Gather your data. Make sure you use solid statistical techniques to gather the data. For example, don't give participants the choice whether to participate or not: that will invalidate your data and is called a "voluntary response sample."
Take a sample from the right population. If you want a general idea of how well U.S. students might perform on a standardized test next year, don't just take samples from inner city schools or suburban schools. Limiting yourself in scope by only sampling a small number of schools will lead to a higher probability your results will be invalid.
Graph your data. When you've gathered your data, a graph is often the best way to predict future trends. For example, if your data looks like a line, you can use a ruler to extend the line upwards. Another possibility is that your data looks like a curve (for example, an exponential curve). Whatever your graph looks like, try to find a line or a curve that best fits your data (statisticians call this "regression.")
Calculate a future trend by referring to the extended section of your graph. For example, if you took results from yearly standardized tests, graphed them and found out that they formed a linear equation (a line), you can extend the graph with a ruler. Read the graph for next year's results by finding next year on the x-axis, tracing up to the graph and reading across to the y-axis.