MBI
Statements
If: Hypothesis
And: test
Then: prediction
and/but: result
Therefore: hypothesis is or isnt supported
Inductive reasoning
Should not be relied on: Ex. All ducks in this pond are white therefore all ducks are white
Hypodeductive reasoning (top down)
If-and-then (comparing results to prediction). Active voice is preferred (I kicked the ball vs the ball was kicked). First person > third person
Scientific Method
Observation
Causal question
Hypothesis
Test
Prediction
Results
Conclusion
Prediction
Logical extension of hypothesis; what were to happen if it were true
Observation test
Purely observable test, no manipulation of variables
Provide support but it's more tentative than controlled experiment
Controlled Experiment
Measures a response variable of interest after manipulating some aspects of chosen test subjects
Ex. Changing the diet of a rat to measure weight gain
Not useful for long term, historical and globally scaled questions
Confounded experiment
Change in independent variable also changes other variables in the experiment (confounding variable)
Models
For questions that cannot be replicated (why did the airplane crash?)
Meta-Analysis
Already completed research studies are analyzed
Descriptive questions
Questions that use a lot of words such as quantify, measure, count, and enumerate
Causal questions
Seek explanations for observed patterns
Proximate hypothesis
Explain how a trait or behaviour works
Testability
Every hypothesis must be testable, so it allows to provide support or against a particular hypothesis
Falsification
Misleading idea that you can never know when something is true, only if it's false.
Science is not bound by this and can show things are tentatively true without needing to find every other possibility false
Sampling methods
Random: every potential sampling unit has an equal chance of being chosen
Systematic: sampling units are selected in a systematic way
Stratified random sampling: random sampling within subgroups where subgroups differ in some characteristic
Basic vs Applied science
Basic: aims to expand fundamental knowledge to understand natural phenomena
Applied: practical applications for scientific knowledge and solve specific problems
Responde (dependent) variable
What to measure
Independent variables
Treatment/factors - what to manipulate
Levels
Number of different treatments
Number of sampling events
How many times you sample
Controlled variable
What is staying consistent between replicates
Sampling error
Difference between a samples characteristics and the characteristics of the entire population it represents
Why does sampling error occur
Because you're only taking a subset of the whole population
As the sample size increases...
Average amount of sampling error decreases
Sampling bias
Data selection is skewed by collection criteria (conscious or unconscious)
ex. Sampling what's easy and supports your hypothesis (only picking tall trees)
Applies to how you analyze and interpret data
Can bias be avoided
Not always, but minimized and acknowledged
Accuracy and precision
💀
Accurate: value close to real value
Precision: values close to each other
Confidence interval
Measure of variation- gives an estimated range of values which is likely to include an unknown population parameter
Statistics
Collection, compilation, and communication of data
Scientific usage: the probalistic analysis of data
NOT THE SAME AS DATA
Statistical hypotheses
Two competing hypotheses (null, alternate)
Goal: determine if there's enough evidence within the sample to reject the null hypothesis
P < 0.05, then we accept the result is "real"
p value
tells you that the probability is real and not due to chance
Null hypothesis
Assumes no effect, no difference, no relationship
Alternate hypothesis
Assumes there is an effect, difference, or relationship
T-tests
Used to compare the means of two groups (two treatment levels)
Includes null and alternate hypothesis
Degrees of freedom: observation rows - # of parameters
Standard deviation
How far apart the numbers are related to each other
Chi-square test
compare observed data to expected data to determine whether there is a significant difference between them
Expected = (row total * column total) / total total
Linear models
Is there an effect of this variable on this thing measured
Regression
Are these two variables related?
Multivariate statistics
How do all of these variables interact to show one thing
Cluster analysis
Group related data into interpretable groups
Discriminant analysis
Put new experimental units into existing groups based on the composition of units
Primary Literature
Original research and/or new scientific discoveries
Secondary literature
Summarizes abd synthesizes primary literature. Usually broader and less current than primary literature
Tertiary Literature
Summarized or condensed version of materials with references to primary or secondary sources (textbooks, dictionaries, handbooks)
Posters
Objective: Present current research, often at conference or meeting
Audience: Interested public, scientists in fields
Authors: active researchers, upper year undergrads
Type: Visual
pros: very current, often first source on new finding, author is present
cons: limited audience
Presentations
Oral, very current, presented newly published paper or even in progress work.
Cons: limited audience, meeting attendees, may present in progress work
Social media
Reaches everyone, promote what you're doing, very current, easy access, lots of material
Cons: info can be misinterpreted, rely on readers going further to get the whole story
Science news articles/news articles
Science news articles are by science journalists and are meant for everybody
- more accessible, great way to understand research
cons:
- clickbait
- reviewed by non specialist/editor with different priorities
- depends on good news source
News articles
- Not written like scientific papers
- Headline- concise and attention grabbing (click bait)
- most important info is presented first
Videos/podcasts
objective: education, reach wide audience, abstract of paper
audience: everyone
authors: everyone
type: video
pros: audience, accessible, creates buzz, promote what you're doing
cons: can be misinterpreted, may not be peer reviewed, relies on readers to get full story