Maggie Johnson

Our Statistics 1040 class was assigned a project that we would contribute to

and assess throughout the semester. Each class member purchased a 2.17 ounces

bag of Skittles, recorded the number of different colors in his or her bag and

compared the data with the rest of the class. The objective was to learn how to

organize and collect data, predict potential probabilities, draw conclusions from the

variations of the results and apply what we were learning in statistics to an

everyday, practical object.

We were instructed to collect data on the number

of candies per color in each Skittles bag. Each person recorded the following

information from his or her bag:

Number of red candies

Number of orange candies
Number of yellow candies
Number of green candies
Number of purple candies

We then submitted the data to our professor who created a dotplot of the
total candies in each bag:

I created charts and graphs I had learned about in class utilizing the Skittles

data. The first example shown is a pie chart displaying the total data from the
class. The second example shown is a pie chart expressing the data from my

individual bag of Skittles.

 We were assigned to groups of 7-8 people where we collaborated on our

Skittles data. These groups remained the same throughout the semester. We

contributed to the group on a discussion board, and also submitted our individual

data to the professor. As a group we discussed

the probability, or chance, that the colors in the Skittles bags were randomly

distributed. We wrote about our findings here:

Term Project Part 2, Group Portion

On the basis of limited information about the manufacturing process we

assumed each color would be roughly equivalent about 20% proportionately. If you

look at the results of the class data set you can see that the variation from 20%

ranges from +0.014 to -0.010. With a sample size of 2,268 that would mean a count

of 454 of each color. We think that the estimate of 20% per color is valid and
relatively accurate. We resisted the temptation to make an inference from our own

bags of Skittles because we considered the sample size to be too small, although

there is no reason to believe it is not random.

If you consider the overall class data as a sample then the population would

be Skittles manufactured. If for experimental purposes you designate only Skittles

purchased by our class as the population then of course every individual bag would

be a sample. In either case, unless there are distribution factors we are unaware of,

each sample is random.

 Count Red Count Count Count Count Total
Orange Yellow Green Purple
Class 464 439 485 449 431 2268
Counts
 Term Project Part 2, Individual Portion

Count Red Count Orange Count Count Green Count Purple Total
Yellow
My Bag 1 12 1 9 16 60
3 0
Class Counts 46 439 48 449 431 2268
4 5

My personal bag was a surprise

to me because I expected to see
more yellows. Yellow is the most
prevalent color and thats why
there are so many banana Runts,
yellow Starburst and lemon
Mambas. They are the throw-
away color. However, I was
pleasantly surprised to see more
purples and reds, as they are
more coveted in the candy
world. (This is a thing.)
There are some outliers,
particularly the bags of Skittles
that had candies totaling in the
forties, while mine was an even
60. This may be attributed to
damage but it makes me wonder
how each bag is weighed 2.17
ounces when the totals are
different.
These differences would affect
the quantitative data and, for
example, histograms and stem
leaf plots would vary quite a bit.
The amount of reds I had in my
bag matched the distribution of
the classs but, as predicted, the
yellows had the highest count in
the class. I had an unusual sort
with yellows being second to
lowest in my bag.
 For our third group project we determined the mean, standard deviation and

5-number summary of the Skittles data. We created a histogram and boxplot to

exhibit the data.

Term Project Part 3, Group Portion

i. Mean number of candies per bag

59.5

ii. Standard deviation of the number of candies per bag

2.9

iii. 5-number summary for the number of candies per bag

53, 58, 59.5, 61, 66

 I went into more detail about these statistical summaries in my individual
portion.

Term Project 3, Individual Portion

1.
The frequency histogram shows the total candies in each bag to be approximately
symmetrical and bell-shaped. The box plot is also symmetric because the
observations are split almost equally at the median. I did expect to see the graphs
as pictured because I assumed Skittles would attempt to be evenly distributed in
each bag to match the weight advertised on the package. In this day and age where
people are measuring 6 Subway sandwiches, Wrigley better get it right! The overall
data agreed with my single bag data.

