First, the goal of the experiment is determined. For example, you may want to find out which freezes first, apple juice or water. This is your goal. you develop a hypothesis, which is a statement. The hypothesis might be "apple juice freezes before water." After you run the experiment -- and collect data -- you determine the hypothesis to be true, or proven to be false. If it's false, your original premise must be re-evaluated.
Based upon the goal, you design the experiment procedure accordingly. According to Science Buddies, a procedure is a series of steps. You outline each step the experiment should take. You must write the procedure very precisely, so another person can duplicate the experiment. Don't use subjective words like "big" or "little." In the procedure, a list of materials is also required. In the apple juice example, a thermometer and a timer or clock would be needed.
As you are running the experiment, collect data. With apple juice freezing, you would take the temperature every five minutes and write down the temperature. The writing down of the temperature, and the time you took the reading is the data collection. With some experiments, data can only be collected at the very end. For example, in a chemistry experiment, you collect the data only after all the chemicals are mixed together.
If the procedure is flawed, the dated collected will be flawed too. This is commonly referred to as "garbage in, garbage out." The flawed procedure is the garbage in. The flawed data collected is the garbage out. For example, suppose you want to determine how many pathogens are in pond water. Your collection jar is unsterilized and full of bacteria. The lab, not knowing the jar was contaminated, states the pond water is full of bacteria.The pond water may in fact be bacteria free, but the data, because of GIGO, is in error. A researcher may reach the wrong conclusion because of GIGO. The "Changing Minds Organization" wrote extensively that validity is all important. Many threats to validity exist, including the method, or procedure, of the experiment itself may not be valid.