NET MIGRATION
IN
SELECTED CENSUS TRACTS OF BOSTON,
MASS.
by
Henry Sucher
Submitted in Partial Fulfillment
Of the Requirements for the Degree of
Master in City Planning
at the
Massachusetts Institute of Technology
1952
Head of Department
Thesis Adviser
Author's Signature
Departmentof City and
Regional Planning
May 16, 1952
-.
-
I
I I EU
-_____________________ -
Cambridge, Massachusetts
May 16, 1952
Professor Frederick J . Adams
Department of City and Regional Planning
School of Architecture and Planning
Massachusetts Institute of Technology
Cambridge, Massachusetts
Dear Professor Adams:
I herewith submit a thesis entitled Net
Migration in Selected Census Tracts of Boston, in
partial fulfillment of the requirements for the degree
of Master in City Planning.
Respectfully,
Henry lSucher
Acknowledgments
I should like to express my thanks and
appreciation for their assistance and cooperation to
the following:
Professor John T. Howard, M.I.T.
Professor Kevin Lynch, M.I.T.
Professor Leo Grebler, Columbia University
Professor A. Solow, M.I.T.
Mr. Jean E. Baril of the Boston City Planning
Board
Mr. Peter H. Nash, Urban Redevelopment
Division, Boston Housing Authority
. Mr. Leon Pollard, without whose able cooperation the method for computing
net migration could not have been
evolved.
To my wife, Cobi, and to Mrs. John Geiger
for their help in the final presentation.
HenrylSucher
Index of Illustrations
Chart:
Table:
1.
Scattergram: Natural Population Change and
Net Migration.
2.
Scattergram:
Average Rent and Net Migration.
3.
Scattergram:
Migration.
Excess Density and Net
4.
Scattergram:
Room Crowding and Net Migration.
5.
Scattergram:
Housing Index and Net Migration.
6.
Scattergram:
Social Index and Net Migration.
7.
Map of the Sampe Tracts.
8.
Bar Graph:
1
Data Sheet
lA
Net Migration by Age Groups in Sample Tracts.
lB
Average Rents and No. of Dwelling Units in
Average Net Migration by Age Groups.
1934 and 1940.
2
Sample Calculation of Net Migration.
NET MIGRATION IN SELECTED CENSUS TRACTS OF BOSTON,
MASS.
Henry Sucher
Submitted for the degree of Master in City Planning in
the Department of City and Regional Planning on May 16,
1952.
Introduction:
I.
Purpose of study is to facilitate prediction of net migration.
General Description of study:
A.
Need for determining and predicting net
migration for neighborhood sized areas.
B.
Methodology:
1.
2.
3.
4.
Four main steps:
Computation of net migration.
Tabulation of factors to be correlated
with net migration.
Coefficients of correlation calculated
between 1 and 2.
Interpretation of results.
C. Limitations of study.
II.
Factors Correlated with Net Migration:
A.
Description and Discussion of Factors:
1. Natural Population Change
2. Average monthly rent
3. Excess density
4. Room crowding
5. Housing penalty score
6. Social Factors ranking
7. Need for major repairs
8. Duration of Occupancy
9. Distance from City Center
10. Age of Structures
11. Owner Occupancy
B.
General Discussion of Factors, special conditions.
1. Changes in the No. of Dwelling Units
The influence of Non-Residential Land
2.
s-ie
3.
4.
5.
III.
The Back Bay Tracts.
The Cycle of Migration--Migration by Age
Groups.
The Role of Decentralization and
Centralization.
Conclusions:
A.
Summary
B.
Suggested Methods for Applying Results
C.
Evaluation
D.
Suggestions for Further research
E.
Specific Conclusions
Appendix:
Description of Method used in calculating
net migration.
Contents
Page
Letter of Transmittal
Acknowledgments
Introduction
1
I. General Description of the Study
3
Need for Determining and Predicting Net
Migration
3
B-.
Methodology
4
C.
Limitations of the Study
6
A.
II.
Correlation with Factors
A.
9
Description and Discussion of Factors
9
1.
9
Natural Change in Population
2. Average Rent
3.
Excess Density for Housing Type
4. Room Crowding
5.
Combined Housing Penalty Score
6. Social Factors Ranking
7.
Need for Major Repairs
13
16
18
20
22
23
8. Average Duration of Occupancy
24
9. Distance from City Center
26
10.
11.
Percentage of Structures over 40 Years
Old
27
Percentage of Owner Occupied Dwelling
Units
28
B.
General Discussion
30
1.
Changes in the Number of Dwelling Units
30
2.
The Inf luence of Non-Residential Land
Use
31
3.
The Back Bay Tracts
32
4.
The Cycle of Migration; Migration by
Age Groups
33
The Role of Decentralization and
Centralization
37
5.
III.
Conclusions
38
A.
Summary
38
B.
Methods Suggestions in Applying Results of
this Study
39
C.
Evaluation
41
D.
Suggestions for further Research
42
E.
Specific Conclusions
44
Appendix
Bibliography
1
Introduction:
The primary purpose of this thesis is to seek
relationships-between net migration and various other
factors in neighborhood sized areas of large central
cities, which will make more accurate population predictions for such areas possible.
At present migration
is usually the component of population change about which
forecasters and planners are least certain.
Natural population change (i.e. the difference
between number of births and number of deaths)which is
the other factor in population change has fairly well
known relationships to the progress of medical science,
to cyclical economic fluctuations, and to the type of
people inhabiting an area.
Especially concerning death
rates rather accurate predictions are possible nowadays,
largely perhaps because of their significance for the life
insurance business*
For cities as a whole rough predictions of net
migration can be made from an examination of their economic
base and outlook, as was done for New York, for example,
by the Regional Plan Association.
But the question still
remains, how large a part of the labor force will live in
the city itself.
1.
The Economic Status of The New York Metropolitan
The RegionaT~Plan Association, Inc.,
Region in 1944.
New York 1944.
Tw"'_
2
This leads directly to the problem of decentralization.
We know that we can expect much slower
growth - if not actual decline - in the central areas
as compared to the environs of large cities.
We do not
generally know, however, what areas of the central city
lose most heavily, or why they do.
It is hoped that some
of the more important causal factors will be brought to
light by this inquiry.
But even if no actual causal
relationship can be established, it is still desirable
to find certain conditions whose presence or absence in
a given area will stimulate or retard, as the case may be,
decentralization activity.
For the purpose of prediction,
correlation rather than causation, is important.
The decade of 1930 to 1940, which was chosen
as the period of study, was one of relatively slight
decentralization activity, due to the great depression.
The correlation coefficient obtained therefore tend to
underemphasize rather than overstate the existing relationships.
