Farm Chicken Flocks in the United States in 1930 and 1935

Farm Chicken Flocks in the United States in 1930 and 1935

Farm Chicken Flocks in the United States in 1930 and 1935 S. A. JONES AND E. S. KIMBALL V. S. Department of Agriculture, Washington, D.C. (Presented ...

441KB Sizes 1 Downloads 15 Views

Farm Chicken Flocks in the United States in 1930 and 1935 S. A. JONES AND E. S. KIMBALL

V. S. Department of Agriculture, Washington, D.C. (Presented at annual meeting, August, 1937; received for publication August 25, 1937)

A

Estimates pertaining to livestock or poultry numbers, in order to be comparable, must be based on some definite date, because of the changes which take place in numbers throughout the year. Since census enumerations have been made at different times of the year, it has been necessary to put the different census data on a comparable basis. A comparison between census enumerations of different dates is useless without these adjustments. One can then readily see why these adjustments of the census data become an important part of the problem of poultry estimates and should be given very careful consideration. In order to get more detailed information on the changes in farm flocks during the period 1930 to 1935, the 1935 United States census data relating to poultry were broken down into size groups so that they

could be compared with the size-group breakdown of the 1930 census. The breakdown was complete for the New England and Middle Atlantic States, Ohio, Missouri, Michigan, Maryland, Delaware, Virginia, West Virginia, South Carolina, Florida, and the Rocky Mountain and Pacific Coast States. In all other states the breakdown was limited to a sample of 10 or more selected representative counties in each state. To assist in analyzing the census figures of 1930 and 1935, and the changes between the two dates, a similar breakdown for a number of states was made of the Department's samples of flocks, which are secured monthly from about 20,000 crop reporters and annually from a different list of about 35,000. Before the April, 1930, and January, 1935, census flocks can be compared, they must be put on a comparable basis; that is, the 1930 census numbers need to be converted to a January 1 base. By studying the changes which took place in the sample flocks of crop and livestock reporters between January and April, 1930, it was possible to make an adjustment of the April, 1930, census figures to a January equivalent. During 1932, these adjustments were made for all states on the basis of the state sample data taken as a whole, but excluding flocks of more than 400 hens and pullets of laying age. For the United States as a whole, the adjustment amounted to about a 25 percent increase in the 1930 cen-

[206]

Downloaded from http://ps.oxfordjournals.org/ at Simon Fraser University on May 30, 2015

LTHOUGH poultry statistics for the • United States date back to the 1880 census which included numbers of chickens and egg production, no attempt was made to make annual estimates by states until 1929. At that time estimates by states were worked back to 1924 upon the basis of a series of reports on numbers in farm flocks belonging to crop reporters. An earlier series of estimates for the United States by geographic divisions from 1920-1924 has a different basis and is not comparable. The basic data for the present series prior to 1930 are rather thin in spots and the estimates prior to that year are not as dependable as those for later years.

FARM CHICKEN FLOCKS IN THE UNITED STATES

geographically distributed in the United States. The census distribution of flocks by size groups as of April 1, 1930, was corrected to a January 1 equivalent by the following method, as carried out in Table 1: (1) For each state the percentage distributions of flocks by size groups was computed for (a) the April 1 census enumeration; (b) the April 1 general schedule farm sample; and (c) the January 1 livestock disposition farm sample. (2) For each flock size group the ratio of this percentage in the January 1 disposition sample to the percentage in the April 1 general sample was calculated as measures of the shift in distribution by size of flock. (3) These ratios of January 1 to April 1 relative frequencies were applied to the respective flock size percentages for the census. (4) The resulting January 1 percentages derived from the census in this manner were adjusted upward or downward by a constant factor so as to total 100 percent. (5) The census total number of flocks was the broken down according to the adjusted January 1 percentages derived from the census. (6) The adjusted number of flocks in each size group was then multiplied by the census average size of flock for that size group to obtain an estimated number of chickens on farms as of January 1, 1930. There are various ways in which information contained in the flock size distributions shown by the January 1 and April 1 samples may be used in adjusting the April census to a January 1 equivalent. None are entirely satisfactory or entirely consistent. In the method outlined above, the two types of samples—the general schedule and the livestock disposition—were used because size of flock detail was not available for the general schedule sample for January 1 and no

