Correlational Research – How and When to Use It

Time to read: 5 Minutes
Correlational-research-Definition

Correlational research is the study of relationships and is used to measure both strength and direction. Learn about correlational research and how to use it in your paper properly below.

Correlational Research – In a Nutshell

  • Correlational research is used to measure the statistical relationship of two (or more) variables.
  • Researchers tend to conduct correlational research when they don’t expect a causal relationship or when manipulation of variables is not possible.
  • You can apply many data-gathering methods, including in-person methods and secondary, archival data collection, to correlational research.

Definition: Correlational research

Correlational research examines the relationship between variables. It is done without manipulating any of the variables themselves and is therefore used to measure the strength or direction of a relationship if one is found.

The direction of this relationship can be positive, negative, or non-existent (zero).

Variables Example
Positive correlation All variables change in the same direction The more exercise you do, the more calories burnt.
Negative correlation Variable change in opposite directions The more you work, the less recreational time you have.
Zero correlation No relation between variables Tea consumption has no relationship to intelligence.

Correlational research vs. experimental research

Correlational research and experimental research are both used to study relationships with quantitative methods.1 However, there are some crucial differences:

Correlational research Experimental research
Purpose Test strength of relationship Test the cause and effect of a relationship
Variables Researcher observes variables without manipulation Researcher manipulates a variable to study its effects
Control Limited control that may not identify additional variables Other variables are controlled so that they can’t impact those being studied
Validity Has external validity so that conclusions can be generalised across other settings Has internal validity, so you can make strong conclusions on your subject(s)

When correlational research is used

Correlational research should be utilized for data gathering when looking for general results in real-world studies.

Example:

Correlational research can help determine whether variables are related within a group, which can then be applied to other situations.2

Investigating non-causal relationships

If you want to discover whether a relationship between two variables exists, but you’re not expecting it to be causal in nature, correlational research can be valuable.

Correlational research is particularly useful for recognizing general patterns, that can then help inform decisions.

Example:

You may investigate the political voting patterns of families that sit down for dinner together.

It’s unlikely that family dinners cause people to vote a particular way (non-causal), but it could be indicative of other ideological or religious issues that can be applied to voting predictions.

Exploring causal relationships between variables

Correlational research can be used to identify causation when experimental research is too costly or unethical.

It may provide the impetus for further research as a base dataset to help develop causal theories.

Example:

You’d like to study the relationship between sewage treatment and gastroenteritis.

While a strong causal relationship can be found, you cannot ethically control the effects of sewage treatment on a population.

Testing new measurement tools

If you’ve developed a new tool for measuring a variable and its effects, you can use correlational research to test its reliability.

Example:

You’ve created a new scale for measuring life satisfaction.

To test the scale, you collect data on life satisfaction using existing tools and your new test scale.

Correlation between the different measurements indicates reliability in the new scale.

Data collection methods in correlational research

There are many data collection methods to choose from in the social sciences. These include everything from surveys and observation methods to historical statistical data.

Choosing the correct method is crucial to creating a solid piece of research. Begin by choosing a representational sample that’s free from personal bias.

Surveys

Surveys are a common method for generating strong data and results. Commonly gathered through questionnaires, you can conduct this research in person or on the street, or with postal, phone, or online surveys where you can’t guide the subject(s).

Surveys are useful because they’re quick and flexible. Still, you must avoid bias when wording surveys and keep data relevant without guiding the answer(s).

Naturalistic observation

Naturalistic observation is a form of field research that involves studying real effects as they happen in real life. It can involve recording events like counting visitors to an exhibit or describing actions as they happen.

While naturalistic observation can be both qualitative and quantitative, the correlation should be analyzed through quantitative data.

Naturalistic observation is crucial for recording real-world events and context, but it can be extremely time-consuming. As it’s done personally, bias is also possible and must be addressed in the methodology outline.

Example:

You may want to study attention spans in college lectures based on gender.

You’d have to record all observations of students checking smartphones, for instance, as an indicator of engagement.

Secondary data

Secondary data is any data that you haven’t conducted yourself. This can be an extremely valuable and effective source of historical correlational research. The more comprehensive and official the data, the more reliable the research.

Example:

Census studies, historic polls, and other records are all potential data sources.

While inexpensive and ready-made, secondary data does lack reliability if you cannot check how data is gathered.

Example:

You could study the effects of alcohol consumption on the health of a population.

For this, you’d rely on historic consumption studies and historic hospital data.

When comparing this data across countries, you do risk comparing unequal data gathering methods, as not all countries record data similarly. However, you can still propose a general correlation.

How to analyze correlational data

Data alone isn’t capable of indicating correlation. You need to analyze it to elaborate on the relationship. You can do this through correlation or regression analysis.

Regression analysis

Regression analysis predicts how much change in one variable affects change in another.

This is visualized with a regression equation that describes the relationship on a graph. This equation can then be used to predict the value of a certain variable based on the value of another.

Only use regression analysis after correlation is confirmed

Correlation analysis

Correlation analysis is used to create a correlation coefficient, where the strength of a relationship and its direction are described by a number. This number expresses the degree or intensity of this correlation.

Most social studies utilize the Pearson product-moment coefficient, where an r is used to assess linear relationships.

FAQs

Correlational research is important because it identifies and provides insight into real-life relationships. This can be used to develop effective policies and theories, or conduct more thorough experimental studies.

Correlation is a mutual connection between variables that isn’t always causal.

At least two variables are traditionally studied in correlational research – but it’s also possible to study more than three.

Sources

1 Lau, Francis. “Methods for Correlational Studies.” in Handbook of eHealth Evaluation: An Evidence-based Approach, 213-27. Victoria, BC: University of Victoria, 2016. https://www.ncbi.nlm.nih.gov/books/NBK481590/pdf/Bookshelf_NBK481590.pdf.

2 Siegle, Del. “Introduction to Correlation Research” October 11, 2015. https://researchbasics.education.uconn.edu/correlation/.