3
I.
General Description of the Study
A.
The Need for Determining and Predicting Net
Migration
Net migrationi for neighborhoods or other
subdivisions of a city must be predicted in many different situations in order to make valid planning possible.
For many of these purposes future age distribution of
the population is also very important, so that a forecast by age groups is often needed.
A sound program of
school construction for instance is inconceivable if such
estimates are not available; the same holds true for
recreation facilities.
Any long range housing market analysis has to
work with population forecasts, taking into account future family sizes and ages of individuals.
Density
standards, zoning and many other planning principles and
tools must take population as one of their starting points.
The embarrassment which the task of predicting
migration causes planners is illustrated by a statement
in a recently published general plan of a large city.
It says that since migration is not subject to direct
forecast, this influence was not included in the calcula1.
Net migration is the difference between total inmigration into an area and total out-migration from
it; net in-migration is positive (+), net out
migration is negative (-).
4
tions of future population.
But even the ability to compute past net
migration for small areas is an important asset.
For
it is possible to draw fairly sound conclusions from
past experience about net migration in the future by
taking into consideration
1)
Wider (e.g. national, regional) trends in
both the migration picture and the economic
situation
2)
Changes in local conditions which are likely
to affect migration trends (this is one of
the reasons why it is important to know
what local factors actually show a correlation with net-migration).
B.
Methodology
The general outline of the method used consists
of four steps:
1)
Net migration between 1930 and 1940 for
twenty-eight selected census tracts in
Boston was computed.
2)
Factors to be examined for their relation
to net migration were tabulated.
3)
Coefficients of correlation were calculated
between 1. and 2.
4)
Some measure of interpretation of findings
was attempted.
5
A somewhat more detailed discussion of the
methods and procedures used will reveal many of the
problems encountered in setting up the study.
The first
tractsl.
step was to select the sample census
Since at first
mainly social and housing con-
ditions in general were thought of in relation to net
migration, the tracts were chosen with the aim of getting
a wide range of housing qualities for each type of social
condition and vice versa.
Wide geographic distribution
within the city was desired, but a somewhat more intensive study of one larger area was thought potentially
useful, and five contiguous tracts in East Boston were
therefore included.
The first difficulty appeared at
this early stage: nearly all census tracts in West
Roxbury and Hyde Park had been split into two or three
between 1930 and 1940, making it practically impossible
to use them because of the added statistical work this
would have involved.
The geographic distribution of
the sample tracts is therefore far from satisfactory, and the
outlying sections are rather poorly represented (see chart
7).
It may be of interest that no tracts were found
1.
It is realized that census tracts are by no means
ideal neighborhood units, but they were the only
units of the size wanted for which presumably
accurate population data were available.
6
in which social conditions were relatively good while
housing conditions were poor; the converse situation
however was found in several cases.
Calculation of net migration in the sample
tracts was part of a regular staff assignment at the
Boston City Planning Board, and was done jointly by
Mr. Leon Pollard and the writer.
However, during the
writing of this thesis certain changes were made in
method.
the
For the 0 - 9 age group an entirely new and more
practical as well as more accurate way of calculating
migration was found.
All net migration figures and
correlations were revised accordingly.
For a detailed
description of the methods by which net migration was
computed see Appendix
The procedures and sources used in tabulating
the factors which were correlated with net migration will
be described in the sections dealing with each factor.
C.
Limitations of the Study
Since this is
essentially a pilot study,
it is
not easy to judge the extent to which the findings are
typical in any general way.
The empirical investigation
is limited to one central city and one decade, neither
of which may be an adequate sample to permit generalized
conclusions.
As a matter of fact, the decade between
1930 and 1940 was in some respects the least "normal"
period in American History to that time.
Also, Boston,
7
more than most other large cities, lacks large quantities
of vacant land suitable f or residential development, a
fact which may influence the migration picture somewhat.
No adequate theoretical framework was available for a study of this type.
Therefore, intuition had
to be used to a considerable extent in
deciding what
factors should be examined for possible correlation with
net migration.
Some factors, such as transportation
facilities or prestige value, could not be used because
of their not readily quantifiable nature.
Because of the f orm in which the information
was available,
only a five point scale was used in tabu-
lating some of the factors.
This probably reduced their
correlations in comparison to those for which a continuous
range of figures was available (e.g. excess density versus
average monthly rent).
Last,
but not least, the entire study is
based
on net migration figures experimentally evolved for a
specific purpose and a certain set of given data.
This
method is in some respects an improvement over those
generally used in the past, but it also contains several
minor shortcomings and inaccuracies, the importance of
which cannot be gauged from this investigation.
shortcomings are mentioned in the appendix).
(These
TECHNOLOGY STORE, H. C. S.
fORM 3 H
40 MASS. AVE., CAMBRIDG, MASS.
fH+-4
H r+t+H
-4-4-+41+1
444-44444 444 44
t
,
!
44
-++
+±Hi4-4-T11
14-i tH
11
1t
1
i 111111 1 +41 1 4- 11i 1±±4++ 1-11-4
4+H+4+I
*
-4
I I I
TP
HH
0
i-f-i-f- 1111
H Iii f0 -
f4+±11-11r44 -H 1-
i
7I
-
I
I
I
I
I
I
7tI1
-~~H
+r-r H
I!
LL
i
-
r
-+
I-P
++-
etH+-+
L
art
4+
Iti
1 t
u-L-
-
- Ii ]lit1
Me 4
Ht HHi
i+
+
1
i
LL-
u
-
+1-
frd4K 1
-
=U
F
4-
-T
4 - -4-4-t-
~4~ft
TF~flT~
LV
i144ft~
F-f
f
tr~
#1 -:
T
H
-
-
-i
4
-
I::
-~
1-4
31
i
H
FrT~
L
jI
§dj
-- -I- -vvi-144-I-MiTh'i-i-4-ii-;
FTT
I,4+
jt
WT
Th
.63
9
II.
Correlation with Factors
A.
Description and Discussion of Factors
1.
Natural Change in Population1
Coefficient of Correlation ( r ) .63 (see Chart
4).
This relatively good correlation of natural increase
with net out-migration seems entirely reasonable:
since
there is very little land suitable for residential development available in most of the areas under examination,
any increase in the population causes a spill-over into
other parts of the city or the metropolitan area.
Con-
versely, where the population is decreasing due to low
birth rates, excess population from other areas can be
attracted and absorbed.
Since most of the areas examined had an excess
of births over deaths, the outward trend of migration was
fairly general.
Four of the five tracts with natural
decrease showed inward migration, while twenty of the
twenty-three areas with natural population increase lost
population due to migration.
