Stats Flashcards
Logical Errors
- Ad Hominem
- Appeal to Authority
- Appeal to Ignorance
- False dichotomy
- Pragmatic fallacy
- Weasel words
- Confusion of correlation and causations
Appeal to Authority
You believe something is true because someone very important said it
Ad Hominem
Attacking the researcher for being disreputable instead of the evidence
Appeal to Ignorance
If you are not certain about your argument, then mine must be true
False Dichotomy
Considering only the two extremes in a continuum of intermediate possibilities
Pragmatic Fallacy
Something is true because something else works
Weasel Words
Use of euphemisms and misleading terminology
- Scientists say that…
- Clinical studies have shown that…
- This medicine may help with…
E.G. Low fat, natural, real fruits, chemical free
Transparency and skepticism
- Challenge existing theories
- Peer review
Authority vs Theories vs Evidence
- Should not rely on authorities
Confusion of Correlation and Causation
Since two things go together, one must have led to the other
Constructs/Concept
- Hypothetical description of something that is not real
E.G. Intelligence, anxiety, motivation
Pre-scientific Constructs
- Cold and hot “energy”
- Spirit forces
- A pinch (of salt)
Scientific Constructs
- Heat energy
- Time in seconds
- Gram
Conceptual Definition
- Describing a construct in terms of what it is and what it is not
- How it might relate to existing theories
Reification
- AVOID
- How someone’s personality reacts to the world
E.G. people belieing a certain gambling machine has greater luck
Falsifiability
- CONSIDER
- If you create something that cannot be measured, there will never be any way to tell if its real
E.G. There are fairies in my garden
Operational definition of construct
- How the construct is measured
E.G.
Motivation = Rate of button pressing
Memory = Number of Things Recalled
Learning = Decrease in Time to Solve Puzzle
Problems with operational definition of construct
- Operational definition is not a construct
- Finding a way to measure it does not make it real
Self-Report
- Are you racist?
- People can answer dishonestly
Social Desirability Scale
- How likely will respondents give answers that sound good instead of answers that are true
Direct Measures
- What’s your favourite painting
Indirect Measures
- Warmness of carpet at specific paintings
Anecdotes
- Interpreted stories of a single occurrence in the past
- Theory and evidence is mixed together
E.G. I was sick so I did X and now I’m better so X made me better
- There were so many other factors that could make you feel better
Case Studies
- Theory and evidence are separated
- Identifying the specific factor that caused you to feel better
Correlational Studies
- At least two variables are measured from each person to find relationship between variables
CORRELATION DOES NOT MEAN CAUSATION
If X correlates with Y, it could be either
- X → Y
- Y → X
- Z → X or Z → Y
Correlation Coefficient: +
As values on one variable get bigger, values on the other get bigger
Correlation Coefficient: -
As variables on one variable get bigger, values on the other get smaller
Independent Variable
The presumed caused and what is manipulated
Dependent Variable
What is measured
Random Allocation
Participants are given an experimental condition at random
Random Selection
Participants are chosen randomly from a population
Replication
Study is repeated with same method and same results are produce
Blinding
Single Blind
- Participants are unaware of which condition they are in
Double Blind
- Participants are unaware of which condition they are in
- Researchers are unaware which condition is being run
Internal Validity
How certain we are that changes in the IV caused changes in the DV
External Validity
Extent to which findings from the study can be generalised to the population at large
True Experiment
- All independent variables are randomly allocated
- Controlled variables
- Establish cause and effect
Quasi Experiment
- Less control over variables or no control variable
- Casual inference
- Lower internal validity but higher external validity
Theory of Cognitive Dissonance
A person comes to believe in what they do to reduce internal conflict
Alternative Hypothesis
Suggests there is a relationship or difference
Null Hypothesis
Suggests that there is no relationship or difference
- If we reject the null hypothesis → We have evidence of an effect
- If we retain the null hypothesis → We have no evidence of anything
Experimental Hypothesis
Explicitly defines the expected change or effect
Mode
Most frequent score
- Advantages: Unaffected by extreme values
- Disadvantages: if all cancers are grouped together cancer is the most common way to die, but if separate cancers are considered road fatalities have a higher rank
Median
Middle score
- Avantages: Unaffected by extreme values
- Disadvantages: Not based on all values
Mean
Average score
- Advantages: Takes all scores into account
- Disadvantages: Affected by extreme values
Standard Deviation
- Count the number of scores
- Add up the scores and find their mean
- Find the deviation scores for every score
- Square every deviation sore then add them up
- Divide by the number of squares
- Take the square root
Inferential Statistics
- Population
- Sample
Sampling Distribution
- Simulate an experiment being run over and over again
- Shows variability across experiments
- Shows likelihood of obtaining a result if the null is true
Raw Score Distribution
- Shows variability across individuals
- Shows likelihood of obtaining a particular score
Inferential Statistics: POPULATION
- Entire collection in which you are interested
- Parameters
Normal Distribution
Two thirds of all scores fall within one standard deviation of the mean
Inferential Statistics: SAMPLE
- Selection from the entire collection
- Statistics
P-Value
Probability of results occuring due to random chance
If the probability is high we retain the null hypothesis
- If p>0.05 retain
If the probability is low we reject the null hypothesis
- If p<0.05 reject
The lower the p, the better, since less likely to have occurred by random chance
Probabilistic Nature of Conclusions
See docs (table)
Statistical Significance
How likely the effect was due to chance
Practical Significance
- How useful is the effect
- Is it worth investigating further
Power
- Probability of finding an effect (rejecting the null hypothesis while it is false)
Type 1 Error
- Probability that the null hypothesis is incorrectly rejected when it is actually true
- Detecting an effect when there isn’t one
Type 2 Error
- Probability that the null hypothesis is retained when it is actually false
- Fail to detect an effect when there is one
Denialism
- Finding a few imperfections in a theory and disproving every aspect of the theory
