I conducted the world’s first scientific taste analysis of all major Caterpillar Cakes. Double-blinded, randomised and order counterbalanced. Results here.

Motivation: M&S’s Colin the Caterpillar has been a UK staple cake for decades, which in recent times has “inspired” nearly every other UK supermarket to create a knock-off. These all have different names in place of “Colin”, but are invariably caterpillars. In early 2021 M&S sued Aldi for their variant, claiming that Aldi (and implying by extension all other pretenders) were “riding on the coat-tails” of Colin’s reputation. I was moved to determine once and for all which is the superior caterpillar cake, given the clear public interest in the question.

Methods: One of each currently-available caterpillar cake (lineup) was purchased from each major supermarket. These amounted to:

  • Colin (M&S)
  • Cecil (Waitrose)
  • Wiggles (Sainsburys)
  • Charlie (Co-op)
  • Morris (Morrisons)
  • Clyde (Asda)
  • Curly (Tesco)

Ironically, the caterpillar which started the fracas with M&S (Aldi’s Cuthbert) is not in production at the time of the experiment so was excluded. An additional 8th cake – Carl (a dairy and gluten-free Tesco cake) was also included for exploratory analysis.

14 participants agreed to take part in the experiment. Middle “abdominal segments” of each caterpillar were cut out and removed of significant garnishes, e.g. smarties or solid-chocolate feet, such that what remained was essentially the fundamental “cake” of sponge, ganache and primary icing. This was to A) reduce visual cues which may assist participants familiar with the cake’s identify in recognising the caterpillar they are eating, and B) to improve reproducibility from bite-to-bite; two samples from the same cake may differ in if they have a smartie or not, and therefore a confounding variable would be introduced. Once treated, the abdominal segments were further cut into bite-sized pieces.

A member of the study team who did not take part in the tasting then randomised the cake identities into arbitrary numbers, presenting the collections of bites in bowls labelled 1-8, with each bowl containing only pieces of that specific caterpillar. Participants were then provided with sheets of paper which instructed them on the order to try each cake (ordering was counterbalanced between participants using the “balanced latin square” method). After eating each, participants rated the taste on a visual analogue scale from the “worst imaginable cake experience” to the “best imaginable cake experience”, with a caveat given on the sheet that this should be done within reasonable cake expectations.

After the experiment was performed, ratings were processed into “raw” scores between 0-10 (i.e. representing the precise position of the mark given on the visual analogue scale, to 1 decimal place), and also converted on a per-participant basis to rankings (i.e., for each participant their worst to best cake was assigned a whole number from 1 – 8, with higher numbers representing greater preference). This was done to control for any participants who may rate all cakes relatively low if they simply do not enjoy caterpillar cakes.

Results: Averages of both the “raw” and “rank” scores were made to find the overall order of preference for the cakes. These data are shown in Table 1, and Figure 1 (for preference by “raw” score) and Figure 2 (for preference by “rank” score).

The overall order of preference is largely set between these methods. For “rank” preferences it is as follows:

  1. Cecil (Waitrose)
  2. Wiggles (Sainsburys)
  3. Colin (M&S)
  4. Charlie (Co-op)
  5. Curly (Tesco)
  6. (Joint) Morris (Morrisons) & Clyde (Asda)
  7. Carl (Tesco Freefrom Cake)

According to raw scores we see a slight reordering, wherein Curly and Charlie swap positions (becoming 4th and 5th, instead of 5th and 4th), and there is no “tie” on 6th place with Morris taking this position proper, and Clyde becoming 7th.

An ANOVA of the converted “rank” scores, to examine for caterpillars which were liked more or less to statistically significant degrees, reveals a model which is overall significant (p<0.001). Tukeys post-hoc testing shows a handful of significant differences between specific cakes, which can all be described with respect to Cecil (Waitrose) being significantly tastier than than Morris (Morrisons), Clyde (Asda) and Carl (Tesco – Freefrom).

No other Caterpillars were better or worse than any others to a statistically significant degree, although notable “trends” (i.e. where p<0.1) emerged wherein Wiggles and Colin were also each ranked better than Morris, Clyde and Carl.

A Spearman correlation between all “raw” scores (effectively making this a comparison of ranked data) against the parent supermarket’s customer satisfaction score (according to Which) shows a significantly positive correlation (p=0.036, r=0.212). This analysis excluded the freefrom cake.

Conclusion: Despite M&S inferring market superiority, in this novel analysis of caterpillar cake preference Waitrose’s Cecil was found to be the best-rated cake, with Sainsbury’s Wiggles coming in a very close second. Nontheless, Colin did still take a respectable third place.

Not only did Cecil take the overall top spot, but this caterpillar was also demonstrated to be the only one which was liked more than others to a scientifically significant degree. This finding can equally be framed as Morris and Clyde being the only “healthy control” caterpillars (i.e. not restricted by freefrom ingredients) to be disliked with a scientific degree of confidence more than other ones on the market. A final exploratory analysis showed that each Caterpillar’s worth is indeed tied quite closely to Which’s customer satisfaction score for its parent supermarket.


Leave a Comment

Your email address will not be published. Required fields are marked *