On the average the loss in
population due to out migration amounted to between one
and one and one half times as much as the gain from natural causes.
This ratio, if confirmed by further research
could be used to predict not only net-migration, but total
population change quantitatively.
1. The figures for natural population changes were obtained in the course of net migration calculations.
10
For Boston as a whole, natural population increase between 1930 and 1945 was 11.E81.
Since total net
out-migration in the sample tracts was about 10% for the
decade of 1930 - 40, it can be assumed that the net outmigration from 1930 - 45 was between 12 and 18%, which
corresponds to the ratio.
This ratio between natural and migratory
population change will vary in all probability according to the city and the period under examination, but
there is no reason why an average ratio cannot be found
in each case by testing some areas in the community and
then using the ratio for prediction.
Changes in national
and regional trends will have to be considered,
effects estimated.
and their
If natural change in population
actually is the most important influence causing decentralization, as seems to be indicated by this study,
questions of great significance arise.
Is the trend towards disappearance of complete
families with children from the central portions of metropolitan districts irreversible?
Or could it be stemmed
by redevelopment on a scale large enough to improve the
environment in these areas thoroughly?
If this is not
possible, redevelopment for either non residential uses
or efficiency and other small apartments will probably be
indicated.
1.
Term Report on "Centrifugal Patterns in the Boston
Metropolitan Region," Geography 121, Professor
Whittlesey, Harvard, February 1949, by P. J. Cusick
Jr, P. H. Nash and R. M. Smith.
11
Experience to date seems to indicate that
families with children can be attracted to large, centrally located redeveloped residential areas (e.g. in
New York City) but the extreme housing shortage may have
been responsible for this.
Tentative plans currently
being prepared by the Boston City Planning Board also
take this trend into consideration and anticipate that
approximately 2/3 of the families to be housed in Boston's
cleared redeveloped areas will consist of one or two
persons.
There is a close connection between this
correlation and the net migration picture by age groups.
The people in their thirties have the largest numbers of
children, and are also the heaviest out-migrants.
The
areas with few families in this group therefore show little
decentralization as well as low birth rates.
1.
Actual population in the 0-9 age group decreased by
10-30% in most sample tracts between 1930 and 1940.
FORM 3
M
TNHNOLOGY STORE, H. C. S,
40
rJ~
m1
Wtfti±i fftltFltt ii f-t-±1
+
'1T~4l-it~4~H
tiTh±i i±H4±~14-1
uL
-n- R
H
W1
H4+
H
{11
1-H -4 H HA
H HiH
~H11
14
'IHHM
1HH
H-4j7
F
4L2
>TW~
I I-l-4,A4tI4
I-H
LI-t'
-r
-4-'
14-
4-T
+1- 1114'L
-ft
H+H+
LHLI,
T__
T7_14 4--1 T4
ffL
1'
4
..
.
..
.
.
-
.
+4
-4-
Ti
I
441+T
Li.
HH+H~
~4jj4~
H+-H
H++H
LI WILL
Tj49LL
1
42
-14-L4
-4-1
I
-H
Ht
lilt -i:V
A
11
4A-
49
MW
1
-t
HT
K
K-
I
LILT
L1
44 14+HH4-!HH-
't+
i
44
~iL
I
4i14iP
i4~Th
I:
244,
HI
~Mt
U4I+~
LI
I
Uj{~L
ni
L4T
11
I
:+4
LLf+[
p1 1
4-
~ttii~
RT T rn-
S
+H+
AVE., CAMBRIDGE, #AA.
RFNT
AVERAGE
CHART 2:
MASS.
+H+h
-K
4--H-1
LLLLLftI
±
4FH
H1±f±t~ ~TT
j~fr
-
-
+
K
-L
It
-r
1'
ji
4.1
T±R
j
FT
1
!1--
i -,ikfl
-L! 71-l!
+
+
i-
It- -k
i
tR -hhI1
-i-I--
IlItItIlIlifIll iT t
It II
IIIIIIIII
i
i_
L
l
TF
.63
I
i
i
i
13
2.
Average Monthly Rent
-
.63
(High rent correlates with in-
migration, low rent with out-migration)
This correlation, while equally high as the
previous one, is somewhat more difficult to explain.
Rent is a rough index of both quality of housing2 and
economic status of the inhabitants.
Seen in terms of
housing quality, it is somewhat surprising that average
rent shows a higher correlation with net migration than
any of the other housing indices tested.
The relation
of rent to the economic status of tenants and to fluctuations in both general economic and specific real estate
market conditions probably tends to make this a relatively less reliable indicator of net migration, especially
over longer periods of time.
For these reasons less
reliance should be placed for the time being on this
correlation than on the previous one.
By comparing 1934 rents with those of 1940 it
was found that the few tracts which showed an appreciable
rise in rent3 during that period also had a preponderance
of inward migration during the decade.
The other three
1.
Figures on average rent taken from the Report on The
Income and Cost Survey of the City of Boston, 1935.
2.
A study (not published to date) by J. E. Baril of the
Boston City Planning Board shows a correlation of .8l
between low rent and lack of private baths, and of .94
between low rent and lack of central heat.
3.
Tracts J 4, K 2, K 3,
and S 4.
14
tracts with net in migration showed no significant change
in rent between 1934 and 1940, which was in line with
the general pattern.
Percentage of change in average rent between
1934 and 1940 is listed for all sample tracts in table lB.
FORM
TECHNOLOGY STORE, H. C. S.
3M
40 MASS. AVE., CAMBRIDGE, MASS.
CHART 3:
I±
t
tt I
14
I
-fr
'I
+4LV
+
i-i-i
U
44
_
LHt
I
-
-A-i-
VSTtI
ut
~FF;
+c4+4
17
?t
4- -
L
F~
FL
0111
I
I
r=.55
16
3.
Excess Density for Housing Type
r = .55 (see Chart 3).
Five classifications of excess density were used:
1)
Not below desirable density standards.
2)
Exceeding desirable density by not more
than 50%-
3)
Exceeding desirable density by 50
4)
Exceeding desirable density by 100
100%-
200%.
5. Exceeding desirable density by more than
200%.
This relatively good correlation confirms the
belief that the search for less crowded conditions is an
important motivating factor in the decentralization movement.
It
indicates too, that while people have different
preferences concerning the type of housing they like to
live in, they are also able to distinguish decent conditions from overcongested ones.
out-migration in
The high percentage of
the 0 - 9 age group from the most densely
built up areas indicates that many families are primarily concerned with providing a good environment for their
children.
The availability of recreation areas, both
active and passive, which is closely related to this
factor was not tested, as it is not easily quantifiable.
1.
