Observational vs experimental study
Observational – you don’t control
Experimental – you control
Spurious correlation – nonsensical correlation – ie bee deaths correlating with the number of students
who enjoy video games.
A/B Testing – split or bucket testing. Method of comparing two versions of a webpage against each other
to see which performs better. Shown to users at random.
Don’t have a lot of variation in the varied app.
**Review A/B Testing video
Experimental Design
Example: IQ Water
What measurement to make (response)- who/what are my units
What condition to study (the treatment)- give IQ water
What experimental materials to use (the units)-
Step 1- Need control and treatment group. Randomly select from population.
Step 2-
Step 3- test the replicability of your experiment. Let others skeptical to run another test to again test
your hypothesis
Double Blind Experiment- Test subjects and test researchers do not know which group is control and
experiment group
Week 3-
Inferential Statistics- experiment and make inference about population
Sampling problem. Different selection bias- online survey, non-response, social desirability.
Central limit theorem – repeating sampling, the sample mean distribution forms a normal distribution.
Mean of the sample mean distribution is the same as the population mean.
Standard deviation devide by sqrt of n = sample sd
T statistic vs z statistic*
Increasing sample size > reduces error > CI is narrower
**ANOVA testing
