# Counting Calories at the BT Young Scientist & Technology Exhibition – the Results…

By Seamus Kelly, Stefania Giammanco and Cronan McNamara

Playing the game

In the game, each participant was asked to enter their age and gender. A picture of a portion of food was then randomly selected and presented to the player (from 41 different available foods) along with two questions:

1. How many calories are in the portion of food shown?
2. How many minutes would you be able to run from the calories in the portion of food shown?

A screenshot of the Creme Global nutrition game. Click on the image or here to play the game.

If both questions were answered correctly then the player was declared a winner and presented with a prize.

Results

The game gathered anonymous data throughout the three days of the exhibition. Overall, there were 1,086 attempts at the game during the event and of those, 119 were winners – i.e. they guessed the answers to both questions correctly [1]. This represents approximately 11% of the game attempts.

Being a data science company, we felt the need to analyse the data from the game and to summarise the results here. We have produced selection of graphs which provide analyses of the results for different combinations of age, gender and food types along with some discussion of the data and the results at the end.

Percentage of correct answers for each gender reported (total n=1,086)

The first graph is a grouped bar chart and compares the success of male and female participants of the games. The graph shows that 11.7% females and 9.7% males respectively answered both questions correctly. So the girls won this one. Or did they? To answer this question, we would have to answer homework question 1. We will leave this as an exercise for the reader.

 Homework question 1: Given that there were 1,086 participants and that the data shows that 11.7% of female contestants and 9.7% of male contestants respectively answered both questions correctly. Assuming that there were roughly equal numbers of male and female participants, would you consider the difference in scores statistically significant within the data sample size? Discuss. Answers on a postcard [2], tweet or email please.

Percentage of correct results by reported age group (total n=1,086)

In the above graph results are broken down by age group in a grouped bar chart. The graph shows that, proportionally, the 18-30 age group were the most successful with 17% answering both questions correctly. In contrast, the 31-60 age group had fewest winners with only 4.7% of the group declared as winners.

We are not sure what this says about the nutritional knowledge of middle-aged adults at the exhibition! But wait, hold on a minute, wasn’t this a student exhibition? Were there many data points in the 31-60 age group? In fact there were 47 data points. [3]

Percentage of correct results for the top 6 food types presented

Now this is interesting, we wanted to have a quick look at the correct answers for various food types in the game. We are interested in the question here: what foods to people seem to have a more or less nutritional knowledge on?

The final graph presents a summary of the foods with most winning guesses. The graph shows that the most correctly guessed foods, in order, were: chocolate chip cookies (30.4%), haddock (30%), cola (23.8%), yoghurt (22.6%), toffees (22.2%) and white rice (19.1%). On the contrary, the foods chicken casserole, chickpeas and stir fry had no correct answers.

 Homework question 2: Given that there were 41 different food portions in the game, and that these were shown randomly, what are the chances of the same food reappearing at least once if you play the game three times? Answers on a postcard [2], tweet or email please.

Discussion

This analysis is intended to be a fun, illustrative exercise in data analysis for the young scientists who attended the event, rather than a rigorous scientific experiment.

There are a few interesting caveats to be aware of in the data, and some interesting lessons in experimental design for the budding scientists, these include:

• The stand was manned by qualified nutritionists during the event, however the age and gender of contestants were self-reported and not supervised or checked by the nutritionists.
• Although the full sample was large (n=1,086) enough to have statistical power, when the data was broken into subgroups of age, gender and food type, some of the data subsets were quite small, so those statistics would have to be treated with caution.
• A number of contestants took a number of repeated attempts at the game until they got a correct answer! They were, of course, presented with a new random food each time, but the same food had a chance of re-appearing if the contestant played for long enough. See homework question 2. Therefore, not all of the 1,086 attempts were independent trials – there may be some correlation in the data as the contestants improved their guesses as they kept trying the game.
• Finally, at times the nutritionists on the stand were persuaded to give a few hints and a little help to contestants, so this may also bias the results a little!

So, it was a lot of fun and games and hopefully some interesting scientific lessons for the students at the BT Young Scientist and Technology Exhibition 2015. We look forward to supporting the exhibition again next year.

 Homework item 3:

References and Footnotes

1. Answers were deemed to be correct if they were within an acceptable error range. The acceptable range was set at 20% for question 1 and 25% for question 2.
2. Homework answers on a postcard to: Creme Global, The Tower, Trinity Technology & Enterprise Campus, Grand Canal Quay, Dublin 2, Ireland.
3. Number of data points in each group: 8-10 years: 152, 11-17 years: 786, 18-30 years: 47, 31-60 years: 64, 60+ years: 37, all age groups / total: 1,086.
4. The Science Behind the Game (pdf).
5. Play the Game here.
6. The European Commission office in Ireland.
7. The BT Young Scientist and Technology Exhibition.
8. The Food4Me Project.