Figures for this factor were obtained by inspection
from a map of the Boston City Planning Board, showing by what percentage the density in each block exceeds those given as desirable by A.P.H.A. for various
housing types. The blocks in each tract were visually
averaged for purposes of this study.
FOM3
H
40 MASS. AV., CAMBRIDG, MASS
TECHNOLOGY STORE, H. C . &
CHART 4
DWELLINGS
:CROWDED
+-H,+f+ ITr
A1
1
4±~Thtt4-i~-Th4Th4
44-4 -44-4-4-4. ......
4.....
1-1-4+44 -4-1
4-1 i Fi
i
-
11.
4
HTH
T
4
1 7i
H
4F
I
L4
_-I_
L
J.
±4+4- -14--fF
- -. ..H.. +. -4---H H-H--414-H44
H
J- I
+.......
--
t
IF--F
21
-1
-
--
-it1H+
UI*Ht
t
F-Ffir~FIF
±4-v riP>
1
-H
TH
+ -1-4
'>44 14-F-
7r-L
4U,
1rF P+'+
-4k4L Th1
7~FFFrmFfl~
44244 ~
T~T4
4
-4H-
-- HA+1H-H4
L
iT
+H-4+1-4
-4--
lW
4-
~1
-I-
41
44
T
4-
-i
11
-- +7
-244-',4444- -4444-
+ 1-I
H",E
1
± A-i-
J,
11±
.
--
4-
+1
>44 I-F-F +4±4--F1-H -I++4
-
++1
T l - rr-i1
+
i
~
~
1
t
-
-
+4
-4-,
-t
4-
'+
4
4- 41
-F
-R
|T
|
|1
H1
1-
i-H-F H-t-F-i-H-H-i-H
i-i H
1aii
H
:-s-
H-
aFae a
S.54
18
4.
Room Crowdi
r = -54 (see Chart 4)
This factor is rather closely related to natural change in population as well as to excess density.
It would appear that this yardstick would measure the
unfavorable results of overcrowding on the individual
family more accurately than excess density for housing
types.
Therefore, it could be argued, the correlation
with out-migration ought to be higher.
On the other hand
it is possible that each family, primarily due to its
particular socio-cultural background, develops a certain
"crowding tolerance" (that of the Italian immigrant
families in Boston's North End being rather high, for
instance).
When the size of the household increases
beyond that tolerance due to natural increase, the
family, or at least parts of it, finally move out.
In any case, the fact that both measures of congestion
show a significant correlation with out-migration,
confirms beyond doubt the importance of this motive for
decentralization.
1.
Percentages of dwelling units with more than one
occupant per room were taken from the Boston Real
Property Inventory, 1934, Vol. II.
TECHNOLOGY STORE, K C. S.
FORM3 H
CHART 5
40
HOUSING
:
MASS.
AVE., CAMBRIDGE, MASS.
SCORE
PENALTY
-t-W±H±i±i++:
' 444
4#
t
~44
+r+{vtIt7
i 442
~Th4~4t~
t+-,+F
+ i ii
Iff
- 1A
+4"+++M
++
4i
r~r
_7
t
-
1
4
r
I
L:4-
-i +? t±F~+IIA +iHHI
4
h 41 V f4'±t-I-~-t44+-
t±HA4-H-i±i-i2
L
--
-
:r
'L
_4> -
-11
f<Ld'L
-i-it:
i
-
I-
L
~t1
L
'l
T4
I~
2
1
4--
T
I
A-1
145
44+
-I
- 4-
-~
T1JY
41
I
H --I
4
ij ~7j
+4jH4
41-
F77L
4-H
_7
~*1-
+TT H±A4+
i4+4++++UH4-
+
-
Y
T
T-
-T
-[fit-
l
l
-1
-- --
--
A
I
"
-11
- - - T~
-
-T
I
r
1 4+
7-
1~
-
- - -
!
+4+r'k+f4+4j±EI
1~ -H
FH+ Kri~I~i±H~H~H±
14741-1-1-4 i44Utt111-
l
I
t
-
- -
T
----
---
--
fL
t
-
-
"
"
"
""
+ 1--T
r -. 53
F
20
5.
Combined Housing Penalty Scorel
r = .53
(see Chart 5)
Three items, all equally weighted, are contained in this housing score: median monthly rent,
physical condition of dwelling structures, and percentage of structures built before 1900.
A score of one
indicates very good conditions, a score of five indicates
poor conditions on all three counts.
The relatively
encouraging results of this correlation led to the breakdown into individual housing factors almost identical
with those comprising the combined index, in order to
determine the relative significance of specific housing
factors.
There are many more factors which determine the
over-all quality of housing, which could not be tested
here.
The A.P.H.A. method for determining the adequacy
of housing, for instance, would be a more satisfactory
index, and might show a better correlation with net
migration.
1.
The figures were taken, with some further condensation, from Mr. Leon Pollards B.C.P. Thesis, M.I.T.,
1950.
FORM 3 H
,TECHNOLOGY
CHART
T
6
40 MASS. AVE, CAMBRIDGE, MASS.
STORE H. C. S.
COMBINED
1V i+
ll-
ifi ............it
HrWTT
+
l
-
+
-l+ I--N-i-t
I
iI: i' 1
iii 4+4 <
4~4A4Th~L
-t
HC~L
+H++
-H+i-
4+H++9++
4+H+- +4
HM++it+H+
*" 44*4* **,4
4t-
HiIN >I+H+++-HH++HiN -H+ -H
Ti
+N
.4
4 - 111
....... 'HH H H !
4
a
-4N
_Hii11tt i i+H+ H++ 44
t
H+ -V ~
i+
1*
++++++-+-t+r
HN-
41
F
4
-+ +NiIl-N iiN++ t4 +H-H-
4±
-4
INDEX
TT
It
1
SOCIAL
1~~~
4tt***##
H++i~
tiw :
Ht
~L
L
t1
V 444--++-F4IH
S H ++H++H
-1-
-
-rt.-H +H+
4t-
44444
H+H+Ht-I
4-
LA
if+~~
ji4lj44ttilz444j4t4z
H-v-I 7ThVV
-ti-H+H±H-i-H-i-iI -i-H-i-I i-i-Ii--i--H-++H±t-i--VH&-H--f-IH-H2-v
4-H-
44-
44fl444444j4j P 4
4V4t 1t
T
411ii4~
14, 14+Hi'- 114
h
a
1
1
La
T u titttittit
+
jIBijJ4l~
Hal
t
44,
I
r.46
22
6.
Social Factors Ranking
r
=
.46 (see Chart 6)
The figures used to indicate social "quality"
of the various tracts are purely relative, having been
obtained by dividing the city in five fifths according
to their ranking.