Downloaded from http://ps.oxfordjournals.org/ at Simon Fraser University on May 30, 2015

sus, which represented a disappearance of about 20 percent in chickens from January to April, 1930. The disappearance ranged from 10 percent in some states to 31 percent in others. The accuracy of the indicated adjustment depended upon the size and representativeness of the sample. For states with small or irregular samples, adjustments were based partly upon trends in adjoining states. The breakdown of the 1935 census material into size groups has made it possible to study the changes not only as a whole but according to size groupings. Sample data from a number of states have been broken down into similar size groups and in states where the farm sample is large and representative, these breakdowns have given a more accurate basis for making the April to January, 1930, adjustments. It would be possible to get a more complete picture of the changes in the industry by size of flock from 1930 to 1935 by making a breakdown by flock sizes of our samples in all states where our returns are sufficient to establish stability in the group averages, and where the flock size distribution in the two samples—livestock disposition survey and general schedule—is substantially the same. Unfortunately, these two requirements are not met in all of the states, and the actual variations from state to state make it impossible to accept with too much confidence the changes in one state as indicative of changes in adjacent states. Table 1 shows the basic data and illustrates the method used in making the adjustments for three North Central States. Since a revised series of poultry estimates will soon be released, only approximate percentage changes are shown in the text of this paper. Corrections for all states were made on the basis of results obtained by this method for a series of selected states

207

208

J*

t O O l O

r H NO T h ON

a 2«

3

qui

3— H S'

CD

1:2 =

3 £S . W} CO

r H CO T h CN

O

O N > 0 T h CO t o

T-H

CM NO CN T h t O O N Q r H CM T h * - - T h

CO

NO

Th t o O

NO

O N H f O N O O CO NO CO T h * — " O r H r H CM T h * - Tt*

r H

CO CM O

O

T-H CM T-H c o O N T h i s . t - O N O N OO NO T-H i o CO NO CM CO O N O N ^ h

3

OO ^ O 0 0 CM CM CO NO NO CM

-Q

5

a (5

>o

00 c o * > - Th O N

•*

CO NO * 0

00

N

f j

Os

rH

^—^

?

T h CO O CM CM NO t o NO t o T h O O N O N NO r H T h ON CO l O

SO



r H CO NO T h

SO

tr>

* 0 ON 0 \ O

o\

O * - CO C M * - O CO NO CM T h t o t O T-H CM T h * -

,—

'

ON



O ] 1

Tf t H

* -

oo o t o NO »o

OO

o

l O N N O H N CN NO CM T h CM rHCMTh

00 I 1

] 1

^

C
oo NO t o t— Th t o oo r H CM T h * —

CO 0 0 CO T h O

CM NO

* — NO * - NO T h T h CM NO CM CO t o * T-H CN T h * -

t—

Th

-

T-H

fc

ON

O

~

CO O N CM

CM NO l O CO 0 0 O N

)H

O

CM O N T h c o O * - c o CO NO CO t o 0 0 * - T H T H CM T h * - CO

0 0 l O T h I - I CO NO 0 0 M N O C N ^ t ^ t > . f N rHCN Th*-Tf<

fc

1?

Cen adju

oo

-

3

3

52

r H T ^ N O f ^

£

*H

w *-<


c o r H 0 0 CO O N 0 0 O N NO O N O N O CO

T-H O N CO O

ent

r of floe

w

«"S~

to Jan

os

8

£

Censu adjust

"5

O

~

3

^ R O

O O " 0 0 0 0 0 NO * c o * — CM c o * - O N » 0 * - - T h NO l O O N CM CO

90.0

Janua per flo

h

00

r H CO l O CO

£ qui

"P1 >>*!

t ^

co CO

CM

rH

CO NO CM T h * - r H © r H C N Th 00 t O

3

NO

oo*-- oo Nor-

T-H T h 0 0 I D

*-

.T—

200,0

P.

