SI 544Introductory

Statistics andData Analysis

Lada Adamic SI 544 Introductory Statistics and DataAnalysis

(Preliminaries)

Lada Adamic

School of Information,University of Michigan

Jan. 3rd, 2007

SI 544Introductory

Statistics andData Analysis

Lada Adamic

Outline

SI 544Introductory

Statistics andData Analysis

Lada Adamic

motivation

Dick DeVaux:"We haven’t evolved to be statisticians. Our students whothink statistics is an unnatural subject are right. This isn’thow humans think naturally. But it is how humans thinkrationally. And it is how scientists think. This is the way wemust think if we are to make progress in understanding howthe world works and, for that matter, how we ourselveswork."

SI 544Introductory

Statistics andData Analysis

Lada Adamic

motivation

Gary King (Department of Government, Harvard University):

"Statisticians will rule the world."

(When discussing the opportunities that availability of massivedata sets will present for addressing questions in social science.)

SI 544Introductory

Statistics andData Analysis

Lada Adamic

how I see things

There is lots of interesting dataWe need to describe what is going onFor this we need descriptive statistics

need to summarizeneed to visualize

We need to tell whether what we are seeing is anactual trend, or part of the random “noise”To this end we need to

Understand probabilityUnderstand probability distributions (what is thelikelihood that this would occur by chance?)

SI 544Introductory

Statistics andData Analysis

Lada Adamic

example (from Dick DeVaux)

A town has two hospitalsLarge hospital, about 100 babies a daySmaller hospital, about 15 babies a day

Over the course of the year, which hospital (if either)would probably have more days in which more than60% of the babies born are male?

SI 544Introductory

Statistics andData Analysis

Lada Adamic

how I see things (continued)

In order to understand what is going on, we need amodelFor example, we want to figure out whether watchingviolent movies is correlated with violent behaviorIt is standard to attempt to reject the null hypothesisthat there is no correlation between the variablesConsideration

data sampleexperimental designconfounding variableschoice of method

SI 544Introductory

Statistics andData Analysis

Lada Adamic

what we’ll cover

ProbabilityDescriptive statisticsInferential Statistics

Sampling distributions: confidence intervals, hypothesistests and p-valuesEstimating population meanComparing population meansAnalysis of varianceUnivariate and multivariate OLS RegressionAnalysis of categorical dataData CollectionExperimental design

SI 544Introductory

Statistics andData Analysis

Lada Adamic

Outline

SI 544Introductory

Statistics andData Analysis

Lada Adamic

basics

This is SI 544I’m Lada Adamic (ladamic@umich.edu)Please respond to survey on office hours:http://doodle.ch/participation.html?pollId=29v979qxx4tdvmr5Make sure you have access to the cTools siteSome materials are available athttp://www-personal.umich.edu/ ladamic/courses/si544w08/

SI 544Introductory

Statistics andData Analysis

Lada Adamic

meetings

every Thursday in West Hall 409every Tuesday in the DIAD (4th floor of Shapiro library)

work on your own laptopwork on lab PC or Mac

SI 544Introductory

Statistics andData Analysis

Lada Adamic

textbooks

Introductory Statistics for the Behavioral Sciences (5thor 6th Edition) by Welkowitz, Ewen, and Cohen

orJohn Verzani: Using R for Introductory Statistics

SI 544Introductory

Statistics andData Analysis

Lada Adamic

R

We will be using Ropen source http://www.r-project.org/

many additional modules available:cran.r-project.org/

on cTools under Resources you can find several niceonline tutorialssteeper learning curve than most other statisticalpackages, but · · ·

free −− so you can use it whatever your job may beprogrammable −− you can create your own modulesyou will be able to switch to other software relativelyeasily

SI 544Introductory

Statistics andData Analysis

Lada Adamic

grading scheme

20% midterm (in class)25% final (take home)25% problem sets

I will drop your lowest problem set score.

20% group project (small)5% news evaluation5% participation

attendancespeaking up in classposting to cTools

SI 544Introductory

Statistics andData Analysis

Lada Adamic

problem sets

turn in only in PDF formatturn in only on cToolslate assignments

can turn in up to 2 days late with a 10% penalityemail me for extensions for legit reasons (medical &family emergencies)include a note about granted extension on cTools whensubmittinggrader Laurel Shipley: email me if there are anyquestions about grading

you are encouraged to collaborate with yourclassmates, but turn in your own workrun your own R code

SI 544Introductory

Statistics andData Analysis

Lada Adamic

cTools participation

You can earn participation points by posting content tocTools

post clarification questions to cToolsThere is no point to being stuck on some syntax in R forhours on end. Just ask for help on the cTools forum.answer others’ questionspost interesting links

SI 544Introductory

Statistics andData Analysis

Lada Adamic

group project

The project is not a major project.At the end you will turn in 3 pages.

form groups of 3-4 peopletime lineJan. 24 form group and select topic (2 pts)March 4th project progress report (3 pts)April 8 project report (10 pts)April 10 presentations (5 pts)

examples of topics from last yearhistory of violent arrests and NBA performancetemperature and beer consumptionnumber of books and ratings on LibraryThingpacing and rowing performance

SI 544Introductory

Statistics andData Analysis

Lada Adamic

news review

5% of your final gradefind news article and critically compare to original study

SI 544Introductory

Statistics andData Analysis

Lada Adamic

exemptions for HCI and IAR stats requirement

http://www-personal.umich.edu/~ladamic/courses/si544f06/statswaiver.html

SI 544Introductory

Statistics andData Analysis

Lada Adamic

Courses you can take instead

Stat. 350 - Introduction to statistics and data analysis(undergrad, cannot count as cognate)Stat. 400 - Applied Statistical MethodsStat. 500 - Applied Statistical Methods (note, has 350as a prerequisite)Biostatistics 510: http://www-personal.umich.edu/~kwelch/510/biostat510.htm

Biostatistics 503:http://www.sph.umich.edu/iscr/caid/display_course.cfm?CourseID=BIOSTAT503

Biostatistics 553http://www.sph.umich.edu/iscr/caid/display_course.cfm?courseID=BIOSTAT553

Sociology 510: statisticsOMS 501: Applied Business Statistics

SI 544Introductory

Statistics andData Analysis

Lada Adamic

Why you should still take this course

neat datasets relevant to HCI & IARwe focus on the relevant skills and critical thinkingwe go easy on the math

no calculus requiredsome algebra helpful

SI 544Introductory

Statistics andData Analysis

Lada Adamic

Outline

SI 544Introductory

Statistics andData Analysis

Lada Adamic

data types

QuantitativeDiscreteContinuous

QualitativeNominal (categorical)Ordinal (rank ordered categories)

SI 544Introductory

Statistics andData Analysis

Lada Adamic

exercise: name the data type

bacteria countoccupations of shoppersUSNWR ranking of universitymarital statustime (in months) since last auto maintenancehandedness

SI 544Introductory

Statistics andData Analysis

Lada Adamic

Counties with highest rates of kidney cancer

SI 544Introductory

Statistics andData Analysis

Lada Adamic

Counties with lowest rates of kidney cancer

SI 544Introductory

Statistics andData Analysis

Lada Adamic

summary

class logistics (questions?)motivation for learning statisticsstarting to gather data!

Next time: R tutorial by Mick McQuaid