Design Your Own Experiment

Overview

You are going to do real science. Not a baking-soda volcano. Not a worksheet where you fill in blanks about the scientific method. You are going to ask a question that nobody has told you to ask, design an experiment that nobody has designed for you, run it with your own hands, collect real data, figure out what the data means, and write it up so that another person could repeat exactly what you did. This is what scientists actually do. The baking-soda volcano is what schools think science looks like.

Here is the difference between this experiment and a school science fair project: you are going to start with a question you actually want answered. Not a question that sounds impressive. Not a question you found on a list of "easy science fair ideas." A question that has been sitting in your brain — something you have genuinely wondered about. Maybe it is "Does my dog actually prefer one brand of food over another, or does she just eat whatever is in front of her?" Maybe it is "Does the temperature of water affect how fast sugar dissolves?" Maybe it is "Do I actually run faster after stretching, or is that just something coaches say?" The question is yours. That is non-negotiable.

The scientific method is not a recipe you follow. It is a discipline for finding out what is true when you cannot simply look up the answer. It exists because human intuition is unreliable. You think you know that plants grow better in sunlight — but do you know that, or do you just believe it because someone told you? You think hot water freezes faster than cold water — but have you tested it? Science is the practice of not trusting your assumptions until you have evidence. That practice will serve you far beyond any laboratory.

By the end of this experiment, you will have a lab notebook documenting every step of a real investigation, from question to conclusion. You will have a data set you collected yourself. You will have a written report that explains what you found and what it means. And you will have learned something that no textbook could teach you: what it feels like to discover an answer through your own effort.

The Deliverable

A completed lab notebook containing: your original question, background research, hypothesis, experimental design with identified variables, raw data tables, data analysis with at least one chart or graph, a written conclusion, and a reflection on sources of error. Plus a final experimental report (2-3 pages) written clearly enough that someone your age could replicate the experiment from your description alone.

Before You Start

You need one thing before Session 1: curiosity about something specific. Spend a few days before this experiment paying attention to the questions that pop into your head. When you catch yourself thinking "I wonder if..." or "What would happen if..." — write it down. Bring your list of questions to Session 1. The best experiments come from genuine puzzlement, not from trying to think of something that sounds scientific.

You also need a parent who is willing to help you obtain materials, supervise any steps that involve heat or sharp objects, and — critically — resist the urge to redesign your experiment for you. The parent's role is to keep you safe, help you think through your logic when you are stuck, and occasionally remind you to write things down. The experiment is yours.

Session 1: The Question and the Design (60 minutes)

Choosing Your Question

Look at your list of questions. Now filter them through these three criteria:

Testable: Can you actually run an experiment to answer this? "Why is the sky blue?" is a great question, but you cannot run an experiment to answer it — you need physics and a textbook. "Does the color of a container affect how fast ice melts?" is testable — you can set it up and measure it.

Measurable: Can you collect numerical data? "Do cats prefer classical music?" is hard to measure because you would need to define and quantify "prefer." "Does playing classical music change how long a cat stays in a room?" is measurable — you can time it.

Achievable: Can you run this experiment with the materials and time you have? "Does fertilizer affect plant growth?" is a fine question, but if it takes six weeks to see results and you have five sessions, you need to pick something faster — or plan your sessions with enough gap between them.

Write your top three questions in your lab notebook. For each one, write a sentence about whether it passes all three filters. Circle the one you are going to investigate.

Background Research (15 minutes)

Before you design your experiment, find out what is already known about your topic. This is not cheating — it is how real scientists work. No researcher starts an experiment without reading what others have done first.

Look up your topic using books, encyclopedias, or supervised internet research. In your lab notebook, write a half-page summary of what you found. Include:

• What is already known about this topic
• Whether anyone has done a similar experiment (and what they found)
• Any information that helps you make a prediction

If you find that your exact question has already been answered definitively, you have two choices: pick a different question, or add a twist that makes your experiment different. "Sugar dissolves faster in hot water" is well established. "Does the type of sugar — white, brown, powdered — affect how fast it dissolves in water at the same temperature?" adds a dimension that makes it your own.

Forming Your Hypothesis

A hypothesis is not a guess. It is a prediction based on reasoning. It takes this form: "If [I do this], then [this will happen], because [this reason]."

The "because" is the part most people skip. It is also the most important part. It forces you to explain your reasoning before you test it. If your reasoning is wrong, the experiment will show you — and you will learn more from a wrong prediction with clear reasoning than from a right prediction with no reasoning at all.

