Do you ever find yourself wondering why it is that somedays you can get home in 10 minutes, other days it takes 12, and sometimes 11? And on rare occaisons it may take an hour and a half? While you might be interested in getting the time as low as possible, a Statistician doesn't really care why it happened. We look at all of this as random variation of a variable. The variable is "the time it takes to get home."
Variation occurs in many different places, and its' one of the primary jobs of the statistician to either control it or describe it. Examples abound. The variation of college students heights, for example, might be different from the variation in their income or their age.
Another example of variation, which you might have noticed in your science labs, is measurement error. Ask ten friends to measure the length of a dollar bill (even the same dollar bill) to the nearest centimeter. Do you think you'll all have the same answer? Usually not, particularly if you don't let anyone else know what the other measurements are until everyone is done. This is because there are a variety of factors that affect our ability to measure things, and these can change from person to person, thus affecting the measurement we record.
A similar thing happens with midterms. A midterm is supposed to
measure how well you learned the material. But you probably know
from experience that you have good days and bad days taking tests, and
so even if you know the material equally well on midterm 1 as on midterm
2, you probably won't get exactly the same score.