OOCUM6.N (*
ROOM
Reprinted from Social, Forces
Vol. 13, No. 4, May, 1935
APR 14 1941
THE CONTRIBUTION OF RESEARCH TO RURAL
T- RELIEF PROBLEMS *
S.vo (UU O.
') Py / j^7v DWIGHT SANDERSON
Federal Emergency Relief Administration
THE Census of Unemployment Relief
Cases made by the Federal Emer¬
gency Relief Administration in Oc¬
tober, 1933, revealed 1,154,960 rural fami¬
lies receiving relief, or 36 per cent of the
total relief load. By July, 1934, this had
grown to approximately 1,500,000 fami¬
lies, rural counties in the drought areas
having the highest relief rate in the
country.
Inasmuch as over most of the United
States there had been no well developed
rural social work prior to the advent of the
emergency relief administrations, and con¬
sequently almost no trained workers, there
were no means of obtaining information
concerning the problems of rural relief as
was possible in cities through established
urban agencies. Furthermore, almost no
data existed as to the extent or nature of
rural relief in previous years, beyond the
barest facts of number and cost of cases in
the reports of state welfare departments.
Obviously, for the intelligent administra¬
tion of so large an undertaking, some basic
facts with regard to the characteristics of
the clients and their needs are highly de¬
sirable. This suggests an important dif¬
ference in the type of research feasible un¬
der an emergency administration and in a
permanent bureau of the government.
Thus the research of the U. S. Children's
Bureau is published primarily for the in¬
formation and education of the public, but
the research of an emergency administra¬
* This paper was read before a joint session of the
Section on Social Statistics of the American Socio¬
logical Society and the American Statistical Associ¬
ation at the annual meeting in Chicago, December,
1934-
tion is chiefly to inform its own adminis¬
trators; it is like the intelligence service of
an army.
Rapid but accurate research concerning
so large a relief population over so wide an
area met three major difficulties: (1) the
inadequate records due to the untrained
and inexperienced personnel in rural
county offices; (2.) the extensive area to be
covered; and (3) the dispersion of the
population, involving high transportation
costs for enumerators.
The first rural study undertook to meas¬
ure the differences between relief families
and their non-relief neighbors as of Octo¬
ber, 1933. A total of 17,700 usable sched¬
ules was obtained from 47 counties in 19
states. The schedule covered household
composition, history of occupation, and
data on the economic status of the house¬
hold. Among the significant facts thus
far revealed by the study is that only 37
per cent1 of rural families receiving relief
operated farms, the proportion varying
from 10 per cent in the Corn Belt to 58 per
cent in the Spring Wheat Area, the per¬
centage being 16 points, or nearly one-
third less than that of the general popula¬
tion as reported by the last census.
The proportion of farm owners receiv¬
ing relief was smaller than that reported
by the 1930 Census, but the proportion of
croppers on relief in the South ranged from
approximately one and one-half to three
and one-half times as great as their pro¬
portion in the census, and the proportion
of other tenants was also larger. Further¬
more, it was the operators of the smaller
1 Preliminary figure.
483
RESEARCH AND RURAL RELIEF PROBLEMS
farms who were on relief. In three of the
most important agricultural areas, farmers
on relief held properties only one-half as
large as those of neighboring non-relief
farmers. Rural relief households aver¬
aged about one person larger in size than
neighboring non-relief households. The
number of dependents per gainful worker
was nearly twice as great in relief house¬
holds as in non-relief households of cor¬
responding size. Obviously most of the
rural relief households were those which
were previously marginal families.
The second study was of relief turnover
for the six months of November i, 1933 to
April 30, 1934. It included all cases in
49 counties in zi states and involved a
total of approximately 99,000 schedules.
The data were gathered by family inter¬
views and included household composi¬
tion, employment status, and record of
last employment for all persons 16 years
of age and over, and reasons for opening or
closing cases.
Among the significant points on which
the study throws light, the following may
be mentioned.
1. The extent to which the rural relief families
coming on and going off relief between
November 1953 and April 1934 were simi¬
lar to -those enumerated in the October
19)3 .Relief Census, particularly in regard
to the over-representatiao of larger fami¬
lies and of children under 16.
l. The relative influence of loss of CWA jobs,
crop situations, etc., in bringing chese
families onto the relief rolls, and of such
factors as CWA, private industry and ad¬
ministration rulings in taking them off the
relief rolls.
3. The extent to which these rural counties
attempted to differentiate between unem¬
ployment relief and other forms of relief,
as shown by the proportions of families
on "unemployment" relief which con¬
tained no gainful workers.
4. The extent of urban-rural differences as
shown by the rate of opened and closed
cases.
An analysis of the rural regions showing
high relief rates with regard to climate,
soil, and other factors, revealed several
areas in which the need for relief, although
precipitated by the depression, is not a
temporary matter, but is based on factors
that give it a semi-permanent or long¬
time character. Six such areas were
studied: the Appalachian and Ozark High¬
lands, the Cotton Area of the Old South,
the Western Cotton Area, the Lake State
Cut-Over, the Short-Grass Winter Wheat,
and the Short-Grass Spring Wheat Areas.
The first three areas are the most impor¬
tant, as they included 40 per cent of the
rural relief cases of the United States in
October, 1933. The study of them was
undertaken with special reference to re¬
habilitation. Data on the basic economic
resources of the county, the reasons for
receiving relief, the type of aid, if any,
probably required in the future, and the
qualification of the head of the household
for rehabilitation in fanning or in other
employment, family composition, eco¬
nomic status, occupation, etc., were
gathered. The study covered 65 counties
in Z3 states. The statistical data were
gathered from the case records in the
county relief offices and from the case
workers, and a rather thorough quantita¬
tive and qualitative description of the
agricultural and economic resources, the
history and methods of relief administra¬
tion, and the programs of rehabilitation
being used or proposed was made.
