Help with Experiment Design using taste and smell

[6Sigma]

This is for a middle school project. The original idea was to have volunteers eat a piece of flavored candy, describe the flavor, and then eat another piece of the same flavored candy but while also smelling a different flavored scented oil - then describe the flavor to determine whether the taste is influenced by smell. Due to requiring volunteers, the project proposal needed to be submitted for approval for human experimentation - and it was submitted several days before topic ideas and hypotheses were due. So the teacher said that the project topic was not measurable and my daughter thought it was back to the drawing board - however, a few days later, the powers that be in charge of the science fair approved the original topic and experiment idea and gave some suggestions along with a blessing. So without further guidance from the teacher, she's decided to go ahead with the original idea, but keeping in mind the teacher's concern regarding the data.

 

So this is the feedback that came back from the "powers that be": To the greatest extent possible, you should try to establish different control groups so that you are sure that the only variable being examined and changed is the flavor of the candy. For example, a good control group would be one that smells an odorless oil but has a, for example, strawberry flavored candy. Then another group should smell the scented oil to see if their descriptions of the candy's flavor does or does not change, relative to the control group. Two other control groups could have a flavorless candy while smelling the scented and odorless oil.

 

OK - this is where some guidance is needed.... It sounds like they are suggesting 4 different groups of volunteers. My concern is two-fold: 1) access to enough volunteers due to time constraints, 2) does having separate groups of people add an additional variable due to differences in individuals? My daughter had thought that using the same population would be more desirable for determining whether taste is influenced by scent. FWIW - she is planning to use cherry flavored gummy candy, lemon essential oil, (most likely) canola oil (for the odorless oil), and plain gelatin to make "flavorless" gummy candy

 

Additionally, due to the teacher's concerns about measuring the data, she is trying to determine what is the best way to get and log the outcomes. Should she ask an open-ended question like describe the flavor? That could lead to variability of responses that may be difficult to portray. Or should she have the volunteers to select from a limited list of flavors? Could this influence the volunteer and skew results?

 

Thank you for your help!

[Prometheus]

My daughter had thought that using the same population would be more desirable for determining whether taste is influenced by scent.

 

Your daughter is smart: if possible have each person take the taste test first with odourless oil, then later with the scented oil. That means each person is his/her own control and reduces variations in your outcome of interest (taste) due to humans just being different (and makes the stats analysis nicer). This is quite a typical feature of dietetic studies - but you then have to think about how eating the candy the first time might influence tasting the candy the second time around. For such reasons some studies might have a wash-out period, allowing any lingering effects of an initial exposure to wear off. In your case it might be a few minutes and a glass of water (i'm just guessing - your daughter could look up things like how long taste lingers, if taste influences future tastes, if water resets taste etc).

 

 

Additionally, due to the teacher's concerns about measuring the data, she is trying to determine what is the best way to get and log the outcomes. Should she ask an open-ended question like describe the flavor? That could lead to variability of responses that may be difficult to portray. Or should she have the volunteers to select from a limited list of flavors? Could this influence the volunteer and skew results?

 

Not an easy answer to this one. Generally quantitative information, which you can measure or categorise somehow, is much more useful than qualitative. Your daughter could look into the pros and cons of each. Maybe google 'quantifying taste for research' or similar.

 

All you actually want to know is whether the taste changes, not what the taste actually is. So you could just ask the question on a scale of 1-10 how similar are the two tastes? But asking this very question could trigger them to seek differences when normally they wouldn't. Even the wording asking for 'how similar' instead of 'how different' could influence people's responses. Just try to make decisions based on what will reduce the amount of bias in the study, justify them if you have good reasons and if there's not a clear answer to a choice just say why its not clear.

 

Like i say, no easy answers. Hope my rambling helps. Nurses do a lot of this kind of research and it sounds like your daughter is asking better questions than many of them.

 

Good luck: sounds like a fun school.

[6Sigma]

Thank you so much for your help! My daughter conducted a first round of tests today (due to uncooperative weather, only about half of her expected volunteers showed up - so hopefully she will be able to do a second round). Based on your advice, the advice of an aunt, and some further research - these are the decisions she ultimately made for the design of her experiment:

 

1) She settled on drinking water and smelling coffee grounds in between tests to "wash out" the flavor and smell.

2) She made the gummy candies so that they were all the same size and shape and the same texture (more on that below).

3) Since the cherry flavoring was red (using red 40 and blue 1), she decided to try to match the color for the unflavored candy using mostly red food coloring and a teeny tiny bit of blue. The color came out really close but she decided to use a blind-fold anyway, just to be extra cautious.

4) Since the cherry flavoring had quite a bit of sugar, she was a little worried about the sweet aspect between the candies. Additionally, research suggested that the texture of gelatin is influenced by the sugar content. Therefore, she decided to add approximately the same amount of sugar to the unflavored candy in the hopes of controlling for the sweet factor AND achieving the same texture.

5) She decided the order of tests would be: plain candy/unscented oil, cherry candy/unscented oil, plain candy/lemon oil, cherry candy/lemon oil. This was based on some early experimenting with siblings using flavored and plain water and determined that the lemon oil can really linger - especially if it accidentally gets on the skin! Also - it is really hard to drink and smell at the same time (which is how she ended up switching to a chewy candy).

6) The questionnaire - way more tricky than she anticipated and quite frankly *almost* made her want to change projects! But after some sleep, she created a questionnaire that contained the following:

a) Participant number

b) three age ranges ( 12&under, 13 to 20, over 20)

c) an open ended question: describe the flavor of the candy (this was asked for each test)

d) on a scale of 1 to 5, rate how sweet the candy tasted (this was asked for each test)

e) on a scale of 1 to 5, rate how sour the candy tasted (this was asked for each test)

f) on a scale of 1 to 5, rate how bitter the candy tasted (this was asked for each test)

g) on a scale of 1 to 5, rate how salty the candy tasted (this was asked for each test)

- she decided to include all of these flavor types in order to not influence the participant

7) After all of the tests were completed and the questionnaire was turned in, she asked a final question that is on a master table that uses participant number to match to the questionnaire later, if necessary - Did any of the candies taste the same? If so, which test numbers?

 

I probably would have added a question about whether the participant was currently or recently ill or congested. I am sure there were probably some other things that could have been included, but it's not my project! I was pretty skeptical of the open-ended question, but the responses were actually pretty interesting. Also - conducting 4 tests on each person seemed really less than ideal in practice - but does seem to be the most sound given the pretty short time frame she was given to complete this project (literally one week from the time her project idea got approval until the data is due - doesn't leave much time for inevitable errors and problems!). The most important thing is that this is her first science fair project and she seems really happy with the results so far So thanks again for all of your advice!

[Prometheus]

Let us know of the results.

 

Could do some simple statistics on the results - maybe save them for a future maths school project?