By By: Dr. Stacy Lynn, Natural Resource Ecology Lab, Colorado State University
*Special thanks to Dr. Paul Strode and Dr. Cecilia Hennessy for encouraging us to update and improve this piece.
Developing a hypothesis is a natural step in scientific inquiry, generally follows the development of a research question and functions as a tentative answer to the question. Some citizen science studies are more observational and not necessarily focused on a research question (e.g., document all species of butterflies observed in my neighborhood), but many studies do take observations a step further to ask a research question that relates those observations to changes in another variable that could be influencing patterns (e.g., plant species diversity or prevalence of native species on different properties in the neighborhood). For such studies, a hypothesis is essentially a specific, tentative, testable, measurable and falsifiable statement made prior to your research. There are two types of scientific hypotheses:
- Generalizing hypotheses: This pattern exists, and with data it might become a future law or relationship.
- Explanatory hypotheses: This mechanism causes this pattern and might become a future theory.
Your hypotheses will be dependent on the conditions under which variables are being measured and the methods you are using to collect your data (Strode 2020).
Developing a well-constructed hypothesis can help guide you to an appropriate research design for your project. If your research question and hypothesis are not well-defined you may have trouble coming up with a project design to test your hypothesis. Alexander Toledo, a PhD in Biomedical Sciences and expert in developing scientific experiments that use the scientific method, claims a hypothesis is crucial for a sound and well-developed experiment and emphasizes how the hypothesis should contribute to the solution or answer to the research question (Toledo et al 2011).
How do data collected by volunteers contribute to addressing my hypothesis?
The goal of your experiment or field study should never be to prove your hypothesis right (or another one wrong), because having such an agenda may impart bias into your approach. Rather, the goal of a scientific experiment is to learn through a structured process more about how the world works. Hypotheses should remain neutral in an effort to keep science as objective as possible. Make sure your hypothesis is capable of being supported or unsupported by the data that you and your participants collect (Cohen 2013).
How do I write a good hypothesis?
A good hypothesis should guide your research and narrow down the variables of interest and the relationship of interest to one that is measurable. Each research question and related hypothesis should address the relationship between one independent variable and one dependentvariable. If there are multiple dependent variables, these should be set up as individual sub-hypotheses (see example #3 below. An independent variable is one which the scientist researcher (you!) changes or manipulates (such as soil amendments or total water applied), or which occurs in categories (such as mid-summer vs. mid-winter, or various distance intervals) or along a natural gradient (such as temperature or rainfall means). A dependent variable is one which you are measuring the values of relative to changes in the independent variable.
Below are a few examples of hypothesis improvements to get you started:
| Original | Improved | |
| 1 | When people water their lawns the grass will turn green. | Water quantity and lawn growth are positively correlated. |
| 2 | Butterflies follow a pattern of activity that follows the sun. | Butterfly activity increases during peak sun and heat during the day in the months of June and July, and decreases in early morning and evening when it is cooler, as demonstrated by timed observations at set points in pollinator gardens. Butterfly activity demonstrates a positive correlation with the interacting variables of heat and sun, where it increases with each alone, but increases more with both heat and sun. |
| 3 | The cement plant upriver is affecting our water quality in town. | The effluent from the cement plant long the river is a source of pollution in the river. |
How do I know I have a strong hypothesis?
My hypothesis is related to my research question
The hypothesis should present a description of a pattern or relationship between two variables that can be further supported or rejected based on the data to be collected during the research.
My hypothesis is specific, and there is a 1:1 relationship between my research questions and hypotheses
Every research question should have its own hypothesis. If a hypothesis has a list of expected outcomes, then think about splitting the research question either into multiple questions, or to list sub-components that can be associated with an independent hypothesis, and easily be tested independent of each other.
My hypothesis is falsifiable
Ask yourself two questions:
- Did I choose/design a data collection protocol that will generate data to address my research question?
- Do I have a clearly identifiable approach to collect and analyze data?
Testing my hypothesis is achievable
I have access to the resources I need to collect my data.
