Content of Sugar and Salt of Global Breakfast Cereal Brands

Source http://www.worldactiononsalt.com/less/surveys/2016/190129.html Original data http://www.worldactiononsalt.com/less/surveys/2016/190144.pdf

require("tidyverse")
require("ggthemes")

Dataset

df <- read_csv("cereal.csv")

Parsed with column specification:
cols(
  Continent = col_character(),
  Country = col_character(),
  `Retailer/ Brand` = col_character(),
  `Product name` = col_character(),
  `Portion or serving size (g)` = col_integer(),
  `Total Sugars (g) per serving` = col_double(),
  `Salt (g) per serving` = col_double(),
  `Total Sugars (g) per 100g` = col_double(),
  `Salt (g) per 100g` = col_double()
)

EDA

summary(df)

  Continent           Country          Retailer/ Brand    Product name       Portion or serving size (g) Total Sugars (g) per serving Salt (g) per serving Total Sugars (g) per 100g
 Length:291         Length:291         Length:291         Length:291         Min.   :25.0                Min.   : 2.40                Min.   :0.020        Min.   : 8.0             
 Class :character   Class :character   Class :character   Class :character   1st Qu.:30.0                1st Qu.: 4.20                1st Qu.:0.210        1st Qu.:14.0             
 Mode  :character   Mode  :character   Mode  :character   Mode  :character   Median :30.0                Median : 7.50                Median :0.280        Median :25.0             
                                                                             Mean   :30.6                Mean   : 7.32                Mean   :0.275        Mean   :23.9             
                                                                             3rd Qu.:30.0                3rd Qu.:10.15                3rd Qu.:0.340        3rd Qu.:31.1             
                                                                             Max.   :40.0                Max.   :17.00                Max.   :0.580        Max.   :56.7             
 Salt (g) per 100g
 Min.   :0.080    
 1st Qu.:0.700    
 Median :0.900    
 Mean   :0.895    
 3rd Qu.:1.130    
 Max.   :1.930    

SUGAR

Are differences in sugar per 100g of the same product by country?

df %>% 
  group_by(Country, `Product name`) %>% 
  summarise( MeanSugarPer100g = mean(`Total Sugars (g) per 100g`)) %>% 
  ggplot(aes(reorder(`Product name`, MeanSugarPer100g), y=MeanSugarPer100g)) + 
  geom_boxplot() +
  coord_flip() + 
  scale_x_discrete(name = "Product Name") + scale_y_continuous(name = "Mean Sugar Content per 100g") +
  ggtitle("Mean Sugar Content of the Same Cereal in \nDifferent Countries") +
  theme_economist()

`panel.margin` is deprecated. Please use `panel.spacing` property instead`legend.margin` must be specified using `margin()`. For the old behavior use legend.spacing

With mean and SD

df %>% 
  group_by(Country, `Product name`) %>% 
  summarise( MeanSugarPer100g = mean(`Total Sugars (g) per 100g`)) %>% 
  ggplot(aes(reorder(`Product name`, MeanSugarPer100g), y=MeanSugarPer100g)) + 
  geom_boxplot() +
  stat_summary(fun.y=mean, geom="point", shape=1,
                 size=0.1, color="red") +
  stat_summary(fun.data=mean_sdl, mult=1, 
                 geom="pointrange", color="red", width=0.01) +
  coord_flip() + 
  scale_x_discrete(name = "Product Name") + scale_y_continuous(name = "Mean Sugar Content per 100g") +
  ggtitle("Mean Sugar Content of the Same Cereal in \nDifferent Countries") +
  theme_economist()

Ignoring unknown parameters: mult, width`panel.margin` is deprecated. Please use `panel.spacing` property instead`legend.margin` must be specified using `margin()`. For the old behavior use legend.spacing

SALT

df %>% 
  group_by(Country, `Product name`) %>% 
  summarise( MeanSaltPer100g = mean(`Salt (g) per 100g`)) %>% 
  ggplot(aes(reorder(`Product name`, MeanSaltPer100g), y=MeanSaltPer100g)) + 
  geom_boxplot() + 
  coord_flip() + 
  scale_x_discrete(name = "Product Name") + scale_y_continuous(name = "Mean Salt Content per 100g") +
  ggtitle("Mean Salt Content of the Same Cereal in \nDifferent Countries") +
  theme_economist() # add geom_jitter(aes(colour = Country)) +  to jitter

`panel.margin` is deprecated. Please use `panel.spacing` property instead`legend.margin` must be specified using `margin()`. For the old behavior use legend.spacing

With mean and SD

df %>% 
  group_by(Country, `Product name`) %>% 
  summarise( MeanSaltPer100g = mean(`Salt (g) per 100g`)) %>% 
  ggplot(aes(reorder(`Product name`, MeanSaltPer100g), y=MeanSaltPer100g)) + 
  geom_boxplot() + 
  stat_summary(fun.y=mean, geom="point", shape=1,
                 size=0.1, color="red") +
  stat_summary(fun.data=mean_sdl, mult=1, 
                 geom="pointrange", color="red", width=0.01) +
  coord_flip() + 
  scale_x_discrete(name = "Product Name") + scale_y_continuous(name = "Mean Salt Content per 100g") +
  ggtitle("Mean Salt Content of the Same Cereal in \nDifferent Countries") +
  theme_economist() # add geom_jitter(aes(colour = Country)) +  to jitter

Ignoring unknown parameters: mult, width`panel.margin` is deprecated. Please use `panel.spacing` property instead`legend.margin` must be specified using `margin()`. For the old behavior use legend.spacing

Salt and sugar

Is any brand with cereal with high levels of sugar and salt?

