IA Tips: How to explain your…DESIGN

Travis DixonResearch Methodology

Follow these three simple steps for an easy explanation of your design.

You can write an excellent explanation of your “design” in just three sentences. You can’t go wrong with What-How-Why (State-Describe-Explain) approach. You can explain any section of the Exploration by showing how it controls for one or more confounding variables

Here are 3 steps to explaining your design:

1. State the design you used.
2. Summarize how it was applied.
3. Give a reason why you used this design.

Step 1. What design did you use?

• Repeated measures
• Independent samples
• Matched pairs*

You’ve most likely used independent samples, since this is the most popular. However, there are often good reasons for choosing a repeated measures. If you’re not sure you can read more about the design types in this post.

Don’t make the mistake of saying your design was “an experiment.” This is a common mistake that loses marks.

Step 2. How did you apply it?

Make it clear that what you’ve said you’ve done is what you’ve actually done.

As an examiner, I want to make sure what students say they have done is what they’ve actually done. Be sure to give a one or two sentence summary of how you applied the design.

If it’s independent samples, make it clear that half your participants were in one condition and the other half were in the other. You can also explain how you allocated your participants (e.g. random allocation) if you used this.

Alternatively, if repeated measures was used make it clear that all your participants experienced both conditions. You probably applied counterbalancing too, so don’t forget to explain this.

Step 3. Why?

To explain the design identify the potential confounding variable you’re controlling for.

• Independent samples = order effects
• Repeated measures = participant variability

Order effects occur when doing one condition of an experiment first affects your results in the second condition. Counterbalancing controls for this in repeated measures, but so does independent samples. Make it clear that you understand your results would be affected if participants did both conditions.

Participant variability refers to the fact that not all participants are identical. The groups might end up naturally different in some regard. For example, you might have older participants in one condition and younger in the other, even though age is not an independent variable you’re studying. Repeated measures controls for this.

Random Allocation: This is a key component of any experiment. If you’re using independent samples then random allocation is straightforward. For repeated measures, you can randomly allocate the order of conditions your participants experience (e.g. randomly select who goes Condition A then B, and who has B then A). This helps control for participant variability and researcher bias.

Example

Here’s an example taken from my IA example in the IA Teacher Support Pack:

We used an independent samples design and had two different groups of participants do each condition. We did this to control for the participant expectancy effect because if participants were to watch both videos and complete the questionnaires, they could have easily figured out the purpose of the study and their answers may have been affected. We could have done repeated measures with each condition on different days, but then there could have been other factors that might affect body dissatisfaction on different days, like what they had done that day or their mood.

Here’s another example based on a fictional memory enhancing drug I called “Rememberol.”

Clearly state the design you used, how you applied that design and why.

Remember that the Exploration section of the IA is about showing you understand the choices psychologists have to make when conducting experiments. This is about the constant balancing act between validity, practicality and ethicality.

*Pro tip: avoid doing matched pairs. Not only is it time consuming and often times a practical impossibility, it gives you a modification to suggest in your evaluation. If you did independent samples, your issue might be participant variability. Matched pairs could address this issue. Similarly, if your procedure was repeated measures you inevitably have an issue with order effects, that can be addressed by counter-balancing but could also be addressed with a matched pairs. Just make sure you are clear about what specific participant characteristic you want to control for (e.g. age, language, gender, etc.) when you explain your limitations and suggest modifications.