Armored 
Medical Research Laboratory 
Fort Knox, Kentucky 
First Supplementary Report 
On 
jmiECT NO. 30 - FIELD TEST OF ACCEPTABILITY AND ADEQUACY OF 
U. S. ARMY C, K, 10-IN-l AND CANADIAN ARMY 
MESS TIN RATIONS 
Subject: Numerical Requirements for Statistically VaMd 
Results in Field Test of Acceptability of Rations 
Project No. 30 
24 March 1945 ARMORED MEDICAL RESEARCH LABORATORY 
Fort Knox, Kentucky 
Project No*. 30 
SFMSA 430.2 
24 March 194$ 
NUMERICAL REQUIREMENTS FOR STATISTICALLY VALID RESULTS 
IN FIELD TEST OF ACCEPTABILITY OF RATIONS 
X. PROJECT NO. 30 - Pield Test of Acceptability and Adequacy of U„ S. 
Army C, K, lO-in-l and Canadian Army Mass Tin Rations. First Supplementary 
Report. 
a. Authority: Letter SPQRD 400.112 (Ration Acceptability, C and K) 
and 4th Ind, A.S.F., Office of the Surgeon General, SPMDO, 4 April 1944* 
b. Purpose: To determine with regard to field testing of acceptability 
of rations: 
(l) Number of men to be used as test subjects 
!2; Number of meals on which data should be obtained 
3) Duration of test 
2. DISCUSSION: 
a. Since field tests must be based on some form of sampling it is 
essential that test subjects be representative of the population to which the 
results are to be applied, and conditions in the field must resemble as closely 
as possible the expected conditions for which the test material is designed. It 
is understood that no test will be satisfactory unless the observers are trained 
and have absolute control of the subjects to enforce isolation from extraneous 
sources of food and to control the activity of the subjects. 
b. Moreover, in setting up a field test of acceptability of rations 
it is important to know the number of men, the quantity of data and the length of 
time required to obtain results which have statistical validity. If such inform- 
ation is not available in advance the test may be carried out on too few men, 
insufficient data may be gathered and the test may last too short a time; thus, 
even the most carefully collected data may give results which are not significant. 
On the other hand, if more than enough subjects are used for longer periods than 
required and the data obtained are more than sufficient, waste occurs in man 
hours of test subjects and observers. 
c. Study of the data obtained in The Field Test of Acceptability and 
Adequacy of Rations* carried out by the Armored Medical Research Laboratory at 
Topside, Colorado in the summer of 1944 permits the definition of size, time and 
quantity of data necessary for this type of field test. 
* A.M.E.L. Report on Project Ho* 30: Acceptability and Adequacy of U. S. Army 
Rations C, K, 10-in-l and Canadian Army Mass Tin Rations, 22 November 1944. 
1 d, It is emphasized that acceptability of rations will be affected by 
a number of variables such as ambient temperature, season, rainfall, stress and 
level of work. Accordingly any inference as to adequacy or techniques of test- 
ing which may be drawn from the Topside study must be interpreted with caution 
when applied to tests of other rations under different conditions. With these 
limitations in mind, however, certain useful generalisation may be made. 
3. CONCLUSIONS: 
a. In short tests (2-3 weeks) little accuracy is sacrificed by collect- 
ing opinion data only every third or fourth day. 
b. In longer tests (2 months) no significant accuracy is lost if data 
are obtained only once a week. 
c. Data from the first few days may show more variations than those 
obtained later on in a 3-we«k test. 
d. Where there are only relatively few items to test, a period long , 
enough for each menu or item to appear only 3 or 4 times will apparently give 
results little different from those obtained when the test components are used 
several times as often. 
e. If acceptability of rations for weeks or months is desired, it is 
necessary to test the rations for the entire period but data need be collected 
no more frequently than every 6 or 7 days. 
f. From the statistical point of view, groups of at least 2 platoons 
(60-30 men) are needed to give reliable information on the relative acceptability 
of items in a single ration* For reasons of organisation, supply, command, morale 
and losses from normal attrition, groups of company size (100-150 men) are more 
desirable. 