Write your hypothesis in your lab notebook. Underline the "because" clause.

Identifying Your Variables

This is where most experiments go wrong, and it is where yours will succeed if you are careful.

Independent variable: The one thing you are deliberately changing. You pick this. In an experiment about water temperature and dissolving sugar, the independent variable is water temperature. You are controlling it — you decide what temperatures to test.

Dependent variable: The thing you are measuring to see if it changed. This is your result. In the sugar experiment, the dependent variable is how long it takes the sugar to dissolve. You measure it — you do not control it.

Controlled variables: Everything else. Everything that is not the independent variable must stay the same across all trials. Same amount of water. Same amount of sugar. Same type of sugar. Same container. Same stirring method (or no stirring). Same starting conditions. If you change two things at once, you will never know which one caused the result.

In your lab notebook, create a table:

Fill in every row. List at least three controlled variables. For each one, write exactly how you will keep it the same. "Same amount of water" is not enough — write "200 mL of water, measured with a measuring cup, for every trial."

Designing the Procedure

Write your experimental procedure as a numbered list of steps. Write it as if you are giving instructions to someone who has never met you and knows nothing about your experiment. Every step must be specific enough to repeat exactly.

Bad step: "Heat the water." Good step: "Heat 200 mL of tap water in a microwave-safe glass measuring cup for 90 seconds on high power. Use a thermometer to verify the temperature is between 70-75 degrees Celsius. If below 70, heat for an additional 15 seconds."

The difference between those two instructions is the difference between science and guessing. Science is repeatable. If someone cannot repeat your experiment and get similar results, the experiment has a problem.

How many trials? At least three trials for each condition. If you are testing sugar dissolution at three different temperatures, that is nine trials minimum (three temperatures times three trials each). One trial proves nothing — it could be a fluke. Three trials start to show a pattern. Write the number of trials into your procedure.

Materials List

Based on your procedure, write a specific materials list in your lab notebook. Include quantities. "Water" is not a materials list entry. "1.5 liters of tap water" is. Review the list with your parent and gather everything before Session 2.

Session 2: Run the Experiment (60-75 minutes)

Setup

Prepare your workspace. Lay out all materials. Set up your data table in your lab notebook before you begin. The table should have columns for: trial number, the value of your independent variable for that trial, your measurement of the dependent variable, and a notes column for anything unusual you observe.

Example (for the sugar experiment):

Take a photograph of your setup before you begin. This goes in your report.

Running Trials

Follow your procedure exactly. Do not improvise. Do not skip steps. Do not say "close enough" when a measurement is off. The value of a well-designed procedure is that it removes the temptation to cut corners.

Record data in pen. This is a real scientific practice. Pencil can be erased, and erasing data — even bad data — is a form of dishonesty. If you make a recording error, draw a single line through it, write the correct value next to it, and initial the correction. Never erase or white-out data in a lab notebook.

Record everything, not just the numbers. If something unexpected happens during a trial — the water was slightly cooler than intended, the sugar clumped instead of spreading out, the dog got distracted by a noise — write it in the notes column. These observations often turn out to be the most interesting part of the data.

Do not stop if the results surprise you. If your first few trials contradict your hypothesis, keep going. The experiment is not failing — it is producing results. Results that contradict your prediction are just as valuable as results that confirm it. More valuable, often, because they mean you are about to learn something you did not already know.

When Things Go Wrong

Something will go wrong. This is guaranteed. A measurement will be inconsistent. A controlled variable will slip. You will realize mid-experiment that your procedure has a flaw. Here is what to do:

If a single trial is clearly contaminated (you spilled something, the timer malfunctioned, the dog knocked over the setup), note it in your data table, mark that trial as invalid, and run an additional trial to replace it. Do not delete the bad trial — document it and explain why you excluded it.

If you realize your procedure has a systematic flaw (you did not control something you should have), you have a choice: fix the procedure and start over, or complete the experiment as designed and note the flaw in your analysis. If you are less than halfway through, start over with the improved procedure. If you are more than halfway, finish and address the issue honestly in your report. Real scientists face this decision constantly.

If your results are inconsistent (the three trials at the same condition give wildly different numbers), that is data too. It means either your measurement technique needs work or there is an uncontrolled variable affecting the outcome. Do not hide inconsistency — investigate it.

