STAT 104: Section 3 21 Feb, 2008 TF: Daniel Moon.

STAT 104: Section 3

21 Feb, 2008

TF: Daniel Moon

TF

Office hour: Th. PM 1:00-2:00, 6th Floor, 601

Agenda of Today Review

Feedback from Hw #2 Examples from 1st Mid-term Exam

Principles of experimental design Control Randomization Replication

Improvements Blocking (Stratification) Placebo Effect

Linear Regression Assumptions

Residuals have en expected value 0. Residuals are uncorrelated. Residuals have the same variance. To check these assumptions:

Residual plot (residuals vs. X): constant variance

Normal Prob. Plot: Normality

Linear Regression Assumptions

<Residual Plot> <Normal Probability Plot>

<Source: Fall 2006 Midterm 1, Q4>

Feedback on HW #2 When writing regression equation, don't forget y_hat

, if you don't include error term.

(Right: "y = a + bx + error" or "y_hat = a + bx") R is unit-free Problem 4.e. What is r^2?  (Y_hat = aX + b)

One unit increase in X leads to "a" unit increase in Y.

(not clear answer: for every $1 of revenue, the value of team goes up $2.59)

(--> (It's better say) for every $1 revenue increase, ")

Examples (2007 Midterm 1)

Examples (2007 Midterm 1)

Examples (2007 Midterm 1)

Review Feedback from Hw #2 Examples from 1st Mid-term Exam

Principles of experimental design Control Randomization Replication

Improvements Blocking (Stratification) Placebo Effect

Sources of Data

Anecdotal Information Available Data Observational Studies Controlled Experiments Randomized Controlled Experiments

Elements of Designing an Experiment

Subjects to be tested Observational or Randomized

Experimental? How many individuals? How will the individuals selected? What kind of variables measured?

Principles of experimental design

Control

Randomization

Randomization Example

Review Feedback from Hw #2 Examples from 1st Mid-term Exam

Principles of experimental design Control Randomization Replication

Improvements Blocking (Stratification) Placebo Effect

Blocking

Blocking Example

Examples (2007 Midterm 1)

Physical Activity Breast Cancer

Examples (2007 Midterm 1)

Observational Study

Lurking Variables

Other good lifestyle habits

Examples (2007 Midterm 1)

Examples (2007 Midterm 1)

Factors that we should consider

Sample Size Individuals Experimental Design Individuals Selected Response Variable

Examples (2006 Midterm 1)

Breakfast Performance

Examples (2006 Midterm 1)

Examples (2006 Midterm 1)

Examples (2006 Midterm 1)