The class totals: 38 bags with 2,268 candies. The mean is

59.7. My candy total: 60

2.
Categorical (qualitative) data classify individuals based on some attribute or
characteristic. The categories do not necessarily need to have logical order. I always
remember the definition of categorical or qualitative data by the word quality, which
is used as a descriptive term. So, to me, categorical = description. Some examples
of categorical data are gender, race, educational level, make and models of cars,
and colors.
Quantitative data give numerical measures of individuals. These values can be
added or subtracted. I associate quantitative with the word quantity, which means
amount or number. The amount can be indefinite or definite. Some examples of
quantitative data are height, weight, size of a room, amount of ingredients in
recipes, and monetary tips at restaurants.
There are several graphs that work well for categorical data. Side-by-side bar graphs
are optimal because they can make comparisons without requiring numerical
information and can give a visual explanation. Frequency tables can list categories
of data along with occurrences for each category, such as colors in a Skittles bag. A
Pareto chart allows the same information to be expressed as a bar graph, only in
descending order. Finally, pie charts can be used to show sectors of categories
divided into proportional frequencies.
Quantitative data have a lot of options for graphs such as histograms, frequency
polygons, ogives, time-series plots, and stem-and-leaf plots. Histograms show the
frequency or relative frequency by using rectangles for each class of data.
Frequency polygons use points connected by line segments, ogives present
cumulative frequency and cumulative relative frequency in the graphs, and time-
series plots show the value of the variable measured at different points in time.
Stem-and-leaf plots are useful for quantitative data because they display raw data,
but they are not helpful with large amounts of data, and would not be applicable to
categorical data because stem-and-leaf require specific digits and not categories.
Histograms, Ive learned, can be confused with bar charts but they are not the
same. Bar charts are used for categorical data because they are used to compare
variables. Histograms are used to show countable aspects of variables.
 For the fourth group project, we each created a confidence interval for

population proportion and population mean of the Skittles bags and added them to

the group discussion board.

Group Project 4

1) n= 2268 (total Skittles), x= 485 (total yellow Skittles), a= .01, t= .005, critical value= 2.576

pp= 4852268 = .2138

E= 2.576 .2138(1.2138)2268 = .0086

Using TI-84 1PropZInt with C-Level 99%: Lower bound = (.1917), Upper bound = (.2360)

2) a= .05, t= .025, critical value for 95% = 1.96, df

Using TI-84 TInt with 2268/bags of candy = 59.5

E= 1.96 .0262(1.0262)2268 = .0066

Lower bound: 59.38, Upper bound: 59.62

I am 99% confident that the population proportion of yellow Skittles lies between .1917 and .2360.

I am 95% confident that true value of the population mean of candies per bag lies between 59.38 and
59.62.
What Ive learned

The Skittles project introduced me to a wide variety of statistical applications.

Not only did I discover that the TI-84 is my best friend, I was able to use critical

thinking skills to break down complicated problems and find solutions. As shown,

the Skittles project utilized many facets of statistics wed learned throughout the

fast-paced semester. I learned a lot visually with pie charts, Pareto charts,

histograms, dotpots and boxplots. This seems to paint a clearer picture of stats to a

broader audience. On a more intricate level, I learned about the goings-on

behind the scenes of something as simple as colors in a Skittles bag. Initially I

suspected the color distribution was based on cost rather than chance, but with

research such as mean, median and mode, probability and confidence intervals, the

data show that Skittles colors appear to be random. I can see the bigger picture now

and, unlike, algebra, I can see myself using statistics on a day-to-day basis.

Reflection Paper

Initially I panicked with the Skittles term project. Not only did I have to show my work,

but I would have to display it to my peers and my teacher. I felt a little vulnerable and was not

sure I was up to the task. When it was disclosed that the project would be ongoing throughout the

semester, I audibly groaned. However, like most projects, there is a learning curve and once that

is overcome it is just another tick in the box.

The Skittles project was surprising. It reminded me that big corporations like Mars are

selling more than sugar in a red package; they are selling accountability. Just like the guys who
sued Subway for selling subs one inch less than were advertised, Mars could find itself in hot

water if it didnt have a normal distribution of color to ratio in each bag. I am not a Skittles fan

but when offered Skittles I will always choose the purple or the red. Yellow is my least favorite.

If I knew that Skittles only offered a few purples or reds per bag I would be far less inclined to

buy them in the future. All it would take is one bag with a higher proportion of the disliked color

to assume that is how all the bags are. How would Mars know which colors are preferred?

Statistics, of course.

I thought about how using statistics with something as simple as a bag of Skittles could

be compared to other forms of statistics in everyday life, such as: relying on the weather forecast

for an outing or a trip, reading quality reviews of products I want to buy online, deciding on the

safety of fluoride for my young son at the dentist office, finding certain brands that arent always

eye-level at the grocery store, and planning which vegetables to plant in my garden, to name a

few.

I wouldnt say the term project was life-changing. It did buoy my resolve to go out of my

comfort zone and allow myself to make mistakes for the sake of education and learning, even

while being scrutinized. I will never look at a Skittles bag the same ever again, not to mention

other candies as well. The biggest thing I learned from the term project is that I have not once

used algebra since my class ended, but statistics is everywhere.