Clinical
Use of
Reports
a Digital Eating
W.
Computer Habit
0.
CASTER, PH.D.*
in Patterns
the
Study
of
W
more
ITt!
THE
it sumption practical this can should
view to
is detailed
aid now
of a modern possible to statistical
digital carry on
Thus the computer out much individual food have with con the extent beentermined which that we the
for tion and Food a intake seven records of daily
situation eating to the which nutrient
was
ideal patterns these intake
for and eating of the
observing measuring habits individual. de
habit
studies
Downloaded from www.ajcn.org by guest on June 6, 2011
records than a few years ago. be accomplished now
finding
would The ease suggests procedures
PROCEDURES notes day were interviews who time after of food to of food items, nutrient composition.’ typically cooked included: and and applesauce, chicken, butterscotch tomato salad, potatoes, spaghetti, cream; and and strawberry and pears. onions, shortcake; cereal, corn included eggs, fritters pudding; salad, orange fruit cocoa and cheese pears juice string apricots, chicken The creamed meat and beans, potatoes, and and juice, milk. syrup, cabbage souffle, and ice cookies; cream; cookies grapeboth Noon ate each typical etc. intake with meal period. compiled between the in servings, This data were kept by fortyfive food the each students the kitchen composition information by basis student and students consumpof these and spent checking of was to a a
explore
out what
these can
with
useful methods.
information The
present which patterns of are paper.
types of nutritionally be obtained by these report directs attention are of of use in individuals are in in the the
Individual on
to those methods the eating habit groups. minimum portions portions of the can Details of and the
defining some and
the
nutritionist
weights
methodology presented Emphasis is directed useful
held
to earlier latter
certain converted table
a
reference
of the paper nutritionally
to a description information that
Breakfast dry menus salad lettuce hash, S. creamed and fruit ice and and
be obtained by Food consumption were derived from program Public intakes and mont.
situation
these procedures. data for these studies a nutritional evaluation Branch relate eighteen boarding in this economic factors of to year school particular the old in the U. food boys school
of the Nutrition Health Service, and of girls Of
was were planning
fourteen in a private interest the
not
to
cream; ice peaches liver and
rice, and
Ver salad,
included: greens
cabbage
evening potatoes, loaf and meals beet
tions
menu terized
fact that the only
consideraguiding the
by
amounts
From University * Public the tional September A merican National This study the
and the the presence of a reasonable
Department of
meals were of almost variety
Physiological
gravy, characpie; sponge unlimited of foods. saltines,
tatoes, fish, beef, of presents nutrients
green
roast
squash,
lamb and
mashed
gravy,
potatoes
brown
and
rice,
apple
beets,
cake
string potatoes, squash, the
and
ice cream;
soup, beans, carrots potatoes mean daily corn sweet and
tomato
on roll ice and dietary the
juice,
cob, and cream; peaches.
cheese
baked white
and
pocake;
celery
Chemistry, Minnesota. Research Fellow
braised and
Table r intake of the
of Minnesota, Health Service Heart was 1960. Journal of Clinical on Institute. presented
Minneapolis, Special
before
the Washington,
Fifth D.
InternaC.,
studied.
Congress
Nutrition,
Not
same
all
date;
the
seven
in
day
addition,
records
some ere w Vol.
were
begun
missed
on and
1962
the
Nutrition
98
10,
February
Digital
Computer
in
Study
TABLE I
of