"neighborhoods",
These rankings were compiled by
each of which coincides fairly well
with an average of two or three census tracts.
The
tracts in the best fifth were assigned the score of
one, and so on.
Sixteen different factors were taken
into consideration in compiling the rankings, all with
equal weight:
Crowding, density, median rent, advanced schooling,
median year of school completed, percentage of
foreign born populationjuvenile court appearances,
cases of the Mass. Society for Prevention of Cruelty
to Children, Dependent Aid cases, Old Age Assistance
cases, Aid to Dependent Children cases, unemployment,
infant mortality, new T.B. cases, T.B. deaths, age
adjusted death rates (all are rates rather than absolute figures, where applicable).
There is a certain amount of duplication with factors
which were separately correlated with net migration
(especially involving housing factors); some of the
factors also are not strictly of a social nature, although
admittedly they may have social implications.
Since the
original figures for individual indices were not available
the entire group was used as described above.
1.
Figures from records of United Community Services.
I-
PWRW
=v
- - -1-1-
23
7.
Need for Major Repairs.
r = .34
The figures listed in the data sheet (Table I)
for this factor represent percentages of structures either
in need of structural repairs or unfit for habitation.
This is
first
correlation coefficient which is not sta-
tistically significant, i.e. it could be accidental.
The depression may be partly responsible for
this low correlation, by forcing people to move into or
stay in slum areas where many structures were in poor
condition.
This same line of reasoning may apply to
many of the other housing and environmental factors.
1.
Figures taken from Boston Real Property Inventory,
1934.
~-pI
24
8.
Average Duration of Occupancy of Rental
Units 1
r = .20
The Real Property Inventory lists duration of
occupancy of the occupied rental dwelling units as of
1934; those durations were assumed typical for the entire
decade, and the duration in months was computed in the
following manner:
Duration
of
Occupancy
(R.P.I.)
9
1 yr.(to 2 yrs)
"f
2 yrs. "3
"
5 "
5-9 " " 10
10 ""
unknown
Number of
Rental D.U.
Units x Dur.
3
3 mos.
0-5 mos.
6-11 "
3-4 "
Average
Duration
Assumed
18
30
48
90
180
"f
"
81
171
144
145
151
98
243
1027
1440
2322
4530
5880
disregarded
14, 500 ; 901:16
Total
rental
d.u.
16 x 3 48 mos
901
Again the Back Bay tracts are the outstanding deviants
from the expected pattern, thereby reducing the coefficient of correlation considerably.
1.
Figures computed from Boston Real Property Inventory,
1934.
NET MIGRATION IN SAMPLE
TRACTS
A
5
0
I
-J
N
Q
Vp'
T-3B
RAILROAD PROPERTY
PARKS & PLAYOROUNDS
CEMETERIES
CENSUS TRACT NUMBERS
CENSUS TRACT 8OUNDARY
MEALTM & WELARE AREA
BOUNDARY
0 -
20-
30% IN
10 - 20%
0-
'a-
1010/
10-
2,0%
"t
20-
30%
"A
30- 40%
I
0
RrolTED taTE
VA
3
2
SCALE IN MILES
CE"L
Trts
~p
J8@4@
10 %OUT
4
S
A
26
9. Distance from City Center
No Correlation
Undoubtedly the lack of correlation is in part
due to the almost complete absence of really outlying
tracts among the sample.
This, as mentioned earlier,
could not be avoided, since most tracts in the southwestern part of the city were split between 1930 and
1940, so that census figures are not comparable.
The
very irregular shape of the city of Boston also is
probably working against such a correlation, since it
has distorted the growth pattern, which in cities with
regular topography is, however imperfectly, concentric,
at-least in the early stages.
1. Figures in miles, obtained from Chart 7, with
center arbitrarily assumed at Haymarket Square.
27
10.
Percentage of Dwelling Structures Forty
or more Years of Age 1
No Correlation
Conceivably forty years is
be regarded as critical in
too low an age to
a city as old as Boston.
It
is also recognized that many other factors besides chronological age influence the obsolescence of a neighborhood.
Nevertheless the complete absence of correlation
was somewhat of a surprise.
It certainly does not mean
that this factor will in no case show any correlation
with net migration.
In many centrally located areas there is a
population increase about the time the district is
converted to rooming house uses, which generally
happens after it has been built up for a good many years.
The fact that large parts of the Back Bay were at that
stage of development during the thirties probably
counteracted whatever correlation would have otherwise
existed.
1.
Figures from Boston Real Property Inventory, 1934.
28
11.
Percentage of Owner-Occupied Dwelling
Units 1
No Correlation
The fact that the highest percentage of
ownership is only about 35% shows that no single family
areas of better quality are included among the sample
tracts.
Since such areas could be expected to show the
least decentralization, the lack of correlation is not
too surprising.
The Back Bay tracts also ran counter
to the pattern here, showing very low percentages of
owner occupancy and a definite in-migratory pattern.
1.
Figures from Boston Real Property Inventory, 1934.
*- *L
I/VEL 4
FP
/
1-q L
/N
/VME
/VC&
/934 -90
E
R-iA C T
CN
M//?T4
A /
-
/I VE
R, REZN T
+
t
/--
f
04
-
E
4/Y
H3
290+3
.50
72
7
-236
+
4
69
+
Y3
~x/
/3j
30
-f
47/.)- /
4630/
I
00
S
'/x
+y
-2435
/143
//
N7
U
S~///-
X
270()0
y~/
b-
/g
OS
0,2
30,
I*
0,3-//
!
03
U+
6
T
~
4
2
&*
24
Of 5
28 v
V
979
130
0
R
16
/
9 99
/C/3
p
30
10
/< 2
4
1
-
98
13 3/3
A'0
4
60
/42-5304
C3
0
-DUL,
j
-Z 2 0
4 3
A 4/
A'
of
q-p
20Iz/0,2
/
40
-HAN6E 11V
s
AA
2
-
REr7
/934-
/AI/
N~
+/39
/I
4
NE T(
* zo
'Z '934'' To ;
UN/ T,
AV/lAER6
+
/7
20
4I
-
113
00
-
(y
/g-
qa+
30
B.
General Discussion
l.
Changes in the Number of Dwelling Units
Apart from tract K 3 where large scale conversion to smaller dwelling units and rooming houses took
place, the count of dwelling units showed few1 significant changes between 1934 and 1940.
In predicting net
migration for any area, the expected change in the number
and type of dwelling units should of course be considered.
Where the change is only in number, the percent of net
migration ought to be somewhat close to the percent of
change in dwelling units.
Where a change in type is
involved too, this is not the case however.
In tract
K 3 for example the net in-migration amounted to 20%,
while the number of dwelling units was more than doubled
between 1934 and 1940.
