The study you point to here is a guideline on conducting studies. It was unfortunately not available online so I can't evaluate their recommended methodology. Looking for actual studies that tried to do randomized feeding trials, I found "A randomized controlled-feeding trial based on the Dietary Guidelines for Americans on cardiometabolic health indexes"[1] as a top hit, which fortunately had the full text [2] available.

Randomized controlled-feeding sounds good, let's check it out. After trudging through this for a bit I came to the meat of the methodology:

> Participants were provided a daily meal checklist (Supplemental Figure 1) that included each menu item with space for documenting the amount consumed; the time each item was consumed; a checkbox to confirm having only eaten study foods; a checkbox to confirm not taking any medications, supplements, or other remedies; space for documenting any adverse events related to eating the meals; and space for documenting any nonstudy foods, drinks, medications, supplements, or other remedies. They were also instructed to return all unwashed packaging; visual inspection was documented by the metabolic kitchen. In addition to the checklists and returned packaging, participants were educated on food safety as well as provided tips on managing challenging social situations while participating in a feeding study. Repeated reinforcement of the value of honesty over perfection was provided. Study coordinators reviewed the returned checklists with the participants to verify completeness.

So ... self reported with some extra steps.

[1] https://pubmed.ncbi.nlm.nih.gov/30101333/

[2] https://www.sciencedirect.com/science/article/pii/S000291652...