I will subset “All bran flakes”. “Cornflakes”, “Froesties” and “Nesquik”, since are available in most countries and will allow a comparison

Is any difference in the content of sugar per 100g by continent?

sugar

Call:
   aov(formula = `Total Sugars (g) per 100g` ~ Continent, data = df2)

Terms:
                Continent Residuals
Sum of Squares        284     12089
Deg. of Freedom         5        85

Residual standard error: 11.9
Estimated effects may be unbalanced

summary(sugar)

            Df Sum Sq Mean Sq F value Pr(>F)
Continent    5    284    56.7     0.4   0.85
Residuals   85  12089   142.2               

Conclusion: Accept H_o

df2 %>% 
  group_by(Country, `Product name`) %>% 
  summarise( MeanSugarPer100g = mean(`Total Sugars (g) per 100g`)) %>% 
  ggplot(aes(reorder(`Product name`, MeanSugarPer100g), y=MeanSugarPer100g)) + 
  geom_boxplot() +
  stat_summary(fun.y=mean, geom="point", shape=1,
                 size=0.1, color="red") +
  stat_summary(fun.data=mean_sdl, mult=1, 
                 geom="pointrange", color="red", width=0.01) +
  coord_flip() + 
  scale_x_discrete(name = "Product Name") + scale_y_continuous(name = "Mean Sugar Content per 100g") +
  ggtitle("Mean Sugar Content of the Same Cereal in \nDifferent Countries") +
  theme_economist()

Ignoring unknown parameters: mult, width`panel.margin` is deprecated. Please use `panel.spacing` property instead`legend.margin` must be specified using `margin()`. For the old behavior use legend.spacing

Salt

df2 %>% 
  group_by(Country, `Product name`) %>% 
  summarise( MeanSaltPer100g = mean(`Salt (g) per 100g`)) %>% 
  ggplot(aes(reorder(`Product name`, MeanSaltPer100g), y=MeanSaltPer100g)) + 
  geom_boxplot() + 
  stat_summary(fun.y=mean, geom="point", shape=1,
                 size=0.1, color="red") +
  stat_summary(fun.data=mean_sdl, mult=1, 
                 geom="pointrange", color="red", width=0.01) +
  coord_flip() + 
  scale_x_discrete(name = "Product Name") + scale_y_continuous(name = "Mean Salt Content per 100g") +
  ggtitle("Mean Salt Content of the Same Cereal in \nDifferent Countries") +
  theme_economist() # add geom_jitter(aes(colour = Country)) +  to jitter

Ignoring unknown parameters: mult, width`panel.margin` is deprecated. Please use `panel.spacing` property instead`legend.margin` must be specified using `margin()`. For the old behavior use legend.spacing

Is any difference in the content of salt per 100g by continent?

salt

Call:
   aov(formula = `Salt (g) per 100g` ~ Continent, data = df2)

Terms:
                Continent Residuals
Sum of Squares       2.74      8.74
Deg. of Freedom         5        85

Residual standard error: 0.321
Estimated effects may be unbalanced

summary(salt)

            Df Sum Sq Mean Sq F value  Pr(>F)    
Continent    5   2.74   0.548    5.33 0.00026 ***
Residuals   85   8.74   0.103                    
---
Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Conclusion: Reject H_o; conclude not all means equal (P = 0.00026) seems that there ARE significant differences in the content of salt for the same product in different continents

TukeyHSD(salt, "Continent")

  Tukey multiple comparisons of means
    95% family-wise confidence level

Fit: aov(formula = `Salt (g) per 100g` ~ Continent, data = df2)

$Continent
                                diff     lwr     upr p adj
Asia-Africa                  0.25042 -0.3220  0.8228 0.797
Europe-Africa               -0.00122 -0.5603  0.5578 1.000
North America-Africa         0.51182 -0.0970  1.1206 0.151
Oceania-Africa               0.21333 -0.4476  0.8743 0.935
South America-Africa         0.08000 -0.5809  0.7409 0.999
Europe-Asia                 -0.25164 -0.4919 -0.0114 0.035
North America-Asia           0.26140 -0.0789  0.6017 0.231
Oceania-Asia                -0.03708 -0.4637  0.3896 1.000
South America-Asia          -0.17042 -0.5971  0.2562 0.852
North America-Europe         0.51304  0.1956  0.8304 0.000
Oceania-Europe               0.21455 -0.1940  0.6231 0.645
South America-Europe         0.08122 -0.3273  0.4898 0.992
Oceania-North America       -0.29848 -0.7729  0.1759 0.450
South America-North America -0.43182 -0.9062  0.0426 0.096
South America-Oceania       -0.13333 -0.6730  0.4063 0.979

Salt and sugar

Is any brand with cereal with high levels of sugar and salt?