The 1930 census did not list
dwelling units by tracts which is the reason why 1934 and
1940 figures were used.
It is believed, though, that the
changes between 1930 and 1934 were relatively unimportant.
The percentage changes between 1934 and 1940 in the
number od dwelling units for the sample tracts are listed
in Table 1B.
1.
In tracts C 3 and R 1 the number of dwelling units
fell by 28 and 24 percent respectively.
-
"'~~
31
2.
The Influence of Non-Residential Land Use
In six of the twenty eight sample tracts the
1
predominant land use was non-residential.
These six
tracts showed an average net out-migration of approximately
16%, as compared to 9.9% for all sample tracts.
In
most of these cases very little encroachment by other
uses on residential land took place during the decade.
This indicates that a relationship exists, although it
does not reveal anything concerning its nature.
No
correlation was calculated between land use and migration,
since exact data were not available.
1.
From Boston Income and Cost Survey,
1935.
32
3.
The Back Bay Tracts
As mentioned earlier, the Back Bay area underwent a fundamental change in character during the early
thirties.
This confused the net migration picture in
these tracts, since a conversion from a high grade single
family area into a rooming house district not only
changes the age distribution of the population thoroughly,
but results in a decided net increase in total population.
The shift in age composition consisted mainly of a sharp
decrease in the number of children and a corresponding
increase in the young adult and old age groups.
Gener-
ally there are many contradictory and variable elements
present in areas of this type, which make the prediction
of net migration more difficult.
1.
Of the sample tracts, the following are located in
what is known as the Back Bay: J 4, J 5, K 2 and
K 3.
Tw
- /!
9
*i
-1
2097
3
-- /
/ //
306
42
-2
237
3 87
1/6
-3s
36
+-
330
5
-
276
+
297
d
+
03
;3
~447
/09
92
9
+
+
12
2031/
+
Pog
+
- 09
6 5 -
+
6.5
997
+
+0
/07
3
+
-156
19
-
-
/69,
3492
-
/0S
65
-
/7U
73
39
-
/02
/ o-
1/5
2/5
-
-
? 7.2
-
.23,2
-
2 34
381
/1
96
73
53
- 33.35 -.26 - 319 Z0 75
6
5
0
6
/26
7
9
-26
3
2
-
/5
-/.
9
93
5
+
/7
+1 61,
/
36
/!go
15
2/.7
-2/o
-3?S
-209
-
+
/61
/22
77
-
-
+
+.9
24
7
-
52
202.
-
5
3
95
+
7/7
+7D
5
&
-/2
2
7
5
+
-/9
-/3
/z
+
/2
-
-253/2
D267
-
32
-
/
5
Q
2
+
+/3
53
+
-520
57
-290
08
-
1120
-5S00
260
67
0o
272
- z/6
190
+ /94
z- 7a
/9
/94
-273
-
/72-
2/
6
6
/-
25
260
7
'
/5
Z
4
5/
37
-
- 13
/95.
3
-
+
-
8
-
29
261
-725-
51+
6
/S
8
+
39
8
t
p0
-
+7' //.
//0
/-
-77
-
-&
-
+
-
-
/
6
/23
/2
3-597
-8
3
-.
-3/3
1
/6
&/1$9
+
2
s5-5
Id /7
6
372
//0
26-o
+
/0,6
-228
377
/
4
+ 96
+11--/7
3-
32
96
q
;
/ Z.2
311
/
-P9.7
+
-1
/6
-
+o
27 ///92
6
Qj
270
-.
209
7
9
-20
,3.2
226
Y
/,53
/3
26
33
T
6/
/ o
72 Y-
.
0
-
+1
/
-6
5
444
4
1
+-
56
-
-
39
3
+j /09
35-
Uy
/
-3 30-39
S
79
717Z+/8
269
-
-
+
12
2z/02 l
269
+
106
/5
M/GRAT/o/V
25
-4230
/66
/
.254z
-,o
/63
.
T
/E
.20 -'24
/3-/19
///3
I
-
/7
+
/3
7
-
60
-227
/8
.29
-4/.o
-275S
-/563 57
+
+
-
6/
9.29
903.
30,
7
2
8
/
+29
901239
6
4/
/2
/6
47/
/0
76
/
0
7
+
+
4
/
33
4.
The Cycle of Migration; Migration by
Age Groups
Sociological theory states that young adults
from rural areas move into the city, often living at
first in the rooming house areas.
more stable residential areas.
Later they move into
After they are married
and start a family they move out into the suburbs if
they are successful enough economically to be able to
(This particular phase explains the sharp
afford it.
outward trend among children under ten years of age).
After the children marry and perhaps either husband or
wife are widowed, he or she moves back to the central
area.
The average, migration in the sample tracts by
age groups (see chart 8) reflects this picture quite
accurately.
Some residential areas of the city, such as
the Back Bay, are predominantly "centralization areas",
while most of the more stable residential districts are
decentralization areas.
In the better areas from an environmental point
of view there are generally people willing to move into
the dwellings vacated by this process.
In the poorer
ones, where no replacements can be found, this movement shows up to its full wxtent as net-out-migration.
Cyclical variations blur this picture to some
extent, but, judging from this study, not enough to make
34
it unrecognizable.
It appears that both centralization
and decentralization activity are increased during
prosperous times and inhibited during depressed eras.
During recent decades the influx of rural
population into the large cities has slowed down to a
This probably explains in part the predomin-
trickle.
ance of out-migration over in-migration, especially
since decentralization still seems to be gaining
momentum.
The migration pattern found in this study with
respect to age group apparently continued with very
little change through the following decade.
A paper1
dealing with net migration in Boston during the forties,
describes an age distribution of migrants very similar
to that found for the sample tracts in this study.
"The decreases are in two specific groups:
1)
The very young; i.e.
boys and girls
between the ages of 0 and 9.
2)
The middle-aged group; i.e. men and
women between the ages of 30 and 65."
"The increases may also be classified into
two distinct categories:
1.
P. H. Nash: Techniques for Calculating Demographic
Changes and Density Standards in Planning, 1IQ,
(American Society of Planning Officials, Chicago,
1951)-
CHART '8 :AVERAGE
I
NET MIGRATION BY AGE
-
GROUPS
r
Lp
-}
-r
-
1
if
t l
!Ii
1i
T-t
-
T--
-T
-
--
I
--
-
- ,
t-
4-
L
-
.....
...
TT
-
--
-
i
4 -
-
-
---
.. .
..
..-..
L ..t L , ...
-
-
--
36
1)
Young people between the ages of 10 and
29, consisting primarily of young girls
(men between 15 - 19 and 25 - 29 indicate
slight losses).
