Cross Validated is a question and answer site for people interested in statistics, machine learning, data analysis, data mining, and data visualization. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have data in an excel spreadsheet of around 500 participants in a taste experiment. The data includes the age of the participant, in addition to one of five locations on their tongue in which they tasted five separate flavours, e.g. (Sweet, Middle), (Salt, Back) - The image below shows the first 20 entries to demonstrate the formatting.

I'd preferably like to be able to use R to see if taste location is dependent upon age, finding the point at which it becomes dependent (if this is indeed the case). Any general methods of doing this would be great.

share|improve this question
What exactly is your dependent / response variable, & what are your explanatory / independent variables here? What is Taster relationship? Eg, you have Child1 in rows 2 & 5, but those kids have different ages. – gung Jul 6 '14 at 15:25
Taster relationship and gender can be ignored - This is a variable used for other analysis within the experiment. It would be easiest to make the taste location the independent variable, and treat the five different flavours as separate analyses. I'm trying to see if older participants are more likely to taste the flavour in one particular part of their tongue, or if it stays seemingly random across all ages. – froseph Jul 6 '14 at 15:31
up vote 2 down vote accepted

This appears to be a multivariate multinomial $Y$ problem, for which I am not familiar with models tailored to this situation. If taster's age were the only variable (i.e., if you ignore taster relationship and gender, and if you assume that all rows represent different persons), you can turn the problem around to see if the different taste-regions relate to age. You could use ordinal or OLS regression to predict age from 5 categorical predictors (one for each taste quality). The global null hypothesis with $2 \times 5 = 10$ degrees of freedom will test whether taste or dominant taste region are associated with age.

share|improve this answer
This worked great - Used OLS regression with 5 categorical variables, resulting in 4 dummy variables ("front", left", "middle", "right) using "back" as the reference variable, and found the regression of age on location. – froseph Jul 6 '14 at 21:13
I computed 10 parameters. Did you run the model separately for region or separately for taste instead of fitting a unified model? – Frank Harrell Jul 6 '14 at 22:46
I ran the model separately for taste. – froseph Jul 6 '14 at 22:51
You can also test some interesting hypotheses by stringing out the data to be tall and thin, with a variable for which taste is represented in the current row. Intra-subject correlation can be accounted for using the Huber-White cluster sandwich covariance matrix estimator or the cluster bootstrap. – Frank Harrell Jul 7 '14 at 12:23

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.