Research day

Question 1

Hypothesis – why is it important?

Answer 1

• Critical thinking, good scientific practise, strong
  inference (Platt 1964)
• Practical aspects: It is important to know what you
  want to know before you start the experiments!
• Benefits of a study with a clear hypothesis:
• minimises the risk of not having the right
  measurements
• saves time (no ‘useless’ measurements)
• makes later interpretation / analysis a lot analysis
  easier

Question 2

5 steps of forming a hypothesis:

Answer 2

1) Observing
2) Asking questions
3) Formulating hypothesis
4) Predicting possible outcomes
5) Doing pilot experiments

Question 3

1) From an observation to a question
- Although observations without a hypothesis are unlikely to produce good data,
an observation can help to ask good questions
Mention 2 types of question types:

Answer 3

Facts – Who? Where? What?

Causation – How? Why? (Mechanistic understanding..)
How? – Nearby causation – proximate causes
Why? – Long-distance causation – ultimate causes

Question 4

Migrating to breed in a new area can have
many benefits: Mention 3

Answer 4

• more food
• less predation
• less parasites

Question 5

2) Formulate hypothesis and predictions
- Questions should be formulated as testable hypothesis: Making predictions of all possible outcomes helps to formulate the hypothesis

For instance:
Why do all the fish swim to the same direction? - Question
Fish swim to the same direction because that protects them from predation - Hypothesis
In the presence of predator, more fish join the synchronously swimming school - Prediction

(intet svar)

Question 6

Hvad er en null hypotese

Answer 6

Null hypothesis assumes that nothing interesting happens unless there is evidence of the contrary.
In scientific research, the null hypothesis is the claim that the effect being studied does not exist. The null hypothesis can also be described as the hypothesis in which no relationship exists between two sets of data or variables being analyzed.

‘Fish swim to the same direction by coincidence’ – 0-hypothesis

Question 7

Null hypothesis should be compared with ?

Answer 7

Null hypothesis should be compared with multiple alternative hypothesis
‘Fish swim to the same direction because that protects from predation?’
‘Fish swim to the same direction because that makes gathering food easier?’
‘Fish swim to the same direction because they all try to avoid the observer?’
‘Fish swim to the same direction because they are all going to the same place?’

Question 8

to help to avoid having a pet hypothesis, you need to …

Answer 8

Formulating multiple hypothesis and predictions help to avoid having a pet hypothesis
‘Fish swimming behaviour is determined by zooplankton swimming behaviour

Question 9

hvad er en pet hypotese

Answer 9

pet project/theory/subject/idea that you particularly like or are interested in

Question 10

• Considering every possible outcome (prediction) for each hypothesis will help to:

Answer 10

• Avoid pet hypothesis and confirmation bias
• Check for the relevance of the question / hypothesis
• Check that the hypothesis is testable

Question 11

Before making the final experimental design, it is often a good idea to do

Answer 11

a pilot experiment

Question 12

Pilot experiment can be used to:

Answer 12

• Better design the final experiment (scale)
• Validate the biological question
• Practise and validate the technique
• Get preliminary data to a better design of statistics

Question 13

1) Formulate a null hypothesis

Answer 13

Fx ‘Fish swim to the same direction by coincidence’

Question 14

2) Formulate alternative hypotheses with predictions of
the outcomes
Giv eksempler på alternative hypoteser ift til en nulhypotese?

Answer 14

Fx
‘Fish swim to the same direction because that protects from predation?’
‘Fish swim to the same direction because that makes gathering food easier?’
‘Fish swim to the same direction because they all try to avoid the observer?’
‘Fish swim to the same direction because they are all going to the same place?’

Question 15

3) Consider if you would need a pilot experiment (and if
so, on what).
Hvad er et pilot projekt?

Answer 15

Udtrykket pilotprojekt kan referere til mange typer af undersøgelser, men som oftest er det en næsten nøjagtig kopi af den rigtige undersøgelse – dog i meget mindre skala. Et pilotprojekt bruges ofte til at teste, om udformningen af den endelige undersøgelse lever op til de kriterier, der er opstillet for den.

