Hypotheses and Scope of Inference

Important - helps create quantifiable, verifiable, replicable and defensible research 

• Research question - Why? - Introduction
• Develop a theory - Your Answer - Introduction
• Identify variables - How? - Materials and Methods
• Suggest hypotheses - Expectations - Materials and Methods
• Test hypotheses - Collect/analyse data - Results
• Evalute results - What does it mean? - Discussion
• Critical review - What does it not mean? - Discussion

Important to care - otherwise method potentially inappropriate and lead to inaccurate results and poor statistical analysis

• Read literature - identify knowledge gaps
• Adress knowledge gap in question
• Forumlate hypotheses 

Hypothesis = clearly stated postulated description of how experimental system works - testable/falsifiable

• Null (H0) - no significant difference/effect, observed difference due to sampling/experimental error
• Test (H1) - statement of what experimental test set up to establish

Test is alternative to null and accepted if H0 is rejected.

Failing to reject H0 does not mean its true but there isn't sufficient evidence against H0 in favour of H1

Study design = no. and spatiotemporal distribution of clearly defined sampling units and manipulations and/or observations made

Provides direct link to research question and hypothesis = efficient estimates of parameters of pop. of intrest

Manipulative study - researcher changes explanatory variable and studies effect on response:

• Focused, subject-related studies
• Provide information about mechanisms
• Complex and invasive/unethical
• Unknown effects of treatments

Observational study - researchers use natural conditions to study variables

• Easy to carry out
• Less inavsive
• Uses natural variation
• Interpretation difficult

Response (dependent) variable - subject intrested in

Explanatory (indpendent) variable - variable change that may affect response variable

Categorical variables - numerical values classifying information and count frequency of observations with each class

Quantative vairales - graph with scales permiting measurements of distance along a continuum, with determinable distance between data

• Discrete - clear categories
• Continuous - any value

Correlation not imply causation:

• Confounding variables - variables researcher failed to control/eliminate, damaging experiemnt internal validity
• Reverse causality - two variables are associated, not in the way expected
• Simpson's Paradox - phenomenon when data groups are combined results unexpected

Well-designed experiments = comparative so controls should be used

Control - experimental factor/treatment where no treatment administered as a reference for comparison

Concurrent controls = take place at same time as experiment

BETTER THAN

Historical controls - lack of propoer controls = false conclusions

Sometimes need >1 control - negative and positive (placebo) control

Inference - conclusion reached on basis of evidence and reasoning about underlying (statistical) population from sample

Scope of inference - population which inference can be drawn based on study

Spatial (georgraphic) dimension to research question

• Defined by study area
• Should be clearly defined - important implications for selecting area and no. sampling units

Temporal dimension to research question - explains how long sampling must occur for in order to study question fully

Ecological dimension - depends on defintion of population and adequancy of sample e.g. representative and intensity

• In order to study multiple species, scope of inference must encompass whole assemblage
• Ecosystems - interactions between unstudied species may affect results

Cannot infer findings beyond the scope of population sampled.