Statistics Through Applications

First Edition

Chapter 1:

How Do We Get “Good” Data?

Copyright © 2004 by W. H. Freeman & Company

Daniel S. Yates, Daren S. Starnes, David S. Moore

​

www.whfreeman.com/sta

1.2 Measuring Data (Day 1)

• Statistics is all about numbers
• Planning a survey or experiment does not by itself produce numbers.
• We must measure whatever characteristics interest us!

But what about the Patients?

• Clinical trials tend to measure characteristics that are easy to put a number on:
• Blood pressure, tumor size, virus concentration in the blood
• They don’t often measure what matters most to patients:
• Does the treatment improve my quality of life?
• Why do you think this is so???

• Read Example 1.11 together
• Identify the instrument, units, variable

• For any Statistical study, ask yourself:
• 1. How is the variable defined?
• 2. Is the variable a valid way to describe the property it claims to measure?
• 3. How accurate are the measurements?

​

​

Know your variables!

• Measuring length is easy, but…
• To measure something harder like “highway deaths” requires you to define what counts:
• Pedestrians hit by cars?
• People in cars hit by a train at a crossing?
• People who die from injuries 6 months after an accident?

Know your variables!

• Measuring length is easy, but…
• To measure something harder like “highway deaths” requires you to define what counts:
• Pedestrians hit by cars?
• People in cars hit by a train at a crossing?
• People who die from injuries 6 months after an accident?
• Gov’t definition: must die within 30 days of an accident to count as a traffic death.
• Valid? Accurate? What do you think?

Let’s talk Unemployment!

Unemployment rate

People are unemployed if they are available and looking for work.

• If you want to work, but are too discouraged to keep looking for a job, you don’t count as unemployed!

Unemployment rate

People are unemployed if they are available and looking for work.

• If you want to work, but are too discouraged to keep looking for a job, you don’t count as unemployed!
• What are different ways that could cause the unemployment rate to go down?

Valid or not:

• Measure person’s height with a tape measure.

Valid or not:

• Measure person’s height with a tape measure.
• Measure college readiness by ACT score.

Valid or not:

• Measure person’s height with a tape measure.
• Measure college readiness by ACT score.
• Measure college readiness by height of student.

Valid or not:

• Measure person’s height with a tape measure.
• Measure college readiness by ACT score.
• Measure college readiness by height of student.
• Measure college readiness by grade point average.

Measuring highway safety

• 1991 → 41,508 deaths
• 2000 → 41,945 deaths
• Did the highways become more dangerous?

Measuring highway safety

• 1991 → 41,508 deaths
• 2000 → 41,945 deaths
• Did the highways become more dangerous?

​

• 1991 → 169 million drivers drove 2172 billion miles
• 2000 → 191 million drivers drove 2750 billion miles
• Did the highways become more dangerous? Justify!

Validity, take 2

Is the SAT a valid measure of “college readiness?”

• Does it consider inborn intelligence, learned knowledge, study and test-taking skills, motivation?
• Is “college readiness” easily defined?

​

​

​

​

​

Validity, take 2

Is the SAT a valid measure of “college readiness?”

• Does it consider inborn intelligence, learned knowledge, study and test-taking skills, motivation?
• Is “college readiness” easily defined? – NO!
• Would a better question be…
• “Do SAT scores help predict students’ success in college?”
• Success is a clear concept – Perhaps.

​

​

​

​

Therefore, SAT scores have predictive validity as measure of readiness for college.

Assignment 1.2 (Day 1)

• Read pg. 24-34
• Do:
• pg. 27:1.25-1.27
• pg. 33:1.29-1.33