Thoughts on: The Mindset of the US Specialty Chocolate Consumer | #PSC 223
OVERVIEW: Episode 223 of #PodSaveChocolate takes a look at a recently-released report from ICCR (formerly, FCCI) about what “specialty chocolate consumers” in the US profess to think – and know – about specialty chocolate.
Links below to watch LIVE and to view the archived episode.
Watch and comment LIVE or view the archived episode!
Subscribe (free!) to the @PodSaveChocolate YouTube channel, like this video, comment, and share this episode to help grow the #PSC community.
Watch and comment LIVE or view the archived episode on LinkedIn.
Join my network on LinkedIn to receive notifications and to refer business to each other.
Watch and comment LIVE or view the archived episode on TheChocolateLife page on Facebook (for 30 days, then watch the archive on YouTube).
Follow TheChocolateLife on Facebook to receive notifications and catch up on other content.
Episode 223
The Institute for Cacao and Chocolate Research (“ICCR”, formerly the Fine Cacao and Chocolate Institute), recently released a summary of a research project they are engaged in:
Citation: López Ganem, José J., Martin, Carla D., and Adamolekun, Dara O. "U.S. Specialty Chocolate Consumer Survey and Market Study." Institute for Cacao and Chocolate Research, July 30, 2026.
In email correspondence with one of the authors, I learned that “[a] detailed analysis [will be] published in a peer-reviewed academic journal.”
Given there is no confirmed date for the availability of the detailed analysis (which will include the survey instrument), either on a pre-print server or by an academic journal, this published summary has to be “relied on” by the organizations that funded the research, specialty chocolate companies, individuals (like me), entities interested in specialty chocolate (such as the FCIA), and by the ICCR itself in its communications and correspondence.
I took this look for many of the same reasons as for other examinations of published reports; as a shared exercise in how to read reports like this: What are some methods you can used to explore what the reports do (and do not) say, and how the authors reach the conclusions they do.
But, don’t take my word for things just because I wrote them. Read the summary, which can be downloaded from the ICCR page above [.PDF] closely – and crucially – critically. This is especially true if you don’t understand the mechanics of market research or statistics.
You might a fault in my work, disagree with my conclusions, or are concerned about my motivations and if I am biased. If so, I urge you to participate in the live stream and/or comment on TheChocolateLife, YouTube, LinkedIn, or Facebook to let me know.
TL;DR Summary
Understanding both the scope and limitations or research like this is important because these summary results will be used to inform (and perhaps justify) the opinions and decisions of many; what might the impact be of not having access to the cross-tabs? That’s why I think it’s important to cast a critical eye on what the report says and how its authors reached the published conclusions.
I am not criticizing the authors or the institutions involved – research like this is hard. Real hard. And essential. Will I critique methods? Yes. Point out what I see as limitations and limits? Yes. Point out statistical indicators and confidence that are not hedged? That, too. Ad hominems? No.
When the detailed analysis is released I can (and will) do a comparison. If I have made a mistake or misunderstood or misinterpreted anything, I will, of course, include that in my follow-up.
BLuF: Caveats and Disclaimers
When I read any research, I take a look at any prefatory remarks the authors include. Those remarks can frame expectations, and it can often indicate what the research was not intended to do/show.
What will be interesting to see – and everyone should check – is how the caveats and disclaimers that preface this publication surface in the published peer-reviewed analysis. In this case, the preface material on the ICCR web page is much more detailed than the preface material in the PDF itself. Furthermore, this “preface” material comes after the embedded PDF and readers may not read past the download for the fine print.
From the PDF: A few important notes on the scope of this research: this study is specific to the United States specialty chocolate consumer. It reflects what the U.S. consumer wants and what motivates their chocolate purchases, preferences, and debates. It does not, in any way, represent the global market. It is complementary to existing studies and it should not be read as an attempt to determine or contradict existing market or academic data. It is, simply, a representative random sample of people who identified themselves as chocolate consumers, capturing their motivations and understanding in 2026.
From the web page: The study is as much an X-ray of what the self-identified specialty chocolate consumer knows (and a test of that knowledge) as it is a validation of what the specialty chocolate subsector believes a consumer should know, should care about, and what the sector spends so much of its energy discussing when it comes to the information found on a chocolate bar, a chocolate product, or a chocolate tour. In other words, it tests theories — mostly anecdotal or company-specific — held by colleagues in the chocolate industry about the validity of the information that is out there: both in terms of its veracity, and its relevance to the consumer.