4. RBGQiatSBDATIOESi 
a. The information contained in this report be made available to agencies 
engaged in ration tests. 
Report submitted by: 
Charles R. Henderson, Captain, SnC 
William B. Bean, Major, MC 
William F. Ashe, Major, MC 
APPROVED 
WILLARD UACm 
Colonel, Medical Corps 
Commanding 
2 Incle. 
#1 - Appendix 
#2 - Figure 1 APPENDIX A 
MATERIALS AND PROCEDURE 
1. For purposes of this study, the following were obtained from the 
original acceptability data for B Company on K Ration for 19 consecutive days? 
a. Tbtal of good, fair, and poor ratings on each item for each day 
of the test* 
b. Total during 19 days of good, fair, and poor ratings on each item 
for each man of the company. 
o* From each of these totals was calculated the percentages of good 
ratings. Working with these percentages, analysis of variance methods were 
used to calculate "date X item discrepance” and “men X item discrepance", each 
within food groups. 
d* With these two statistics available, it is possible to estimate 
the reliability of results obtained from different numbers of men for varying 
periods of time. 
,.'•■4 ,. t - • 
. >/, v* • ' - • 
2. RESULTS 
a* It should be made clear that the discussion to follow is concerned 
primarily with comparison of different items of the same general type when eaten 
concurrently by the same men. The comparison between items to be used is the 
average percentage of "good" ratings by which one item exceeds another. If 
only a portion of the available data is used, the advantage one item has over 
another will not be exactly the same as when all data are used. The probable 
differences between results using the data for all men of E Company for nineteen 
(19) days on the K Ration and results using data from fewer days and data from 
fewer men are tabulated in Tables 1 and 2. 
1 
Incl. #1 TABLE 1 
Humber of Days 
Data Used 
Mavi.nmjn Rrror in Difference 
Between Means of any Two Items.* 
(Per Cent Rated Gk>od) 
1 
15o3 
2 
IOl.5 
3 
8.3 
4 
7.0 
5 
6.1 
6 
5.3 
7 
4.7 
8 
4.2 
9 
3.8 
10 
3.4 
11 
3.1 
12 
2.8 
13 
2*5 
14 
2o2 
15 
1.9 
* Mean Item A (19 days)—Mean Item B (19 days)—Mean Item A (n days) 
Mean Item B (n days)* P> .95 
TABLE 2 
Humber bf Men’s 
Data Used 
Maximum Probable Srror in Difference 
Between Means of any Two Items* 
(Per Cent Rated Good) 
10 
33.7 
20 
23.8 
30 
19.4 
40 
16.8 
50 
15.1 
60 
13.7 
70 
12c7 
do 
Ho9 
2 
Inclo #1 b. Illustrating the use of the above tables, coffee and bouillon were 
rated good t>7% and kl% of the time respectively by B Company the period 
studied, a difference of 20/6, Had data from only four days been picked at random 
for Company E for these two items, the difference between the two items would 
probably (P> fall in the range 20$ ±1% or \3% to The odds are better 
than even that it would fall between \T% and If data had been selected 
from only one day, the difference would probably (P> 95%) fall in the range 5% 
to 35%• From Table 2 it may be seen that using the 19 days* data from 50 of the 
110 men would be likely to give results as far from the actual results observed 
as would be the case when caaly 1 day's data on all men were used. Similarly the 
use of 2 days* data from all men would give results somewhat more accurate than 
would use of 80 men*s data for the entire 19 days, 
c. Figure 1, on which are plotted the day by day acceptability of 3 
beverages, illustrated why little accuracy is sacrificed when only a relatively 
small proportion of the different day's data is used. After the first 3 days, the 
rank of the items remained the same. Minor fluctuations in absolute ratings are 
not considered to be importanto 
d. The wide differences among men (even though in the same company) in 
their opinions of the relative merits of different items are illustrated in 
Tables 3 and 4—Table 3 shows the ratings given by three different men to three 
different The discrepancies among them are almost identical with 
those for all men of the company. 
