What We Accept

Make your hypothesis before you start collecting or analyzing data, not after! Otherwise, your results can influence your hypothesis (which isn't honest). 

You should also be honest when you evaluate your hypothesis after collecting data - it’s ok to find out your hypothesis was not supported by your data, but it’s not ok to be dishonest about it or to change your hypothesis to match your data.  

Unlike some journals, we don’t require your research to be completely new to science. However, you must honestly not know the “answer” to your hypothesis before you start, based on the background information you have. For example, “We hypothesized the world is round” is not an original-to-you hypothesis. 

You also must come up with the hypothesis and experiments yourself, not because someone assigned it to you (as part of an in-class lab, for example). 

Your hypothesis should clearly state your prediction: if you change a certain variable, what do you think will happen? If your hypothesis includes a comparison, you should state what you’re comparing to. 

You must also be able to test your hypothesis using your experiments and analysis. For example, let’s say you want to know whether planting a certain species of grass will decrease global warming. Unless you can actually measure global warming with your experiments, you would need to make your hypothesis more specific: “We hypothesize that X species of grass can reduce atmospheric carbon dioxide concentrations.” 

Your hypothesis should also include a measurable outcome; for example instead of saying “plants will grow better”, you can say, “plants will grow taller and gain more biomass”. 

For eiRxiv, we ask that you start your hypothesis with “We hypothesized..” For example, “We hypothesize that the type of leavening agent used will impact cookie density.”

Some people also find “if/then” statements to be helpful. For example, “We hypothesize that if pine tree saplings are grown in soil with and without added peat moss, then those grown in peat moss will grow taller.” 

Another way to assess whether your research is hypothesis-driven is by analyzing the experimental setup. What variables in the experiment are independent, and which are dependent? Do the results of the dependent variable answer the scientific question? Are there positive and negative control groups?

Your hypothesis needs to be specific to your research, but not so specific or complex that it becomes hard to evaluate. 

Too broad: “We hypothesize sugar will impact cell growth.” 

What kind of cells? What type of impact? How is growth being measured?

Too narrow: “We hypothesize adding 50% more sugar to the environment will cause Clostridium difficile cells to growth 75% bigger than they would with just water.” 

What if the cells grow 76% bigger? Does that still count? What about other concentrations of sugar? 

Too complex: “We hypothesize increasing environmental sugar will increase the rate of cell division and that cells dividing faster will have higher rates of mutation.” 

Hypothesis statements that contain words like “and” and “or” are ‘compound hypotheses’. What if one part of the hypothesis is supported, but the other isn’t? If you have a compound hypothesis, make sure every experiment you do addresses all parts of the hypothesis. 

Too explanatory: “We hypothesize increasing environmental sugar availability will increase the rate of cell division for Clostridium difficile because the cells will have more fuel for metabolism.”

This is another kind of compound hypothesis. Unless you are explicitly testing the “because”–in this case, whether sugar drives metabolism for this species, and whether metabolism rate drives the rate of cell division for this species–then it’s best to leave out the “because” section. Instead, you can use your Discussion section to propose ideas for why you got your results. 

Just right: “We hypothesize increasing environmental sugar availability will increase the rate of cell division for Clostridium difficile.” 

This hypothesis states what will be measured (rate of cell division), what will be manipulated (environmental sugar availability), and what the researchers predict will happen (increased rate). It is not too broad or too narrow. It is a single hypothesis and does not include an explanation as part of the hypothesis. 

Descriptive research collects information or describes observations without a particular hypothesis in mind. This can include environmental surveys, new species descriptions, inventions, and more. 

Discovery research analyzes large datasets to find new patterns or correlations. This can include data mining, machine learning, molecular docking studies, and more.

Although these types of research can provide background knowledge for creating hypothesis-focused experiments or techniques, they are outside of eiRxiv’s scope because they do not focus on testing hypotheses. 

That said, eiRxiv would accept manuscripts where authors make a hypothesis before they start their research and then use surveys, observations, inventions, algorithms, data analysis, etc. to test it. 

eiRxiv does not accept manuscripts that just introduce, optimize, or compare an invention, a method, a machine/deep learning or AI algorithm, or other similar tool or strategy. 

Common examples of invention, method, or algorithm hypotheses that would not be accepted include: 

• I hypothesize that my invention/method/machine learning model will work
• I hypothesize that I can build this invention/method/machine learning model
• I hypothesize that my machine/deep learning or AI model will be effective and yield accurate results
• I hypothesize that my machine/deep learning or AI model will be more effective than previous/other models
• I hypothesize that my machine/deep learning or AI model will be more effective when changing a particular variable or factor

In all of these hypotheses, the invention/method/algorithm is the subject of the hypothesis. 

However, you can use new inventions, methods, or algorithms to test a hypothesis: For example: 

• Unacceptable: “We hypothesize machine learning can be used to predict bone spur severity" would not be accepted.  
• Acceptable: "We hypothesize childhood calcium consumption impacts bone spur severity", and the researchers then use machine learning to test this hypothesis. 

In short, you can use a new or optimized invention, method, or ML/AI to test a hypothesis, but it should not be the subject of your hypothesis. 

The hypothesis should be about the experimental system, not about the scientist. 

For example, “We hypothesized that we could build a model to predict how well different fabrics absorb sound” is about what the researchers can build. 

Similarly, “We hypothesized that determining whether Quercus albus has genes for iron metabolism would help us better predict local soil quality” is about how the researchers would take the data and use it. 

Sometimes, these hypotheses can be changed to be about the science and experiment instead (for example, “We hypothesized that materials with a lower density to depth ratio would absorb sound better”). 

If you’re struggling to state the hypothesis without mentioning the scientists, a model, or a method, this may be a signal that you have a hypothesis that is invention-based or discovery research.