Also from the web page: Some readers may ask how a sample of 1,052 respondents can be called representative. The answer lies in how the sample was built. Fielding was conducted by Dynata, an independent survey provider, which selected respondents according to ICCR's specifications: self-identified specialty chocolate consumers representative of the general demographics of the United States and spread across all states and territories of the union. A carefully constructed sample of this size is a well-established standard in national consumer research — most nationwide polls in the United States rely on samples of roughly one thousand respondents, which yield a margin of error of approximately ±3 percentage points at a 95% confidence level. What matters most is not sheer size but the quality and independence of the selection, and this approach gives the findings a solid, credible foundation.
I am not ascribing motivations here – merely making an observation that any reader should take into consideration: relative placement (or lack) can have an influence.
Some Initial Thoughts
In fact, I read the entire preface, and, noting what was said, I formulated a number of questions.
- How is specialty chocolate defined … or was it left up the survey respondents? (Spoiler alert: it was left up the survey respondent.)
- Is there an unintentional bias in this following: “… tests theories — mostly anecdotal or company-specific — held by colleagues in the chocolate industry about the validity of the information that is out there”? (Spoiler alert: it’s likely.)
- The preface mentions a margin of error and a confidence level. Were those included in the PDF? (Spoiler alert: No.)
- The preface mentions that the survey respondents are “ … representative of the general demographics of the United States and spread across all states and territories of the union.” Is that supported in the PDF? (Fact-check: only partially.)
And, most importantly – to me:
- How does the “self-identified specialty chocolate consumer” framing influence what the published numbers actually indicate. (Hint 1: There is no such thing as a plain reading of the text. Hint 2: For this research, “specialty" is whatever the 1,052 people in the Dynata panel decided it meant to them.)
Three Examples of Methodological Rigor to Explore
I can only respond to what was published. But two question readers of the published summary should be considering are around the claims of:
- How the survey respondents were selected.
- How the demographics of the respondents are “representative of the general demographics of the United States [and] spread across all states and territories of the union.”
Respondent Selection (recruitment)
- Respondents were “recruited at random through a survey provider whose reach accounted for all states and territories of the country.” (The provider is named as Dynata.)
“Recruited at random” is doing a lot of unexamined work here. Dynata is an opt-in access panel; people sign up to take surveys for incentives. That’s a fundamentally different thing from a probability sample of the U.S. population. Panel-based “random” typically means random selection from within the panel pool, not random selection from the U.S. population.
That in mind, companies like Dynata employ screening criteria and quotas during the selection process to compensate (as best they can; quotas have an impact on selection) for this fact.
Demographics
| Generation | Ages (2026) | % of adult pop. | Target out of 1,052 |
|---|---|---|---|
| Gen Z (18+) | 18–29 | 19.2% | 202 |
| Millennials | 30–45 | 26.0% | 274 |
| Gen X | 46–61 | 24.5% | 258 |
| Boomers | 62–80 | 24.5% | 258 |
| Silent | 81+ | 5.8% | 61 |
- The preface says the respondents [paraphrase] represent the demographics of the US population and geography.
📈 Statistical notes about age distribution:
On slide 2 of the report, the number of Gen Z respondents is listed as comprising 2.9% of the respondents. Fact check: Gen Zers represent about 21% percent of the US population. That’s complicated by the fact that a percentage of Gen Zers (born 1997–2012) are under the age of 18. Assuming only those 18+, we’d expect there to be something closer to ~200 Gen Z respondents. (2.9% is about 30 people.)
This is an inconsistency that the detailed analysis should address.
🌐 Statistical notes about geography:
It would be extremely difficult to proportionally represent five separate territories (Guam, the U.S. Virgin Islands, American Samoa, Puerto Rico, and the Northern Mariana Islands) with combined populations less than 3.6 million in a sample this size without deliberately oversampling small units beyond their population share.
Combined, California, Texas, and New York would contribute supply roughly 280 of the 1,052 respondents, more than a quarter of the entire national sample, to hit proportional representation.
This is another inconsistency that the detailed analysis will have to address.
Open-Ended (unaided recall) vs Closed-list Choices
- Slide 7: What Style and Format of Chocolate Do U.S. Specialty Chocolate Consumers Like Most? A caption reads (in part): “Multiple selections allowed.” What this means is that respondents were picking from a list of choices.
