RESEARCH STATISTICAL TESTS Flashcards
Aka two-sample t-test. It is a statistical test that determines if there is a significant difference between the means of two independent or unpaired groups. Its variable is continuous.
Independent Samples T-Test
Groups in which cases or participants in each group are different.
Unpaired/unrelated groups
Requirement and Assumption for Independent Samples T-Test
- Random sample data from population
2 Two groups are unrelated - Dependent variable must be continuous and normally distributed in each group
- No significant outliers
- Variance of dependent variable across groups should be equal. (Homegeneity of variance)
Methods in Checking Outliers
- IQR Method
2. Boxplot
IQR Formula
Q3-Q1
Lower-bound formula
Q1-(1.5 x IQR)
Upper-bound formula
Q3+(1.5 x IQR)
True or False. Any points lesser than lower bound and higher than upper bound are outliers.
True
True or False. If outliers exist, we must leave them in.
False. Delete the outliers.
Methods for Checking Normality
- Distribution Plots
- Q-Q Plots
- Shapiro-Wilk test
True or False. For Q-Q Plots, if all points are close to the diagonal reference line, then it is normally distributed.
True
For Q-Q Plots, if points sag above or below the diagonal reference line, then there is a problem with the ________.
kurtosis
If points snake around the diagonal reference line, then there is a problem with the ________.
skewness
It is used to test null hypothesis that variances in different groups are not equal.
Levene’s Test
In Levene’s Test, the variances are heterogenous if ________.
p-value is less than or equal to alpha
In Levene’s Test, the variances are homogenous if ________.
p-value is more than alpha
In Levene’s Test, if variances are heterogenous then we ________.
Reject the null hypothesis
In Levene’s Test, if variances are homogenous then we ________.
Fail to reject null hypothesis
Used to compare differences between two independent groups when the dependent variable is ordinal or numeric, but not normally distributed.
Mann-Whitney U Test
Assumptions of Mann-Whitney U Test
- Two groups should be independent
2. Dependent variable must be numeric or ordinal and is not normally distributed
This test is used to compare means that are from the same related units. The 2 groups typically represent 2 different times. It aims to determine whether or not the mean difference between paired observations is significantly different from zero.
Paired Samples T-Test
Assumptions of Paired Samples T-Test
- Random sample data from population
- Two groups should be dependent or paired
- Dependent variable must be continuous.
- No outliers in either of the 2 groups.
- The difference of observations of the paired groups should be normally distributed.
In Paired Samples T-Test, “the difference between group 1 and 2 is equal to zero” represents the ________.
null hypothesis
In Paired Samples T-Test, “the difference between group 1 and 2 is NOT equal to zero” represents the ________.
alternative hypothesis
This is the nonparametric alternative test for the Paired Samples T-test. It is used to compare 2 sets of variables that are of either ordinal level of measurement or numeric that are from paired observations, usually, from the same participants.
Wilcoxon Signed Rank Test
Assumptions of Wilcoxon Signed Rank Test
- Two groups should be dependent or paired
2. Dependent variable must be numeric or ordinal
Used to compare the means of 2 or more independent groups in order to determine whether there is a statistical evidence that the associated population means are significantly different.
Analysis of Variances (ANOVA)
Requirements and Assumptions of Analysis of Variances (ANOVA)
- Independent variable should be categorical.
- Dependent variable must be numeric and normally distributed.
- The variances of the dependent variable across groups should be equal. (Homogeneity of Variances)
Used if in ANOVA, the null hypothesis is rejected.
Post-Hoc Tests
Three common Post-Hoc Tests:
- Least Significant Difference (LSD)
- Tukey’s Honestly Significant Difference (HSD) Test
- Scheffe’s Test
Used to compare the means of 2 or more independent groups in order to determine whether there is a statistical evidence that the associated population means are significantly different. And if normality is not satisfied.
Kruskal-Wallis Test
If the test assumes normality then it is _______.
Parametric
If the test do not assume normality then it is _______.
Nonparametric
This Post-Hoc test is good for exactly 3 groups.
Least Significant Difference (LSD)
A Post-Hoc test that is good when Homogeneity of Variances is satisfied.
Tukey’s Honestly Significant Difference (HSD) Test
This Post-Hoc test is good when sample sizes are different.
Scheffe’s Test
