Nutritional balance of kinds of breakfast cereals

Perezgonzalez assessed the nutritional balance of breakfast cereals in 2012 (see b³,c⁴,d⁵,e⁶,f⁷,g⁸). The same data can be analysed further, to ascertain whether features such as being sold under a generic brand or having children as the target customer may (unintentionally) inform about the overall nutritional balance of above products.

Illustration 1 collates information per group within each kind of product, namely group size, as well as median and interquartile range for nutritional balance (BNI). Illustration 2 collates information about mean ranks per group, as well as tests for equality of ranked distributions for independent groups (Mann-Whitney U tests and equivalent Z areas).

Illustration 1: Group size, medians & interquartile ranges
Kind	'No'			'Yes'
	n	median	( IQR )	n	median	( IQR )
Children	64	37.27	( 41.31 )	15	72.65	( 48.24 )
Generic	63	39.51	( 49.72 )	16	36.89	( 61.00 )
(Medians closer to 0.0 indicate greater balance)

According to these results, it can be concluded that:

Cereals grouped according to having children as their customer base or not show ranked group distances so extreme that, under normal circumstances, such distances would only occur about 4 times in 100. As cereals not targeted to children have a lower median (37.27) than cereals targeted to children (72.65), it can be inferred that cereals not targeted to children are more balanced nutritionally. Indeed, the absence of such targeted marketing alone would identify some 23% of more nutritionally balanced breakfast cereals (see rho values in illustration 3).
Cereals grouped according to whether they are marketed under a generic brand or not show ranked group distances within the expected normal range, and relatively similar medians. Thus, it can be inferred that breakfast cereals marketed under generic brands have a similar nutritional balance than breakfast cereals marketed under specific commercial brands. Therefore, this grouping is not helpful in identifying different nutritional balance among breakfast cereals.

Illustration 2. Mann-Whitney U tests
Kind	m.rank 'no'	m.rank 'yes'	U	Z	( p )
Children	37.47	50.80	318.0	-2.03	( .043 )
Generic	40.10	39.59	497.5	-.08	( .937 )
(Mean ranks closer to 0.0 indicate greater balance. Sig ≤ .05, 2-tailed)

Illustration 3 collates information about a multiple regression model for assessing the potential combination of both kinds of breakfast cereals groupings. The regression model (R, 'stepwise' procedure) retains only the 'targeted to children' category as a significant variable, capable of identifying the nutritional balance of about 23% of breakfast cereals in the sample. The 'usability' of such regression model for predicting the nutritional balance of breakfast cereals not in the sample is, approximately, 20% (adjusted R).

Illustration 3. Regression model on ranked BNI
	rho	( p )	β	( p )
Children	.229	( .042 )	.229	( .042 )
Generic	-.009	( .937 )

		R =	.229	( .042 )
		Adj.R =	.200
(Dependent variable = BNI. Sig ≤ .05, 2-tailed)

Research approach

Exploratory study.

Sample

The research sample included 79 breakfast cereals (ie, oats, wheat, rice, corn, bran and muesli). Although the research aimed for a stratified sampling procedure, products were nonetheless collected in a convenient manner (see Perezgonzalez, 2012a²).

Materials

The main (or dependent) variable was the BNI classification for each food as assessed using the Balanced Nutrition Index™ (BNI™) technology (see Perezgonzalez, 2011¹).

Grouping (or independent) variables were two: generic brands (also known as supermarket brands or low-cost brands), and products targeted to children. A product was classified as a generic brand if it was a well-known supermarket brand or one of a wholesaler's generic brands. A product was classified as targeted to children independently of whether it was verbally targeted to children or marketed in a childish manner (eg, with cartoons); products without such direct or indirect references were not classified as targeted to children, even if they looked similar to the former or was suspected children were the most probable consumer target.

