Two-Sample T-Test Calculator

This advanced tool performs a comprehensive Two-Sample T-Test to compare the means of two groups of data. Whether you are analyzing scientific research, A/B testing results, or student grades, this calculator determines if the difference between your groups is statistically significant or likely due to random chance.

It is designed for professional accuracy and supports the three primary variations of the t-test:

1. Independent T-Tests (Unpaired): Used when comparing two separate, unrelated groups (e.g., "Control" vs. "Treatment", "Men" vs. "Women").
   • Student’s t-test: The classic test assumes both groups have equal variances.
   • Welch’s t-test: The modern, robust default (also known as Uthe nequal Variances t-test). It is more reliable when group sizes or variances differ.
2. Paired T-Test (Dependent): Used for "matched-pair" or "repeated measures" data (e.g., "Pre-test" vs. "Post-test" scores for the same students). This test analyzes the difference within each pair.

Beyond basic p-values, this calculator includes automatic assumption checking (Shapiro-Wilk, Levene’s Test), effect size calculation (Cohen’s d), and a suite of interactive visualizations.

Key Features

• Complete Hypothesis Control:
  • Test Selection: Easily toggle between Independent and Paired modes.
  • Variance Assumption: For independent tests, choose between Student's (Equal Variance) or Welch's (Unequal Variance) logic.
  • Hypothesis Direction: Supports Two-Tailed (Non-directional) and One-Tailed (Greater/Less than) alternative hypotheses.
  • Significance Level (α): Adjustable alpha (e.g., 0.05, 0.01) for strict research standards.
• Automatic Assumption Checks: Don't just guess—verify your data's validity.
  • Shapiro-Wilk Test: Checks if your data follows a normal distribution (normality assumption).
  • Levene’s Test: Checks if your groups have equal variances (homogeneity of variance).
  • Smart Warnings: The tool will alert you if assumptions are violated, suggesting when to use Welch's t-test or a non-parametric alternative like the Mann-Whitney U test.
• Publication-Ready Statistics:
  • Cohen’s d: Calculates the Effect Size (magnitude of the difference), crucial for modern reporting.
  • Confidence Interval (CI): Shows the range of the true difference between means (e.g., "We are 95% confident the difference is between 2.5 and 5.1").
  • Standard Error (SEM): Reports the precision of the means.
  • Core Stats: t-statistic, p-value, Degrees of Freedom (df).
• Advanced Visualizations: Move beyond simple bar charts with four interactive plots:
  • Box Plot: Best for summary statistics (Median, IQR, Outliers).
  • Violin Plot: Visualizes the full probability density (shape) of the data.
  • Jitter Plot: Shows every single raw data point to prevent "over-plotting."
  • Difference/Overlap Plot: For Paired tests, displays a histogram of differences. For Independent tests, shows how the two distributions overlap.
• Big Data Ready: Capable of analyzing datasets up to 10,000 data points (5,000 rows per group) using a high-performance cloud backend.
• Export & Reporting:
  • PDF Report: Generate a full landscape PDF containing the summary table, interpretation, and key charts.
  • High-Res Images: Download charts as PNG or JPEG in both Light and Dark themes.

How to Use the Calculator

1. Enter Data: Paste your dataset into the "Group 1" and "Group 2" columns. You can manually type data, paste from Excel, or use the "Import CSV" button.
   • Note: For Paired tests, ensure rows match (Row 1 in Group A corresponds to Row 1 in Group B). For Independent tests, groups can be different sizes.
2. Configure Analysis:
   • Test Type: Select "Independent" or "Paired".
   • Hypothesis: Choose your alternative hypothesis (e.g., Group 1 ≠ Group 2).
   • Assumption (Independent Only): Uncheck "Assume Equal Variances" to perform Welch's t-test (recommended), or check it for the classic Student's t-test.
3. Run Analysis: Click "Calculate Results" to perform the test.

Interpreting Your Results

• Summary Tab:
  • Interpretation Box: A color-coded message tells you immediately if the result is Statistically Significant based on your chosen alpha.
  • Assumption Warnings: Look for yellow warning notes here. If the Shapiro-Wilk p-value is < 0.05, your data may not be normal. If Levene's p-value is < 0.05, you have unequal variances.
  • Cohen's d: Use this to understand the "real world" impact. Generally, 0.2 is a small effect, 0.5 is medium, and 0.8+ is a large effect.
• Visual Tabs: Switch between the Box, Violin, and Difference tabs. A Box Plot where the notches or boxes do not overlap is a strong visual indicator of a significant difference.

Comparison to Other Tools

Excel vs. Learnbin Lab T Test Calculator

While Excel's "Data Analysis ToolPak" can perform t-tests, it requires manual configuration for each specific type (Paired, Homoscedastic, Heteroscedastic) and produces static, text-only output. This tool automatically calculates Confidence Intervals and Effect Size (which Excel lacks) and instantly generates professional visualization plots that would take significant time to build manually in Excel.

When to use a different test?

• One Group: If you are comparing a single group against a known standard (e.g., "Is our average height 175cm?"), use the One-Sample T-Test Calculator.
• More than Two Groups: If you have 3 or more groups (e.g., Placebo vs. Low Dose vs. High Dose), you must use the One-Way ANOVA Calculator to avoid statistical errors.
• Non-Normal Data: If the Shapiro-Wilk test warns you that your data is not normal, consider using the Mann-Whitney U Test (for independent) or Wilcoxon Signed-Rank Test (for paired).