If I need special permissions, tools or other considerations to collect data, I’ve figured out how to get them.
My hypothesis is neutral.
I’ve done my best to avoid inherent biases and leading language when asking and answering my research question.
The purpose of my experiment is to objectively look for a relationship between two variables
My hypothesis is simple and clear.
If project participants need any prior knowledge, training or equipment to collect data, I’ve given thought to how to provide it: what form resources need to be in to best serve their needs and be accessible to a diverse group of participants.
I know how I will analyze the data we collect.
A Quick Word about Causation, Correlation, and Sample Sizes
When you observe correlation between independent and dependent variables, it can be tempting to assume the relationship is causal (that one variable causes the other). Be careful not to fall into this trap! Correlation does not necessarily mean that the relationship is causal. There could be something else at work that influences the relationship. Determining causation can take multiple studies that consider multiple potential alternative explanations and control for those.
Sample size is also an important consideration as you plan your study. The more data you collect, the more reflective your sample and results are likely to be of reality. A small sample size – just a few observations – is “observational” and not enough to support or refute your hypothesis to answer your research question(s). Generally, the larger the number of measurements observed, the more observations you need to make to identify patterns.
Making citizen science place-based and relevant
Citizen science is often naturally “place-based”, or rooted in a local perspective. Even when a citizen science project is interested in larger scale than the local, your participants will be connecting with the local processes in the place where they live and/or collect data. So larger projects may string together data from many localities, each of which is grounded in the local context, to see what the relationship between variables is over space and/or time.
Creating projects that are founded on local issues and circumstances that are relevant to your anticipated participants and which may potentially have impacts on land use, management, or policy decision-making can bring both enthusiasm and a sense of making a difference (Newman et al 2016). Volunteers naturally like to make a difference with their time and efforts, and learning something new fits into the model of lifelong learning that draws many people to citizen science. Volunteers learn something new while contributing to a greater group effort that would not be possible without their contributions and the contributions of many others.
How do I handle unanticipated results?
Remember, it’s okay if your data and results are unanticipated and do not support your hypothesis. All results are interesting! If the results of your study are not what you expected, that could generate more questions for you to answer. Continue to build your data, and over time you will be able to narrow down the relationship between your variables of interest. It is also possible to run repeat studies multiple times over time, and to look at patterns in the data over seasons and years, and a variety of different conditions (e.g., total precipitation, “shock” events such as wildfire, etc.).
Important Takeaways
The purpose of a hypothesis is to voice your expectation of what you will find as the relationship between two variables, based on knowledge and previous experience.
A good hypothesis is one that is specific, tentative, testable, measurable, and falsifiable.
A good research project design (either experimental or observational) is Specific, Measurable, Achievable, Relevant, and Time-bound.
When research results do not align with your hypothesis, that is interesting too!
References
Cohen, M. F. (2013). An introduction to logic and scientific method. Read Books Ltd.
Newman, G., M. Chandler, M. Clyde, B. McGreavy, M. Haklay, H. Ballard, S. Gray, R. Scarpino, R. Hauptfeld, D. Mellor, J. Gallo (2016). Leveraging the power of place in citizen science for effective conservation decision-making. Biological Conservation. 10pp. htpp://dx.doi.org/10.1016/j.biocon.2016.07.019.
Strode, Paul. (2020). Science: It’s Not Always Hypothetical. The American Biology Teacher. 82(8): 513. https://doi.org/10.1525/abt.2020.82.8.513
Teaching the Hypothesis. Paul K. Strode, November 2, 2014
The Writing Center (2017) How to Write a Research Question, George Mason University, https://writingcenter.gmu.edu/guides/how-to-write-a-research-question
Toledo, A. H., Flikkema, R., and L. H. Toledo-Pereyra (2011). Developing the Research Hypothesis. Journal of Investigative Surgery, 24:5, 191-194, DOI: 10.3109/08941939.2011.609449 https://doi.org/10.3109/08941939.2011.609449
Cover Photo: Sarah Newman
CitSci.org Blog 