2)
Old men and women over 65 years of age."
The only differences between the above and the
findings of this study are: the twenty-eight sample tracts
showed slight out-migration in the 10 - 14 group.
Also,
the age groups were not broken down by sex in this calculation.
The percentages of migration in
each tract by
age group are listed in Table lA, and the average migration by age groups for all sample tracts is
graphically in chart 8.
shown
37
5.
The Role of Decentralization and
Centralization
It is unfortunate that practically nothing is
known about the origins and destinations of migrants.
Such knowledge would indicate much more clearly what
part of migration can be ascribed to decentralization
and concentration.
As this study is designed, there
is no way, for instance of separating people who move
into (or from) the suburbs from those making interstate or inter-regional moves.
Thus a variety of forces
influencing migration could not be isolated, so that
their quantitative effects were not measurable.
The
only thing which can be done under the circumstances is
to assume that inward and outward movement of other
types cancel each other, so that what remains is
"concentrated" decentralization and centralization.
T B
ousin
n0
r
r-,
+
,
rTtre
n
PS
o
DnIty
r,0
a'e
rwdd
Penalty,
-,
i
A 2
73
A
3
1413
*
E
z72
4
+
A" z
A' 3
30
-
2
0 //5
0 3
0
17
3
7Y2
3
Z/
/?~
81
2g- +
4- 725
R /143,
7
22
26/
T6/8+
U-
y
2Z4
/6
.2 f
/
96
.200
0 g
27
31
2
3
36 +-432-02
43
3
62
54
17
4
0/
34
64
oA1.2
3
31
/494iO
/7S
0
FT
3
4
3
76
305
3
34 6
3
60
S.2
IV~~~~~~
57
3
3
2
.63
.(v
/8,
3
4
2/
7
/42
9
29
/4
Z.2
3/694559z
0
6
48
41
25
33
0
/Jz
34q341
3//4
/b3
b7
34S
/
3
3
/
3
4q.4
9 /3/
3/
)-61/
3893.
Z3K
047.2
4/
/9'
3
2.
22
7
0y4/
222
/ 28f1
7 0
6,6+103
29
22
q5
3 Y,2
5$7
6
41449gSI31
77
+
+
9
/Z9
/9 q54/
/0
2'
U S // /
V &3
X
V
5
260
)-2
2
3
3
6
0
03+
(
r
12-/
441/7
+/991
y
a jorovr4
44/
92.-
2
/V
Fa ctor I-
u
for
9
4i-
7
eec-
7
/3~~~Z
y +
Social
3
20
2321
3.
43
8)2-
NE
2G
V2
C 3
T/
aneai
Score
/9
.232
/3
4132
A
3
A
E
f
33
'/
3
/
|
EAA
38
III.
Conclusions
A.
Summary
Eleven correlations were performed between net
migration and various housing, social, economic and environmental factors; the following coefficients of correlation
were obtained:
1. Natural population change:
.63
2. Average monthly rent
.63
3. Excess density
-55
4.
-54
Room crowding
5. Housing penalty score
6. Social factors ranking
.46
7. Need for major repairs
.34
8. Duration of occupancy
.20
9.
Distance from city center
none
10.
Age of structures
none
ll.
Owner occupancy
none
-53
The first five of these correlation coefficients are
significant at the 1% level , the sixth one is significant only at the five percent level, and
7 through 11
are not statistically significant.
The Data Sheet, Table 1, may be referred to
for a complete listing of all the figures used in the
correlations.
1.
i.e. it can be stated with 99% certainty that the
correlation found was not due to accident or faulty
selection of the sample.
39
B.
Methods Suggested in Applying Results of this
Study
The correlation coefficients obtained in this
study show that at least six factors have a definite relationship to net migration.
In order to predict net migration for a given
area, it
is
therefore advisable to test these factors
both for the degree of their presence or absence,
and
for their trend.
If
net migration is wanted for just one area
of a community,
it
work with trends.
will generally be more practical to
Working by incidence would require the
study of other districts for purposes of comparison, in
a way similar to the procedure followed here with the
sample census tract, but will probably give more accurate results.
With either method past net migration should
be computed first, and city and regional as well as
national trends should be considered in determining what
changes can be anticipated.
Changes in average rent which are out of line
with the general pattern, changes in the number and kind
of dwelling units, changing land use.patterns, changing
population are all factors which should be examined for
their influence on the migration picture.
40
Comprehensive study of migration in a whole
city will in most cases yield more accurate results,
even for specific small areas, than isolated examination of such sub-areas, since the element of relativity
and comparison can be utilized only over wider areas.
The method for calculating net migration will
have to be varied from the one here described depending
on the availability of better, poorer, or similar data,
and the accuracy of the results will vary accordingly.
41
C.
Evaluation
While the correlation coefficients here ob-
tained are not high enough to permit accurate quantitative forecasts to be made on their basis alone, they
should be quite helpful if used in conjunction with
other information as outlined in the previous section
(III B).
Even if the correlations were considerably
higher, one single pilot study would not be reliable
enough to permit general application of results without
further research.
Among the factors which in this study showed no
statistically significant correlation with net migration,
some may very well have such a relationship in general,
though it was obscured by special conditions in this
particular case.
Further data are needed on the time dimension
of migration trends, in order to determine whether conditions in one decade will permit prediction of migration
during the succeeding decade.
The overall significance of this study lies
more in the new approach it points to for population
prediction and computation than in the specific information it conveys regarding the migration picture in
Boston between 1930 and 1940.
42
D.
Suggestions for further Research
It was stated earlier that the same or similar
correlations would have to be performed for other cities
and other periods in order to test the universality of the
findings here obtained.
Apart from this, several questions
came up during this investigation, which for one reason
or another, could not be adequately resolved, but which
seem well worth investigating:
1.
Are net migration trends in areas of
various sizes of sufficient duration to
base their prediction on conditions in
an
earlier period?
2.
What influence do cultural and ethnic
factors in the population exert, both on
the net migration pattern, and on the
sensitivity to certain correlated factors?
3.
What influence do additional factors, such
as changes in employment sources, ease of
access, traffic conditions, social reputation of the area, changes in social or
ethnic composition of the population etc.
have on the migration pattern?
4.
To what extent do city, state or nationwide migration trends correlate with and
influence local trends?
5.
What effect do cyclical economic fluctua-
43
tions exert on migration trend in areas
of various types?
6.
What influence do planning measures and
controls of various types have on the
migration pattern?
7. Can adequate explanations be found for the
deviation of some areas from the normal
pattern of correlation?
8.
What can be learned from correlation of
migration in specific age groups with other
factors?
