Scatterplots, Association and Relationships - PowerPoint PPT Presentation

1 / 18
About This Presentation
Title:

Scatterplots, Association and Relationships

Description:

Shows the relationship between two variables measured on ... Lurking Variables. Lurking Variables are Hidden Variables other than the explanatory and response ... – PowerPoint PPT presentation

Number of Views:58
Avg rating:3.0/5.0
Slides: 19
Provided by: harrison7
Category:

less

Transcript and Presenter's Notes

Title: Scatterplots, Association and Relationships


1
Scatterplots, Association and Relationships
  • Chapter 7

2
Variables
  • Response Variable measures the outcome of a
    study (y variable, dependent variable, the
    result)
  • Explanatory of Predictor Variable attempts to
    explain the observed outcomes (x variable,
    independent variable, the cause)

3
Scatterplot
  • Shows the relationship between two variables
    measured on the same individual.
  • Graph the explanatory variable on the horizontal
    axis. (x list)
  • Graph the response variable on the vertical axis
    (y list)

4
Interpreting Scatterplots
  • Overall pattern
  • Direction (increasing or decreasing ?)
  • Form (linear, exponential?)
  • Strength of relationship (narrowness of curve)
  • The scatter or width of the hallway
  • Outliers
  • What does not fit the pattern
  • Falls outside the usual values or either variable

5
Direction (Think Slope)
  • Two variables are positively associated when the
    above average values of one are associated with
    the above average values of the other.
  • Two variables are negatively associated when the
    above average values of one are associated with
    the below average values of the other

6
Form
  • Form is the general shape of the dots in the
    scatterplot
  • Linear, exponential, logarithmic, . . .
  • Correlation is ONLY relevant with linear data.
  • Curved data must be straightened before we
    can use correlation

7
Strength
  • How much is the data lined up.
  • The closer to a straight line the stronger the
    relationship
  • Correlation is a measurement of the strength of
    the relationship between predictor and response
    variables

8
Correlation
  • How strong is the relationship

9
Calculation of Correlation(The Theory, thank TI
again)
  • Calculate the means and standard deviations of
    both the xs and ys
  • Standardize both the x value and y value for each
    data point (calculating z-scores)
  • Multiply the z values of each point
  • Add up all of the products of the zs
  • Divide the grand total by n 1 where n is the
    number of points

10
Calculation Example
11
Correlation
  • The average of the product of standardized xs
    and ys

12
Another way to calculate correlation coefficients
?
13
Correlation Facts
  • It makes no difference which variable is the x
    and which is the y
  • Positive r indicates a positive association
    between the variables and negative r indicates a
    negative association
  • -1 r 1 Values near zero indicate a weak
    association. Values near 1 or -1 indicate strong
    association

14
Correlation facts 2
  • There are no units on r so it is immune to
    changes when units change.
  • Correlation measures the strength of ONLY linear
    relationships. Correlations do not describe
    curved relationships
  • Correlation is greatly affected by extreme values

15
Correlation Facts 3
  • Outliers can make a strong relationship look weak
  • Outliers can make a weak relationship look strong
  • Report the correlation both with and without the
    outlier(s)

16
Conditions for Correlation
  • The quantitative condition Both variables must
    be quantitative.
  • The linear condition The form of the
    relationship must be linear.
  • The outlier condition Outliers greatly affect
    correlation. Report the r value both with and
    without the outliers factored in
  • Correlation Demo

17
Some things to watch for
  • Dont say Correlation when you mean
    Association
  • Association is a vague term used when describing
    the relationship between two variables
  • Correlation is a very precise term describing a
    linear relationship between quantitative
    variables
  • Beware of Outliers
  • Dont confuse correlation with causation
  • The explanatory variable does not cause the
    response to change

18
More to watch for
  • Lurking Variables
  • Lurking Variables are Hidden Variables other than
    the explanatory and response variables that we do
    not see on the scatterplot.
Write a Comment
User Comments (0)
About PowerShow.com