UNIT 1: CH4 - Ecological Methods Flashcards
3rd Week material:
Steps of the Scientific Method?
a process that evaluates ideas with observation and analysis
ex: a theory is submitted, a prediction is tested by gathering data, and if it’s supported by data (good), and if it’s not, it’s revised or rejected
WHY? We search for a CAUSAL explanation, in order to make sure there is a ‘mechanism’ that can explain the correlation. Internal consistency is important!
Sample?
Population?
Infer?
- Sample : a portion of the population (gather descriptive statistics. Then, infer parameters.) Who we actually measure.
- Population: a collection of subjects or events that share a common characteristic
- Infer : use samples to infer (predict and apply
When would an inference be inaccurate even if the data results were clear?
Sample could be too small
It could be a biased or skewed sample.
Not randomly selected.
Why would these be important?
1. Replication
2. Randomisation
3. Representation
- it means the basic findings can be reproduced(repeats the research to see if the basic results can be obtained again), confidence in the results increases, it confirms the original results, original results are more reliable(if different researchers get consistent results from testing the hypothesis)
- Used to minimize confounding variables (variables that could skew or account for the results instead of the indep variable) such as pre-existing differences among the experimental/control groups, it’s by chance, randomly assigns participants to either experimental or control group. The 2 groups will be the same, on average, for all variables.
- make sure that the sample actually will represent the population as a whole. ‘Random Sampling’ will help with this variable.
Types of Study Design:
- Descriptive (observational, comparative)
- Experimental
- Using Models
- This design is simply gathering data. Does not manipulate variables. Comparative generally shows correlations, but doesn’t explain the causation. This design studies the WHAT/HOW.
- In order to find this causation, an experiment must be done. To discover cause & effect relationships. Random assignment is used to minimize confounding variables, operational definitions are established, Independent (manipulated to study it’s effect) and dependent (what is measured) variables are set up. Experiments study the WHY.
- Models are simply conceptual(not empirical) & abstract. representation of a real system. It helps people know, understand, or simulate a subject the model represents. It could be a physical model, too. Qualitative or Quantitative. It can predict how dep. variable will change if indep variable changed.
THE ASSUMPTIONS of a Model? (don’t always predict accurately!)
- Has ‘fixed regressors’ – has fixed numbers so that the model can be used over and over again with consistent results.
- No correlation should exist. Add more variables so that it follows real circumstances (increases validity too)
- There should be constant parameters so that the data is being collected by the same process. (ex: if 1 sq km was being studied for a species and this was the parameters for the model, but now, this year, the area allowed was 5 sq km. So the parameters aren’t constant. Need to add more variables in, to account for the increased area.
- That it is a linear model. As one thing changes, the other thing changes.
- Normality : the data will be normally distributed.
Compare Experimental vs. Observational/Comparative.
In order to find the ‘mechanism’, what needs to be done?
- Need an experiment
- Need to collect the data from the experiment
- Needs to be replicated
- Establish cause & effect relationship
A well-designed experiment can help identify causal mechanisms
Key things to watch for designing an experiment:
* confounding variables
* randomization
* operational definitions
* replication
* validity
Correlation vs Causation?
- shows the strength between 2 variables and how they are related (positive or negative or neutral)
- shows which variable is causing an outcome. The outcome is dependent on the cause.
Types of Data:
Numerical vs Categorical?
What is meant by testing for ‘interactions’?
An interaction is observed when the effect of one independent variable depends on the level of another.
In statistics, an interaction may arise when considering the relationship among three or more variables, and describes a situation in which the effect of one causal variable on an outcome depends on the state of a second causal variable(Source: Wiki)
ex: phenotypic plasticity
Phenotypic plasticity?
Phenotypic plasticity can be defined as ‘the ability of individual genotypes to produce different phenotypes when exposed to different environmental conditions’ (Pigliucci et al. 2006).