Article by: Perry Romanowski
As a cosmetic chemist, one of the most basic things you need to be able to do is to evaluate formulas and new raw materials. This is one reason we highly recommend that formulators try EVERY product they make. But just haphazardly trying a product will only give you so much information. If you really want to know if a cosmetic formulation is different, you’ll have to conduct a triangle test.
What is a triangle test
Simple put, a triangle test is a study in which you compare 3 samples and pick out the one that is different. Of the three samples, one is actually different while two are the same. If you can consistently pick out the different one, there is a pretty good chance that there is actually some difference that you’re picking up on. You might not always know what the difference is but sometimes that doesn’t matter. The key is that if you can notice a difference then there is probably something different about it.
When to conduct a triangle test
Triangle tests are useful whenever you need to determine whether a change in the formula is noticeable or not. So, they work well for…
- Fragrance evaluations – When you have an odor change and you want to see if there is a noticeable difference.
- New raw materials – To see if there is some performance effect by using a raw material
- Cost savings – Remove a raw material and see if anyone can tell a difference
- Formula development – See if changes you’ve made actually improve your results
Triangle test basics
A triangle test can be done by one person or by an entire panel of people. The former is best when you are making new prototypes and evaluating raw materials. The later is better for making final decisions as it will give you some statistics and numbers to show your boss.
Step 1 – Design your test
The first thing you have to do is figure out what test you’ll run to compare the samples. For odor evaluations this can be a simple “sniff test”. Just put your formulas in a small jars, label them and smell. I like to keep my eyes closed for odor evaluations or do it in a dark room so you can’t be influenced by color. For other products you can use any number of tests such as foam tests, moisturizing tests, combing tests, etc. It really depends on what you are testing and what characteristic you want to notice.
If you are doing a test in which you want to get some useful statistics, you need a panel of about 30 people before you get meaningful data. However, if the differences are significant you’ll be able to make decisions on many fewer test subjects (say 15).
Step 2 – Make your samples
To do a triangle test you need to make two batches of a product. Be sure to create enough to pour off multiple samples. One batch will be your control while the other will be your test. When you are making your samples, flip a coin to decide which batch will be the one you take two samples from and which will be the odd one out. Flipping a coin (or other random process) is important because you don’t want to always conduct the test in which the odd sample is always the test sample.
Ideally, if you are going to test the samples, you should have someone else make and fill the samples. At the very least you should have someone else fill and label them. That way you can be singly blinded and won’t be able to trick yourself.
Step 3 – Test your samples
After the samples are made & labeled, run the test. Evaluate the products in the order they are presented and give an initial guess as to which is the odd sample. It is useful to write down your answer at this point.
Once you’ve tested the samples, it is helpful to re-run the test. This time you should randomly mix the samples and evaluate them without knowing which sample is which. Try to hide the labels. If you are unable to pick out the same sample then the first time was probably a fluke and you’ll want to rerun it a few more times until you can consistently (or not) pick out the same sample.
Step 4 – Reveal the codes
Once you are certain you’ve got the odd sample, look at the codes to see if you picked out the odd sample. If you did, then you can have some confidence that there really is a difference. If you didn’t, there probably isn’t any real difference and the new raw material or the formulation change you made was likely inconsequential, at least for the characteristic you tested.
Step 5 – Determine significance
If you were conducting the test with a panel of people you need to determine if there were statistically significant differences. For people who like statistics you can figure out the Chi squared distribution. X2=? (|O-E|)2/E, where O=observed and E=expected. Or you can use this handy chart. Just find the number of people who have done the test on the left side and see if the number of people who got it correct is on the right.
For example, if you did a test with 15 panelists, 9 of them would have to have picked the correct odd sample for you to decide that there was a significant difference.
The triangle test is one of the key tools that a cosmetic chemist has at her disposal. You should practice it often because the more you use it, the better you’ll get and the more useful it will become.