Hypothesis and Experimental Design

Jump to: Activity Examples | Resources


Two important elements of The Scientific Method that will help you design your research approach more efficiently are “Generating Hypotheses” and “Designing Controlled Experiments” to test these hypotheses. A well-designed experiment that you deeply understand will save time and resources and facilitate easier data analysis/interpretation. Many people reading this may be working on a project that focuses on designing a product, or discovery research where the hypothesis it is not immediately obvious. We encourage you to read on however as the exercise of generating a hypothesis will likely help you think about the assumptions you are making in your research and the physical principles your work builds upon.    

These activities will help you … 

  • Begin formulating an appropriate hypothesis related to your research. 
  • Apply a systematic process for designing experiments. 

What is a Hypothesis? 

A hypothesis is an “educated guess/prediction” or proposed explanation” of how a system will behave based on the available evidenceA hypothesis is a starting point for further investigation and testing because a hypothesis makes a prediction about the behavior of a measurable outcome of an experiment. A hypothesis should be: 

  • Testable – you can design an experiment to test it 
  • Falsifiable – it can be proven wrong (note it cannot be “proved”) 
  • Useful – the outcome must give valuable information 

A useful hypothesis may relate to the underlying question of your research. For example: 

“We hypothesize that therapy resistant cell populations will be enriched in hypoxic microenvironments. “ 

“We hypothesize that increasing the number of boreholes simulated in 3D geological models minimizes the variation of the geological model results.”  

Some research projects do not have an obvious hypothesis to test, but the design strategy/concept chosen is based on an underlying assumption about how the system being designed works (i.e. the hypothesis). For example: 

“We hypothesize that decreasing the baking temperature of the photoresist layer will reduce thermal expansion and device cracking”  

In this case the researcher is troubleshooting poor device quality and is proposing to vary different fabrication parameters (one being baking temperature). Understanding the assumptions (working hypotheses) of why different variables might improve device quality is useful as it provides a basis to prioritize what variables to focus on first. The core goal of this research is not to test a specific hypothesis, but using the scientific method to troubleshoot a design challenge will enable the researcher to understand the parameters that control the behavior of different designs and to identify a design that is successful more efficiently. 

In all the examples above, the hypothesis helps to guide the design of a useful and interpretable experiment with appropriate controls that rule out alternative explanations of the experimental observation. Hypotheses are therefore likely essential and useful parts of all research projects. 

Suggested Activity – Create a Hypothesis for Your Research 

Estimated time: 30 mins  

  1. Write down the parameters you are varying or testing in your experimental system or model and how you think the behaviour of the system is going to vary with these parameters.  
    • (Alternative) If your project goal is to design a device, write down the parameters you believe control whether the device will work. 
    • (Alternative) If your project goal involves optimizing a process, write down the underlying physics or chemistry controlling the process you are studying. 
  1. With these parameters in mind, write down the key assumption(s) you are making about how your system works. 
  2. Try to formulate each one of these assumptions into a hypothesis that might be useful for your research project. If you have multiple aims each one may have a separate hypothesis. Make sure the hypothesis meets each of the three key elements above.
  3. Share your hypothesis with a peer or your supervisor to discuss if this is a good hypothesis – is it testable? Does it make a useful prediction? Does it capture the key underlying assumptions your research is based upon? 

Remember that writing a good research hypothesis is challenging and will take a lot of careful thought about the underlying science that governs your system. 

Designing Experiments 

Designing experiments appropriately is very important to avoid wasting resources (time!) and to ensure results can be interpreted correctly. It is often very useful to discuss the design of your planned experiments in your meetings with your supervisor to get feedback before you start doing experiments. This will also ensure you and your supervisor have a consistent understanding of experimental design and that all the appropriate controls required to interpret your data have been considered. 

The factors that must be considered when you design experiments is going to depend on your specific area of research. Some important things to think about when designing experiments include: 

Rationale: What is the purpose of this experiment? Is this the best experiment I can do? Does my experiment answer any questionDoes this experiment help answer the question I am trying to ask? What hypothesis am I trying to test? 

Will my experiment be interpretable?  What controls can I use to distinguish my results from other potential explanations? Can I add a control to distinguish between explanations? Can I add a control to further test my hypothesis? 

Is my experiment/model rigorous? What is the sensitivity of the method I am using and can it measure accurately what I want to measure? What outcomes (metrics) will I measure and is this measurement appropriate? How many replicates (technical replicates versus independent replicates) will I do? Am I only changing the variable that I am testing? What am I keeping constant? What statistical tests do I plan to carry out and what considerations are needed? Is my statistical design appropriate (power analysis, sufficient replicates)? 

What logistics do I need to consider? Are the equipment/resources I need available? Do I need additional training or equipment access? Are there important safety or ethical issues/permits to consider? Are pilot experiments needed to assess feasibility and what would these be? What is my planned experimental protocol and are there important timing issues to consider? What experimental outputs and parameters need to be documented throughout experiment? 

This list is not exhaustive and you should consider what is missing for your particular situation. 

Suggested Activity – Design an Experiment Using a Template 

Estimated time: 45 min 

  1. Explore the excel template for experimental design (Resource 1) or modelling (Resource 2).  A template like this is very useful for keeping track of protocols as well as improving the reproducibility of your experiments. Note this template is simply a starting point to get you thinking systematically and should be adapted to best suit your needs.  
  2. Fill out the template for an experiment or modelling project you are planning to complete soon. 
  3. Consider how you can modify this template to be more applicable to your specific project. 
  4. Using the template document, explain your experimental design/model design to a peer or your supervisor. Let them ask questions to understand your design and provide feedback. Alternatively, if there is a part of your design that you are unclear about this is a great starting point for a targeted and efficient discussion with your supervisor.  
  5. Revise your design based on feedback. 

Activity Examples

Hypothesis Generation
Example Hypotheses Several examples of valid hypotheses.
Experimental Design
Experimental Design Template Examples Examples of completed experimental design templates.


Resources

Templates for Experimental Design
1Experimental Design Template – Wet Lab Template for experimental design.
2Experimental Design Template - Modelling Template for modelling design.