Jamovi Data Exploration (Descriptives)

The screenshot shows the Descriptives analysis module selected under the Exploration menu in jamovi.

🔹 Left Panel: Analysis Setup

• Available Variables: StudentID, Classroom, Gender, and MathScore .

• MathScore has been moved into the Variables field → descriptive statistics will be computed for this variable.

• The Split field is empty → results are not stratified by group (e.g., Gender or Classroom).

• The Statistics and Plots sections (collapsed) allow customization—e.g., enabling mean, median, SD, skewness, or generating histograms/boxplots.

🔹 Right Panel: Output

The table displays summary statistics for MathScore:

• N = 20, Missing = 0 → complete data for 20 observations.

• Mean = 79.0, Median = 79.5 → similar values, suggesting a roughly symmetric distribution.

• Standard deviation = 12.4 → typical deviation from the mean.

• Min = 55, Max = 99 → full observed range.

In jamovi, this output updates dynamically—if Gender were added to Split, separate tables would appear for each group (e.g., Male/Female). Likewise, checking options like Skewness, Kurtosis, or Confidence interval for mean would expand the results.

This analysis is foundational: it validates data integrity (e.g., no unexpected missing values or implausible scores) and informs decisions about subsequent analyses (e.g., suitability for parametric tests).

---

During Exploratory Data Analysis (EDA), the decision to examine additional statistics—such as skewness, kurtosis, or the confidence interval for the mean—depends on the analytical objectives, sample size, and the nature of subsequent statistical procedures.

For a variable like MathScore with a sample size of N = 20, deeper exploration beyond basic measures (mean, median, standard deviation, range) is often beneficial but should be approached with appropriate caution.

Skewness and kurtosis provide numerical summaries of distributional shape. Skewness quantifies asymmetry, while kurtosis reflects tail weight relative to a normal distribution. In small samples, however, these statistics can be unstable and sensitive to individual observations. Therefore, their primary utility lies in triangulation with graphical tools—such as histograms, boxplots, or Q-Q plots—rather than as standalone diagnostics. In jamovi, enabling these statistics (under the Statistics dropdown in the Descriptives module) adds minimal effort and supports more informed interpretation, particularly when assessing assumptions for parametric tests.

The confidence interval (CI) for the mean—typically the 95% CI—is highly recommended during EDA, especially with modest sample sizes. Unlike a point estimate (e.g., mean = 79.0), the CI conveys the precision of that estimate. A wide interval signals greater uncertainty, which may influence decisions about data collection, modeling choices, or interpretation of group differences. In jamovi, this option is readily available and computationally straightforward to include.

A structured EDA workflow would prioritize the following steps in sequence:

1. Verification of data completeness and range (e.g., no missing values, plausible min/max).

2. Computation of central tendency and dispersion (mean, median, SD, IQR).

3. Visual inspection via histogram and boxplot to detect skewness, outliers, or multimodality.

4. Supplemental numerical indicators (skewness, kurtosis) to corroborate visual impressions.

5. Reporting of the 95% confidence interval for the mean to contextualize inferential intent.

Formal normality tests (e.g., Shapiro–Wilk) are generally deferred to the assumption-checking phase of hypothesis testing rather than included in initial EDA, as they tend to lack power in small samples or become overly sensitive in large ones.