9.
How are additional deaths distributed among
age groups and sexes in areas which have
higher death rates than the city as a
whole?
10.
Can normative ratios be set up between
natural increase and out-migration for
specific communities and specific periods?
11.
Would the change or trend in
the factors
which were correlated with net migration
perhaps be more significant than their
incidence?
44
E.
Specific Conclusions
1.
The net-migration picture in the sample
tracts was dominated by decentralization
and centralization according to age groups.
2.
The city was divided into decentralization
areas with net out migration, and centralization areas with net in-migration.
3. Family growth was the most important factor
in out-migration.
This conclusion is sup-
ported by the good correlation with factor
1 and by the higher fertility rate of outmigrants.
4.
Families with growing children were leaving
the central city area if they were able to
afford it.
The question remains whether
this trend can be stopped by redevelopment.
5.
Poor living conditions cause decentralization probably to a greater extent than high
taxes and similar considerations.
46E
er 1stir
62()4/2~
suatO
I
z
170
4203
22/3
z 26
25--29
-
S
6
V
Zf6
33
2372-
z7
2
320
t
JY
06
26
4q
3 7
7
9
32
/2e
/2
54
[4
90
3
4
34
7-S
Z/i
2
/1
91
7
12
o
z.
8.
z27-
z
2 o
er-v 9'
-7
2
260
|ti3$2
9
33
q
2Z
3 214339
3F
2/6
T27
2
/
-3V
3
3
9
/,3
79YO Y
2-
2
993//1/3?Y
190
22 YSb
49774
22
-
9
(
24g
2/
Z27
327
f-j 3?-/
-/y 392926
9
Z7
.,zoo
A
2f
/Go
6-v
a
:I Iu
c-Is
o3
/09
/7
t
4--
Z3
3
6
423/
9 46
5/
f7
7
377
7
17,4 qg6
2o4-o93
7
3 31
8R
40/0
h
(^
etz
rC,
Ir-/
rat
amet
V
t
/67-/
2
/4
b
1s'9v
2.k
0--
~
O-4
43
e
3
Z
3
7
20-yZ75 - 2F
30-31
35 -37
40- Vy/
q(0
/7
20
3o
3,5
/0
97
20
22
0
b&
2
4
35
79
q
375Z
22
J'||x2
21(l/
lg Iz
z7 Z0,7
2Y
21 7
I~3Y 1/
ZS-1
3167
Ne/
Z03
= e2
y68/
4
/L
n9
t
9
-
/0/
9
20
332.
10*Z43
7
/
64q
45
Appendix:
Computation of Net Migration
This was by far the most time consuming and
troublesome part of the investigation, primarily because
no satisfactory method was available.
The procedure
finally adopted was corrected and modified several times
during the process of its evolution, and, as described
here, should be regarded as an initial effort rather
than a finished product.
The objective was to calculate, as accurately
as possible, net migration by census tract and age groups,
from the following given data:
Population in 1930 by age group;
Population in 1940 by age group;
Actual number of births for each year of the
decade;
Actual number of deaths for most years of the
decade (but not broken down by age group);
City wide fertility rates;
City wide survival rates.
The following method finally appeared to be most satisfactory for this purpose:
(see sample calculation, Table
II.)
a)
1930 census population
b)
1940 census population
c)
1935 (average) population
46
d)
Five year (1935-1939) city wide survival
ratesi
e) Five year hypothetical 2 cumulative deaths
(1935-39) in tract : d) applied to c)3
f)
Ratio between actual deaths (l0a) and
hypothetical deaths in tract (IDh ) established, to be used in calculating 1940
survivors in the tract.
g)
Ten year city survival rates.
h)
Tract deaths during decade, if city and
tract rates the same.
i)
Tract deaths during decade(where tract rate
different):
city rate times ratio
of
actual to hypothetical deaths applied to
1930 population.
j) 1940 survivors in tract above ten years of
age ( a-i, carried forward by ten years)
k)
Difference between actual births in tract
during decade and census count for 0-9 age
group doubled, giving net migration in
that age group.
m)
Net migration for other age groups equals
j - b.
All survival and fertility rates taken from population
study of Boston City Planning Board, with adjustment
where necessary for differences in time period covered.
2. Hypothetical, as used here, means according to city
wide rates.
3. Death rate = 1000 - survival rate.
1.
47
Step k requires some explanation:
it is based on the
assumption that migrants have half their babies in the
tract, half of them outside; since both out-migration
and birth rates were generally higher during the first
half of that decade, the inaccuracy introduced by this
assumption should be quite small in most cases.
By comparing the net migration figures obtained
by the above method with those arrived at by a method
involving the assumption that migrants had the same fertility rates as non-migrants, it was discovered that such
an assumption would have introduced an important error.
The fertility
rates of out-migrants were in most cases
considerably higher than those of non-migrants in the
same tract.
This confirms the close correlation found
between natural population change and net migration.
One point to be kept in mind when interpreting the net migration figures by age groups is this:
The actual age of the migrant is up to ten years lower
than appears from the tables, since they show the age
at the end of the decade, while migration took place at
any point during the decade.
Generally the assumption that births, deaths
and migration are evenly distributed throughout the
period is implied in the method used here for computing
net migration, although in this particular case the inaccuracy is reduced as explained in connection with step
k above.
48
Finally,
it
should be noted that all
percent-
ages of net migration are based on the average natural
population for the decade i.e. the average between 1930
census population and 1940 expected survivors.
Bibliography
General Plan for Boston, Preliminary Report, Boston City
Planning Board,
1950.
Edwards, A. L., Statistical Analysis,
Rinehart, New York,
1946.
Firey, W. T., Land Use in Central Boston, Harvard University Press,
Gist, N. P.,
Cambridge,
and Halbert,
1947.
L. A.,
Urban Society,
Crowell,
New York, 1948.
Income and Cost Survey of the City of Boston, Boston City
Planning Board, 1935.
or
Jaffe, A. J., Handbook of Statistical Methods _f
Demographers, U. S. Bureau of the Census,
Washington,
1951.
Nash, P. H., Techniques for Calculating Demographic Changes
and Density Standards,
Planning 150, American Society
of Planning Officials,
Chicago,
Ratcliff, R, Urban Land Economics,
1951.
McGraw Hill, New York,
1949.
Real Property Inventory of the City of Boston, Vol. I, II,
Boston City Planning Board, 1935.
15th U. S. Census, Bureau of the Census, Washington, 1932.
16th U. S. Census, Bureau of the Census, Washington, 1942.
Wkelpton, P. K., and others, Forecasts off the Population
off the U. S. _
-_1l'a,
U. S. Government Printing
Office, Washington, 1947.