Eating
Habit
Patterns
99
Average
Daily
Nutrient
Intakes
of
Boys
and in
Girls These
in this Intake
Study Data
Together
with
Statistical
Measures
of
the
\‘ariation
Girls
Boys
Mean
01
r1
rd
Mean
ad
r1
rd
Calories Carbohydrates
2,236 290 91 73 1.12 1.50 12.8 8,180 1.16 1.90 11.5 130
455
268 38 10 8 11]
237 [10] 25 10 0.16 0.18 1.6 5,050 [0.04] 0.32 1.0 [40] 0.65
3, 300 412 144 117 1.83 2.27 18.7 0.68 0.37 11,060 1 .69 2.90 17. 1 96
496 86 32 17
0.47
170 [1] 11 14 0. 71 0.37 2.3 2,200 0.17 0.50 2.4 [271
145 [2] 31 18 [0.11] 0.36 2.6 6,100 0.20 0.55 2.8 30
(gm.)
Fat (gin.) Protein (gin.) Calcium (gm.) Phosphorus (gin.) Iroii (mg.) \‘itamin A (lU.) Thiamine (ing.) Riboflavin (mug.) Niacin (mg.) Ascorbic acid
99
13 0.33 0.30 3.4 3,440 0.27 0.41 3.2 144 [0.
0.55 0.37
0.42 0.69 0.57 0.30
0.48 0.55
0.19 2.3 1,720 0.16 0.24 1.7 70
0.32
0.33 3.6 5,160 0.29 0.33 4.0 50
0.55 0.35 0.58
Downloaded from www.ajcn.org by guest on June 6, 2011
0.69
0.73
(mg.)
some factor twentyeight days used. were studies. available
were analysis
incomplete. studies students represented
For only and a complete portions of the coefficients methods.2’3
the the the
correlation records four block of the
and for
di:Terences
the the numbers were ships data may
in
are be
nutrient studies.
studied, obscured
intake When
the from
figures large
visual systematic
shown groups
relationinspection
by of
statistical
consecutive of data total
that Other used
larger for some
analysis
were
of
variance
by
Correlation computer
obtained
by the obliterated and habit
shear by
mass of the a combination variations. therefore, it
data, of
or may methodologic eating necessary
be
tor
analysis
THE DIGITAL
procedure
COMPUTER
was
The centroid that of Thurstone.4”
IN DATA PROCESSING
fac
physiologic patterns, and within
In studying will be
to measure variation
discuss the various the data before of general relationships. of using a digital work stem directly which four results advantages procedures and
sources going into computer from can can
of a
In measure, which those hand ciples ventional
the are which
computation a digital
of computer
any
given uses
statistical procedures
mathematically equivalent with would be employed in a traditional Hence, the are the interpretation completely general of conapplicable
detailed study The advantages for statistical speed prin tained. apparent. and
the
calculation. applied to statistics
ease with In practice, (1) Statistical
be obbecome more In the involved,
to computer results. method of computation, the data upon which procedure can only those consistent relationships which However, is usually observe effect in the which in
A conclusion, by can be no better than statistical experimental it is based. A statistical be expected to point out followed the time differences or systematic are implicit in the is defined, raw data data. obtain group. it andmore
effectively guide because of anypast,
experimental the time
work. work
studies have tended to follow work, and occasionally have it by as much as several years beyond at which it would be possible to data It from the same experimental to such of is and hand between several comuse as now becomes practical statistical procedures,
further (2) elegant
once such a relationship possible to turn to the some detail itself. study are as the manner manifests present foodstuffs This a means
responsible
analysis, in routine studies in which the factor data. The cost comparison process is used tional fifteen minutes of computer time of determining of expertly supervised for the specific manweeks
nutri
100 putation.
can
Caster I2/k Vi =
cr2id

