Memilih uji statistik mana yang akan digunakan ( Choosing which statistical test to use)

 Choosing the appropriate statistical test is crucial for obtaining accurate and meaningful results from your data analysis. The choice depends on various factors, including the study design, type of data, and the specific research question you're trying to answer. Here's a guide to help you select the right statistical test for common scenarios:

1. Nature of Data:

a. Parametric Data (Numeric, Normally Distributed):

1. One Sample:

   • Test: One-sample t-test
   • Use: Compare the mean of a sample to a known value.

2. Two Independent Samples:

   • Test: Independent samples t-test
   • Use: Compare means of two independent groups.

3. Two Related Samples:

   • Test: Paired samples t-test
   • Use: Compare means of two related groups (e.g., before and after treatment).

4. More than Two Independent Samples:

   • Test: Analysis of Variance (ANOVA)
   • Use: Compare means of multiple independent groups.

b. Non-Parametric Data or Ordinal Data:

1. One Sample:

   • Test: Wilcoxon Signed-Rank Test
   • Use: Compare the median of a sample to a known value.

2. Two Independent Samples:

   • Test: Mann-Whitney U test
   • Use: Compare medians of two independent groups.

3. Two Related Samples:

   • Test: Wilcoxon Signed-Rank Test for Paired Samples
   • Use: Compare medians of two related groups.

4. More than Two Independent Samples:

   • Test: Kruskal-Wallis test
   • Use: Non-parametric alternative to ANOVA.

2. Categorical Data:

1. Two Categorical Variables:

   • Test: Chi-square test
   • Use: Assess independence between two categorical variables.

2. Association between Two Variables (one categorical, one numeric):

   • Test: Point-Biserial Correlation or Spearman's Rank Correlation
   • Use: Assess the relationship between a categorical and a continuous variable.

3. Association between Two Categorical Variables:

   • Test: Cramér's V or Phi coefficient
   • Use: Measure the strength of association between two categorical variables.

3. Experimental Design:

1. Before-and-After Study:

   • Test: Paired samples t-test or Wilcoxon Signed-Rank Test
   • Use: Compare measurements taken before and after an intervention.

2. Matched Pairs Design:

   • Test: McNemar's Test (categorical) or paired samples t-test (numeric)
   • Use: Compare related pairs in a study with matched pairs.

3. Repeated Measures Design:

   • Test: Repeated Measures ANOVA or Friedman Test
   • Use: Assess the effect of an intervention over time with multiple measurements.

4. Regression Analysis:

1. Linear Relationship between Variables:

   • Test: Pearson correlation coefficient
   • Use: Measure the strength and direction of a linear relationship.

2. Predicting an Outcome:

   • Test: Simple linear regression or multiple linear regression
   • Use: Predict the value of one variable based on the value(s) of one or more other variables.

5. Comparing Variability or Dispersion:

1. Comparing Variances:
   • Test: Levene's Test
   • Use: Determine if variances are equal across groups.

6. Non-Parametric Equivalent:

If assumptions for parametric tests are violated, consider non-parametric alternatives. For instance, Mann-Whitney U test instead of the independent samples t-test or Kruskal-Wallis test instead of ANOVA.