Introduction to Statistical Computing in Clinical Research

1
Introduction to Statistical Computing in Clinical
Research

• Biostatistics 212
• Lecture 1

2
Today...

• Course overview
• Course objectives
• Course details grading, homework, etc
• Schedule, lecture overview
• Where does Stata fit in?
• Basic data analysis with Stata
• Stata demos
• Lab

3
Course Objectives

• Introduce you to using STATA and Excel for
• Data management
• Basic statistical and epidemiologic analysis
• Turning raw data into presentable tables, figures
  and other research
• Prepare you for Fall courses
• Start analyzing your own data

4
Course details

• Introduction to Statistical Computing - 1 unit
• Schedule 7 lectures, 7 lab sessions, on 7
  Tuesdays in a row
• Dates July 31, August 7,14,21,28, September
  4,11
• Lectures 115-245
• Labs 300-400
• All in China Basin (CBL 6702, 6704)
• Final Project Due 9/18/06

5
Course details

• Introduction to Statistical Computing
• Grading Satisfactory/Unsatisfactory
• Requirements
• -Hand in all five Labs (even if late)
• -Satisfactory Final Project
• -80 of total points
• Reading Optional

6
Course details, cont

• Course Director
• Mark Pletcher, 514-8008
• mpletcher_at_epi.ucsf.edu
• Teaching Assistants
• Jennifer Cocohoba
• jcocohoba_at_nccc.ucsf.edu
• biostat212_section1_at_yahoo.com
• Sharon Chung
• sharon.chung_at_ucsf.edu 
• biostat212_section2_at_yahoo.com

• Folashade Jose
• josef_at_peds.ucsf.edu
• Lab Instructors
• Mandana Khalili
• Alan Bostrom

7
Overview of lecture topics

• 1- Introduction to STATA
• 2- Do files, log files, and workflow in STATA
• 3- Generating variables and manipulating data
  with STATA
• 4- Using Excel
• 5- Basic epidemiologic analysis with STATA
• 6- Organizing a project, making a table
• 7- Making a figure with STATA or Excel

8
Overview of labs

• Lab 1 Load a dataset and analyze it
• Lab 2 Learn how to use do and log files
• Lab 3 Import data from excel, generate new
  variables and manipulate data, document
  everything with do and log files.
• Lab 4 Using and creating Excel spreadsheets
• Lab 5 Epidemiologic analysis using Stata
• Last 2 lab sessions dedicated to working on the
  Final Project
• - Labs 3 and 5 are significantly longer and
  harder than the others

9
Overview of labs, cont

• Official Lab time is 300-400, but we will start
  right after lecture, and you can leave when you
  are done.
• Lab sections led by Jennifer Cocohoba and Sharon
  Chung
• Mac session
• Labs also staffed by Fola Jose, Alan Bostrom,
  Mandana Khalili, and I

10
Overview of labs, cont

• Labs are due the following week prior to lecture.
  Labs turned in late (less than 1 week) will
  receive only half credit after that, no points
  will be awarded. However, ALL labs must be
  turned in to pass the class (even if no points
  are awarded).
• Lab 1 is paper
• Labs 2-5 are electronic files, and should be
  emailed to your section leaders course email
  address biostat212_section1_at_yahoo.com
  (Jennifer) or biostat212_section2_at_yahoo.com
  (Sharon)

11
Final Project

• Create a Table and a Figure using your own data,
  document analysis using Stata.
• Due 1 week after last lab session, 20 points
  docked for each 1 day late.