(3) The use of computer methods (3) greatly decrease the probability that arithmetic errors may influence results and conclusions. This is particularly true when (4) lengthy hand calculations are involved. For one thing, process of matter of
Large
T’/nk
+
ka’,
=
ni
D2/n
Vd a2j +
fla2d
=

T’/nk
1
k

a total complete a few minutes
of
recheck of recalculation by
experimental
results is
data can
by only methods. be
the (o) a

.
:y2
02jd

D2/n
nk


I2/k

+ +
1
T2/nk
\id
k
n
computer
( 4)
operation
samples
handled tively the will and The
with
time,
little
than
small
samples
more effort, or computer is needed for the relaof data represented The upon the making
where, nutrient
days,
a given table of data showing the intake values, y, for n individuals on k the sums of the nutrient intake values the individuals I,, 12, II summing are T D1, is the D2, over this period I,,, respectively; all individual D3 grand
Dk,
in
present study. be dependent
program used.
exact upper limits the specific computer for use of a digital
for each of in days are daily each
of
totals day of and y.
values reof all total
first
step
toward
computer procedure is to intake data to punchedcard on magnetic or perforated
the nutrient These terms are measures of: or record it o,, the within individual (intraindividual) input tape variation. This is a pooled value for day to in the manner specified by the computer day variations for each of the individuals within center. The computer operation may occupy the group and represents the average day to only a few minutes. This operation requires day variation to each individual about his both the input data and a “program” conown mean. sisting of some hundreds or thousands of wc1, the within day (interindividual variamachine commands that control the computer tion. This is a pooled value for the k days of during the entire operation. It is this program the deviation of individual values on a given that is the “brain” of the operation. It may day from the group mean value on that day. require a statistical mathematician with a O,d, the interaction term, is a collection of background in computer technics some months error terms. It includes the random variaof work to develop and test this program. tions introduced into the data by method errors Once tested it is available to use over and over and by the fact that individuals are not again in the solution of many problems. The perfectly consistent from day to day. Contribsolution to a problem is usually obtained, uting to the method errors are at least two directly, or after a separate tabulation step, effects: (1) discrepancies between the precise in printed form on large sheets of paper. amounts of each foodstuff consumed by the
METHODS OF EXPRESSING RESULTS
transfer form paper
spectively, nk values
Downloaded from www.ajcn.org by guest on June 6, 2011
subjects
and
the
intake
estimates
recorded
in
the data, and (2) the errors inherent in the use There are a number of terms and abbreviaof food consumption tables. tions which are used in expressing the results u,, that part of the interindividual variation of these various statistical procedures. In the (not attributable to random error) which is discussion of the analysis of variance results, consistent for the individual from day to day. the variances ofOid, Oj, cr2, and #{176}wd are oi, that part of the day to day “general related to the mean squares, Vd, V, Vd, up and down movement” of the group values Vwd, and V,, as indicated in the following which is over and above that expected from equations: random “error.”

(1)
Vwj
=
2wi
=
tT2id
+
The

intraclass are
correlations, expressed and
as
defined computed
by as
a,j
=
nk

n
Snedecor,6 follows:
=
(2)
Vd
=
cr2wd
=
cr2id
+
cr2
=
nk

n
(Vwd

Vid)/(\wd)
=
(cr2)/(o2j
+
O2jd)
Digital The individuals, range how each values value from different
set.
Computer with varying to
in respect
Study to a tells from
of
Eating
Habit
Patterns
101
intraclass r, is
correlation an index
over
©
of values from zero clearly and consistently other are of different of r,, any Here to determine the longterm individuals.
+ 1 .1), that separated
nutrient intake The larger the extent is as to which distinguishably data in the the pat
the greater the one individual that it will which the
V,d)/(V,)
from
of all other be used individual person’s
individuals
extent terns
rd
=
a measure of eating habit nutrient intake. + with
a2id)
#{174}‘
,
,
\
\
#{174}
of the relationvariables.
FIG. ship for
1. discussion
A diagrammatic four highly
representation correlated
between
See text
(V,

(a2d)/(a2(j)/(a2d
The
days,
intraclass
rd, is a
correlation
similar index that
respect shows the
to the in
manner the Now
shown origin if variable
in by variable S
Figure an but T a
1 as angle has
two whose a
vectors cosine
extent to distinguishably of the other as an index preparation also control high school Another ship between nutrients, the usual cient.
=
which
the different
data from This
for
any given day the data for all will be which the
joined at is 1)80. =
relation
Downloaded from www.ajcn.org by guest on June 6, 2011
high smaller
cor
with
much
days. of the staff, the students. method two designated product daily
measure
extent to by controlling nutrient expressing groups x and moment of
for
with variable R, it is quite possible foodrelationship between these three the menu, could be shown by a plot such as intake of these Figure 1 in which the point representing able T lies in (on very close to being the relationplane defined by the origin and the individuals means coeffi demonstrate or representing of can now tation. diagrammatic
can
no
used one
that the variables shown in vanin) the points U will
y, is by correlation
variables be defined another If rRs = ritu = representation be expressed be 1 is of
R in aspect
rsu
and S. a fashion of
=
Variable that represen
this
_____________________
\/(Nx2