Session 3: Analyze the Data (45 minutes)

Organize

Transfer your raw data from the data table into a clean summary. Calculate the average (mean) for each condition. If you tested three temperatures with three trials each, you should have three averages.

To calculate the mean: add the three trial values and divide by three. Write the calculation out — do not just use a calculator and write the answer. Showing your math is part of the record.

Also calculate the range for each condition (highest value minus lowest value). The range tells you how consistent your results were. A small range means your trials were consistent. A large range means there was significant variation — which is worth investigating.

Make a Graph

Data in a table is hard to see patterns in. A graph makes the pattern visible. Use graph paper or draw axes with a ruler.

Choose the right type of graph:

• If your independent variable is numerical (temperature, time, amount), use a line graph or scatter plot. Independent variable on the horizontal axis, dependent variable on the vertical axis.
• If your independent variable is categorical (brand A vs. brand B, music vs. no music), use a bar graph.

Label everything. Both axes need labels with units. The graph needs a title. If you plotted averages, note that. If you included individual data points along with averages, even better — this shows both the trend and the variation.

Draw the graph by hand. This is deliberate. When you plot each point yourself, you see the data in a way that a computer-generated graph does not give you. You notice which points are outliers. You feel whether the trend is strong or weak. After you have drawn it by hand, you can make a digital version if you want — but the hand-drawn version comes first.

Interpret

Look at your graph. What does it show? Write a paragraph in your lab notebook answering these questions:

1. Is there a clear pattern in the data? Describe it in plain language.
2. Does the pattern support your hypothesis, contradict it, or show no clear relationship?
3. How strong is the pattern? Are the differences between conditions large and consistent, or small and variable?
4. Are there any outlier data points — results that do not fit the overall pattern? What might explain them?

Be precise. "The data supports my hypothesis" is not enough. "Sugar dissolved an average of 47 seconds faster in 75-degree water than in 25-degree water, which supports my hypothesis that higher temperature increases dissolution rate" is precise.

If your hypothesis was wrong, say so clearly and explain what the data actually shows. A wrong hypothesis with honest analysis is excellent science. A right hypothesis with sloppy analysis is not.

Session 4: Sources of Error and the Report (60 minutes)

Sources of Error

Every experiment has errors. Not mistakes — errors. These are factors that may have affected your results and that you could not perfectly control. Identifying them honestly is one of the most important parts of scientific thinking.

In your lab notebook, write the heading "Sources of Error" and list at least three. For each one, explain:

• What the source of error was
• How it might have affected your results (in which direction — would it make your measured values too high, too low, or just more variable?)
• What you would do differently to reduce this error if you ran the experiment again

Common sources of error to consider:

• Measurement precision (your ruler measures to the nearest millimeter, but the actual value is between two marks)
• Timing inconsistency (you cannot start and stop a stopwatch at exactly the right instant)
• Environmental variation (the room temperature changed between morning trials and afternoon trials)
• Sample variation (the sugar crystals were not identical in size; the plant seeds were not genetically identical)
• Observer bias (you might unconsciously time things differently when you expect a certain result)

That last one — observer bias — is the hardest to control and the most honest to acknowledge. In professional science, this is why experiments use "blinding": the person measuring the result does not know which condition they are measuring. Think about whether your experiment could have used blinding, and whether it would have changed anything.

Writing the Report

Your experimental report is the final deliverable. It follows a standard structure that scientists have used for centuries. Write it on clean paper or type it up. It should be clear enough that someone your age — someone who was not there — could read it and repeat your experiment.

Title: A specific description of what you investigated. Not "My Science Experiment." Something like "The Effect of Water Temperature on Sugar Dissolution Rate" or "Do Dogs Show a Preference Between Two Brands of Kibble?"

Question: The question you set out to answer, in one sentence.

Background: A short paragraph (3-5 sentences) summarizing what you learned during your background research. Cite your sources.

Hypothesis: Your prediction and your reasoning, in the "If... then... because..." format.

Materials: Your complete list, with quantities.

Procedure: Your numbered steps, exactly as you followed them (updated to reflect any changes you made during the experiment).

Data: Your data table and graph. Include both the raw data (every trial) and the calculated averages.

Analysis: Your interpretation of the data. What pattern did you find? Does it support or contradict your hypothesis? How strong is the evidence?

Sources of Error: Your list from earlier in this session.

