Sidebar Menu
Main Menu EN
Statistic Test
Guidelines for Descriptive Statistics Test
- Details
- Written by Ade Setiawan
An important step in determining the appropriate statistical test procedure is to know the type of variable we are observing. What is the independent variable and the dependent variable? After the variable is determined, then determine how to measure the variable (variable measurement scale)? If we already know/determine this, it will be easy for you to determine the appropriate statistical test procedure.
Guidelines for Predictive/Regression Statistical Test
- Details
- Written by Ade Setiawan
Predictive Statistics Test
An important step in determining the appropriate statistical test procedure is to know the type of variable we are observing. What is the independent variable and the dependent variable? After the variable is determined, then determine how to measure the variable (variable measurement scale)? If we already know/determine this, it will be easy for you to determine the appropriate statistical test procedure.
Guidelines for Association/Relationship Statistics Test
- Details
- Written by Ade Setiawan
Measuring Relationships (Association and Correlation)
An important step in determining the appropriate statistical test procedure is to know the type of variable we are observing. What is the independent variable and the dependent variable? After the variable is determined, then determine how to measure the variable (variable measurement scale)? If we already know/determine this, it will be easy for you to determine the appropriate statistical test procedure.
Guidelines for Comparative Statistics Test
- Details
- Written by Ade Setiawan
An important step in determining the appropriate statistical test procedure is to know the type of variable we are observing. What is the independent variable and the dependent variable? After the variable is determined, then determine how to measure the variable (variable measurement scale)? If we already know/determine this, it will be easy for you to determine the appropriate statistical test procedure.
How many sample groups (populations) will be compared?
One Sample Group (Population)
What is the Variable Measurement Scale?
How many Categories?
- 2 categories
- Binomial Test
- purpose
- >= 3 categories
- Uji Chi-Square goodness of fit
- Statistical Test: → 1 sample sign test (1 sample Sign Test)
- Statistical Measurement: → Median
- Note: The Wilcoxon one-sample test measurement was rejected because the distribution is not symmetric
What is the shape of the distribution curve?
- Normal
- Population variance is known:
- One sample z test
- Statistical measurement: Mean
- Population variance unknown:
- One sample t test
- Statistical measurement: Mean
- Population variance is known:
- Symmetrical
- Statistical Test: 1 sample Wilcoxon Test
- Statistical measurement: Median
- Note: The 1-sample t-test cannot be carried out because the distribution is not normal
- Other
- Test mark 1 sample
- Statistical measurement: Median
- Note: The Wilcoxon one-sample test measurement was rejected because the distribution is not symmetric
- Don't know
- View from the histogram: compare the normal distribution line with the histogram
- Formal test: Saphiro-Wilk Normality Test. If the probability of p < 0.05, it means that the sample is not normally distributed
Two Sample Groups (Population)
2 samples are independent (independent)
=> What is the measurement scale?
- 2 categories
- F Uji test
- Statistical measurement: Test the difference in proportion or odds ratio
- >=3 categories
- Uji Chi Square
=> What is the shape of the distribution?
- Both are identical
- Statistical Test: Man Whitney Test
- Statistical measure: Median Hodges-Lehman . Difference
- Note: The t-test cannot be carried out because the distribution is not normal
- Different shape
- Statistical Test: Mood's Median Test
- Statistical measurement: Median Difference
- Note: The Man Whitney test cannot be carried out because the distributions of the two are not identical
- Don't know
- Visual: Box-Whisker Plot to see the shape of the distribution
- Formal test: Saphiro-Wilk Normality Test. If the probability of p < 0.05, it means that the sample is not normally distributed
- Both Normal :=> View SD
- The two population variances are known
- Statistical Test: z . Test
- Statistical Measurement: Mean
- Homogeneous variety
- Statistical test: t test for independent samples with homogeneous variance (equal variance)
- Statistical Measurement: Mean
- Heterogeneous Variety
- Statistical Test: Welch t test for independent samples with heterogeneous variance (unequal variance)
- Statistical Measurement: Mean
- Don't know
- Formal Test:
- F test: Sensitive to samples whose distribution is not normal
- Forsythe-Brown test: tolerance to abnormal distribution
- If the probability of p < 0.10, it means that the variance of the two samples is homogeneous
- Formal Test:
- The two population variances are known
- The shape of the distribution is identical
- Statistical Test: Man Whitney Test
- Statistical measure: Median Hodges-Lehman . Difference
- Note: The t-test cannot be carried out because the distribution is not normal
- Different shape
- Statistical Test: Mood's Median Test
- Statistical measurement: Median Difference
- Note: The Man Whitney test cannot be carried out because the distributions of the two are not identical
- Don't know
- Visual: Box-Whisker Plot to see the shape of the distribution
- Formal test: Saphiro-Wilk Normality Test. If the probability of p < 0.05, it means that the sample is not normally distributed
2 paired samples: →Measurement scale?