TABLE 3 
Percentages of Time Different Men Eated Different Beverages Good 
Subject 
Lemon Juice 
Bouillon 
Coffee 
EL01 
50 
0 
59 
£L03 
> 20 
97 
79 
2112 
7 
0 
36 
Table 4 illustrates the wide difference of opinion among Company- 
mates as to the rank of each of 3 beverages. 
TABLE 4 
Percentage of Men Ranking Bach of Three Beverages 
First, Second and Third Choice 
Beverage 
First 
Second 
Third 
Lemon Juice 
33 
29 
38 
Bouillon 
24 
22 
54 
Coffee 
43 
49 
8 
Incl. #1 To illustrate further the differences which may complicate interpretation 
of acceptability questionnaires, some of the differences among platoons in the 
same company and among companies are presented in Tables 5 and 6j 
TABLE 5 
Average Percentage of Good Ratings of Confections by Different 
Platoons in the Same Company 
Platoon 
Sugar 
Gum 
Fruit Bar 
Caramels 
Sweet Chocolate Bar 
1 
91 
92 
55 
89 
82 
2 
82 
76 
59 
75 
81 
3 
94 
93 
72 
87 
83 
4 
89 
90 
75 
89 
88 
The platoon by item discrepance is statistically significant. 
TABLE 6 
Differences Among Companies in Acceptability Scores* Given 
Experimental C Ration Item Groups 
Item Group 
G Company 
X Company 
T Company 
Me ate 
2.74 
2.87 
2.59 
Cereals 
2.91 
2.94 
2.80 
Beverages 
2.76 
2.89 
2.74 
Confections 
2.97 
2.97 
2.93 
Biscuits 
2.91 
2.87 
2.64 
* Score calculated by the aquation 3 (No, of good) +2 (No. of 
Pair) 4l(No, of Poor) divided by the total number of ratinge. 
e. The results of this study indicate that the mcst promising way in 
which to reduce the labor of obtaining ration acceptability information is 
through reducing the number of days of data collection rather than through re- 
ducing the number of test subjects* 
f. If information regarding the relative acceptability of items for a 
short period of time (say 3 weeks) is desired, there appears to be little value 
in obtaining data more often than every 3 or 4 days. If one suspects that there 
will be a trend in acceptability upwards or downwards during a period as short as 
3 weeks men should be kept on the ration for that length of time. Collection 
of data only on each third or fourth day would give results very little different 
from data collected daily, 
g. If information is desired on long time acceptability, the only apparent 
way to obtain it is to feed the ration for a long time, Extrapolation of results 
from a short test is a dangerous procedure. If a long test is planned, even longer 
Incl. #1 intervals between data collection days would be possible. (Five to 7 days 
should give valid results), 
h. In contrast to the small variability from day to day in the relative 
acceptability of ration items, different men in the same company differ widely 
in their opinions of the relative desirability of different items. For example, 
the odds are even that any 2 men of accompany picked at random will differ in 
their opinions as to how much better one item is than another by nearly 30^, 
This being the case, the use of too small numbers of men in evaluating a ration 
is likely to result in data which are not representative of the population in 
which one is interested, 
1, Furthermore, the fact that platoons differ within companies and that 
companies themselves differ in their opinions of the relative merits of rations 
and ration items indicates that great care must be exercised in selecting test 
personnel representative of the larger group for whose feeding the ration is 
proposed, 
j, If it is desired to compare different rations, the test plan should 
be set up with at least 2 platoons or better a company for each ration; and in 
which groups are systematically rotated from one ration to another to smooth 
the differences among groups. On the basis of this study, it would appear that 
an entire company, or at the least two platoons, are needed to test ration items 
adequately for acceptability. 
5 
Incl. #1 PERCENT GOOD RATINGS 
FIG. I 
DAILY FLUCTUATION IN ACCEPTABILITY OF K RATION BEVERAGES 
Incl. #2 
FIG. I