Data Analysis-
1: Designing Experiments/Collecting Data
Statistical Questions - Before designing an experiment, be sure that a question has a variable answer. Example, "How old am I?" versus "How old are students in my school?"
Types of Variability (definitions)
Measurement:
Natural:
Induced:
Sampling:
We will mostly use sampling.
Population vs. Random Sample
Definitions:
In the picture the can says "Mixed Nuts: Less than 50% are peanuts" and Billy says "I'm making sure less than half of these nuts are peanuts."
Question 1: Is Billy taking a random sampling?
Question 2: If the can is not less than half peanuts, is the manufacturer in trouble for fraud in advertising?
Designing a Question: You want to explore the topic of the war in Iraq. Write three different questions.
1. Write a question that obviously biases the person answering toward being pro-war.
2. Write a question that subtly biases the person answering toward being anti-war.
3. Write a fair unbiased question.
6. A student asks if the precision with which manufacturers must calibrate their tools is at all related to statistics. How do you respond?
7. Mariah read that deaths due to stampedes occurred primarily in non-U.S. countries at rock concerts and argued with her parents that she would be perfectly safe at the running of the bulls in Pamplona, Spain. If you were her parents, how would you reason with her?
More examples
Example #1
From MEDCity News (http://medcitynews.com/2012/03/9-out-of-10-doctors-wouldnt-recommend-acareer-in-healthcare/)
- 9 out of 10 doctors wouldn't recommend a career in healthcare
- The results come from a 5,000-physician survey from The Doctors Company, a medical liabilityinsurer for doctors.
- Sixty percent of respondents said health reform would have a negative impact on patient care,
while just 20 percent said it would have a positive impact.
Is this a good sampling? What isn't known that you wish was known?
Example #2
The New York Times (http://www.nytimes.com/interactive/2013/02/04/science/girls-lead-in-scienceexam-but-not-in-the-united-states.html?smid=tw-share)
Is this a good sampling? What isn't known that you wish was known?
2: Displaying Data
Continuous vs. Discrete Data
Continuous Data is:
Discrete Data is:
More Ways to Represent Data
Line Graphs
1. What does the line graph show us?
2. How frequently were measurements taken?
3. Why does it make sense to connect the dots with lines?
4. If you see a graph with a connected line like this, what does it imply about the data?
Scatterplots-
1. What information does the scatterplot show us?
2. Why does it not make sense to connect the dots?
3. Even if we don't connect the dots, can we draw a conclusion from the picture?
Positive, Negative, and no Association
Given a scatterplot the points either have positive association (a general positive slope), negative association (a general negative slope), or no association (can't even guess at a trend line).
Multiple Line Graphs-
1. What information does the line graph show us?
2. Why is it useful to have the multiple line graphs in one picture?
3. Why does it make sense to connect the dots like this?
How the heck do we know when to use what?
Bar graph
Histograms
Stem and leaf plot
Scatterplot
Line graph
Circle graph