Uploaded by Muluken Ayenew

13 chapter 12

advertisement
Chapter 12
Quantitative Approaches
© 2017 Taylor & Francis
Which Test to Use?
You must identify:
1. The number of variables in the analysis
2. The level of measurement of the
variables
3. Whether variables are independent or
dependent
4. Whether the relationship is assumed to
be linear
© 2017 Taylor & Francis
Hypothesis Testing
Two basic elements:
1. The test is based on the ratio of percentage of
explained variance to the percentage of
unexplained variance
•
•
Variance is the basis of all analysis
The variance explained is the variance in the dependent
variable (explained by the independent variable)
2. The relationship tested is generally a linear
relationship
© 2017 Taylor & Francis
Two Basic Approaches to Analysis
• Group Differences
• Group differences use analysis of variance (ANOVA)
• Variable Relationships
• Within group variable relationships use correlation and
regression
© 2017 Taylor & Francis
Translate the Measures
Prepare the data set for a software program
• A rectangular matrix presents variables in
columns and subjects in rows
• A column represents scores on a particular
variable
• A record may refer to a single line of data
• A case (often the same as a record) is the set of
data that represents a single subject
© 2017 Taylor & Francis
Group Difference Tests
Generally, group differences have:
• Independent variables with a nominal or ordinal
level of measurement, and
• A dependent variable measured at the interval or
ratio level
© 2017 Taylor & Francis
Group Difference Tests
•
•
•
•
T-test
One way ANOVA
ANOVA
MANOVA
© 2017 Taylor & Francis
T-test
• The simplest of differences of means tests
• This difference of means test is an ANOVA
(analysis of variance) with a single independent
variable with only two levels and a single
dependent variable; thus, there are only two
groups to compare
• T to test explained variance
© 2017 Taylor & Francis
One Way ANOVA
• A one way ANOVA (analysis of variance) can be
run wherever a difference of means t-test can be
run
• This difference of means test has a single
independent variable with two or more levels and
a single dependent variable; thus, it can compare
two or more groups
• F to test explained variance
© 2017 Taylor & Francis
ANOVA
• The general form of ANOVA; any t-test or one
way ANOVA can be run as a general ANOVA
• This difference of means test can have more than
one independent variable with two or more levels
and a single dependent variable
• F to test explained variance
© 2017 Taylor & Francis
MANOVA
• Multivariate Analysis of Variance
• An ANOVA that allows more than one dependent
variable
• Beyond the scope of this text
© 2017 Taylor & Francis
Variable Relationship Tests
• Variables need to be interval or ratio level of
measurement
•
•
•
•
Correlation
Simple correlation
Multiple correlation/regression
Canonical correlation
© 2017 Taylor & Francis
Correlation
• The most common associative measure, the
Pearson Product Moment Correlation, is a
description of the linear relationship between two
variables
• No assumption that the two variables are
independent or dependent
© 2017 Taylor & Francis
Simple Correlation/Regression
• Correlation with two variables; one designated
independent, the other dependent
• b, β, a, 𝝰
© 2017 Taylor & Francis
Multiple Correlation/Regression
• Multiple regression is probably the most used
statistical procedure in the world
• More than one independent variable and a single
dependent variable
• R
© 2017 Taylor & Francis
Canonical Correlation
• Analogous to MANOVA, canonical correlation is
multiple regression with multiple independent
variables and more than one dependent variables
• Beyond the scope of this text
© 2017 Taylor & Francis
Descriptive Statistics
Measures of central tendency:
• Mean: the arithmetic average
• Median: the middle score of the group (ordinal)
• Mode: the most frequent score
• The definition of a normal distribution:
mean = median = mode
© 2017 Taylor & Francis
Descriptive Statistics
Measures of dispersion:
• Range: calculated by taking the difference
between the highest number in the group and the
lowest
• Standard deviation: the average deviation a set
of scores are from their mean
© 2017 Taylor & Francis
Statistical Difference
• Do the means likely come from different groups?
• Is there a real (statistical) difference between the
groups being compared?
• We need a hypothesis and a test
© 2017 Taylor & Francis
Download