The basic-science hypothesis
The hypothesis of a basic-science paper is written as an explanation, in the present tense and with confidence. A sentence called a hypothesis can turn out to be a prediction, a claim about impact, or a hypothesis stated for research that has none, and a writer or editor can judge which.
What a hypothesis is
In basic science a hypothesis is a statement of explanation for an observation, which François Jacob described as the invention of a possible world. It guides experiments and is supported or rejected by their outcomes. The hypothesis is a key concept in the scientific method, which is a continuous process: a hypothesis informs predictions, predictions inform experiments, experiments produce observations, and observations modify the hypothesis.
In a worked case, bacterium X makes mice sick and many of its genes appear to code for toxins. A broad hypothesis is that toxins made by bacterium X make mice sick. A more specific and more directly testable one is that gene A, a suspected toxin gene, produces a toxin that makes mice sick. One prediction is that without gene A the bacterium will not make mice sick, the experiment is to inactivate gene A and infect mice with the altered bacterium, and the hypothesis is supported if the mice stay well.
Basic research deals with how things work, and clinical research aims to draw inferences from its findings. All hypotheses, basic or clinical, arguably share the same anatomy and essential purpose.
Tense and confidence
A hypothesis explains something happening here and now, so it is written in the present tense. In a good example from the Journal of Bacteriology, the authors hypothesise that S. aureus "utilizes" fatty acids present within lipoprotein particles, then describe the experiment that follows from that explanation. A second example from the same journal moves from the observation that the known SpoVG-binding sites share no obvious consensus sequence to the hypothesis that SpoVG "may interact" with nucleic acid structural motifs instead.
A hypothesis should also be stated with confidence, regardless of the outcome. The modal verb "may" in the second example can express possibility, but it can also communicate a lack of confidence. Where the authors do not appear confident in their hypothesis, the reader has less reason to trust it, the experimental design or the results. A paper in PLoS Pathogens hypothesises that a cytokine's role "may critically depend" on a strain's resistance to neutrophil phagocytosis, and "depends critically" would convey confidence. Hedging is appropriate when drawing conclusions from results, but a hypothesis is made to be tested, and disproving one is an important part of the scientific process and nothing to fear. When authors tell their story with confidence, the reader has more confidence in them and their data, and so in how they interpret their results.
Hypothesis and prediction
Many authors phrase their hypothesis as a prediction, although predictions are made and experiments performed on the basis of a hypothesis, which is itself based on observations. A passage from the Journal of Bacteriology says that the effects of ybeY loss on cellular RNAs are well documented in other bacteria, and so "we hypothesized that deletion of ybeY would lead to changes in mRNA levels in B. abortus". It moves from observation to prediction to method and arguably contains no hypothesis at all. An editor can turn the prediction into a hypothesis: the revision "we hypothesized that YbeY modulates mRNA levels in B. abortus" is inferred from the stated observation, prediction and approach, and the author should be asked to confirm that inference.
In a paper in Free Radical Biology and Medicine, the statement that cells unable to repair oxidative lesions "would be more sensitive" to paraquat is a prediction, as "would be" shows. It stands in for a hypothesis stated only later in the paper, that paraquat mutagenesis depends on the generation of reactive oxygen species and oxidative-stress-induced DNA damage, which is easy to identify as the hypothesis because it describes a mechanism by which paraquat damages cells. Stating that hypothesis first and deriving the prediction from it helps guide the reader through the narrative.
Claims about impact
Authors often hypothesise about a study's impact. An abstract in Medical Hypotheses on Duchenne muscular dystrophy hypothesises that precise CRISPR-Cas9 editing of induced pluripotent stem cells, coupled with myogenic progenitor differentiation and autologous transplantation, "can lead to safe and effective muscle repair". That is not an explanation but a prediction about the outcome of the proposed work, and one strategy is to remove the term hypothesis and introduce it as "we predict". The abstract goes on to say that the hypothesis may provide an optimal treatment approach, which is illogical: a hypothesis cannot provide a therapy, though the experimental results may guide research toward one.
Research without a hypothesis
A hypothesis cannot always be stated, as in hypothesis-generating research. An unwritten and undeserved rule seems to hold that such work is not valuable, which possibly prompts authors to state a hypothesis that is not really there. Authors in Microbiome who surveyed the caecal microbiome of chickens stated the hypothesis that its microflora would vary significantly between breeds and lines. The authors did the work to make an observation, to define a phenotype that can be investigated further, so their hypothesis amounts to saying that they will make an observation, which goes without saying. Only after observing could they generate a hypothesis to test.
Editorial help with the hypothesis
PhD-level investigators often struggle to write a hypothesis, although their research is inherently hypothesis driven. Deficits in mentoring and in graduate-level education in research methods, and lax publishing standards, could explain this. Writers and editors cannot fix those causes at the root, but they can help investigators state their hypotheses clearly and concisely.
If the hypothesis is important to people doing research, it should be important to people writing about it. A writer or editor can usually infer the author's hypothesis and revise the text, with a comment asking the author to check it, and one who is uncomfortable with the subject can still restructure the text so that the author fills the gaps. Writers are not experts in every subject, but they can judge whether a stated hypothesis is really a hypothesis, independently of whether it is a good one, and this skill can go a long way in helping authors report clear, robust research to their audience.