NO CN NO r H NO

CN CN r H T J t ^ OO • * r H T ^ « * J>- * - NO T > ^ 00-^rH-HH CN

<3

T-H

O N O H 0 \ O O rt*OO^NO>00

mber

ave

< 53

00

o

s •a

5? >< -

co o o NO co Th

M i O N H N O f O O O N O Q O r t l / ) O c o t o T h T h * - - NO

H N O N t o T h T-H O NO CO T-H NO NO 0 0 > 0 CM CM CM NO O CO CM

t^. t-~

r H O0 C O ^ CM NO NO l O r H

200

ckens 1

o

r H CN CM NO

O0

O N NO CO CM CO

o

Th to

-HH l O T h r H

so

*-<

CO » 0 t O CM

SO

»>• CO CO vO O CM co

^h

* - O N 0 0 O T H T h CO T-i CO O N O O CM T h 0 0 NO CM CN CO r H

l O t ^

l O

H H ^ O O r H •HH UO t O r H

t ^ SO

l O T h > 0 NO \ © r H NO SO CM 0 0 t O 0 0 r H t O > 0 CO

o

CO * -

rH

8

r h i o Th t o CO CO CM

t ^ 00

»o >o SO ^-1

*-H

so

^ O O N O N O N T-H T-H CM CM

OS

*-

CN 0 0 c o t O O N N O r H r H C S i O N C N O O N CM r H I O CO O rH

to * - ON Th 00 NO O 00 CO t o * -

0 0 CM O N 0 0 CO T h CM

rH rH tO

CO O

1

r H CM CO r H

Tt<

CN

l O * — t O CM

OOONOtOCM

SO ON

CM CO CM

l O

SO Os

O • «

ft

ere

NO CM NO CO O O ^ O N CO CO CM

CM T-H

H

ft

r H CM t O r H

o

O

00 ThON 00

rH

t O O CN O

R

O

1

CM CO CO r H

T-i

8 "•

ent

tock sition pie .1

Cen Apri

3 _

T h CN CM O N T H T h 00

III* 1

to

1 0 ^ 0 , ^ CN T*1 T-H

o

t ^

t O O N r H NO r H

8

O

CO r H CN r H

o

t ^ . CO CO r H CO CM r H

8

Th l O O 00 CM CO CO

rH

CN COCOTH

O

rH O

8

T h NO t O O CO r H CM CO CM

o

CO \ 0 l O CM CN CM

CO T h * -

o

tO 1

8

8

3 O O N O N O N - T S O N O N O N O N PH

a i O N T—I c o NO O N

r-.

o

a

O 3 g to O N O N O N O N -75 •S i. O N O N O N O N CI* *3 S f > T H r ^ i v O O i o

fe

s

O Q tO

"S u, S

o |

3

O ON ON ON "Z* O O N O N O N P-

QjONrHCONOON,-,

!aj

8

Jtal

tO H

ON * — r— T h r H r H CM CO r H

inne Und 50-9 1002004007001}00(

O

©

rH

Total

l O O N NO ^ t * CO T-H CM CO T-H

CD

iscon Und 50-9 1002004007001,00(

CO » 0 NO 0 0 T h CM CM

a AH

Tot

ent

JT3

no Und 50-9 1002004007001,0(X

PL,

en i-H

General sample April 1

rcen t of flock

~ ,g

roup

os ,H

H

Downloaded from http://ps.oxfordjournals.org/ at Simon Fraser University on May 30, 2015

mber o

at en

qui

3

(fl

'S fi
Os

3

nds

y-i

T h T h CO NO O

Os rH O

H o

o

00

9.2

3

r H

t O O N r H Th

348

tS

CO

O N CM * -

1,859 0,486 3,098 7,332 4,352

3a

O f O O M O i O ^ N

lO O N ^O vO • * "O

9.4

? « K &

5,5

Census djusted o Jan. 1 ol.6X7

April as a percent of January number Col. lO-Hll

S. A. JONES AND E. S. KIMBALL

209

FARM CHICKEN FLOCKS IN THE UNITED STATES

It is desirable to test comparability of all the disposition and the general samples, in so far as their flock size distributions are concerned. This may be done by comparing these two samples as of January 1. The detailed tabulation by all flock sizes was not available for the January 1 general sample, but a coarse breakdown into three flock size groups, 0-399, 400-999, 1,000 and over, had been made for all states and it was possible to test the agreement of the January 1 disposition and the January 1 general schedule with respect to these three broad classes. It would be more desirable to make the test for finer classifications since by far the greater number of flocks lie in the 0-399 group, but the results of these tests are presented for what they are worth. The method selected to test the agreement of the two samples was Karl Pearsons' Chi-square test as applied to test whether two samples may have been selected from the same universe. The actual calculations involved were carried out according to a scheme due to Brandt and Snedecor and described by R. A. Fisher in his Statistical Methods. This test was applied to the samples from the New England States combined and from 12 other states. The Chisquares obtained are shown in Table 2 below with the corresponding values of the probability that worse disagreement might

