We explain what types of hypotheses exist and the characteristics of descriptive, causal, correlational and more.
What is a hypothesis?
A hypothesis is a proposition or statement that we wish to corroborate or contradict, through a research. In other words, a hypothesis is a idea that we presuppose and that we wish to subject to the rigor of a research method, as is the case of scientific method, for example, or that we wish to contrast by means of the experience.
The hypotheses are provisional, tentative statements that may or may not turn out to be true and demonstrable, but that initially serve us to establish what it is that we want to investigate and allow us to find our ideal verification method. That is why it is said that the hypothesis is the link between the theory and the observation. All research, therefore, necessarily begins with the formulation of a hypothesis.
However, it is possible that an investigation raises more than one hypothesis, and that these are of a different nature. Of course, some of them will turn out to be valid (when verified), while others will turn out to be invalid (when refuted). But next we will see a more or less general classification of the hypotheses.
Types of hypotheses
descriptive hypotheses
Those that establish the relationship between the variables that are being studied, without worrying about their causes and without making comparisons between them. They are limited, as their name indicates, to describe and anticipate the variables, values and qualities of matter.
As an example, suppose a group of scientists studies the recurrence of a disease in the population of their country. They decide, as a working hypothesis, to assume that the disease is equally distributed among all the ethnic groups that make up the total population, but as they complete their research, they realize that some ethnic groups are more affected than others.
Correlational hypotheses
Also called joint variation, which, as its name indicates, propose a correlation between the variables studied, that is, they state the way and the degree in which one affects the other. Depending on how this relationship is, these hypotheses can be of three types:
- Positive correlation hypothesis, when the increase in one variable brings with it the increase in the other. For example, if the scientists who study the disease propose that the older the patients, the greater the possibility of death when infected.
- Negative correlation hypothesis, when the decrease in one variable brings with it the decrease in the other. For example, if scientists studying the disease propose that there are fewer infected patients when the age of the population is lower.
- Mixed correlation hypothesis, when the increase or decrease of one variable brings with it the decrease or increase, respectively, of the other. For example, if scientists studying the disease propose that earlier treatment leads to fewer deaths from the disease.
Causal hypotheses
Predictive hypotheses project the cause and effect relationship into the future.
Those who explore the relationship cause effect between the variables studied, proposing some type of specific meaning. According to how this sense is, we can speak of:
- Explanatory hypotheses, which propose a verifiable cause and effect relationship between the variables, such that one can be explained by the other.For example, returning to the case of the disease that scientists are studying, once it has been verified that it does not afflict all ethnic groups equally, the hypothesis could be raised that the disease affects more people of a certain ethnicity because they have greater abundance of a specific protein in the blood.
- Predictive hypotheses, which pose a probable cause and effect relationship between the study variables, projecting it into the future. For example, again with the case of the disease studied, scientists could hypothesize that the greater affectation of certain sectors of the population will soon cause a change in the genetics of the infectious agent.
Statistical assumptions
Those that refer to sets of variables and express their relationships in percentage or proportional terms, instead of absolute terms. They are very common in probabilistic, population or predictive studies. This type of hypothesis can be classified, at the same time, in:
- Statistical estimation hypotheses, which allow the researcher to evaluate the value of some statistical variable for a population and a set of previous information. For example, if the scientists investigating the disease state that, of the infected patients, 70% present a certain symptom, so this should be considered a main symptom.
- Statistical correlation hypotheses, which seek to establish in statistical terms some correlation between the variables. For example, if the scientists who investigate the disease consider that its mortality has to do mainly with the socioeconomic level of the patients, since 80% of the serious cases come from popular neighborhoods.
- Statistical hypotheses of differentiation of means, which pose a relationship between the statistics of two human groups.For example, if scientists who study the disease consider that men are 40% more likely than women to suffer from it.
null hypotheses
A null hypothesis is one that refutes what is established in a research hypothesis, be the latter of any type. Therefore, the null hypotheses are the reverse of the research hypotheses, and can be of the same type as any of them (any of those we have listed so far).
For example, if scientists studying the disease seek to demonstrate that the severity of the disease has nothing to do with the sex of the patients.
Inductive, deductive and analogical hypotheses
Any of the above hypotheses can be inductive, deductive either analogous, based on the logic used to establish the relationship between the variables studied. This is expressed in the very way of presenting the relationship, as follows:
- Deductive hypotheses or hypotheses that operate by deduction, those that pose a relationship from the general to the particular, using as a starting point other previous hypotheses that have already been demonstrated. For example, if the scientists who study the disease verify that it affects a certain ethnic group more than another, they can then deduce that it affects more those who present a certain genetic component, since the latter is dominant in said ethnic group.
- Inductive hypotheses or hypotheses that operate by induction, those that pose a relationship from the particular to the general, that is, contrary to the deductive ones, based on the intuition from what is observed. For example, if scientists studying the disease do not find any serious cases among people of a certain ethnic group, they can argue that there is a genetic component in it that makes it immune.
- Analogous hypotheses or that operate by analogy, those that pose a relationship between the variables inspired, copied or transferred from another field of knowledge in which it was verified. That is, they assume that if said hypothesis was valid in another field, it can also be valid in theirs. For example, if scientists studying disease posit that since a different but similar disease was treated with a specific antibiotic, it is possible that this new disease will respond in the same way.