Group Title: spatial analysis of the dimensions of economic health in the southeastern United States (1950 and 1960)
Title: A spatial analysis of the dimensions of economic health in the southeastern United States (1950 and 1960)
CITATION PDF VIEWER THUMBNAILS PAGE IMAGE ZOOMABLE
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/UF00097777/00001
 Material Information
Title: A spatial analysis of the dimensions of economic health in the southeastern United States (1950 and 1960)
Physical Description: ix, 101 leaves. : illus. ; 28 cm.
Language: English
Creator: Romsa, Gerald Harry, 1969
Publication Date: 1969
Copyright Date: 1969
 Subjects
Subject: Economic conditions -- Southern States   ( lcsh )
Geography thesis Ph. D   ( lcsh )
Dissertations, Academic -- Geography -- UF   ( lcsh )
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
 Notes
Thesis: Thesis--University of Florida.
Bibliography: Bibliography: leaves 97-100.
Additional Physical Form: Also available on World Wide Web
General Note: Manuscript copy.
General Note: Vita.
 Record Information
Bibliographic ID: UF00097777
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000559424
oclc - 13489737
notis - ACY4880

Downloads

This item has the following downloads:

spatialanalysiso00romsrich ( PDF )


Full Text







A SPATIAL ANALYSIS OF THE DIMENSIONS OF
ECONOMIC HEALTH IN THE SOUTHEASTERN
UNITED STATES (1950 AND 1960)









By
GERALD HARRY ROMSA


A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY






UOF F LIBRARIES




UNIVERSITY OF FLORIDA
1969































To my parents and my wife

Mary













ACKNOWLEDGMENTS


I wish to express my gratitude to my advisor Dr.

Clark I. Cross and the other members of my committee, Dr.

Stanley D. Brunn, Dr. James R. Anderson and Dr. John E.

Reynolds for their guidance, thoughts, constructive criti-

cisms, and patience during the preparation of this disserta-

tion.

In addition, I want to acknowledge the University

of Florida Computer Center's facilities and services which

were utilized for the data analysis in this study.


1ii














TABLE OF CONTENTS


Page


. .iii


. vi


. ix


ACKNOWL',%UEDGMENTS . . . . . . . . .


LIST OF TABLES . . . . . . . . .


LIST OF FIGURES . . . . . . . . .


CHAPTER
I THE CONCEPT OF POVERTY . . . . .


Introduction . . . .
Objective . . .
Need for Such a Study . .
Poverty in the United States
Some Approaches to the Study
Review of the Literature .
The Problem . . . .


STUDY AREA AND METHODOLOGY .


The Setting . . .. .
Historical Background . . . .
Recent Trends in the Economy . .
Outline of the Analytical Model . .
The TModel. . . . . . . .
Variables Used . . . . . .
Summary . . . . . . . .


THE DIMENSIONS OF ECONOMIC HEALTH IN THE
SOUTHEASTERN UNITED STATES (1950-1960)
1950 Factor Analysis . . . .
The 1950 Dimensions . . . . .
1960 Analysis . . . . . .
The 1960 Dimensions . . . . .
Comparison of the Two Time Periods .
Regression Analysis . . . . . .
The Hypotheses . . . . . .
Summary . . . . . . .


. . 27
. 29
. . 30
. . 35
. 36
. . 37
* . 40
. . 42


. . 45
. . 46
. . 48
. . 60
. .62
* . 75
.* 78
* . 79
* . 84


. 1


. . . .
. . . .
. . . .
. . . *
of Poverty
.
. . e .


III


. . .









Page


CHAPTER

IV SUMMARY AND CONCLUSIONS . . . . . . 86
1950 Analysis . . . . . . . 87
1960 Analysis . . . . . . 89
Comparison of Changes in the Two Time
Periods Studied . . . . . . 92
Conclusions. . . . . . . .. 93
Future Research Efforts . . . . . 94

BIBLIOGRAPHY . . . . . . . . 97

BIOGRAPHICAL SKETCH . . . . ...... . ..101












LIST OF TABLES

Table Page

1 Per Cent of Families Earning Less than
$3,000 a Year by Regions (1960) . . . 8

2 Distribution of Poverty by Residence (1960) 9

3 Median Income by Residence (1960) . .. . 9

4 Variables Used . . . . . . . 41

5 Percentage of Total Variance Explained by
Each Dimension in 1950 . . . . . 47

6 Percentage of Variance (Communality) of
Each of the 26 Variables Accounted for
by All Six Components in 1950 . . . 47

7 Dimension I Urban Industrial Counties
in 1950 . . . . . . . . . 49

8 Counties with Extreme Factor Scores on
Dimension I in 1950 . . . . . . 50

9 Dimension II Tenant and Small Farm
Poverty in 1950 . . . . . . .. 52

10 Counties with Extreme Factor Scores on
Dimension II in 1950 . . . . . 52

11 Dimension III Wealthy Urbanized Counties
in 1950 . . . . . . . 55

12 Counties with Extreme Factor Scores on
Dimension III in 1950 . . . . 55

13 Dimension IV Rural VWhite Counties in 1950 57

vi








Table Page

14 Counties with Extreme Factor Scores on
Dimension IV in 1950 . . . . . . 57

15 Dimension V Locational Advantage in
Relation to National Market (New York)
in 1950 . . . . . .. .. 59

16 Counties with Extreme Factor Scores on
Dimension V in 1950 . . . . . . 59

17 Dimension VI Locational Advantage in
Relation to Local, Regional and National
Markets in 1950 . . . . . . . 60

18 Counties with Extreme Scores on Dimension
VI in 1950 . . . . . . . . 60

19 Percentage of Total Variance Explained by
Each Dimension in 1960 . . . . . 61

20 Percentage of Variance (Communality) of
Each of the 26 Variables Accounted for
by All Seven Components in 1960 . . . 61

