In our local school districts, children must be of the required age on September 15th. The cutoff dates were created to keep parents from putting kids in school too early. But nowadays the trend seems to be starting school later. Several friends and coworkers are holding their children out to let them mature or not subject them to the bureaucratic rigors of seat assignments and timed periods for work and recess until they're a bit older.
Starting school is a major decision, running the gamut of life's questions: Will he know his letters and numbers? Will he be too small for soccer? Or too big? Will he like being the first kid in his class with a driver's license? Or the last? And when he begins school determines when he graduates and moves on to real life. There are no pat answers.
When I started school, the cutoff date was January 1st. I was born December 15th so I was always the youngest kid in my class. My brother was born a little over a year later on January 2d. He was always two grades behind me and one of the oldest in his classes. We both had our share of successes and problems in school. Our age extremes didn't seem to matter that much. And because I had an early growth spurt, I was one of the largest kids in 9th grade (the same height I am now but a whole lot lighter).
The trouble with predicting what will happen with a single human being is that we base most of our assumptions upon statistics: the average person at this age is this tall and capable of these thought processes. Statistics are descriptions of populations. That's why the word is usually plural. In order to determine an average, you must have at least two people or two occurrences or two other data points. And usually you get many. There are several methods for determining the appropriate sample size (one that will give you a reasonably accurate view of what the entire population may do).
Probability is what is used to describe the behavior of one entity in the population. For instance, statistically, after 100 coin flips, 50 (plus or minus a small amount) should be heads. That matches the probability of 1/2 that the next flip will be heads. But don't bet on it.
I can take the assumptions about populations, say that a ten-year-old boy will generally be able to do fifth grade math or that a sixteen-year-old can write an essay that will satisfy an 11th grade English teacher, and assume that chances are Matthew will be able to do okay at all those tasks during all those years.
But there's no guarantee. Neither statistics nor probability prepare me for making the real life decisions. We want to make the right choice, to make sure that he is not tested too severely by difficulties but is challenged enough to become resourceful and independent. So we fell back on the safe choice. We hedged our bet and put him in the four-year-old class. If it works out, fine. If not, then we can always switch teachers or preschools and start in a different four-year-old class next year.