About experiments

In social sciences, experiments are associated chiefly with social psychology. However, they can also be used in other disciplines like economics, political science, sociology, or educational sciences.

An experiment is a procedure that aims to investigate something by inducing change. Depending on its scope and type, an experiment can either investigate a cause-and-effect relationship or explore something that has not been previously explored. Based on logical reasoning and a solid procedure that can be repeated, conclusions are formed. Usually, an experiment includes an intervention induced by researchers to investigate its consequences. Then, observation researchers determine if it was their manipulation or other factors that generated the change. There are two sides to an experiment: the researchers and what they do, and the reaction that the experiment has provoked. The former is called an independent variable, and the latter is called dependent variable. Thus:

  • Independent variable: a variable that a researcher manipulates to induce a specific reaction or test a hypothesis. It is independent from the research participants. It is possible to have more than one independent variable.
  • Dependent variable: a variable that changes due to the independent variable. The researcher measures it.

While designing your research experiment, it is essential to consider whether the independent variable influences the dependent variable. Experiments should have high:

  • Validity – an experiment should measure what it was supposed to measure, i.e. nothing but the independent variable should affect it. Validity can be secured by controlling all conditions of the experiment. However, in social sciences it is not easy to control all the conditions. For example, in the educational sciences academic progress can be measured, but for example it is impossible to maintain a stable family influence which can affect knowledge acquisition.
  • Reliability – the measurement is consistent and does not change over time (if the experiment is repeated).

There are two types of experiments:

  • Laboratory experiments – when all the conditions are controlled. Those experiments are undertaken in a lab, or similar conditions, provided that they are controllable. Due to the controlled environment, the laboratory experiments have high validity. However, as the experiment is conducted in a laboratory, it tends to be unnatural for participants.
  • Field experiments – are conducted in an environment that is natural for research participants (at school, at work etc.). Due to this, research participants feel more at ease about the experiment and behave naturally. However, the researchers have limited opportunities to control the experiment setting.

New means of communication made it possible to develop online experiments. Online experiments allow researchers to gain access to more participants compared to the traditional experiments, and use online data collection techniques. Huber and Gajos (2020) list benefits of conducting experiments online, such as access to a diverse pool of respondents with different demographic profiles, low cost of research, and fast tracking verification of hypotheses. When it comes to challenges, they point to ensuring reliability and control of the environment.

Experiments online are not significantly different from those conducted offline. Hooley, Marriot and Wellens (2012) indicate that both laboratory and field experiment can be conducted online. However, researchers should consider three critical issues:

  • Research context: researchers should be aware of the characteristics of their research sample. For example, they can invite people to visit their website and use the tools available (e.g., a virtual lab). It is also possible to perform a field experiment online (e.g. on FB, Instagram, TikTok). Researchers should always pay attention to the ethical aspects, as the rights and privacy online also need protection.
  • Choosing tools: you can design your research tool yourself, if you have relevant skills. However, many tools are already available (some examples are given in the "Additional readings" section below). Sometimes sophisticated tools are not necessary to design an experiment, just a clever idea, Google Docs or Digital Research Tolls Wiki will do.
  • Recruitment: you have to find respondents ready to participate in your experiment without remuneration, or find an alternative way to thank them, like providing feedback after completing the task, or issuing a certificate of participation. Remember to consider methods of ensuring control of the environment, e.g. check whether respondents read the experiment instruction.

The considerations listed above might create an impression that experiments are difficult and complex. However, even if you do not study psychology, you still can try your skills and design a simple experiment. In the following part of the module, you will learn more about:

  • using experiments in the practice of scientific research,
  • experiments that you can replicate,
  • experiments in which you can take part.