Let's say we have two auto-rickshaw drivers, Ram and Shyam, and we want to see if Ram drives faster on average for short trips than Shyam. We record their speeds (in km/hr) for 3 trips each.

Ram's speeds: 40, 45, 50
Shyam's speeds: 30, 35, 40

---1. Calculate the observed difference in means:
Ram's average = (40 + 45 + 50) / 3 = 135 / 3 = 45 km/hr
Shyam's average = (30 + 35 + 40) / 3 = 105 / 3 = 35 km/hr
Observed difference = 45 - 35 = 10 km/hr

---2. Combine all data:
All speeds = {40, 45, 50, 30, 35, 40}

---3. Create permutations (shuffled groups). Let's do a few manually:
Permutation 1: Randomly pick 3 for Ram and 3 for Shyam from the combined list.
Ram: {30, 40, 45} (Average = 38.33)
Shyam: {35, 40, 50} (Average = 41.67)
Difference = 38.33 - 41.67 = -3.34 km/hr

---Permutation 2:
Ram: {35, 40, 50} (Average = 41.67)
Shyam: {30, 40, 45} (Average = 38.33)
Difference = 41.67 - 38.33 = 3.34 km/hr

---Permutation 3:
Ram: {30, 35, 40} (Average = 35)
Shyam: {40, 45, 50} (Average = 45)
Difference = 35 - 45 = -10 km/hr

---4. Repeat this many, many times (usually thousands or millions) using a computer. For each permutation, calculate the difference in means.

---5. Count how many times the absolute difference from the shuffled groups is greater than or equal to our original observed difference (10 km/hr).

---6. If this happens very rarely (e.g., less than 5% of the time), it suggests that our original difference of 10 km/hr is probably not due to random chance, and Ram might actually be faster. If it happens often, the difference might just be random.