*Thyne, Clayton
*Intro to Stats, Chapter 10 examples

	clear
	set mem 80m
	*cd [your directory]

*  NOTE: the above command asks you to change the directory.  The directory will depend on how you've set up
*  the folders on your computer.  For example, my command is...
*  cd F:\teaching\Fall_2008\PS572-401-baby_stats\do_file_directories\chapter10
*  This is necessary so you know where stata is saving things.

****Part 1: t test and one-way anova
	use http://www.uky.edu/~clthyn2/intro_stats/machines_example.dta
	drop if machine==3
	ttest outputs, by(machine)
	oneway outputs machine, tab
	*Note the P value for the t test is identical to that for the anova

****Part 1b: you can also seperate them to make the analagous to previous do files
	gen var1=.
	replace var1=47 in 1
	replace var1=53 in 2
	replace var1=49 in 3
	replace var1=50 in 4
	replace var1=46 in 5
	rename var1 machine1
	gen var2=.
	replace var2=55 in 1
	replace var2=54 in 2
	replace var2=58 in 3
	replace var2=61 in 4
	replace var2=52 in 5
	rename var2 machine2
	
	ttest machine1=machine2, unpaired

****Part 2: examining summary stats for each group (school)
	clear
	use "http://www.uky.edu/~clthyn2/intro_stats/GPAs_example.dta"
	sort school
	by school:sum GPA

****Part 3: examining "explained" versus "unexplained" variance
	clear
	use "http://www.uky.edu/~clthyn2/intro_stats/GPAs_example.dta"
*compare
	graph hbar (mean) GPA, over(school)
*, which doesn't tell us much, so...
	graph hbox GPA, over(school)

****Part 4: using anova
	clear
	use "http://www.uky.edu/~clthyn2/intro_stats/GPAs_example.dta"
	oneway GPA school, tab

****Part 5: using anova with sample sizes of different N
	clear
	use "http://www.uky.edu/~clthyn2/intro_stats/GPAs_example.dta"
*first, we add in 4 new observations for UK (IA and TN stay at 4)
	set obs 16
	replace school = 2 in 13
	replace school = 2 in 14
	replace school = 2 in 15
	replace school = 2 in 16
	replace GPA=2.1 in 13
	replace GPA=2.3 in 14
	replace GPA=2.4 in 15
	replace GPA=2.2 in 16

	oneway GPA school, tab

****Part 6: using the Scheffe test
	clear
	use "http://www.uky.edu/~clthyn2/intro_stats/GPAs_example.dta"
*recreate the same sample as before, then test...
	oneway GPA school, scheffe

	*So, we can see that IA/TN are significantly different.  IA and UK are close, but not sig.
	*This seems pretty obvious by just looking at the means...
	sort school
	by school: sum(GPA)

****Part 7: two-way anova; also looking at interactions
	clear
	use "http://www.uky.edu/~clthyn2/intro_stats/GPAs_example.dta"
	gen race=.
	replace race = 1 in 1
	replace race = 2 in 2
	replace race = 3 in 3
	replace race = 4 in 4
	replace race = 1 in 5
	replace race = 2 in 6
	replace race = 3 in 7
	replace race = 4 in 8
	replace race = 1 in 9
	replace race = 2 in 10
	replace race = 3 in 11
	replace race = 4 in 12
	label define race 1 "white" 2 "black" 3 "hispanic" 4 "other"
	label values race race

	oneway GPA school, tab
	anova GPA school race
	anova GPA school race school*race
      *the interaction above says that races may be doing better/worse at different schools

****Part 8: anova and regression
	clear
	use "http://www.uky.edu/~clthyn2/intro_stats/GPAs_example.dta"
	gen race=.
	replace race = 1 in 1
	replace race = 2 in 2
	replace race = 3 in 3
	replace race = 2 in 4
	replace race = 2 in 5
	replace race = 2 in 6
	replace race = 2 in 7
	replace race = 1 in 8
	replace race = 1 in 9
	replace race = 3 in 10
	replace race = 2 in 11
	replace race = 1 in 12
	label define race 1 "white" 2 "black" 3 "hispanic"
	label values race race

	anova GPA school race school*race
	oneway GPA school
	gen UK=1 if school==2
	replace UK=0 if UK==.
	gen IA=1 if school==1
	replace IA=0 if IA==.
	regress GPA UK IA race
	*Note some of the same info from ANOVA to Regress, such as the total SS