be due entirely to random sampling fluctuations. A Chi-square of less than 5.00 is generally accepted as indicating that the samples are drawn from the same universe and there is no significant difference between the samples. The New England area is the only one TABLE 2.—Chi-square tests of agreement of January 1 general sample and January 1 disposition sample, with respect to number of flocks falling in three flock size classes: 0-399, 400-999, and 1,000 and over

New England New York Pennsylvania Ohio Wisconsin Iowa Kansas North Carolina Tennessee Texas Colorado Washington California All States

Chisquare

Probability that worse disagreement could result from random sampling fluctuations

5.07 1.07 1.51 1.64 4.39 2.90 4.36 .94 1.04 .06 1.95 4.33 .36 1.73

.08 .59 .48 .45 .11 .24 .11 .64 .60 .99 .38 .11 .94 .43

of the group of states which exceeds the 5.00 Chi-square. However, ordinary sampling fluctuation gives some large Chisquare values which are not significant. The true distribution of Chi-square for three frequency classes is known and so we may make a Chi-square test of the agreement of the distribution of the 13 Chi-square values themselves with the distribution we should expect from ordinary random sampling fluctuation alone. For this purpose we group the 13 Chi-square values with three groups: 0-1.99, 2.00-3.99, 4.00 and over, as shown in Table 3. The result of this test indicates that only 13 out of 100 times could such a set of 13 Chi-square values group as heavily in the large values through random drawings of pairs ojF samples each pair from the same universe. This result does not indicate dis-

Downloaded from http://ps.oxfordjournals.org/ at Simon Fraser University on May 30, 2015

disposition sample is taken in April. This leads to the question which has been raised: Are these two types of samples sufficiently comparable so that differences in their flock sjze distributions may be taken as measuring actual shifts in the distribution of all flocks between the two dates, or is a large part of the indicated shift in distribution due to the fact that the disposition and the general schedules are drawn from dissimilar sub-groups of all flocks? In applying the correction method described above, this comparability was assumed.

210

S. A. JONES AND E. S. KIMBALL

agreement between the general and the disposition samples for January 1 and, therefore, there is no significant difference between the samples. Two points should be noted. First, the test loses some significance in this application because of the extremely coarse grouping into only three classes, and, second, for some states the group frequencies were smaller in the largest flock class than desirable for satisfactory application of the

taken place in each size group, it is observed that the bulk of the increase in the number of flocks has occurred in the small flocks with less than 50 chickens. In the New England States, the census shows about a 12 percent increase in number of flocks from 1930 to 1935. After making allowance for adjustment of the 1930 census to a January base, it is evident that most of the increase in flocks and chickens has occurred in those with less than 50

, Chi-square Group

Expected random distribution Percent

0-1.99 2.00-3.99 4.00 and over TOTAL

Number

Observed distribution Number

/

/'

Difference between observed and expected


P/f

.2 2.0 2.2

.04 4.00 4.84

.004 1.33 2.69

d-f'-f

63.2 23.3 13.5

8.2 3.0 1.8

8 1 4

100.0

13.0

13

4.02

Chi-square = 2
Chi-square test. On the whole, however, these tests confirm the assumption of comparability upon which the census correction from April to January was based. From this size group study, certain conclusions have been drawn. The United States census of 1935 shows a considerable increase in the number of all flocks compared with the 1930 census. This increase has been so great in some states as to raise doubt as to its validity, at least to its full extent. This doubt is associated with a similar question as to the fact of the extreme decrease in number of flocks shown in some states by the census of 1930. Along with an increase of 8.6 percent in the number of flocks in the United States between 1930 and 1935, from 5,372,597 to 5,833,079, the census shows, when the 1930 figures are adjusted to a January 1 base, a decrease of 16.8 percent in the number of chickens on farms. In studying the changes which have

chickens, with some increase in flocks above 200. Flocks with 50-199 chickens decreased slightly. While the original census figures show an increase in the total chickens on farms of almost 20 percent from 1930 to 1935, the adjustment increase was less than 2 percent. The increase of chickens in the large and very small flocks was almost wiped out by the decrease in the flocks of intermediate size. While flocks of under 50 increased greatly in number, the average size of these small flocks was decreased. In the three Middle Atlantic States, the unadjusted census figures show the increase in the number of small flocks with less than 50 chickens was more than the total reported increase of flocks for the states as a whole. Flocks with 50-199 chickens decreased. Flocks of above 1,000 chickens increased markedly. These figures, if adjusted, would probably accentuate these differences, but would not change their

