Experimentation is a research method through which knowledge is derived or verified by means of performing experiments and analyzing their outcomes. Researchers often perform experiments to test hypotheses, check the possible deviations of a theory, or as an alternative to a theoretical approach. At times, they are also performed to derive or understand relationships between variables in a study.

Here are some circumstances where you might choose to design an experiment for your research project:

  • The research question or question requires an experiment to demonstrate or validate the hypothesis  
  • Your area of research demands you to use lab work as there is little or use for theoretical work 
  • A theory may require more empirical evidence to back it up 
  • Existing research may no longer be valid and needs to be updated 
  • You may just prefer to test your hypothesis with an experiment


Types of experiments

Depending on the area of research, experiments are categorized as social or non-social. The primary difference between these two experiments is who the subject of study is. But more on that later. 

Social experiments are done on humans or with human beings as the sample set. Surveys and behavioral studies are types of social experiments that are designed to understand how humans act or think when subjected to certain conditions. Social experiments are more likely to be found within disciplines such as psychology, sociology, economics, and other social sciences. 

Since the emphasis here is on understanding human behavior, there are many ethical considerations involved. If you plan on conducting a social experiment, you must seek approval from your supervisor and the university’s ethics committee (or its equivalent) before proceeding with the experiment. Prepare a consent form for your participants.. 

A non-social experiment, as you might have guessed, is one that does not involve human participants. This type of experiment is typically done in a lab, in order to test physical phenomena. Although they don’t involve humans, it is nevertheless important to adhere to ethical practices and norms followed in your field of study. 


How to design and conduct an experiment 

Regardless of whether you are conducting a social or non-social experiment, there are some basic steps common to both. Let’s take a look at the elements an experiment should contain: 

A well-defined and testable objective

As a rule of thumb, it is really important to be clear about the objectives of the experiment you are about to perform and how it helps you arrive at a conclusion about your hypothesis. Having a clearly well-defined objective helps you consider approaches to how you can solve the research problem and what variables will help you do that. 

More specifically, this helps you identify the independent and dependent variables that guide your experiment.

Designing the setup

Once you have identified the kind of experiments you can perform, the next step is to devise a step-by-step strategy to perform the experiment. While devising the strategy, ensure that the design of the experiment would involve the least possible error. It becomes easy to strategize the experiment by clearly defining the variables that you would like to obtain experimentally and the ones that can be assumed. 

For example, if you are assessing the health risks of serious Covid-19 patients by their average income, important variables you are likely to consider are the number of people admitted, their yearly income, and the proportion of the total number of people who survived. In such an experiment, how you choose to define the term “serious” in this particular context also helps you determine variables. 

Always remember that it’s the variables that impact the conclusion of your study. So it goes without saying that you should choose them carefully and then strategically plan your experiment. 


The protocols of the experiment 

Once you have determined the variables of the experiment, you must devise the protocols (the process) you will be following while conducting the experiment. It’s important to have this planned thoroughly since deviations are likely to make your observations prone to errors.  


Data tabulation 

While this step may seem very obvious, it’s a step that’s surprisingly easy to miss! Note down data points as clearly and precisely as possible. Keeping a detailed record of your observations makes your conclusions more substantial, which eases the dissertation writing process tremendously. 

On an excel sheet, clearly mention the variables and the unit that you are using to study it. It’s also good practice to note down the date and time of observations taken. 

It’s also a good idea to use more visually appealing formats such as pie-chart graphs, bar graphs, histograms, and line graphs to showcase data. While a table does cover the data you have gathered, these formats show the bigger picture, and it’s helpful in deriving meaningful inferences. Trends, and connections from the data points. 


Curve fitting 

When you design an experiment, you may start having a vague idea of what model or patterns you’re likely to see in your observations. While testing a hypothesis with an experiment, the general procedure is to fit the data points with a geometrical function. For example, if you’ve studied crime rate in relation to the population of a city and you’ve discovered that they are directly proportional, you can use a linear curve to show the results. 

There are various softwares that help us to determine the best fit curve. For example, MS Excel has various options such as linear, polynomial, and exponential models to help us determine the best fit of the data. The element of curve fitting helps us to validate our hypothesis and also gives us an estimate of the amount of deviation we may have from the results obtained.


A lab diary

A lab diary (or lab notebook) is a rough and informal record of the process and data collected in the experiment. A lab diary has details about the primary observations, your thoughts of the experiment, an account of obstacles you have encountered in the way, and the error. This record not only helps you connect dots later on but also serves as an aid for your memory (especially when you’re writing up the dissertation). You may also, through the course of your experiment, come up with related ideas that you can’t work on immediately. Recording these helps you later on, if you eventually decide to work on them!  

Did you know that a lab diary is also proof of intellectual property? They’re often used to counter plagiarism charges, as proof of original research work! 


Fake results: don’t fudge data! 

Imagine this: A mobile phone company, before launching their new phone model, commissions a study to see how well the phone receives telecom signals and the results show that the phone catches the signal weakly. But since they now have to sell their huge stock of phones, they report that the phone signals are excellent. Do you think they are doing a good job? 

It’s important to remember that the results of an experiment must be reported the way they were obtained and should not be changed for any reason. Fudging data goes against scientific integrity, is considered an unethical practice, and ultimately compromises the knowledge you are trying to gain! Data manipulation is also something you should avoid and can be easily found out if a fellow researcher decides to replicate your experiment

It’s important to be honest with the results obtained in the experiment and not manipulate them for any reason whatsoever. It’s tempting to do this if your results contradict your original hypothesis. If this happens, you’re better off explaining the deviations and why they happened, rather than producing false results. 

This kind of contradictory result is perfectly acceptable in scientific research, because the ultimate aim is to find new knowledge, however you discover it. 

One way of dealing with such “erroneous” data points is by discarding them from plotting, after you discuss the same thoroughly with your dissertation supervisor/guide. This is common practice, especially when a few out of many data points do not align with the trends that your data is showing. But even after this, you must clearly specify this in your dissertation and justify why you discarded them. 


Errors and deviations

No result obtained in any experiment is 100% accurate. In fact, there is always a window of deviation in the obtained result. Due to human error, changing circumstances, and other unavoidable factors, it’s almost impossible to replicate a study exactly as the original experiment. This possibility of error is also something you have to account for in the dissertation. Describe how the results differ from the conclusion, and what might have led to that. 

For example, let’s say that you’re doing an experiment that examines how temperature affects the shelf life of milk kept in the open and you conclude that milk above 30oC goes bad after being kept for 6 hours. But if someone else were to do the same experiment at a different time, the results may vary: perhaps it will take 4 hours, or even 8. 

Such a window provides a detailed meaning to our experiment and also establishes its validity over a range of values. It is possible that certain variables of our experiment may be highly prone to errors than the others. While designing the experiments, make sure that these kinds of variables are measured with due care.

Legal bindings

Legal binding is the limitation that the law puts on the extent of the experiment being performed. Before you perform an experiment, ensure that you are not breaking any laws. If your experiment involves animals or humans, for instance, you must assess any legal consequences it might have. 

If you’re not too sure about what this might mean for your experiment, it’s always a good idea to have a discussion about this with your supervisor before you move forward with the experiment.