Question 16

What is a manipulation study? (experiments)

Answer 16

Manipulation means that something is purposefully changed by the researcher in the environment. Control is used to prevent outside factors from influencing the study outcome. When something is manipulated and controlled and then the outcome happens, it makes us more confident that the manipulation “caused” the outcome.

Question 17

Hvad er Benefits of experimental studies:

Answer 17

• Controlled – No confounding factors, less noise, clear cause-effect links
• Practical – No need of complicated logistics
• Easier to interpret than correlational studies

Question 18

Hvad er Important to take into account i experimental studies

Answer 18

• Relevance of the treatment
• Effects of the manipulation itself
• Simplified – interpretation?
• Are the results representative? Can they be generalised?

Question 19

Hvad menes med “control”, altså at lave en “kontrol” prøve

Answer 19

Control is used to prevent outside factors from influencing the study outcome.

Control is a reference against which the results of an experimental manipulation can be compared

Question 20

Nævn nogle eksempler på kontroller

Answer 20

• Negative controls (no manipulation)
• Positive controls (also control group is manipulated, but differently from the treatment group)
• Historical control (using previous experiments as a control)
• No control (if treatment groups are compared to each other)

Example:
‘Species richness decreases with chemical pollution’
– tested in a mesocosms set up with a gradient of chemicals
Negative control: No chemicals added
Positive control: A chemical of known effect/toxicity is added
Historical control: Using the control from a previous experiment
No control: A range in chemical additions that are compared to each other, no baseline

Question 21

Hvad er blind procedures og double blind procedures?

Answer 21

Blind procedures make the controls most effective
* Blind procedure: The person taking measurements does not know the treatment group
of the subject
* Double-blind: Neither the subject nor the person taking the measurements knows the
treatment group of the subject

Question 22

Hvad er blind procedures og double blind procedures gode for?

Answer 22

• Avoids confirmation bias
• Someone should know the treatments groups though!

Question 23

Other things to take into account in designing the experiments:
* Balanced design: hvad er det?

Answer 23

Equal number of replicates in each treatment group

Question 24

Taking measurements (experimental execution)
Calibration: Both instruments and persons
Reducing imprecision and inaccuracy
Hvad er imprecision and inaccuracy

Answer 24

Imprecision due to e.g., imprecision of
an instrument (random noise)

Inaccuracy due to e.g., a systematic
error in measurement (bias)

Question 25

Avoiding experimental artefacts:

Answer 25

Observer effects

Question 27

Reducing inaccuracy
Reduce the subjectivity of the measurement:
Continuous vs. categorical variable

Hvad er continuous variables?

Answer 27

continuous variables: Age, length, weight, temperature

Question 28

Reducing inaccuracy
Reduce the subjectivity of the measurement:
Continuous vs. categorical variable

Hvad er categorical variables?

Answer 28

Behaviour, looks, dead/alive, species

Question 29

Reducing inaccuracy
Reduce the subjectivity of the measurement:
Continuous vs. categorical variable

What do you need to have / do to reduce inaccuracy in experiments in this context?

Answer 29

• As few categories as possible
• Objective division between categories
• Clear criteria
• Automated measurements (image analyzer)

Question 30

Hvad er correlational studies?

Answer 30

A correlational research design investigates relationships between variables without the researcher controlling or manipulating any of them. A correlation reflects the strength and/or direction of the relationship between two (or more) variables. The direction of a correlation can be either positive or negative

Question 31

Hvad er fordelene ved correlational studies

Answer 31

• Often easier to carry out than experiments
• No stress or handling effects, no contamination
• No experimental artefacts
• Biologically relevant variation

Question 32

hvad er en experimental artifact?