Nxv

xv________ y2

0.80, then the of the total data in three to give If express a single dimena perfor this model, the plane 5, and resulting depend projected. T. any rein
longer
precisely
(x)2)
(N
(y)2)
When volved, relations variable plicity, directly. which providing
many the with such Factor is designed a simple result showing
different is the a
variables massive correlation In to table
are of of
plane sions. incorspective
but must Figure view were
expressed an attempt this situation. to
all of the others. a table is difficult analysis to aid with pictorial
it each reason its multi lationship then point interpret device defined byThis of error. upon the care and
necessary
in a strictly U could be origin some in
two dimensional projected onto and points distortion R, and U it would was
is a statistical this problem representation
by the involves The exact manner
position which
of
the relationship in this group factor analysis
between each of correlated plot, the higher the
of the members variables. In the correlation closer
between any two variables, their points will lie. For a more detailed procedure variables, vectors arbitrary between consider the
interpretation case of four
a With cision, variable together variable of this This
there would be little loss of prethe representation would show that U is closely correlated with both R and variable S. smallscale example demonstrates must factor be kept in mind in analysis results. of N variables may It give a is the lower
arbitrary
R, S, T and U as being represented of unit length extending from origin. variables If rRs R
=
several points which interpretation of bythe A representation an precise require problem as of many factor as
0.80, S can
the be
relation represented
and
N dimensions. analysis to
102
Caster
‘C
/
A
Prot
Cd
B1
Fe Nia B2
/
/
Garb Prot CAL Fat
/
Nia
/
Fe
#{149}\ CAL
B,
c,J
Garb
Downloaded from www.ajcn.org by guest on June 6, 2011
2A FIG.
2B plots showing the relationship between the intakes of different nutrients in this study.
2.
Two
factor
analysis
dimensional (preferably sional) representation
as
of accuracy Fig. in
twothese 3) all two (see
little
loss problem
of (see between
amount of or threedimensame data with tal variables, proximately as possible. A
residual
information.
Experimen
represented unit length,
as are
vectors plotted loadings at the
of on
apthis
“neat” relationships represented sions. provide between sentation representation In the solve jecting different discussing
is one variables
in which can
the axis system coordinates be vector. is roughly
using the factor to locate the point In Figures 1 and perpendicular
as the end of
Other an or
accurately problems a solution.
or three Fig. 2) Here the
dinien the do not axis choice
2 the first factor to the page and
as simple
inexact threedimensional reprean exact but incomprehensible in four or more dimensions. 2A more an attempt adequately model hyperplanes different
is extends from the origin to the reader’s eye. The second and third factor axes are horizontal and vertical, respectively. In Figure 2 (right) the fourth factor replaces the third as the axis. analysis These There and differ are different procedures largely in is art, for the methods expressing manner iii the of the any is, of
case this the
of Figure problem
fourdimensional threedimensional these two
vertical is made to by pro factor results. into two and threedimenwhich present
the axis system state of the that
constructed. In any interpretation
sional approximations. A second point may considering the example relates to the fact that used nature in factor of the
also be discussed of Figure 1. the coordinate system at best, doubtful. For present purposes it is better to ignore the axis system and analysis is related only to themuch focus attention upon the relationship between computational procedure, and are introduced or each variable and all method of factor NUTRITIONAL Thurstone,4 an orI summarizes is constructed in Table data are which associated of listed this the indicate with study. three the the of the others.
the data byrelationship axis This one
is highly dependent upon between the data points and system (as opposed to another)
will change as new variables deleted. In the centroid analysis as described by thogonal coordinate which the first axis line passing through axes axes) are perpendicular and extend in
RESULTS
system is a least squares straight the data. Subsequent the to it (and direction of
the After standard nature measurements
nutrient each mean deviation of the that
intake value figures con
all previous the largest
variations
Digital
Computer
x x
in
Study
of
Eating
Habit entitled respect the of a
rd,
Patterns r to extent the given the (the intraclass individual) to which individual nutrient, intraclass provide
103 corprovide
The columns relation with a measure of habit pattern his
columns
the eating determines whereas correlation the of the
