Epidemiology and Beyond: November 2012

Tuesday, November 27, 2012

Survival analysis using Stata

Survival analysis using Stata

Survival Analysis with Stata - Stephen P. Jenkins
What is the relationship between baseline hazard and baseline hazard contribution?
Cox's Proportional Hazards Model and Cox Regression - Princeton University
How to get the hazard and survival functions for a specific time point?
Introduction to Duration Models - Matt Golder

Monday, November 19, 2012

Stata Programming

Stata programming

Stata Blog: Programming an estimation command in Stata: A map to posted entries
Manage the ado programs
- Stata Syntax-highlighting for Notepad++ or download from the Notepad++ website
- Syntax Highlight Packages (Dynamic documents in Stata: MarkDoc, Ketchup, and Weaver): Synlight, Weaver, MarkDoc
- Determine build-in command or ado-file: which command
- Check codes: viewsource command or type "directory\xxx.ado"
- list of the packages: ado dir
- Install: ssc install package
- check update: adooupdate
- update: adoupdate, update
- uninstall: ado uninstall package|[#]
- re-run profile.do: do profile.do
- How to edit an ado file, '- ADOEDIT -': module to edit ado file in Stata's do-file editor
Programming Stata of Stata Tutorial - Princeton University.
The University of Wisconsin provides nice websites for Stata for Researchers and Programming in Stata. The new Stata Programming Essentials is better for new programmers and the new Stata Programming Tools is better for more advanced persons.
Macro: global and local macros: macro dir/macro drop
Example Macro with `' and/or "":

for both local and global macro, braces { and } can be used to form nested constructions.

global pth "c:\abc\"
di "${pth}def\" // ==> c:\abc\def\

local x 2+2

di "x" // ==> x

di `x' // ==> 2+2 ==> 4

di "`x'" // ==> 2+2

local x=2+2 // ==> local x 4

di `x' // ==> 4

di "`x'" // ==> 4

di "`=2+2'" // ==> 4

di "`2+2'" // ==> N/A

local x 2

di "`=`x'-2'" // ==> 0

local pth "c:\project"

di "`pth'\data\" //==> "c:\project\data\"

global pt "c:\project"

di "$pt\data\" //==> "c:\project\data\"

di "${pt}data\" //==> "c:\projectdata\"

local a 2+3

local b 7

display `a'+`b' //==> 10

display "`a'+`b'" //==> 2+3+7

display "`a'"+"`b'" //==> 2+3+"7" invalid name

display "`a'""+""`b'" //==> 2+3+7

regress mpg weight

local rsqf e(r2)

local rsqv = e(r2)
di "R-squared_1f=`rsqf'" //==> R-squared_1f=e(r2)
di "R-squared_1v=`rsqv'" //==> R-squared_1v=.6515312529087511
di "R-squared_2f=" `rsqf' //==> R-squared_2f=.65153125
di "R-squared_2v=" `rsqv' //==> R-squared_2v=.65153125
di "R-squared_3f=" "``rsqf''" //==> R-squared_3f=.6515312529087511
di "R-squared_3v=" "``rsqv''" //==> N/A

Stata Blog: Programming an estimation command in Stata

Example 1: Storing and extracting the result of an extended macro function

local count : word count a b c
display "count contains `count'" ==> count contains 3

Example 2: Using gettoken to store first token only

local mylist y x1 x2
display "mylist contains `mylist'" ==> mylist contains y x1 x2
gettoken first : mylist
display "first contains `first'" ==> first contains y

Example 3: Using gettoken to store first and remaining tokens

gettoken first left: mylist
display "first contains `first'" ==> first contains y
display "left contains `left'" ==> left contains x1 x2

Example 4: Local macro update

local p = 1
local p = `p' + 3
display "p is now `p'" ==> p is now 4

Example 5: Local macro update

local p = 1
local ++p
display "p is now `p'" ==> p is now 2

Friday, November 16, 2012

The rule of thumb: AIC preference/signficance levels

The rule of thumb: AIC and BIC preference/signficance levels

The Akaike information criterion (AIC) is a measure of the relative goodness of fit of a statistical model, which was developed by Hirotsugu Akaike in 1974. It was not much appreciated until 21 century. Now it’s one of the most used fit statistic output for statistical modeling. The preferred model is the one with the minimum AIC value; however, there are no statistic test for model choice based on this criterion. I found a table in Joseph M Hilbe’s book (Negative Binomial Regression, 2^nd Ed, 2011) which may help us to choose a better model. Below is a table I modified based on his confusing table:

AIC = -2(log-likelihood) + 2(number of predictors including the intercept)

========================================

AIC of (Model A – Model B) , if AIC(A) > AIC(B)

----------------------------------------------------------------

< 2.5 No difference in models

2.5 – 5.9 Prefer B if sample size n > 256

6.0 – 9.9 Prefer B if sample size n > 64

≥ 10.0 Prefer B

========================================

Another criterion measure is Bayesian information criterion (BIC). Raftery AE (1995) gave the scale for relative preference of two models (I modified the table in Hilbe's book):

BIC = -2(log-likelihood) + (number of predictors including the intercept)*(ln(sample size))

========================================

BIC of (Model A – Model B) , if BIC(A) > BIC(B)

----------------------------------------------------------------

< 2.0 Weak

2.0 – 5.9 Positive and prefer B

6.0 – 9.9 Strong and prefer B

≥ 10.0 Very strong and prefer B

========================================