What do people really eat? Consolidating datasets to account for intake from composite foods, misreporting and outdated survey data.
This paper presents a method to consolidate diverse datasets and update German food intake data from 2006 to 2025, creating a consistent, bias-reduced reference of status quo diets that accounts for composite foods and misreporting to better inform health, environmental, and economic policy.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
To understand how a nation eats, scientists usually look at two different things. One is the supply side: the massive amounts of grain, meat, and fruit that arrive at ports, factories, and supermarkets, measured in tons. The other is the personal side: what individuals say they ate when asked in a survey. For years, these two pictures have rarely matched. The supply numbers often suggest people are eating far more than they report, while the survey numbers often seem too low to explain the health trends we see, such as rising obesity rates. This gap creates a problem for anyone trying to design better diets or predict how changes in farming will affect what ends up on a plate. If the data is flawed, the advice given to the public and the policies made by governments are built on shaky ground.
A team of researchers in Germany has developed a new way to stitch these two pictures together. Instead of treating supply data and survey answers as separate facts, they built a single, consistent model that links the market to the mouth. They started with the known quantities of food available in the country and worked backward, using recipes and cooking processes to figure out what must have been eaten to make those numbers work. By forcing the data to fit together logically, they could spot where the surveys were lying, not because people were trying to deceive them, but because of the natural human tendency to underreport foods that are seen as unhealthy or overreport those seen as virtuous. The result is a clearer, more accurate map of what people actually consume, including the hidden ingredients inside complex dishes like lasagna or cake that surveys often miss.
The researchers began by acknowledging a simple reality: most people do not eat raw ingredients. They eat foods made from them. A survey might ask if you ate a piece of cake, but it rarely asks how much flour, sugar, butter, and egg went into that cake. This makes it difficult to know how much of the raw agricultural products are actually being consumed. The team created a system that connects the market balance—the total amount of wheat, sugar, or milk available for human use—with the specific foods people report eating. They used thousands of recipes to break down composite foods into their ingredients. This allowed them to trace the intake of items like plant oils or eggs not just when eaten directly, but when they are hidden inside bread, sauces, or pastries. They found that when you account for these hidden ingredients, the amount of oil people consume is nearly eight times higher than what the survey numbers suggested on their own.
Once the ingredients were linked, the researchers turned their attention to the errors in the survey data itself. It is well known that people often misreport what they eat, usually eating less than they say or claiming to eat more of "good" foods. The researchers used a statistical method to adjust the survey numbers until they matched the physical reality of the market supply. They treated the market supply numbers as a fixed anchor, assuming that the total amount of food available is a hard fact, while the survey answers were the variable that needed correction. The adjustments revealed a clear pattern. People were significantly underreporting their intake of sugar, dairy, and meat products. In fact, the average adult was underreporting their daily energy intake by about 620 calories. Conversely, people were overreporting their consumption of fruit and juices, likely because these foods are socially viewed as healthy choices.
The study also tackled the problem of outdated information. The last major national survey in Germany was conducted in 2006, making the data nearly two decades old. Food supplies change every year, but surveys are expensive and rare, leaving a long gap in knowledge. The researchers used their new model to update the 2006 data to reflect the food supply of 2025. By feeding the latest market numbers into their system, they generated a synthetic picture of what people are eating today. This update showed that as the availability of certain foods changed, so did the way they were combined. For instance, as poultry became more available and red meat less so, the model showed a shift in how these meats were used together in dishes, with fewer meals containing both. This ability to update the data without waiting for a new survey provides a much more current reference for policymakers.
The implications of this work extend beyond just counting calories. The researchers found that the corrected data changes the entire story of dietary recommendations. For example, current guidelines suggest people should eat more oil. However, when looking at the unadjusted survey data, it seemed people were eating very little oil, suggesting a massive increase was needed. But once the hidden oil in processed foods and the underreporting were accounted for, the data showed that people were already consuming far more oil than recommended. This means the advice should actually be to eat less, not more. By providing a more accurate baseline, this method helps ensure that dietary guidelines are based on what people really eat, rather than what they say they eat or what is simply available on the shelves.
This approach does not claim to be a perfect measurement of every individual's diet, as no single method can be. It does not capture food grown in home gardens or food that is wasted before it is eaten. However, it offers a much stronger foundation than any single source could provide on its own. By reconciling the supply of food with the reported intake, the researchers have created a tool that can help simulate how changes in agriculture or policy will actually affect what ends up on a plate. It turns a collection of conflicting numbers into a coherent story, revealing the true shape of the modern diet and offering a clearer path toward healthier and more sustainable eating habits.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.