 1b
Sub-issue 1c
Sub-issue 2a
Open question
Issue 2
Sub-issue 2b
Sub-issue 2c
Precise issues can be tested by hypothesis
Precise issue
Sub-issue 3a
Issue 3
Sub-issue 3b
Sub-issue 3c
Issues are independent and complete
15
Issue trees an example
What ? or How ?
What can we do about climate change ?
16
Issue Tree applied to an engineering problem
What ? or How ?
composition
oxidant
temperature
pressure
reactant properties
composition
temperature
fuel
the state of the fuel / air mixture in each burner
pressure
pilot / main flame configuration
burner face
flow shear at injection point
aero-dynamics
down-stream
position of cooling air
velocity of cooling air
the introduction of energy to the fuel / air
mixture in each burner
What influences the re-light and light-round
characteristics of an aeroplane engine
type of spark
burners
position of spark
energy of spark
duration of spark
spark
timing of spark
the design of the network of burners in the
engine
The distance between burners
Axial view of combustion chamber in an aeroplane
engine
The order in which they are turned on
17
Contents
Hypothesis-driven and Investigative Experimental
Design
18
Hypothesis trees
Why?
Tertiary hypothesis 1a
Secondary hypothesis 1
Tertiary hypothesis 1b
Tertiary hypothesis 1c
Tertiary hypothesis 2a
Primary hypothesis
Secondary hypothesis 2
Tertiary hypothesis 2b
Tertiary hypothesis 2c
All hypotheses must be precise statements
Tertiary hypothesis 3a
Secondary hypothesis 3
Tertiary hypothesis 3b
Statement 3c(i)
Tertiary hypothesis 3c
or
Statement 3c(ii)
All the secondary hypotheses must be true for the
primary hypothesis to be true
19
Hypothesis trees - example
Why?
I can make money by blackmailing someone.
20
(No Transcript)
21
(No Transcript)
22
Hypothesis trees - example
Hypothesis 1.4
Conv. Unst. flame
amplitud\e of response
amplitude of forcing signal
Abs. Unst. flame
amplitud\e of response
amplitude of forcing signal
23
Hypothesis trees - example
Hypothesis 1.4
Experimental results
Conv. Unst. flame
amplitud\e of response
amplitude of forcing signal
Abs. Unst. flame
amplitud\e of response
amplitude of forcing signal
24
(No Transcript)
25
(No Transcript)
26
(No Transcript)
27
Contents
Hypothesis-driven and Investigative Experimental
Design
28
Application to experimental design

• 1. Start with an issue tree (ask 'what?' or
  'how?') e.g.
• What questions do I have?
• What do I want to show?
• What is my PhD about?

What ? or How ?
2. For each sub-issue, think of an investigative
experiment. Then ask 'so what?' does this tie
in with a theory?
So what ?
3. For each relevant sub-issue, develop a
hypothesis tree (ask 'why?' or 'what has to be
true for this hypothesis to be true?') Develop
the analysis to test the sub-hypotheses.
Why?