Experimental design is the backbone of robust research. It's a systematic approach to testing a hypothesis, allowing you to establish cause-and-effect relationships with confidence.
It is a meticulous technique researchers use to create causal relationships between variables in their study and collect empirical evidence. They manipulate independent variables to establish and measure the extent of modifications in dependent variables.
Moreover, they thoughtfully plan and organise data collection conditions to check hypotheses’ validity and reveal valuable patterns. A good experimental design will consider all of the particular components of your research system. It develops trustworthy and relevant data for the study’s major challenges and is always context-oriented.
Whether you're a student, a budding scientist, or just curious about how things work, understanding the basics of experimental design is a valuable skill.
Specifically, if you need to “do my dissertation”, grasping how the experimental design works will prove to be a handy tool helping you in your research.
Experimental design is a group of procedures that enables you to change independent variable(s) while controlling the experiment’s conditions. It helps you examine causal relationships by systematically experimenting with a thesis statement or argument.
You need to first comprehend the system you are analysing to create a good experimental design.
Let us give you an example to help you easily understand it. Suppose you need to test a new fertiliser for plants. You take two groups of plants and apply it to one of them to examine the effect of the fertiliser.
Then, you found the growth in the first group of plants and no growth in the latter. Sounds like it was the impact of the fertiliser. Right?
But here is the catch!
The second group of plants were in a shadier spot.
Now, how can you be sure that the fertiliser and not the light caused differences in growth?
This is where experimental research design comes in.
It helps you minimise bias and isolate the true effect of what you're testing.
You run any experiment for a reason. So, ask yourself what you want to achieve through this experiment. How will you get an answer to this question?
Firstly, determine independent and dependent variables. After defining them, think of ways to control them in your experiment.
Some people believe that exercise is related to our moods. You will make that into a research question that says, “Does regular exercise improve mood in adults?”
Changes in teaching methods sometimes impact students' performance. You want to analyse it. Thus, you create this research question: “Does the type of teaching method (traditional lecture vs. flipped classroom) affect students’ performance in mathematics?”
Here are the independent and dependent variables for both.
| Research Question Examples | Dependent Variables | Independent Variables | Extraneous Variables |
|---|---|---|---|
| Exercise and Mood | Self-reported mood ratings | Duration and intensity of regular exercise | Exercise at different times of the day |
| Teaching methods and students' performance | Grades in the maths course | Traditional lecture-based teaching style | Support available at home |
In the preceding step, you have determined all the variables. Now it's time to craft a specific hypothesis that can be tested according to your research question.
| Research Question | Null Hypothesis (H0) | Alternate Hypothesis (H1) |
|---|---|---|
| Exercise and Mood | Exercising daily does not correlate with the changes in mood in adults | Regular exercise boosts mood in adults |
| Teaching methods and students' performance | Teaching methods do not impact students' performance in maths exams | Changes in teaching methods can improve or lower students’ performance in maths exams |
You have determined the hypothesis for your research questions. Now you will need to create a controlled experiment. Here are the essential elements to consider:
Here, you decide how much you will manipulate your independent variable, as it may impact your experiment’s external validity.
Now, you may ask, “What is external validity?”
Well, it means to what degree you can apply findings from a certain research to other situations, events or groups.
Let's look at both of our examples and how we can vary the independent variable in those situations.
You can vary regular exercise to:
Moreover, you might have to pick how granularly you need to manipulate your independent variable. Sometimes, your experimental system selects it for you, but mostly you will have to choose, and this will influence the extent to which you can draw conclusions from your findings.
You can look at teaching methods as:
You need authentic and trustworthy findings for your experiment to be successful. And for that, you need to understand how to implement your experimental treatments to your study’s participants or test subjects.
The first step is to look at the sample size — the number of test subjects you have for your experiment. Generally speaking, your test’s statistical strength directly correlates with your study’s participants, which in turn establishes the amount of trust you can have in the outcomes.
The second step is to add your participants to different experiment groups randomly. You will treat each group differently (like no change in teaching method, slight change in teaching method, extensive change in teaching method)
Also, don't forget to add a control group which gets no treatment. It determines the results without any experimental treatment.
Further, you need to make two primary decisions while assigning your study’s participants to groups.
You can randomise a test completely or within blocks, or strata.
Fully Randomised Design: You assign every participant or subject to a treatment sample haphazardly.
Stratified Random Design or Randomised Block Design: You first classify subjects considering a trait they have in common and then arbitrarily allot them to treatments within those samples.
| Research Topic | Design Type | Description |
|---|---|---|
| Exercise and Mood | Fully Randomised Design |
|
| Randomised Block Design |
|
|
| Teaching Methods and Students' Performance | Completely Randomised Design |
|
| Randomised Block Design |
|
Randomisation isn't always moral or doable, thus scholars develop non-random or partially random designs. An experiment with a non-random design is known as a quasi-experimental design.
People get just one of the potential extents of an experimental treatment in a between-subjects design (also called a classic ANOVA design or an independent measures design).
If you are studying social sciences or something related to healthcare, employ matched pairs within your between-subjects design so that every treatment group has a similar range of participants or experiment subjects in the same amounts.
Every person gets each of the treatments in the test consecutively while you note their reactions to each treatment in a within-subjects design (also called repeated measures design).
In this experimental design, an impact happens gradually, and the subject’s reactions are gauged gradually to note this impact as it happens.
Counterbalancing refers to randomising or back-pedalling the order of treatments among participants or subjects. It is typically utilised in within-subjects designs so that the order of treatment implementation doesn’t impact the findings of the test.
| Research Topic | Between-Subjects Design | Within-Subjects Design |
|---|---|---|
| Exercise and Mood | You arbitrarily allot an intensity of exercise (none, less, or extensive) and adhere to that intensity of exercise throughout the test. | You arbitrarily allot participants repeatedly to none, less, and extensive intensities of exercise throughout the test and randomise the order in which they adhere to these treatments. |
| Teaching Methods and Students' Performance | You arbitrarily allot teaching method treatments to students, and you keep getting students taught using these methods throughout the test. | You get every student taught using each teaching method treatment (no change, slight change, extreme change) repeatedly during the whole test, and you randomise the order in which they get taught through those methods. |
Lastly, you want to choose the methods you will use to gather data on your dependent variable results. You need to get authentic and trustworthy results that reduce error or bias to a minimum.
You can objectively consider some variables, such as the teaching method, with students’ opinions. You might need to operationalise others to convert them into measurable observations.
You can check the performance of your dependent variable in one of two ways in your experiment about exercise and mood.
You want to gauge your dependent variable accurately, as it impacts the types of statistical analysis you could use on your data.
Experiments always hinge on their contexts, so a decent experimental design will consider all of the distinct aspects of your research system to create data that is both authentic and related to your thesis statement or hypothesis.
Our academic experts understand the nitty-gritty of experimental design as they have completed research in their respective fields. They have helped write this blog using their knowledge and expertise. They also prepare different types of academic papers for students in the UK, including HND assignments. Their authentic and research-based content has helped students achieve their academic goals.
