As a researcher, you can only find out the accuracy of your research if there are no factors to dispute your finding. The confidence in the outcome is what is referred to as internal validity. In this post, we will explore the concept of internal validity, its importance, and how to test it.
Internal validity is the concept of how much confidence you have in the result of your research. This is because the lesser the possibility of confounding variables in research, the greater the internal validity and the more confident a researcher can be of the research.
What this means is that internal validity is the degree to which you can establish the cause-and-effect association between your treatment condition and your research result. It is how best you can avoid traps that can alter your result as a researcher.
Internal validity also shows that knowing the standard data, helps a researcher to take out insignificant results in the research. With an accurate selection of sample groups and correct measurement, the result of the relationship between the data will be valid.
There are 4 primary types of validity, which are
In this type of validity, the focus is on whether the test measured what it was intended to measure. This wants to check whether the objective of conducting the test was satisfied.
Content validity examines whether the test truly supports what it wants to test. This refers to the testing method to be used. Content validity checks if it is the most appropriate method for the study.
Face validity explains whether the content or data of the study can truly represent the aims of the test. This checks whether the data gathered is solid enough to conduct the test.
Criterion validity addresses whether the outcome of a test matches with the result of another test that has the same data. This means that when you conclude your test, it compares the result with the standard outcome.
What Internal validity does for research is it makes the researcher's findings credible. This is done when an experiment shows a cordial connection between two data. A high internal validity clears further doubts that may arise as it relates to the study.
An example is when you decide to run a hypothesis test on whether an apple a day improves good health. You then arrange for volunteers to take part in your research. You selected an even number of middle-aged people, grouped them, scheduled them in morning and evening participation groups. The morning group is the treatment group while the evening group is the control group.
To prove your hypothesis, you administer an apple each to the treatment group and give a carrot each to the control group. Also, both groups were tested to determine the state of their health. After collecting the test results and analyzing them, you realize the treatment group tested better than the control group. Is this enough to conclude that an apple a day improves good health?
To draw this conclusion, you must first cancel out other reasons that could explain your research result.
Internal validity is important to prove the trustworthiness of your research findings.
We can establish internal validity if the three necessary conditions are present. To identify internal validity in research, these three conditions must be present to determine if there is a relationship between your treatment (independent variables) and your control measures (dependent variables).
Now, if you look at the research example cited above, two conditions out of the three stated conditions that should be established in an internal validity have been satisfied. Let's look at how they apply in the example.
The number three condition that hasn't been met in the example is the time when the apple was eaten.
Because eating the apple happens in the morning, there might be the probability that the sleep through the night and the calmness of the morning influenced the improvement.
Therefore, you cannot be certain that eating an apple in the morning is solely responsible for the health improvement recorded in the treatment group.
The absence of this final conclusion means your research does not have high internal validity. Hence, you cannot confidently conclude on a cause-and-effect relationship between an apple and good health.
There are three categories of threats to internal validity. The three categories are;
Single group threats to internal validity refer to threats that can arise when your study treatment focuses on one group. This refers to the criticism or confusing variable that can arise when your research focuses on one sample group.
Multiple group threats to internal validity refer to threats and criticism that can arise when your study is focused on multiple groups.
Social threats to internal validity imply threats that occur because of the reaction of people to what affects them and others.
It is important to identify the threats to internal validity in a research study. This will help the researchers develop a control measure for the study.
Let's consider these 12 threats to internal validity;
1. Attrition: Attrition is bad for your research because it leads to a bias. When your participants leave your research, your findings will rely on those participants who decided otherwise about dropping out of the research. This will certainly affect the credibility of your research because you have lesser inference to draw from.
2. Confounding variables: When your research has an extra variable related to the treatment you applied to your sample group that affects your results, then that leads to confusion. Confounding variables reduces your study's internal validity.
3. Diffusion: This is a tricky one. Diffusion is when the treatment in research spreads from the group you administer treatment to the control group. This happens when there's interaction and observation among the two groups.
Diffusion poses a threat to your research because it can lead to what researchers call resentful demoralization. This is when the control group is less motivated because they resent the group they are in.
4. Researcher's bias: This is when the researcher behaves differently to one group. (This can be in favor of the group or harshly against the group). This bias from the researcher can affect the results of the study.
5. Historical events: A past event may directly or indirectly influence the result of research through the participants. Some things like natural disasters or politics can influence the coordination of the research participants and how they perform.
6. Instrumentation: You can guide participants of your research into acting in some type of ways through the research method to use. This might result in your participants acting in ways different from how they would initially have.
7. Maturation: This means that time should be considered an important variable in research. If your participants become older or went through a biological (time) change in the cause of your research, it may be difficult to prove that the results of the study were not influenced by time.
8. Statistical regression: This is a threat to internal validity as participants at extreme ends of treatment can naturally fall into a direction because time is passing by and not because of the treatment administered.
9. Repeated testing: Testing your research participants with repeated measures will influence your research findings. It is only natural that when you give a particular test to someone repeatedly, they'll become accustomed to the test.
10. Selection of subjects: This is the bias that may result from the selection of study groups. If your sample size is small, it can cause Simpson's paradox
11. Selection-maturation interaction: When you select the study group, the time difference between administering treatment to the groups can cause confounding results and a researcher may conclude that the treatment administered influenced the result. This can be a wrong conclusion.
12. John Henry effect: The John Henry effect refers to when a man named John Henry outperformed a machine in an experiment because he was aware that the experiment was between him and a machine.
Manipulating an experimental design can alter or reduce many threats to internal validity. Here are some ways this can be done:
Here are three examples of internal validity. We'll look at them one after the other. These examples show how due processes improve the chance of internal validity.
A researcher formed a hypothesis that word puzzle apps will counter negative thoughts. To test the hypothesis, the researcher selected participants randomly and equally distributed the participants into two groups. The treatment group (those that will engage the app) and the control group.
Besides randomly assigning participants, the researcher ensured the participants were unaware of the research objectives and the other test group.
An insurance company wants to find out if adopting flexible working hours will provide job satisfaction among its employees.
To know this, they selected two groups from their employees to participate in the experiment.
The first group being the treatment group with flexible working hours and the control group with fixed working hours.
The experiment is to run for 9 months. To get an accurate result of the research objectives, all participating employees filled a survey describing their working hours’ satisfaction before the test, representing the pre-test. They also filled a survey form to describe their experience after the test. Which is the post-test.
A study is to be carried out by a researcher. The aim of the research is to test whether having a stress ball on the office desk improves the productivity of the marketing team of an IT firm. The researcher handed a stress ball to all the employees participating in the study which is to last 3 months. All the participants were made to fill a survey form indicating their productivity level before the stress ball administering and after the research.
The examples cited describe how confounding and threats to internal validity can be reduced if due processes are followed.
We have been able to establish in this post that internal validity shows the credibility of a research finding by eliminating other explanations that can affect the result of the test.
You may also like:
Hypothesis testing is as old as the scientific method and is at the heart of the research process. Research exists to validate or disprove ...
Simpson’s paradox, also known as the Yule-Simpson effect, is a fascinating phenomenon that shows the importance of causal reasoning. The ...
Before undergoing a life-changing medical procedure, it's a patient's right to understand the risks involved, so they can adequately ...
As learning progresses, teachers use subtle methods to track how much students know and to discover any challenges learners have with the ...