Every claim in the Short, and the sentence behind it
C1: The survey included three invented foods
The researchers added three invented names to estimate how often people respond to a word that sounds like food or answer without reading.
were added to the survey list to provide an estimate of how much someone will respond to a word which sounds like a food name or will answer without reading
Source: Method, "The Survey," printed page 4 (PDF page 12).
C2: The survey went to personnel from four US services
The report says it surveyed about 3,800 personnel from the Army, Air Force, Navy, and Marine Corps during fiscal years 1973 and 1974.
3800 personnel of the United States Army (USA), Air Force (USAF), Navy (USN), and Marine Corps (USMC), during FY 73 and 74
Source: Abstract, printed page i (PDF page 5).
C3: The fake names sat in a list of 378 foods
The survey presented 378 food names in random order.
The 1972 Food Preference Survey consists of 378 food names arranged in a random order
Source: Method, "The Survey," printed page 4 (PDF page 12).
C4: Most respondents marked "never tried"
The three fake names were the least familiar items in the list. The report records 84 percent for funistrada. For braised trake, the figure was 82 percent. For buttered ermal, it was 80 percent.
The least familiar items, i.e. those with the highest percentage of "never tried" responses were the three nonsense items
Source: Results, "Armed Forces Food Preferences," printed page 54 (PDF page 62).
C5: Some respondents rated each fake item
For each fictitious name, 16 to 20 percent of the sample gave a rating instead of marking "never tried."
From 16 to 20 percent of the sample did respond to one of these three fictitious word names
Source: Results, "Armed Forces Food Preferences," printed page 54 (PDF page 62).
C7: The instructions said how to handle an unknown food
The form told respondents to mark NEVER TRIED if they had never tried or heard of an item, and to leave the rest of that line blank.
If you have never tried the food item or have never heard of it, fill in the circle in the first column labelled NEVER TRIED and leave the rest of the line blank
Source: Appendix A, "Food Preference Survey Instructions," printed page 140 (PDF page 148).
C8: Staff were not supposed to explain the foods
If someone asked what an item was and described it incorrectly, they were advised to mark "never tried."
However, respondents were not told what any food item was. If they described it correctly, their view was corroborated; otherwise they were advised to check "never tried"
Source: Method, "Survey administration," printed page 4 (PDF page 12).
C9: A score of 5 meant neutral
Respondents rated liking on a 1-to-9 scale. Five meant neither like nor dislike.
Notice that the rating of 5 is neutral, meaning that you neither like nor dislike the food
Source: Appendix A, "Food Preference Survey Instructions," printed page 140 (PDF page 148).
What the survey doesn't say
The form had no separate question about eating. Its NEVER TRIED column did that job: people were told to rate only foods they had tried. So a rating implied having tried the food, but the report only says these people responded.
The report's stated purpose gives two possible explanations: a word sounded like food, or someone answered without reading. It does not tell us which explanation applied to each person. This result is not evidence that people lied.
The printed Results page calls one item "braised trade." The Method and survey form call it "braised trake." Table 2 uses that spelling too. The page-image check confirms that "trade" is printed in the report, not an OCR error. This page uses "trake," the spelling in the method and survey form.
The 164-page report has no erratum page. We found no correction or retraction of it.
The source
Meiselman, H. L., Waterman, D., and Symington, L. E. (1974). Armed Forces Food Preferences. Technical Report TR-75-63-FSL, Food Sciences Laboratory, US Army Natick Development Center. DTIC AD-A110512, DOI 10.21236/ada110512. Archived report PDF.
The report is a US government work. Its cover and documentation page say "Approved for public release; distribution unlimited." No Table 2 figures are used on this page; its scanned figures were not included in the claims manifest.
The Short
Watch it on YouTube: https://youtu.be/3MnZWbyZ37I