intake respect
of entitled to
with
days,
a measure
extent to determining In the some
x
which the menu is the factor the intake of that nutrient. case of calories, carbohydrates B vitamins listed, sample from with in the O’id. intake day no intraclass are indicating that the variations to day were the nonsystematic Neither menu whims of extreme intraclass furnish had these
largely and in this between small in
of the
correlation
values population individuals comparison included food
upon
0
C)
variations changes nor effect in A with Vitamin best this some
preference the
a measurable nutrients
Downloaded from www.ajcn.org by guest on June 6, 2011
0
0
study. At the other of the largest A intake data
to day analysis student basis of and their plot all showing of the the others intakes. relationship when comof a day
is vitamin correlations. one of the
examples
l’IG.
3. on
A
factor the
between pared
each
nutrient
of
respect obtains
the
menu to this
effect, reflecting in controlling nutrient. A the boys a menu, intake with this,
the importance the intake with similar situation of fat and meat in the heavy meat meat item
tributed is a physiologic are between Whenever
d d)
to
that
mean. of the in the of
is
The
first
of and
measure errors
methodologic data, while o,
these, proteins, and diet. eaters. appeared ferences and the differences subboys. difference whoprotein
with respect to largely associated Some of the Whenever on the in protein correlated In in the a
intakes with the were large occurred. there
substantial
measures individuals o or ad may in degree patterns, are the under menus
the consistent and days,
large (in comparison
differences respectively. with to of a
individual difSimilarly, were marked of the different the boys, then, individual in the calcium, case of phos
one
nutrient stantial habit which plans One from results
are
conclude question by or and either (2) control
that is
the controlled (1) to of the
intake
the milk intake total data from substantial up not only in the case of
individual day person
eating one variations
finds
between
day
showing but also
of the problems one nutrient from the fact expressed inwidely it I.U.). For is desirable The
the food. phorus and riboflavin. Among the girls, two items in particular comparing the figures with those from the next stand out as being determined largely by the that the different nutrients individual eating habit pattern. These are different units (calories, vitamin C and iron. There is a small but in such a comparison, to remove the use of the intraclass data there positive from intake correlatake, of correlation (r= these nutrients. is associated 0.36) The with between vitamin the the C tendency in
prepares
grams, fore, these
units.
of course,
tions r and rd provides one simple method for to ingest large quantities of fruit juice and doing this. To simplify the presentation, all salad. The marked differences between girls the small and nonsignificant intraclas cor with respect to their intakes of either one of relations have been dropped out of Table i these nutrients can be seen quite readily upon so that only the larger effects are evident. examination of the raw data. Take for ex
104 ample ular and the pair similar nutrient of girls, caloric intake both intakes, the entire of seven per cent between C and all of the clearly For girls the shown two same The lie, the Thus with accompany is here calories fat. different points closer higher far more methods. record with and period of successive more the other iron of one normal it the
Caster particwill of this weight startling. study are useful Indeed, if with this raised the field, approach. lies in the an objective but the are not results particularly were widely past might of of obproto ex
found that over one girl on each gested the The iron, can be factor other about girl. relationship pictured analysis 50
be at variance in study, perience days inproperly be didthis statistical the results tamed by of
conclusions of some questions the The importance
concerning
validity were
than intakes
fact that they and quantitative used with situations equal in both
vitamin
nutrients cedure that can be describe other dietary by the use of studies. this purpose and animal7 will be contwo hemiin Figure In the three tical plotted groups toare tend
INDIVIDUAL
ease human
the data for both sidered together. spheric 2 can factor be considered
boys and As indicated, plots as of the space. points them. together
DIFFERENCES
analysis
preceding
section
two
different
statis
dimensional views in fourdimensional gether correlation is shown
trients particular
methods have of interrelated to be
(2)
been used nutrients by (1) planning To this the to define differ
to define the whose intakes
Downloaded from www.ajcn.org by guest on June 6, 2011
controlled the the person food. and help individuals The
eating
habit
patand data in in
any
