Article by: Perry Romanowski

Yesterday, I wrote about what makes a good scientific study and how you can prevent fooling yourself. The key was to remain skeptical and create studies that are blinded.

Here is a video which shows just how one can be done. Granted this video is not about selecting a cosmetic ingredient level. However, it does illustrate the proper way to do a scientific study.

[youtuber youtube=’’]

Here’s what they do.

1. Start with a hypothesis. In the video, the hypothesis is that people with different immune systems will be more attractive to each other than people with similar immune systems.

2. Figure out a way to test the hypothesis. There are a number of implications of this hypothesis that can be tested. In this video, they look at the relationship between natural odor preference and genetic makeup.
In the video, the hypothesis is that a woman’s natural odor will be prefered by a man based on their mutual genetic composition. If the hypothesis is correct, the man should pick the woman who has the least common genetic immune system.

3. Do a blinded test. The evaluator is unaware of which sample is which so he is blinded and therefore, unbiased. Of course, this isn’t a perfectly blinded test because the evaluator does have some information about the study (like why he should pick a difference). A better study would have been to have evaluators that are completely unaware of the purpose of smelling the t-shirts.

However, in cosmetic formulating it’s rare that you could use unaware panelists to help in this way.

This study is a bit simplified and for real science you need to be more rigorous (e.g. have a lot more panelists) but overall, this type of study is a nice example of how you should create studies for evaluating new raw materials in your cosmetic formulas.



  1. Avatar

    What a lovely video, and about smelling too (always a winner for me) 🙂
    I agree with you that in cosmetic formulating one could hardly use unaware panelists.
    Also when formulating I personally don’t have a hypothesis other than ‘there is a/no difference in performance (one can quantify this more specifically) between product A containing ingredient X and product B that doesn’t’. Do you, Perry?

    1. Avatar

      What you describe would be considered a hypothesis. It’s typically the only one cosmetic chemists have because they don’t always know what an ingredient is going to do in the formula. Sometimes we do but often we don’t.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.