Lesson 1: Intro to Statistical Research Methods

• To have some level of confidence in survey results, you would need to know, at least:
  • how many people were surveryed
  • who was surveyed
  • how the survey was conducted
• Constructs
  • Things that are generally difficult to measure precisely eg:
    • Happiness
    • Hunger
    • Health
• Operational Definition
  • Systems to measure constructs

• Samples & population

  • Average across population is called "population parameter". Denoted by 'mu'
  • Within population there's a "sample" of people.
  • Average for the sample is called "sample statistics". Denoted by 'x-bar'
  • Difference between mu and x-bar is called "sampling error"

  

  • Obviously, the bigger the sample, the closer the mu is to x-bar population
  • Visualize relationship
  • Helps to find correlations between data
  • Correlation does not imply causation
  • "No two countries with a McDonald's have ever gone to war since opening the McDonald's" -- Thomas Friedman
  • Casual inference
  • To observe pattern between two variables, have to consider "lurking variables"
  • To show relationships, you can do observational studies or surveys
  • To show causation, you'd need to do a "controlled experiment"
  • Downsides to surveys
  • Untruthful responses
  • Biased responses
  • People not understanding the questions or refusing to answer
  • Controlled Experiments
  • Placebo
    • Allows for a comparison 'control group' to the 'experimental group'
  • Blind
    • Ensures everyone thinks they're the 'experimental group'
  • Double blind
    • Researchers observing results shouldn't know who's in control or experimental to prevent judgement bias
  • Independant variable = variable that is changed to test effect on dependant variables
  • Dependant variable = variable being tested