STA 3381 Lecture Notes - Lecture 12: Scatter Plot, Linear Regression, Cadency
23 May 2018
School
Department
Course
Professor
X ffactor → type
Y → response?
Differences from ch.8:
● POI (step 1)
● Step 3
● Step 4: df is different
● Step 5: add one extra assumption
Under t Test table on JMP, be careful to note that it is pharmaceutical minus computer, not the
other way around. If it is the other way around, can go back to step 1 and switch miu1 and miu2,
to make less confusing.
9.3
● Have dependencies between two sample sets
○ Ex twin example: each row is a pairing, so dependent
● Step 3 on twin ex: if the first twin is nicer, then the difference should be LLARGER
● Step 5: the difference between the data is normal. Not the individual data.
○ Make a separate column in JMP for the differenced data.
○ Then look at normal quantile plot
● Step 6; Test mean: test it against zero
9.3 is dependent, 9.2 is independent. If independent, need to do pool value
--
One from group one matched with one from group 2 → paired, so dependent.
● If sample size of groups aren’t the same, you can’t form pairs.
● Good way to know no pairs, so independent.
Ch 12: Linear regression
● Dependent variable → response variable (y)
● Independent variable → predictor variable (x)
● Make scatter plot first
○ Need to be straight ish
Document Summary
Under t test table on jmp, be careful to note that it is pharmaceutical minus computer, not the other way around. If it is the other way around, can go back to step 1 and switch miu1 and miu2, to make less confusing. Ex twin example: each row is a pairing, so dependent. Step 3 on twin ex: if the first twin is nicer, then the difference should be llarger. Step 5: the difference between the data is normal. Make a separate column in jmp for the differenced data. Step 6; test mean: test it against zero. One from group one matched with one from group 2 paired, so dependent. If sample size of groups aren"t the same, you can"t form pairs. Good way to know no pairs, so independent. Create 2 columns with x and y, and do three steps on the slides. If you don"t have values near zero, don"t interpret the y-intercept!
