Transcript view presentation online

CHAPTER 12
Becoming
Acquainted With
Statistical Concepts
Why We Need Statistics
• Statistics is an objective way of interpreting a
collection of observations.
• Types of statistics
- Descriptive techniques
- Correlational techniques
- Differences among groups
Univariate and multivariate
How Computers Are Used
in Statistics
• Frequently used in offices, labs, and homes for
statistical analysis
• Types of software for statistics
- Biomedical Series (BIMED)
- Statistical Analysis System (SAS)
- Statistical Package for the Social Sciences
(SPSS)
Measures of Central Tendency
and Variability
• Central tendency scores
- Mean: Average
- Median: Midpoint
- Mode: Most frequent
• Variability scores
- Standard deviation
- Range of scores
Categories of Statistical Tests
• Parametric
- Normal distribution
- Equal variances
- Independent observations
• Nonparametric (distribution free)
- Distribution is not normal
• Normal curve
- Skewness
- Kurtosis
Normal Curve
Skewness
Kurtosis
Statistics
• What statistical techniques tell us
- Reliability (significance) of effect
- Strength of the relationship (meaningfulness)
• Types of statistical techniques
- Relationships among variables
- Differences among groups
Interpreting Statistical Findings
• Probability
- Alpha: false positive (type I error)
• Typical: p < .05 or p < .01
- Beta: false negative (type II error)
• Meaningfulness (effect size)
• Power: Probability of rejecting the null hypothesis
when it is false
Truth Table for the Null Hypothesis
H0 true
H0 false
Accept
Correct decision
Type II error (beta)
Reject
Type I error (alpha)
Correct decision
Alpha & Beta
• Alpha = p-level in statistical tests
• 1 - Beta = the power of the statistical test
Ways to  Statistical Power
•  alpha (often preset to .05 or .01)
•  beta (often preset to .20)
• N
Statistical Power and Effect Size
• Effect size is invariant
• Overpower = greater N than needed to statistically
detect the effect (detect trivial effects)
• Underpower = not enough N to statistically detect
the effect (can’t detect meaningful effects)
• Appropriate statistical power is achieved from an a
priori power analysis
Power Analysis
• Effect size = statistical power
• With the info of effect size, alpha, and beta,
power analysis can tell us what N we need
for the study
• Tables, computer programs, and math
equations