Scientific Method and Hypothesis Testing - PowerPoint Presentation

Scientific Method and Hypothesis Testing

“Whereas other types of human speculation are based on mere opinion, science pursues and sticks to the facts” Hypothesis and Perception: The Roots of Scientific Method By Errol E. Harris

What is science? The world is understandableScientific ideas are subject to changeScientific knowledge is durable Science cannot provide complete answers to all questions

The backbone of science The process of “doing” science What components are involved?

Scientific Method From: Martin and Bateson

My version of the flow chart Martin and Bateson forgot the two most important steps

Ask Collect and analyze dataPreliminaryobservations Choose measuresChoose recording method Hypothesize predict design experiment analyze Interpret OBSERVE Write Up

Reality of doing science No scientist has a flow-chart on the wall The truth about the scientific method

OBSERVE AskCollect and analyze data PreliminaryobservationsChoose measures Choose recording method Hypothesize predict design experiment analyze Interpret Write Up

How does science REALLY work? http://undsci.berkeley.edu/article/scienceflowchart Coalition for the Public Understanding of Science

Science is self-replicating… Observation: Caddis in lab unpalatable Question: Do they have ttx ? Answer: yes….?

In a way, science is self-replicating… New Obs : small = more ttx , large = less New Q: Why? Answer: One big difference in behavior

In a way, science is self-replicating… New Obs : Fewer caddisflies at top New Q: Do female newts take advantage? Answer: Yes, females lay eggs at top

In a way, science is self-replicating… New Obs : Females lay eggs high New Q: Does this translate into fitness advantage? Answer: Yes!

Falsifiability – part 1 Example of scientific hypothesis Example of non-scientific hypothesis

Falsifiability – part 2 Media headlines: “Science Proves That Beer Is The Perfect Human Beverage”“Science Proves That You Should Un-Friend Your Ex On Facebook”“Science Proves That Your Cat Hates You” “ Science proves that cannabis prevents cancer tumors ” “ Science proves that smoking causes lung cancer ”

"Philosophy of science is about as useful to scientists as ornithology is to birds .“Richard FeynmanAmerican physicist

Hypothesis

Fish Exercise Observation: Some minnows appear to act strangely when a predator is nearbyQuestion:Hypotheses:H 0HAPrediction:Experimental Test:

Harvard First Law of Animal Behavior“A student of ethology meticulously controls all aspects of an experiment so that the animal can give only clear and illuminating responses, whereupon it does as it damn well pleases.”

Type of Question http:// www.youtube.com/watch?v=JIrqirr6jBY

Type of Question Pair knowledgeable “tutor” with naïve Then test naïve alone Do they acquire the antipredator behavior

Choosing a study organism Easily seen in nature Readily available Does it acclimate well Are life-history characteristics known Background information Is species practical for study When is it active

Type of Study Two general types: Causality is sometimes difficult to inferExample: lung cancer and cigarette smoking in humansCan we assign treatments?Observational study → Increased risk of cancer correlated with smoking Conclusions:

Type of Study

Where to study? What are some pro’s and con’s for each?

Experimental Studies May be simple or complexSimple experiments:Tinbergen 1962Made “artificial” gull eggsPlaced broken eggs at different distances Recorded number of eggs eaten from different nests

Alternative Method Compare the dependent variable between individuals that get manipulated (treated) and those that don’t (control)Compare predation rate for eggs vs no eggs

Problems with Manipulation Assumption:Examples:Egg shells different sizesNests in different positionsTemperature differences between nests Tests/trials run at different timesPersonal ExampleGoal?

How do deal with these factors? Temp slightly different at different nestsSome closer to the edge Individuals tested at different times

How do you control for the Harvard 1 st Law?Most powerful control for inherent variability in ethology Use individual in control and experimental conditionsExample: hellbender behaviorAnother way:

Example 1: Gull egg experiment looking at manipulated vs unmanipulated nestsExample 2: Do salamanders change where they lay eggs when insects are present Example 3: Does cutting the olfactory nerve effect the homing ability of fish compared to controls A problem with manipulated treatments

What effect 2+ factors have on behaviorExample: body size and alarm cues Gall and Mathis 2011, CopeiaMultiple Independent Variables If we had not looked at body size, what would the graphs look like?

Never collect data on only 1 dependent variable Example:Lines crossedTime movingNumber moves Gall and Mathis 2011, Copeia Multiple Dependent Variables

Acclimate Hides/blindsOne-way glassMirrorVideo camera Observer Influence

Quantifying Behavioral Data Watch, record, and catalogue behaviorsFocus on discrete behaviors Assign each behavior to 2 categoriesWhy not just use a functional description from the beginning?

Types of measures

Sampling - who Often necessary to observe groups Must decide who to record (impossible to watch everyone)

Two parts to determining when to make your observations Sampling - when

Design

Construct a visual representation of your data Can we tell if there is a difference between treatments?What information do we need before making an inference? Once your data are collected

Once your data are collectedTo answer whether the difference in your data are meaningful need statistics

Statistical Tests ParametricDepend on meeting assumptionsRobust to violationsTransformations can correct departures Nonparametric Does not rely on assumptions Converts data to ranks Less robust than parametric tests Types of Tests