Illustration 4: Breakfast cereals from generic brands and targeted to children
Product100g	BNI	Generic	Children
Be Natural multi grain porridge	6.07	No	No
Harraways organic rolled oats	6.83	No	No
Traditional Porridge brown sugar & cinnamon	6.89	No	No
Be Natural cashew, almond, hazelnut & coconut	8.25	No	No
Homebrand traditional muesli	9.77	Yes	No
Creamy porridge	12.21	Yes	No
Harraways organic wholegrain oats	12.21	No	No
Harraways traditional wholegrain oats	12.21	No	No
Harraways rolled oats	12.21	No	No
Harraways Scotch oats	12.21	No	No
Oat Singles plain	12.21	No	No
Home Brand Quick Oats	12.26	Yes	No
Be Natural vanilla almond porridge	12.64	No	No
Oat Singles chocolate	16.64	No	Yes
Simply Reduced Fat Muesli cranberry & vanilla	18.04	No	No
Vogel's rolled oats plus	18.24	No	No
Simply toasted muesli original	19.69	No	No
Pams natural muesli	20.25	Yes	No
Big Breakfast toasted muesli	20.26	No	No
Simply toasted muesli 5 golden grains	20.35	No	No
Woolworths Select muesli morning crunch	20.83	Yes	No
Pams toasted muesli	24.16	Yes	No
Be Natural pink lady, flame raisins & fruit	24.76	No	No
Oats Smooth & Tasty vanilla	26.26	No	Yes
Pams reduced fat toasted muesli	29.53	Yes	No
Uncle Tobys Plus omega 3	29.81	No	No
Sanitarium puffed wheat	31.88	No	No
Vogel's manuka, honey & apricot porridge	33.54	No	No
Oats Smooth & Tasty strawberry	34.13	No	Yes
Simply toasted muesli apricot	34.40	No	No
Oats Smooth & Tasty honey	34.51	No	Yes
Weet-Bix	35.26	No	Yes
Oat Singles honey & golden syrup	35.55	No	No
Sustain original	36.00	No	No
Pams wheat biskits	36.32	Yes	No
Oat Singles morning berry	36.74	No	No
Harraways fruit harvest	37.24	No	No
Mini Wheats mixed berry flavour	37.31	No	No
Homebrand wheat biscuits	37.46	Yes	No
Mini Wheats blackcurrant flavor	38.79	No	No
Bran & berries	39.51	No	No
Oat Singles apple, sultana & cinnamon	39.68	No	No
Bran & apricot	40.89	No	No
Sanitarium Honey Puffs	40.90	No	Yes
Fibre Life bran flakes	41.40	No	No
Be Natural 5 whole grains	43.11	No	No
Pams Lite & Fruity apricot fruit	44.65	Yes	No
Freedom Foods rice flakes with psyllium	44.94	No	No
Just Right original	54.64	No	No
Kellogg's sultana bran	55.58	No	No
J.R.Fields choc scoops	56.21	No	Yes
Special K	56.50	No	No
Budget tropical muesli	57.29	Yes	No
Alison Holst decadent delight	57.82	No	No
Bran & sultana	60.33	No	No
All-Bran Wheat Flakes honey almond	61.40	No	No
Heart 1st	67.46	No	No
All-Bran Wheat Flakes original	68.58	No	No
Nutri-Grain	70.08	No	No
Alison Holst rise & shine	72.63	No	No
J.R.Fields corn scoops	72.65	No	Yes
Kellogg's corn flakes	74.90	No	No
J.R.Fields rice pops	80.35	No	Yes
Coco Pops	81.29	No	Yes
Budget cocoa puffs	81.39	Yes	No
Homebrand cocoa puffs	81.39	Yes	No
Rice Bubbles	81.69	No	Yes
Homebrand corn flakes	81.93	Yes	No
Frosties	82.76	No	Yes
J.R.Fields honey wheats	84.04	No	Yes
Homebrand rice pops	84.05	Yes	Yes
Sanitarium Ricies	84.63	No	Yes
Crispix honey	87.65	No	No
Protein 1st	90.43	No	No
All-Bran original	97.55	No	No
Healtheries bran flakes	128.74	No	No
Brantastix	129.23	No	No
Homebrand natural brand	131.80	Yes	No
San Bran	156.01	No	No
(Source: Perezgonzalez, 2012a²)

Analysis

The distribution of the BNI variable deviated from normality in a significant manner (see illustration 5), some of the claim groups were too small in size, and not all variables were related linearly with the dependent variable (BNI) (see illustration 6). Thus, a non-parametric approach was used for statistical analysis⁹.

Illustration 5. Assessment of normality of the BNI variable
Variable	Mean	Median	Skew.Z	( p )	Kurt.Z	( p )	K-S	df	( p )
BNI	47.97	38.78	3.76	( .000 )	1.74	( .082 )	.149	79	( .000 )
(Sig ≤ .05, 2-tailed)

Illustration 6. Curve estimation for kind of cereal on the BNI
Kind	Generic	Children
Linear fit (p)	( .984 )	( .123 )
Best fit curve	compound	compound
(p)	( .910 )	( .042 )
(Dependent variable = BNI. Sig ≤ .05, 2-tailed)

Main analyses: tests of significance (Fisher's approach) based on sig ≤ .05 or more extreme probabilities, 2-tailed.
Main tests: Mann-Whitney U test for equality of ranked distributions of independent groups, Spearman's rho correlations, and multiple correlation on ranked data.
All analyses were carried out with SPSS-v18 (PASW Statistics 18).

Generalization potential

Most breakfast products were produced locally or imported and exported from or to Australia, respectively. Thus, the results of this study may be generalizable to the following populations (in order of decreasing generalization power):

Australia.
Internationally, assuming breakfast products to be of approximately similar nutritional composition anywhere.

Author

Jose D PEREZGONZALEZ (2012). Massey University, Turitea Campus, Private Bag 11-222, Palmerston North 4442, New Zealand. (JDPerezgonzalez).

Want to know more?

BNI™ journal (2012, issue 3) - Breakfast cereals: This issue of the Balanced Nutrition Index™ journal collates all BNI™ nutrition information for the original sample in a single book. You can also access information about each individual breakfast cereal on the BNI™ database and aggregated information on Wiki of Science (nutritional balance of food)
Wiki of Science - Marketing claims and the nutritional balance of breakfast cereals: Two Wiki of Science pages provide summary and detailed information, respectively, about a research related to the one here discussed.