Tags & Description
Statistically Significant
Unlikely to occur by random chance
Strength of Evidence
How sure we are it's not chance
Sample
The set of our observational units
Sample Size
Number of observational units (n)
Statistic
Summary of the sample (p-hat)
Parameter
The long-run proportion (π or p)
Test of Significance
Measuring the strength of evidence against the chance model (3-S Strategy)
Null Hypothesis
Random chance explanation
Alternative Hypothesis
Special situation explanation
Null distribution
Distribution of simulated statistics from simulating the null hypothesis
3-S Strategy
Statistic, simulate, strength of evidence
Steps of a statistical investigation
Research question, collect data, explore data, draw inferences, conlusions, look back and ahead
Data
Recorded from observational units
Variables
Recorded characteristics of the observational units
Quantitative Variables
Numerical values
Categorical Variables
Category designations
Distribution
The pattern of the data
Random Process
A process that can be repeated an infinite amount of times with predicted outcomes
Simulation
Artificially re-creating a random process to estimate the probability
Anecdotal Evidence
Personal experience or striking example
Dot Plot
Each dot is a statistic
Standardized statistic
(statistic - mean) / (standard deviation)
P-hat
The summary number of the sample - what you really saw - statistic
π
Your prediction for the whole population - not just your sample
P-value
Evidence against the prediction (π) - how unusual was it? - probability
Too unusual to be random (%)
< 5% (p-value)
Binary Variable
only two outcomes
Distribution
how often it happens
Statistic
summary number
Strength of evidence
against random chance
Two-Sided Test
looking for a difference, bigger or smaller
When do we use the two-sided hypothesis?
When there is no prior assumption of the study
What does the standardized statistic specifically measure?
It measures how many standard deviations it is away from the mean
What EXACTLY is the p-value
proportion of statistics in null distribution as extreme as the statistic.