Make sure the tested variable is the only thing that varies.

For example, if you want to test what effect Miracle Gro has on plant height, make sure that the amount of Miracle Gro is the only thing that is different among all of your trials.  If you put half of the plants by the window and the other half on your dresser in a dark corner of the room, you have not isolated the variable you are testing.  The amount of sunlight is another  factor that varies among your groups and invalidates your results.

Choose the best and most precise instruments possible for making exact measurements.

We've seen in class that beakers are good for a rough approximate measure of volume, but they are not very good for accuracy.  When making a measurement, decide if the measurement needs to be exact and remember:

• Look at the markings on the instruments.  The instrument with the smallest interval between markings will be more precise.
• If two instruments have the same intervals marked, choose the "skinnier".  Most often, this will be the more accurate instrument because there will be a larger distance between markings.  This enables it to be read with a higher precision.
  

Make sure you are using the best sample size

Use a sample size that is appropriate for the instrument that you are using to measure it.  If you are measuring a sample that is about as small as the smallest markings on the instrument (or smaller), than you are measuring something beyond what the instrument is designed to measure.  Either choose a larger sample size or a better instrument.

Make sure your sample is a true representative of the tested item.

For example, if you are testing the amount dissolved oxygen in a pond, do not just take samples from very top layer of water. Be sure to take samples from all different layers in the pond.  The top layer will always have a higher oxygen level, so if that is all you are sampling, your results will not show an accurate measure of the pond as a whole.  

Also, be sure to take your samples from different regions of the pond.  It is possible you might be taking your samples by a drain inlet where fresh water from a nearby stream is entering the pond.  This water will be much higher in oxygen, and if you only take your samples from this spot, will not be a true representation of the water in the pond.

Perform multiple trials and average together for Experimental Answer.

Any experiment should exhibit good precision to ensure its reliability.  This means that the results should be repeatable.  The golden rule is to always perform at least three trials to verify your results. If you do not get similar results in all three trials, more trials should be done.  If possible, it is usually a good idea to try to vary your sample size in each of your trials.

To determine your experimental answer, take the average of all of your trials. Do not simply choose one trial that looks or feels right.  The empirical information from each trial is sacred and should not be ignored or discarded.

Trust your Data.  Do not "fudge" or throw away data just because it looks or feels wrong.

As mentioned above, the data collected during experimentation is sacred and should not be overlooked, changed,  or discarded because it doesn't look or feel right.  If a trial result does appear to be quite different from all of the other trial results, you may perform the Q-Test to see if the data is an anomaly and can statistically be discarded.

However, if a "Blunder" (spilled some of the sample, etc.) is made during the trial the results will obviously be invalid and therefore discarded.