12
Orientation to binder

• Course Overview
• Final Project
• Lectures and Labs (just in time)
• Other handouts

13
Getting started with STATA

• Session 1

14
Types of software packages used in clinical
research

• Statistical analysis packages
• Spreadsheets
• Database programs
• Custom applications
• Cost-effectiveness analysis (TreeAge, etc)
• Survey analysis (SUDAAN, etc)

15
Software packages for analyzing data

• STATA
• SAS
• S-plus, and R
• SPS-S
• SUDAAN
• Epi-Info
• JMP
• MatLab
• StatExact

16
Why use STATA?

• Quick start, user friendly
• Immediate results, response
• You can look at the data
• Menu-driven option
• Good graphics
• Log and do files
• Good manuals, help menu

17
Why NOT use STATA?

• SAS is used more often?
• SAS does some things STATA does not
• Programming easier with S-plus?
• Complicated data structure and manipulation
  easier with SAS?
• Epi-info (free) is even easier than STATA?

18
STATA Basic functionality

• Holds data for you
• Stata holds 1 flat file dataset only (.dta
  file)
• Listens to what you want
• Type a command, press enter
• Does stuff
• Statistics, data manipulation, etc
• Shows you the results
• Results window

19
Demo 1

• Open the program
• Load some data
• Look at it
• Run a command

20
STATA - Windows

• Two basic windows
• Command
• Results
• Optional windows
• Variable list
• History of commands

• Other functions
• Data browser/editor
• Do file editor
• Viewer (for log, help files, etc)

21
STATA - Buttons

• The usual open, save, print
• Log-file open/suspend/close
• Do-file editor
• Browse and Edit
• Break

22
STATA - Menus

• Almost every command can be accessed via menu

23
Demo 2

• Enter in some data
• Look at it
• Run a couple of commands

24
Menu vs. Command line

• Menu advantages
• Look for commands you dont know about
• See the options for each command
• Complex commands easier learn syntax
• Command line advantages
• Faster (if you know the command!)
• Closer to the program
• Only way to write do files
• Document and repeat analyses

25
STATA commandsDescribing your data

• describe varlist
• Displays variable names, types, labels
• list varlist
• Displays the values of all observations
• codebook varlist
• Displays labels and codes for all variables

26
STATA commandsDescriptive statistics
continuous data

• summarize varlist , detail
• obs, mean, SD, range
• , detail gets you more detail (median, etc)
• ci varlist
• Mean, standard error of mean, and confidence
  intervals
• Actually works for dichotomous variables, too.

27
STATA commandsGraphical exploration continuous
data

• histogram varname
• Simple histogram of your variable
• graph box varlist
• Box plot of your variable
• qnorm varname
• Quantile plot of your variable to check normality

28
STATA commandsDescriptive statistics
categorical data

• tabulate varname
• Counts and percentages
• (see also, table - this is very different!)

29
STATA commandsAnalytic statistics 2
categorical variables
30
STATA commandsAnalytic statistics 2
categorical variables

• tabulate var1 var2
• Cross-tab
• Descriptive options
• , row (row percentages)
• , col (column percentages)
• Statistics options
• , chi2 (chi2 test)
• , exact (fishers exact test)

31
Getting help

• Try to find the command on the pull-down menus
• Help menu
• If you dont know the command - Search...
• If you know the command - Stata command...
• Try the manuals
• more detail, theoretical underpinnings, etc

32
STATA commandsAnalytic statistics 1
categorical, 1 continuous
33
STATA commandsAnalytic statistics 1
categorical, 1 continuous

• bysort catvar summarize contvar
• mean, SD, range of one in subgroup
• ttest contvar, by(catvar)
• t-test
• oneway contvar catvar
• ANOVA
• table catvar , contents(mean contvar)
• Table of statistics

34
STATA commandsAnalytic statistics 2 continuous
35
STATA commandsAnalytic statistics 2 continuous

• scatter var1 var2
• Scatterplot of the two variables
• pwcorr varlist , sig
• Pairwise correlations between variables
• sig option gives p-values
• spearman varlist , stats(rho p)

36
Demo 3

• Load a STATA dataset
• Explore the data
• Describe the data
• Answer some simple research questions

37
In Lab Today

• Familiarize yourself with Stata
• Load a dataset
• Use Stata commands to analyze data and fill in
  the blanks

38
Next week

• Do files, log files, and workflow in Stata
• Find a dataset!