two between grouped
theterns or preparing each nutrient to the other nuhelp it in the at this study. which carboeating obtain of student
the menu extent the the from manner others
characterize school, some
nutritional
situation
which diet associated protein and +0.90).
tend to which with and
under we The find
Closely hydrate,
calcium
intakes
(r
min
=
phosphorus In one
are highly correlated of the plots they
habits. information and the
obvious concerning manner in
next each which
step is to individual he fits into
closely might
associated with fat, protein and B2, the group of nutrients which expect to find in milk. Likewise,
arethis picture. Figure 3 shows a factor analysis which provides a direct representation of vita plot relationships. one the interindividual the To obtain Figure 3 the same basic data were from were intakes day as a different viewpoint. The considered as the variables, of the different nutrients over considered all data correlation nutrient nutrients correlated 2, this 3, etc. fashion the obwere norprogram, intake data on each with The was a
enrichment complex, vitamin B1, niacin and iron are highly correlated tions ranging from =r 0.60 to 0.80). Two nutrients, vitamin C and are notable for intake This their (r = low poor +0.02 correlation total caloric respectively). tant which may control Hence, vitamin fivefold and
considered vitamin B2, (correla individuals and the vitamin A, the four +0.07, imporwith servations. malized and then (fortyeight 4 days)
correlation is
period were To be precise, required by the column of items = twelve Subject
the
because it are poorly vary most
is precisely related to and are
those nutrients caloric intake that of most susceptible
for
1 was
in by individual eating habit patterns. while longterm average vitamin A and subjected to factor analysis. Since the major part of the information in this analysis can be C intakes can vary as much as fourto between different people eating at the expressed in two dimensions, the representaof the data is fairly simple. The points same table, it is difficult to find, within thistion 3 represent the twentyeight different particular school situation, individuals who in Figure teenage boys and girls who were the subjects have the same total caloric intake and yet study. Those individuals whose nuhave consistent differences of more than two of this intakes most resemble each other are fold between them with respect to any of thetrient other Up nutrients. to this point, the results and conclusions represented The points as points which with an lie “0” close are together. the girls indicated
to comparable correlation
set from Subjects matrix obtained
Digital and those in lie with an “X”
Computer are the boys.
in
Study
of
Eating
Habit
Patterns subsisting under
105 quite intakes both (1)
be seen points with the ward
Figure 3 that about in a single cluster
twothirds which is
a curved line. Points representing girls are a substantial distance away the bottom) and three of the boys removed from the the top). then are the people deviate in a substantial
average. Looking
It will concerning other dietary of the different In the present enclosed were characterized six of (tothe with are cluster tion amount daily between and intake
groups situations. study with
the nutrient respect to
somewhat (toward These habits
group
main
type of variation figures and (2) each nutrient and
associated the correlaall of the signifsurvey the in deciding food effirela
rest. whose eating icance way from thework. diet records of Figure (averaging tive represented
These data have for the planning As sizes Anderson8 of q1 and efficient data. can method Since
a very practical of nutritional has be indicated, used of the gathering statistical
the at girls hand C
we points had over were
find in very 200 also
that
the
three
the lower left high vitamin mg. high per (averaging
day).
On several days one over 1 quart of orange or some other fruit represented portion vitamin by points of the graph C levels did but
ciency of a study is inversely proportional to size of the variance, it is generally most intakes the to sample where the variability is over 15 mg. per day). efficient greatest. For example, in the case of the of these girls consumed A intake data it is found that the juice or grapefruit juice vitamin juice. The three girls day to day variation, aj, is particularly large. Their iron in the lower not had have quite consistently as
corner intakes
the most by consumption 3
Downloaded from www.ajcn.org by guest on June 6, 2011
central Hence, if one is interested in obtaining a good of the average vitamin A intake of a high estimate high group of people, it is much more efficient to obtain records over a large series of days and vitamin A levels (in excess of 10,000 I.U.). a relatively small number of people rather In general then, these six girls had a strong from to and fruit intakes. iron intakes phosphorus eat large juice, Their were intakes amounts and tended vitamin high while tended of C, salad to have vitamin mate than to obtain for the records from a large number if one is a survey rial milk low people of attempting A a few days. to interpret Alternatively, the data from
