Chapter #6 Hypothesis Testing Flashcards
Hypothesis
The use of statistical procedures to answer research questions
Null Hypothesis
The assumption of no difference in a hypothesis statement
Two types of statistical procedures?
Parametric and Non-Parametric
Parametric
Data is assumed to come from a distrobution
Non-Parametric
Data is NOT assumed to come from a distrobution
Parametric tests are best for what data?
Interval and Ratio
Non-Parametric tests are best for what data?
Nominal and Ordinal. BUT some limited use in all four types: Nominal Ordinal Interval and Ratio.
ANOVA
Analysis of Variance
Goal: Determine if an IV has a significant affect on a DV or determine if different test conditions yield different results.
Difference between Significant and Not Significant
Significant: Data is due to testing conditions
Not Significant: Data is due to chance
statistical significance is claimed if
p < .05
NOT statistical significance is claimed if
F > 1.0 and p> .05, or F < 1.0 and p = ns
When do you do a posthoc test?
Where there are more than two conditions in a ANOVA
Degrees of Freedom, for ANOVA
n = # of test conditions
m = # of participants
Effect: n-1
Residual: (n-1)(m-1)
update: Residual apparently: (m – n)
Post Hoc Comparison Test
If F is significant then at least one test condition is significantly different from another. Hence Post hoc is used to determine which differ from one another significantly
Which Post Hoc do we use?
Scheffé Post Hoc
Two Way ANOVA
Experiment with two independent variables
Chi-square Test
Focuses on Nominal data
Compares relationships between categorical data or nominal data
Directly compares the observation data with expected data
Significant if X^2 is above critical value (There’s a chart based on alpha)
Usually alpha is .05
Degrees of Freedom for Chi-Square Test
df = (r – 1)(c – 1)
r = row
c = column
Test for Normality
Tests if the data follows a normal distribution
Test for Normality, ways to do it
Kolmogorov-Smirnov (K-S)
Lilliefors
Kolmogorov-Smirnov (K-S)
Test for normality that requires population mean and standard deviation
Lilliefors
Test for normality that requires sample mean and standard deviation
M (not mean) needs to exceed the VC(critical value) for it to be significant
Null hypothesis is rejected if
p< .05
Null hypothesis is NOT rejected if
p > .05
Non-parametric Tests for Ordinal Data, four type
Mann-Whitney U and Kruskal Wallis
Wilcoxon Signed-Rank and Friedman
Non-parametric Tests are for what kind of data?
Ordinal
Mann-Whitney U
Non-Parametric test
2 columns between subjects design
U, z, p
Wilcoxon Signed-Rank
Non-Parametric test
2 columns within subjects design
z, p
Kruskal Wallis
Non-Parametric test
3+ columns between subjects design
H, p
A Post hoc can also be done on this data
Friedman
Non-Parametric test
3+ columns within subjects design
H, p
A Post hoc can also be done on this data
Non-parametric tests can also be used on what?
multi-factor experiments and ratio-scale data