- 2 categories
- Statistical Test: → McNemar Test
- Statistical measurement: → Test the difference in proportion or odds ratio
- >= 3 categories
- No suitable test ??
- Statistical Test: → Sign test
- Statistical measurement: → Median Difference
- Note: → The Wilcoxon sign test cannot be carried out because the distribution is not symmetric
- Normal
- Population Variety Known
- Test Statistics: → Test z
- Statistical Measurement: → Mean
- Population variance unknown
- Statistical Test: → T test for two paired samples
- Statistical Measurement: → Mean
- Population Variety Known
- Symmetrical
- Statistical Test: → Wilcoxon Signed Rank Test
- Statistical measures: → Median Hodges-Lehman Difference
- Note: → The t-test cannot be carried out because the distribution is not normal
- Other
- Statistical Test: → Sign test
- Statistical measurement: → Median Difference
- Note: → The Wilcoxon sign test cannot be carried out because the distribution is not symmetric
- Don't know
- View from the histogram: → compare the normal distribution line with the histogram
- Formal test: → Saphiro-Wilk Normality Test. If the probability of p < 0.05, it means that the sample is not normally distributed
Paired observations on one subject
- 2 categories
- Statistical Test: → McNemar Test
- Statistical measurement: → Test the difference in proportion or odds ratio
- >= 3 categories
- No suitable test ??
- Statistical Test: → Sign test
- Statistical measurement: → Median Difference
- Note: → The Wilcoxon sign test cannot be carried out because the distribution is not symmetric
- Normal
- Population Variety Known
- Test Statistics: → Test z
- Statistical Measurement: → Mean
- Population variance unknown
- Statistical Test: → T test for two paired samples
- Statistical Measurement: → Mean
- Population Variety Known
- Symmetrical
- Statistical Test: → Wilcoxon Signed Rank Test
- Statistical measures: → Median Hodges-Lehman Difference
- Note: → The t-test cannot be carried out because the distribution is not normal
- Other
- Statistical Test: → Sign test
- Statistical measurement: → Median Difference
- Note: → The Wilcoxon sign test cannot be carried out because the distribution is not symmetric
- Don't know
- View from the histogram: → compare the normal distribution line with the histogram
- Formal test: → Saphiro-Wilk Normality Test. If the probability of p < 0.05, it means that the sample is not normally distributed
Three or more Sample Groups (Population)
Free samples
- 2 categories
- Statistical Test: → Chi-Square Test
- >= 3 categories
- Statistical Test: → Chi-Square Test
- The shape of the distribution is identical
- Statistical Test: → Kruskal-Wallis Test
- Note: → Anova test cannot be carried out because the distribution is not normal
- The shape of the distribution is different
- Statistic Test: → Medial Mood's . Test
- Note: → The Man Whitney test cannot be carried out because the distributions of the two are not identical
- Normal
- Homogeneous Population Variety
- Statistical Test: → One Way Anova Test
- Statistical Measurement: → Difference Mean
- heterogeneous population variety
- Statistic Test: → 1-Way Between Welch Anova
- Statistical Measurement: → Difference Mean
- Don't know
- Formal Test:
- Bartlett Test: → Sensitive to Normality
- Forsythe-Brown test: → tolerance to abnormal distribution
- If the probability of p < 0.10, it means that the variance of the two samples is homogeneous
- Formal Test:
- Homogeneous Population Variety
- The shape of the distribution is identical
- Statistical Test: → Kruskal-Wallis
- Note: → Anova test cannot be carried out because the distribution is not normal
- The shape of the distribution is different
- Statistic Test: → Medial Mood's . Test
- Statistical Measurement: → Difference Mean
- Note: → The Man Whitney test cannot be carried out because the distributions of the two are not identical
Paired sample
- 2 categories
- Statistical Test: → Cochran Test
- >= 3 categories
- No suitable test ??
- Statistical Test: → Friedman Test
- Note: → Anova test cannot be carried out because the distribution is not normal
- Normal
- Test Statistics: → One Way Repeat Measure Anova
- Statistical Measurement: → Difference Mean
- Other
- Statistical Test: → Friedman Test
- Note: → Anova test cannot be carried out because the distribution is not normal
Observations / repeated measurements (repeated measure)
- 2 categories
- Statistical Test: → Cochran Test
- >= 3 categories
- No suitable test ??
- Statistical Test: → Friedman Test
- Note: → Anova test cannot be carried out because the distribution is not normal
- Normal
- Test Statistics: → One Way Repeat Measure Anova
- Statistical Measurement: → Difference Mean
- Other
- Statistical Test: → Friedman Test
- Note: → Anova test cannot be carried out because the distribution is not normal
Search
Download Smart Excel Add-in dan Sample Data