Downloaded from http://ps.oxfordjournals.org/ at Simon Fraser University on May 30, 2015

TABLE 3.— Test of agreement of 13 Chi-square values with random sampling distribution of Chi-square

FARM CHICKEN FLOCKS IN THE UNITED STATES

direction. The number of flocks of 200-999 probably did not change greatly. While the original census figures show more flocks in this group in 1935 than in 1930, they might, when the 1930 figures are adjusted, show less. Although there was an increase of about 5 percent in the total number of flocks in all groups, there was a decrease of about 15 percent in the number of chickens on farms according to adjustments made in 1932.

about 11 percent in number of flocks which all occurred in the small flocks with less than 50 chickens. All other groups decreased. As a result of this shift in flocks, total adjusted chickens on farms decreased about 22 percent. In the Pacific States the increase in flocks was about 14 percent, which is also mostly accounted for in the small flocks, with a probable decrease in all other groups, especially in the large commercial flocks with more than 1,000 chickens. The adjusted number of chickens on farms decreased about 29 percent because of this heavy decrease in large flocks. CONCLUSIONS

After considering the changes which have taken place in flocks of different size groups in the various geographic regions of the United States, the following conclusions may be drawn: (1) That the increase in the number of flocks as shown by the 1935 census as compared with the 1930 census, adjusted to a January base, is probably due to two factors, (a) a more complete census of small flocks in 1935, and (b) an increase in small subsistence farm flocks which developed during the recent depression. We have no present means of measuring how much each contributed to the increase in total number of flocks. The extreme increases in small flocks were partly due, of course, to the decrease in the general size of farm flocks, resulting in a shift into the smaller and particularly into the smallest group. (2) That generally, throughout the United States, the average size of the farm flock has been getting smaller because of (a) the universal absolute increase in the number of small flocks, and (b) in the Middle West and Far West to a marked decrease in the number of larger flocks,

Downloaded from http://ps.oxfordjournals.org/ at Simon Fraser University on May 30, 2015

In the North Central States, there has been a decided increase in the number, as well as the percentage of small flocks with less than 50 chickens, with a tendency toward a decrease in all other groups in most of these states. Decreases are particularly marked in the flocks of 1,000 or more birds. The census total number of flocks increased about 5.5 percent, while the number of chickens adjusted decreased about 24 percent as a result of this large increase in the percentage of small flocks at the expense of the larger flocks. In the South Atlantic States, there has been a decided increase in the reported number as well as the percentage of small flocks, with a general slight increase in flocks with more than 50 chickens, except in Deleware, Maryland, West Virginia, and Florida, which show a tendency to decrease in flocks of more than 50 chickens. The reported increase in total flocks was about 10 percent while the decrease in total chickens adjusted was about 13 percent. In the South Central States, the census shows the same decided increase in small flocks. While there was a tendency for larger flocks to increase in some states, there was an apparent tendency to decrease in others. The increase in total number of census flocks was about 10 percent, while the decrease in the adjusted number of chickens was about 20 percent. The Mountain States show an increase of

211

212

S. A. JONES AND E. S. KIMBALL

especially the larger commercial groups. (3) That the New England and Middle Atlantic States and parts of the South Central groups show some increase in number of larger flocks, but a far more pronounced increase in the number of small flocks of less than 50 birds. New England is the only major geographic division that shows by the adjusted figures a relative increase in chickens and the increase there is only 2 percent.

(4) That the decrease in larger flocks in most areas more than offsets the increases in these groups in most of the Atlantic Coast area and in a few states in other areas. (5) That the number of chickens on farms has decreased generaly throughout the United States as a result of these predominating shifts to smaller flocks, notwithstanding the increase in total numbers of flocks. Downloaded from http://ps.oxfordjournals.org/ at Simon Fraser University on May 30, 2015