Conclusion: One paragraph answering your original question based on your evidence. State clearly what you found, how confident you are in the finding, and what question you would investigate next.

Session 5: Reflection and Presentation (45 minutes)

Present Your Findings

Explain your experiment to someone — a parent, a sibling, a friend, a grandparent. Walk them through the question, the design, the data, and the conclusion. Use your report and your graph as visual aids. Let them ask questions. If they ask something you cannot answer, write it down — it might be a good question for a follow-up experiment.

The ability to explain your work clearly to someone who was not there is a separate skill from doing the work. Scientists call this "communicating results," and it matters as much as the experiment itself. If you cannot explain what you did and what you found, the knowledge dies with your lab notebook.

Reflection

In your lab notebook, answer these final questions:

1. What was the hardest part of this experiment? Not the most time-consuming — the hardest. The part that required the most thinking.
2. If you had unlimited time and resources, how would you improve this experiment? Be specific — what additional conditions would you test, what equipment would you use, how many trials would you run?
3. Did your results raise any new questions? What would you investigate next?
4. What is the difference between what you did and what you see in a school science textbook? Which one felt more like real learning?
5. Think of a decision you make in everyday life — what to eat, how to study, which route to take. How could you apply the experimental approach (hypothesis, controlled test, data, analysis) to that decision?

Archive Your Work

Put your lab notebook, your report, your photographs, and any physical artifacts from the experiment into your folder. Date the outside. This is your first piece of real scientific work. It has value not because it will be graded, but because it represents a question you asked and answered through your own effort.

Common Failure Modes

"I can't think of a good question." You are overthinking it. The best experiments come from small, concrete questions, not grand ones. "Which paper towel brand absorbs the most water?" is a perfectly good experiment. So is "Does my bike go faster with higher tire pressure?" Start with something you can observe in your kitchen, your yard, or your daily routine.

"I changed two things at once and now I can't tell what caused the result." This is the single most common experimental error, and it is why Session 1 spends so much time on variables. If this happens, you have two options: rerun the experiment controlling one variable at a time, or acknowledge the flaw in your report and explain what you would change. Do not pretend the flaw does not exist.

"My results are all over the place." High variability means something interesting is happening. Either your measurement technique is imprecise (practice it and rerun), there is an uncontrolled variable (figure out what it is), or the thing you are testing genuinely produces variable results (which is a finding in itself). Messy data is not failed data — it is data that demands investigation.

"My hypothesis was wrong." Congratulations. You just experienced what most of science actually feels like. The majority of hypotheses in professional research turn out to be wrong or partially wrong. The value was never in being right — it was in the rigor of the test. Write up your actual findings clearly and explain what you think is really going on.

"The experiment is taking longer than I expected." Experiments almost always take longer than planned. If observation time between sessions is the bottleneck (waiting for plants to grow, waiting for something to dry), adjust your session spacing. If the procedure itself is too long, simplify — reduce the number of conditions or trials. Three conditions with three trials each (nine total) is a solid experiment. You do not need twenty conditions to do good science.

"I am bored during data collection." Data collection is repetitive by design. That is what makes it reliable. If you are bored, you are doing it right. Put on music if it does not affect your experiment. Take breaks between trial sets. But do not skip trials or rush measurements because you are tired of the process. Discipline during boring repetition is one of the core skills of any serious endeavor — science, athletics, music, craft.

Extensions

• Run a follow-up experiment. Your conclusion almost certainly raised a new question. Design a second experiment to answer it. This is how real research programs work — each experiment opens the door to the next one.
• Replicate someone else's experiment. Find a published experiment (in a science magazine, a book, or an educational website) and try to replicate it exactly. Did you get the same results? If not, figure out why. Replication is the backbone of science — a result that cannot be replicated is not a result.
• Add statistical analysis. Learn to calculate standard deviation for your data sets (look it up or ask a parent to teach you). Standard deviation tells you not just whether there is a difference between conditions, but whether that difference is meaningful given the amount of variation in your data. This is the beginning of real statistical thinking.
• Enter your experiment in a fair or competition. Science fairs exist because they give young researchers an audience and a deadline. Find a local or regional science fair and submit your work. The feedback from judges — who are often working scientists — is invaluable.
• Teach the method. Find a younger child (a sibling, a neighbor, a cousin) and walk them through designing their own experiment. Teaching someone else is the deepest form of understanding. You will discover gaps in your own knowledge that you did not know existed.