tendency
and and
low side. One of the in this same direction.
boys also On the
has been limited to one or two days, he their calcium that to be on theshould be aware of the fact that vitamin A had a tendency intake figures, of all the nutrient intake figures, other hand, themay be the farthest away from representing a toward longterm fond C data source amounts intakes of mized As average are of for these people. Vitamin large be mini. results C intakes their low allows by the beas quite different variation which by in that is the a needs to analysis vitamin because intake completely other found nutrients, or some hand, to exist such between questions survey contribution the individuals as the variables.
three boys whose points were displaced the top of Figure 3 were not particularly of salads and vegetables but ate large of about meat and 2 quarts potatoes, per day. and had milk
by sampling indicated
methods. the factor
COMMENTS
of Figure 2A, are of particular few the data incidence no attempt or correlation thethem to be
vitamin A and importance with total On which of and vitamins, possibility most caloric more the were controlled
On will be nutritional eating student of this procedures
the
basis made to
of
these discuss of
habit
effect patterns
specific peculiarities in a general high Rather, it is the the study
individual whim. in school high correlations purpose tween the intakes
population. presentation that eating The relate can
certain phosphorus raises
to describe be used to habit patterns following only to the
statistical between calcium and char the B complex concerning andmethodology. Probably the the
acterize the and groups. conclusions
of individuals discussion specific
of simplifying interesting
group
studied, although are capable of
the statistical providing similar
methods used of these methods information within the group
occurs when are considered
106 Even the with present the study, “normal factor group” analysis considered allowed
Caster
survey in unit wish Ahn assistance, of and of supported the the financial in under to and Mrs. and University assistance part by of a University the thank the direction Miss Judith staff of to P. of of Martha Dr. Arthur for for project. from the S. Mr. their their This techtechniwork Graduate Analysis Keats.
one I
also
J.
Thomason, Numerical
to obtain (1) the names of those eating habits differ from those part of the group, and (2) some the extent individual. tion which to correlate physiologic and
persons whose nical of the major Center indication of cal
was School
Philip
Hoffman the this
Minnesota grant
nature of this difference in each These are the types of informaare particularly sought in attempts dietary states peculiarities and clinical
SUMMARY
Minnesota.
with conditions.
different
1. BOYD, Food U. 2. MCGOWAN, multiple the GUERIN, analysis Program 4. 5. THUR5TONE, 1935.
HOLZINGER,
REFERENCES E. Value S. Public D. F., EADS, Table Health C.,
ROSE,
M. for
G. Calculation
and SANDSTEAD, of Washington,
KENNEDY,
H. Diet L. program
R. Records, 1947. A for
Service,
Available computer
statistical technics, can
methods now be
that, applied
by
digital quickly
J.,
1103 1956.
and
regression 9: RR127, M. program 417 L. University K. PF L. Multiple
and
correlation Central
Univac
Scientific,
Exchange and 704. Mind. correlation SHARE
and easily to large volumes of data, to take the usual food consumption group of individuals and to derive fairly clear picture of the of the different individuals They within differ nature also ning allow one to locate whose the group in a significant
allow data from
one for 3.a it a
Newsletter,
regression for the 1959. of the Press. H. University IBM
Downloaded from www.ajcn.org by guest on June 6, 2011
eating habit patterns within that group. the specific individuals
CR1, Vectors of
Chicago, Factor of Chicago 2436.
Chicago HARMAN, 1941.
J.
Chicago,
and
H.
eating way, and these
habit patterns to point to the They plandata.
6.
Analysis. Press. SNEDECOR, Ames, of the the Nutrition, G. 1946. W. fatty R.
and extent of provide suggestions and interpretation
differences. concerning the of dietary survey
W. Iowa
Statistical Methods, p. State College Press. R. T. between dietary andblood components markets. J.
7. CASTER,
0. and acids 73: 337, L. of hog
HOLMAN, of heart
Statistical study linoleate and lipids. J. in the An:.
relationship
ACKNOWLEDGMENT I Branch permission wish of to the to use thank U. S. these my Public data colleagues Health which in were the collected Nutrition their by a 8.
1961. in 1947. two
ANDERSON,
Use of variance prices
Service for
analysis Statistical
A., 42: 612,