The results of this study are now being
tabulated. They are of especial useful¬
ness to the administration in indicating
the following sorts of data:
i. The extent to which the rural relief cases
reside in the open country, as contrasted
with villages and small towns,
a. The usual occupations of these persons, par¬
ticularly the proportions of farm operators
and farm laborers, the total proportion
usually engaged in agriculture, etc.
SOCIAL FORCES
484
3. The extent to which the heads of the relief
households must be adjudged incapable
of self-support, due largely to the propor¬
tion of aged persons among them, and the
extent to which this varied between the
Negro and the white groups.
4. The extent to which "broken families,"
chiefly women with young dependent
children, are found.
5. The extent to which the recent urban-rural
migrations have caused variations in re¬
lief rates.
6. The extent of insufficient schooling of the
heads of relief households, and the pas¬
sible implications of this factor for a re¬
habilitation program.
As a sequel of the first survey of rural
relief and non-relief families, the same
counties were studied in June, 1934, as of
May, to determine the distribution of rural
relief households by residence, the prob¬
able re-employment prospects and the
probable trend of relief during the summer,
the resources available for the rehabilita¬
tion of farm families, and the prospects for
rehabilitating village families on the land.
Experience with this survey led to the
conclusion that, if such a study could be
conducted over a large enough sample of
representative counties and the results
could be quickly assembled and reported,
it would be of considerable practical value
and that a repetition of such a study in the
same counties two or three times a year
would give an intimate picture of the
changes and trends in the rural relief
situation not otherwise obtainable.
From experience with this and previous
studies, it appears that the most important
basic problem is to obtain a valid sample
of the counties of the United States. The
county was taken as the unit, for it was the
only one for which uniform statistical data
could be obtained either from the census or
from relief statistics. The whole proce¬
dure has been described in a recent article2
* A. R. Mangus. Sampling in the field of rural
relief. Journal Amirican Statistical Association. De¬
cember, 1934.
by the analyst who had charge of the
sampling. It seemed evident that, if all
counties in the whole country were taken
as a universe, the heterogeneity of the as¬
pects would be so great as to make a
sampling process fruitless. The country
was therefore divided into areas within
which there was sufficient socio-economic
homogeneity to use given factors as a basis
for sampling: The East Cotton Belt, the
West Cotton Belt, the Corn Belt, the Hay-
Dairy Region of the Northeast, the Self-
sufficing Farming Area of the Appala¬
chian-Highlands, the Wheat Growing
Region, the Lake States Cut-Over Region,
the Range Region, and the Pacific Coast.
The criteria used are the percentage of
land in the major agircultural products,
in some instances the proportion of farm
products consumed on the farm (the self-
sufficing farm area), or the percentage of
land in farms (the range area). The
Pacific Coast Area, however, is merely a
geographic region. The regions defined
include about 70 per cent of the counties
and of the rural population of the United
States, the chief omission being the At¬
lantic and Gulf Coast counties and Florida.
In selecting the sample the procedure used
was to select counties which were repre¬
sentative of the area for a combination of
certain independent variables, such as per
capita land value, residence (i.e., farm,
rural, village, or city), tenancy, and the
geographical distribution of the units.
This was done on the assumption that if
one set of facts is related to a second set,
then a sample representative of the one
set will at the same time be representative
of the other.
For the purposes of the survey proposed
it was decided that a 7^ per cent sample
was the largest practicable for field ad¬
ministration within the time and resources
available, but the sampling scheme used
was such that it might readily be adjusted
485
RESEARCH AND RURAL RELIEF PROBLEMS
to allow the selection of almost any size
sample desired.
The representativeness of the sample was
shown to be high not only with respect to
the variables which were directly con¬
trolled in selecting the sample, but also
with respect to several related variables,
such as per cent of all rural negro families,
per cent of all rural farm and non-farm
families and per cent of all gainful work¬
ers in agriculture who are wage workers.
The crucial question was whether the
sample selected upon the basis of these
rural population factors would also be
representative of the rural relief popula¬
tion, the basic assumption of the proce¬
dure. This could not be fully tested due
to lack of information, but the October
1933 Relief Census gave the total rural re¬
lief load, and the total Negro rural relief
load by counties, which made it possible
to test the representativeness of the sample
as to these factors. This was done and it
was shown that the counties selected were
as representative in these respects as for
the independent variables previously men¬
tioned.
It is appreciated that no one sample will
be valid for all purposes, but it is felt that
an important beginning has been made in
establishing a method whereby a sampling
of statistical data for the counties of the
United States, valid both from logical and
statistical reasoning, and having applica¬
tions in many fields of rural social research,
can be had.
By this method about 140 counties were
selected in 33 states and a survey has just
been made as of October 1934, along the
lines of the survey of last May. This
seeks to determine the place of residence
and usual occupation of the heads of rural
relief households; to obtain a picture of
the r61e of the Rural Rehabilitation Pro¬
gram; to obtain an estimate of the number
of rural cases which will require relief in
February next; and to discover any tend¬
ency for open-country relief families to
move into villages and towns or vice versa.
Research cannot of itself determine poli¬
cies, but it may obtain the facts which
will reveal the need for certain policies or
programs, and may measure their prod¬
ucts through accurate description of the
changes in the amount, kind, and reasons
for relief in representative localities.