Answer 32

Artifact (error), misleading or confusing alteration in data or observation, commonly in experimental science, resulting from flaws in technique or equipment

Question 33

Example:
Hypothesis: The abundance of blue whales in Antarctica is decreasing due to food limitation.
Nævn eksempler på 3 alternatives (for at samle data):
Observations,
data collection,
natural experiments

Answer 33

Observational approach: We COLLECT EXISTING information on whale and krill spottings (‘Blue whales are often spotted in the same place as krill schools’)

Data collection approach: We collect samples of krill and whale in as many areas as possible (‘The abundance of blue whales and the abundance of krill are correlated’) and analyse that data

Natural experiment: We choose areas with lots of krill and areas with few krill and set up a program for WHALE OBSERVATIONS in these
different places (‘The whales will be more abundance in the area with lots of krill than in the area with no krill’)

Question 34

In our example the amount of blue whales correlates with the amount of theirfood (krill), but does that imply causation??

Is the amount of blue whales determined by the amount of their
food?
If yes, why do you think so?
If not, why not??

Answer 34

whales are limited in their size by their metabolism, which gets faster as they get bigger, effectively putting a limit on growth

Question 35

The challenges in concluding causation from correlation:

Answer 35

• Third (confounding) variable
• Needs a mechanistic understanding of the cause – consequence link
• Scale (both time and space)
• Multiple web of causation

(A confounding variable is a third variable that influences both the independent and dependent variables. Failing to account for confounding variables can cause you to wrongly estimate the relationship between your independent and dependent variables.)

the multiple web of causation is the relationship between multiple different factors that all contribute to the cause of for example decreasing populations.

Question 36

hvad er the multiple web of causations

Answer 36

the multiple web of causation is the relationship between multiple different factors that all contribute to the cause of for example decreasing populations.

An observed effect can be caused by a multiple net of causations – biological (and human) systems are complex!

Question 37

What is a Third (confounding) variables

Answer 37

a variable which affects both variables in a similar way – therefore it will look as if variable A would influence variable B, while in reality variable C influences both A and B

Our example: What third variable could cause a decrease in both krill and blue whales?

It is difficult to be sure that the observed patterns are due to the
factors which we have measured and not some other unmeasured
variable
Another example: Amount of ice-cream which is sold in England and drowning accidents correlate with each other. Why?

Det sidste eksempel er det nok mere temperatur og varme som påvirker dem begge fx end de hænger sammen med hinanden

Question 38

Mechanistic understanding of the processes
Forklar Reverse causation:
Nævn eksemplet på hvordan reverse causation kunne være i bluewhale/krill eksemplet

Answer 38

A can affect B, but it could also be that it is B which affects A.

Our example: Could it be that it is not the amount of krill that affects the amount of blue whale, but it is the amount of blue whales which affects the krill?

Question 39

There might be a time-lag between the cause and the effect
Problem with the scale
- Cause and consequence can work in different temporal and / or geographic scales
Nævn hvad spørgsmålene her kan være ved krill / blåhvaler?

Answer 39

Our example: Do the whales and krill reproduce in the same time scale? Do they move in similar water volumes?

Question 40

Causes of the effect can be either deterministic or probabilistic. Forklar deterministic

Answer 40

(the same effect always follows from the cause)

Our example: Decreased abundance of krill could be only one reason for decreased abundance of whales. Others could include e.g., temperature influencing the ocean currents that influence whale migrations, noise from cruise ships influencing whale behavior etc.. Mostof these effects would be likely to be probabilistic.

Question 41

Causes of the effect can be either deterministic (the same effect always follows from the cause) or probabilistic. Forklar probabilistic.

Answer 41

(the cause-consequence link is likely but does not happen every time)

Our example: Decreased abundance of krill could be only one reason for decreased abundance of whales. Others could include e.g., temperature influencing the ocean currents that influence whale migrations, noise from cruise ships influencing whale behavior etc.. Mostof these effects would be likely to be probabilistic.

Question 42

Hvad vil det sige at Evaluating the evidence til sidst efter et forsøg?

Answer 42

Most of the time you will need to weight different types of evidences against each other to proceed – a combination of different approaches might give best results