Types of hypotheses

A hypothesis is a possible explanation for a research problem. In science, the research process proposes answers that may or may not be true, but must be verifiable.

There are different types of hypotheses, which depend on the type of research and the different authors. Below we present the different types of hypotheses according to Roberto Hernández Sampieri and his collaborators.

Types of hypothesesDefinitionExample
Research hypothesisAssumption that seeks to relate two or more variables.
  • Descriptive: “Admission to UNAM will grow 20% in 2021”
  • Correlational: “Depression in women increases at menopause”
  • Group difference: “Plants in greenhouses are more productive than in fields”
  • Causes: “Bullying and aggression among adolescents increases school dropout”
Null hypothesesAssumption opposed to the research hypothesis.“Bullying and aggression among adolescents does not cause higher school dropouts:
Alternative hypothesesOptional assumption different from the research hypothesis and the null hypothesis.“Bullying and aggression among adolescents causes less school dropout”
Statistical hypothesisExpression of the hypotheses in statistical terms.
  • Estimated:
    Income 2021 = Income 2020 + 20%
  • Correlation: r depression , menopause ≠ 0
  • Difference of means:
    % CDMX women ≠% Monterrey women

Research hypothesis

Research hypotheses attempt to explain how two or more variables are related. They are also called working hypotheses, since based on them the researcher conducts the activities to solve the research problem. They are symbolized as Hi.

Within the research hypotheses, four subtypes are distinguished: descriptive, correlational, group difference and causal.

Descriptive hypothesis

This hypothesis attempts to predict a data or value in one or more variables to be measured or observed.

For example : Hi: “The admission of students to the Autonomous University of Mexico in 2021 will be 20% higher than in 2020.”

Correlational hypotheses

These hypotheses seek to associate or relate two or more variables, without necessarily establishing which one causes the other. That is, when a variable is present, another variable also appears.

For example : Hi: “Depression in women increases at menopause . “

Group difference hypothesis

This hypothesis seeks to compare groups.

For example : Hi: “Plants grown in greenhouses produce more fruit than plants grown outdoors.”

Causal hypotheses

The hypotheses establish which variables have an effect on the others, that is, a variable X causes the variable Y.

For example : Hi: “Bullying and aggression among adolescents causes higher school dropouts”.

Null hypotheses

The null hypotheses are the counterpart of the research hypothesis. They serve to reject or refute the working hypothesis. They are symbolized as Ho.

For example : Ho: “Bullying and aggression among adolescents does not cause higher school dropouts.”

Alternative hypotheses

They are optional explanations other than the research and null hypotheses. They are symbolized as Ha.

For example : Ha: “Bullying and aggression among adolescents causes less school dropout”.

Statistical hypothesis

Statistical hypotheses represent in statistical and mathematical terms what are the null and alternative research hypotheses. They can be classified into: statistical hypotheses of estimation, of correlation and of difference between groups.


They are the statistical hypotheses that translate the descriptive hypotheses that a data predicts.

For example , the descriptive hypothesis “The entrance of students in the UNAM in 2021 will be 20% higher than the year 2020” is statistically translated as:
Hi: Income 2021 = income 2020 + 20%


The statistical hypothesis of correlation aims to translate a correlation between two or more variables into statistical terms. The symbol for a correlation between two variables is “r” (lowercase) and between more than two variables “R” (uppercase).

For example , the hypothesis “depression in women increases at menopause” is translated as:

Hi: r depression-menopause ≠ 0; the correlation between depression and menopause is different from zero.

Of mean differences

This hypothesis compares a statistic between two or more groups.

For example , we want to compare the percentage of women in Mexico City with the percentage of women in Buenos Aires:

  • Hi:% WomenCDMX ≠% WomenBA (the percentages of women are different)
  • Ho:% WomenCDMX =% WomenBA (there is no difference between the percentages of women)

It may interest you:

Add a Comment

Your email address will not be published. Required fields are marked *