Becoming Acquainted With Statistical Concepts

1
PEP455 Part 2 Statistical Concepts

• Becoming Acquainted With Statistical Concepts

2
Second Half of PEP455

• We are going to spend the rest of the class
  discussing statistics
• Why do we need them?
• Descriptive techniques (what do the data look
  like?)
• Correlational techniques (how are two variables
  related?)
• Differences between groups
• Assessment of how well your tests function
• Not all statistics are done after a study is over
• How many subjects do you need to conduct a study?
• Preferably, there is a way to determine how many
  you need without guessing

3
Descriptive Statistics

• What do your data look like?
• Measures of central tendency what is the
  middle?
• Mean (average)
• Median (middle score)
• Mode (most frequently occurring)
• Measures of variability how spread out are the
  data?
• Range (high and low)
• Standard deviation
• Variance
• These statistics tell you some interesting
  things, but not everything you might want to know

4
Normal Curve
5
Skewness
6
Kurtosis
7
Correlational Techniques

• How are two variables related to each other?
• As one variable goes up
• Does another one go down? go up? not change?
• You can learn some very interesting things just
  by graphing a series of data
• Shape of the relationship can be helpful
• Correlational techniques can also provide you
  with quantitative information
• Correlation coefficients and coefficients of
  determination

8
Example of Correlation

• This is a really lousy graph why?

9
Differences between groups

• If you have two groups of scores, are they
  different?
• Because the means are the same, doesnt mean they
  are
• Have to take into account how variable they are
• Statistics actually deal in probabilities

10
How well do tests function?

• When administering tests, you cannot assume they
  work very well
• You need to be concerned with
• Reliability
• Validity
• Objectivity
• In reality, you do not rubber stamp a test, but
  understand there are varying degrees of these
  issues
• You use statistics to quantify the degree to
  these tests work

11
How do you pick your subjects?

• Statistical techniques play a role in designing
  studies
• One way to make a statistician mad is to not
  include him/her in study designs!
• Picking your subjects
• Random sampling
• Stratified random sampling
• Systematic sampling
• How many subjects do you need?
• Depends on how variable the data are
• Too few can ruin the results
• Too many is a waste of resources

12
Categories of Statistical Tests

• Parametric
• Normal distribution
• Equal variances
• Independent observations
• Nonparametric (distribution free)
• Distribution is not normal
• Normal curve
• Skewness
• Kurtosis