A good beginning point for curious minds is in planning for tests which can be convincingly fair. Young students usually have a healthy respect for fairness. Doing tests scientifically also involves fairness.
Clearly and simply state the question you would like to answer . This is the most difficult step. In science, it is not the right answer that is most difficult but getting the right question. Good questions will focus your work and make finding answers easier.
After good question is clearly and simply stated there is no accepted way all scientists proceed with scientific experiments. However there is an accepted way to present findings. Look at the following outline. It also gives guidance in accepted standards for scientific experimenting.
When writing about your findings use the following outline:
Clearly and simply state the question you are investigating.
Write a hypothesis which clearly states a claim for the expected outcome of the question. This should have a simple accept/reject conclusion after the observations are collected and organised.
Next define your variables.
manipulated variable: the variable you plan to experiment with.
responding Variable: the change you plan to measure and how you will measure the change.
Controls: State how you plan to either control all other variables or minimise bias in populations to insure fair testing.
Method: clearly describe what was done
Observations: Includes what you observed measured data is very important. Find a way to make observations by using numbers. Real data measurement is not very neat to the untrained eye. It always has some uncertainty which may make graphs look uneven or untidy. This kind of data is more believable to a scientist than exact and evened out data.
Conclusions: Get the data together in a table to look for trends and similarities. Graph the results. Try several kinds of graphs and choose the one which best suits the data for clear presentation. Make a accept / reject claim about your hypothesis. Support your choice and indicate your confidence in the choice from the evidence collected in the observations.
For more detail in experimenting read Scientific Attitude Check (Science Skills and attitudes.)
Scientific Attitude Check:
Science Skills and Attitudes
Students who produce a quality experimental science project develop scientific attitudes. They follow certain observation rules or attitudes in their work. They strive to make their observations objective and unbiased, reproducible and accurate.
 A good scientist does not prejudge results or filter data through expectations to either discard data which does not appear to fit expectations or accept "flyers" beyond their significance . A scientist is expected to be objective,unbiased. Have faith in the assumption, "Nature is reliable." Unexpected results, data which does not fit perfectly, and apparent contradictions are important. These should not be discarded but rather investigated further.
Real data does not look perfect to the untrained eye. Points on a graph do not either line up perfectly or without exception fall in an expected area. In fact, perfect data summaries are likely to be be less believable to a judge. There is always some experimental uncertainty in the data. This often called experimental error. However do not connect this idea of experimental error with "mistake". This "error" is the natural "uncertainty" in every measurement.
 Measurements must be reproducible.
The basic assumption here is that measurements will be made more than once and can be repeated by others to verify the results. This means similar measurements must agree.
Agreement among similar measurements is called precision, agreement or reliability. Significant figures and an uncertainty as '±'( range of error ) show the reliability of the measurement. Every measured value has some error or uncertainty. This error is uncertainty, not mistake. it is your "guesstimate" in the last figure of the measurement. Others will assume the last figure in a measured number has opinion in it therefore do not round off this digit with opinion. All measurements are assumed to have one digit with opinion.
 Scientists must be accurate.
The experimenter must be reliable. Experimental accuracy represents how careful you are to measure the true value. Be careful to measure what you say you are measuring.
Often when experimenting there is no accepted value to measure your work against. Therefore you must prove your work to be true(valid) to make conclusions which will be seen as defendable by others. This means doing multiple trials and or using large samples to provide data for analysis.
Statistics are a quantitative method used to decide the reliability of measurements and the soundness of conclusions. The "best" work at either a national or international level of project participation is judged based on whether or not statistical analysis was done on the data collected to form conclusions. There are several statistical tests used on data, each specific to the type of data and claim to be made.Statistical tests can be applied to most well designed projects.
Measured Numbers and Significant figures:
Rule for measuring: When measuring include all the figures known beyond doubt and one estimated figure (estimate tenths of the smallest division on the measurement tool)
Include all the measured digits known beyond reasonable doubt and one with uncertainty or opinion. Measured numbers are therefore rounded based on the "degree of confidence of extent of trust" . The degree of confidence an experimenter has in his results and devices is often determined by repeating the same experiment many times under exactly the same conditions.
All instruments must be calibrated with agreed standards if resulting information is to be comparable. Measurements are made by comparing an unknown quantity to an accepted standard.
A simple calibration, for example, is to check a thermometer with boiling water. At sea level the thermometer should read 100oC when pure water boils. Similarly it should read 0o in a mixture of water and ice at the freezing/melting point. You should never assume an instrument is linear outside the calibration points. Therefore the more calibration points used to standardise the instrument the higher the degree of confidence you are likely to put in the instrument.
Experiment starting point:
It has been said, "The answers are easy it is finding the right question that is difficult" The starting point of every experiment is finding the right question. The outcome of a good question can be answered by either a simple yes or no which is based on measured test results.
The hypothesis is your opinion of the outcome of the question you wish to investigate.
A good hypothesis should have the following characteristics:
-It must have a yes/no answer
-It must be measurable
-It must be realistic given the resources of the experimenter.
-The number of observations required to give the "yes/no" answer is governed by the size of the change to be measured, the number of trials (population) and the inherent variability in the technique, population or experimenter.
The experimenter develops a research plan. In the research plan the experimenter must outline a "fair test" for the hypothesis. In this plan the variables should be identified and the measuring plan for the variables clearly stated. It is important to use numbers when you measure and to indicate the confidence you have in the measurement. Example 98±3 m this would mean the measured value is between 101m and 95m
Identify the following:
Manipulated variable: State the range of change planned for the manipulated variable.
Responding Variable: State how the responding variable will be measured.
Controls: This should be a description of how you attempted to insure the test was fair by assuring the conditions of each test were the same and compared to an acceptable standard.
Method or Procedure: The method will state clearly what was done to test the hypothesis.
It is often necessary to practice tests to which are designed for an experiment to master the test and remove operator ()/ error.
If your results are to be comparable and reproducible you must calibrate instruments Whether the instruments are bought from a manufacturer or homemade they must be calibrated or standardised to read in standard units.
Simple statistics do not make results better however they do help to clarify the trends, meaning and degree of confidence in the data.
It is important to collect enough data. Enough data rules out variations which may be caused by operator, technical, equipment, population and daily variations.
Some simple rules to follow are:
Do as many tests as possible, as closely together as possible
Do not do tests on one day and controls on another.
If working with humans take care that you do not bias your results by letting either the subject or the administer of the test know which group they belong to.
Collect data as you go along. Do not leave data collection till the last deadline of your time frame. It may be too late to deal with variability resulting from poor methodology!
Analysis of Data Collected:
Create a spreadsheet. This need not be electronic. Just group the data together. See if the data looks similar in different groups. Electronic spreadsheets do however have their advantages especially for graphing and statistical analysis.
Once a spreadsheet has been created the data should be graphed. The choice of graphic expression will determine the acceptability and usefulness of the results. Clarity in presentation will aid in the interpretation of the data. try a few different ways to graph plots. Don't be afraid to use a log10 axis if the data is not normally distributes (very common in population data.) Seeing all the results of the different groups or conditions will give you a good idea if there are any differences and if the hypothesis is proven.