Questions: 1) What criteria were used to arrive at these choices? 2) Is the choice list exhaustive? 3) How were the include/exclude decisions made? 4) Do the choices make sense (e.g., is a peanut butter cup a candy bar)? - Slides 15/16:
(s15) Two different measurements are used on the same slide, which could hinder comprehension. Recognizing and understanding that different measurements are being used and presented side-by-side requires familiarity with survey methodology; many readers won’t have the experience to catch it.
(s16) Specialty brand recognition was from a list of choices (multiple selections allowed). Questions: 1) Who curated the list? 2) What were the selection criteria for inclusion on the list? 3) Does the list make sense – Fran’s and Blue Stripes on the same list as Amano and Bonnat … how and why?
📊 Statistical bias note: It could be interpreted that these two pages are the survey defining what “specialty” (as opposed to “Mass Market” and “Premium”) means by example/exclusion. - Slide 17: I was unable to locate a separate “International Chocolate Challenge.” It’s not clear (there is no text on the slide) if the respondents were presented a list of choices or this was unaided recall and respondents are conflating “International Chocolate Challenge” with “International Chocolate Awards,” or if they are referring to the awards program run by the International Chocolate Salon, or something else.
On a statistical aside, given the self-reported familiarity about awards programs, it would be interesting to know how self-reported familiarity correlated to recognition. There is no count for “Extremely Familiar” and Cacao of Excellence and Sofi are not awards programs I would expect a member of the general public with low self-reported knowledge of specialty chocolate to be familiar with.
Where Do Things Sit, Then?
Knowing the exactitude of the numbers published in this summary (e.g., how closely the respondent population matches actual US demographics, Gen Z and geographic representation) will have to wait for the detailed analysis, but there is no mention of this in the PDF.
In fact, there is no indication, in the summary, of a margin of error or statistical confidence. Nor is there any hedging of the methodology, no place where the authors indicated that the interpretation stated is subject to unstated factors. For a good example of communicating limitations of the methodology and how that affects interpretation, watch The Surprising Things We Discovered In The Cost Of A Cappuccino by James Hoffmann.
An Aside
The study’s authors name wine industry colleagues “who provided us with key advice on consumer research survey design beginning in 2019.” The names mentioned suggested an avenue of additional research.
Wine consumer research, I learned, has a well-established body of literature on how self-reported expertise correlates poorly with objectively measured knowledge or tasting ability.
The literature explicitly acknowledges the Dunning-Kruger effect in wine consumers, where less knowledgeable tasters tend to overestimate their own expertise while highly skilled tasters tend to underestimate theirs relative to peers.
SOURCE: Aqueveque, C. (2018). “Ignorant experts and erudite novices: Exploring the Dunning-Kruger effect in wine consumers.” Food Quality and Preference, 65, 181–184. [Note: this is a single-author study with n = 193 wine consumers published in a peer-reviewed journal.]
Aqueveque's paper doesn’t warn wine (and by inference for this study, chocolate) researchers about something incidental to wine. It’s warning them about a measurement problem that shows up specifically in complex, knowledge-dependent product categories where subjective self-reporting is used to sort people into expertise tiers.
IF the ICCR survey’s methodology was adapted from a field that has spent years documenting exactly this failure mode in self-assessed expertise, it could mean the same known vulnerability was imported into the chocolate study rather than corrected for.
We won’t know until the detailed analysis is published.
In the Episode
I will explore some of the things you should be considering when looking at the numbers in this report and interpreting the results while waiting for the detailed analysis to be published.
Future Episodes
August News and AMA
Ceremonial Cacao “Certifications”: WtAF?
Cocoa and Health – new research, with regular ChocolateLife contributor, Keith Ayoob.
Celebrating(?) the 25th anniversary of the signing (on Sept 19, 2001) of the Harkin-Engel Protocol with confirmed special guests Terry Collingsworth and Miki Mistrati.
Episode Hashtags
#cocoa #cacao #cacau
#chocolate #chocolat
#PodSaveChoc #PSC
#LaVidaCocoa #TheChocolateLife
#PodSaveChocolate and #TheChocolateLifeLIVE Archives
To read an archived post and find the links to watch archived episodes, click on one of the bookmark cards below.
Clay Gordon
Episodes 201-300 (links to older PSC archives are in the top nav).
Clay Gordon