21 Dimension I Urban Industrial Counties
in 1960 . . . . . . . . . 63

22 Counties with Extreme Factor Scores on
Dimension I in 1960 . . . . . . 64

23 Dimension II Rural Negro Counties . . 66

24 Counties with Extreme Factor Scores on
Dimension II in 1960 . . . . . . 67

25 Dimension III Tenant and Small Farm
Poverty in 1960 . . . . . . 67

26 Counties with Extreme Factor Scores on
Dimension III in 1960 . . . . . 69

27 Dimension IV Locational Advantage in
Relation to National Market (New York)
in 1960 . . .. . . ....... 70
vii








Table Page

28 Counties with Extreme Factor Scores on
Dimension IV in 1960 . . . . . 71

29 Dimension V Wealthy Urbanized Counties
in 1960 . . . . . . . . . 72

30 Counties with Extreme Factor Scores on
Dimension V in 1960 . . .. . .. 72

31 Dimension VI Locational Advantage in
Relation to National, Regional and
Local Markets in 1960 . . . . . 73

32 Counties with Extreme Factor Scores on
Dimension VI in 1960 . .... 74

33 Dimension VII Areas of Declining Prosperity
in 1960 . . . . . . . . 74

34 Counties with Extreme Factor Scores on
Dimension VII in 1960 . . . . 74

35 Expected Correlations Between Median
Family Income and the Derived Dimensions. 79

36 Normality of 1950 Factor Scores . . . 80

37 1950 Correlation Matrix . . . . . 80

38 1950 Summary Table . . . . . 81

39 Normality of 1960 Factor Scores . . . 81

40 1960 Correlation Matrix . . . . 81

41 1960 Summary Table . . . . . .. 82

42 Comparison of the Beta, Correlation
Coefficients and the Coefficient of
Determination Between the Two Time Periods 82


viii













LIST OF FIGURES


Figure Page

1 Study Area; the Southeastern Core . . . 28

2 Urban Industrial Counties, 1950 . . . 51

3 Tenant and Small Farm Poverty, 1950 . . 54

4 Urban Industrial Counties, 1960 . . . 65

5 Tenant and Small Farm Poverty, 1960 . . 68













CHAPTER I

THE CONCEPT OF POVERTY


Introduction

The increasing attention given by geographers to the

problems of poverty, economic health and underdeveloped

areas is a healthy sign that they are progressing with the

temporal needs of today's society. This concern with areal

differentiation in economic growth and levels of economic

wellbeing is not a new geography. It is, instead, a more

specialized topical outlook on the net result of economic

activity as viewed in a spatial framework.1 This approach

attempts to relate regional inequalities in levels of liv-

ing with man's economic endeavours as they are tied to his

cultural, technological and resource environment.

The trend to look for factors other than resource

endowment is perhaps the major criterion which separates

present geographical research from previous regional stud-

ies, Ullman has pointed out that natural resources and cul-

tural differences are not sufficient to explain the dispar-
2
ity in economic growth. Differences in social conditions

and advances in technology often provide additional informa-

tion in explaining the areal variation in prosperity. The

inclusion of additional variables to explain this
1









differential in economic broth has led to the use of multi-

variate statistical techniques as analytical tools. The

usefulness of these techniques lies in their applicability

in the analysis of large numbers of indices; in their isola-

tion of significant variables and in the testing of hypoth-

eses. These are important contributions when dealing with

interrelated data.

The realization by social scientists that the dis-

tribution of wealth and poverty is uneven has focused their

attention on the need for more regional studies. National

and state data often tend to hide small pockets of high

deviation in levels of income. This has led to economic

regionalization based on counties,and, where possible,

smaller units.

Spatial interaction is another geographic concept

which is increasingly being realized as playing an important

role in explaining the difference in levels of progress.

This concept was introduced by Ullman and refined by Hager-

strand at Lund University. Hagerstrand used simulation

models in an attempt to measure the degree of resistance

that cultural and physical barriers imposed upon the accept-

ance of new ideas among contiguous areas in Sweden.

Hartshorne points out that the comparison of eco-

nomic growth among underdeveloped regions should be rela-

tive rather than concrete.4 A direct comparison between two

culturally different regions may not show the disparity in









economic achievement accurately unless the indices utilized

are standardized to represent with less bias the contrast-

ing cultural and political values. This also is true of

temporal studies. Although this problem at present has not

been resolved, studies within a similar cultural and polit-

ical unit based upon certain multivariate techniques are

feasible.


Objective

The purpose of this study is to identify the salient

socio-economic dimensions or characteristics of five south-

eastern states (Mississippi, Alabama, Georgia, South

Carolina and North Carolina) for two time periods, 1950

and 1960. These characteristics will be examined to ascer-

tain if they reflect any areal patterns or changes through

time. If areal variations do exist, they will be analyzed

in regard to their stability, location, effects of urbaniza-

tion and the importance of push and pull factors.

In addition the extracted dimensions will be tested

individually and as a group for their validity in explain-

ing the high deviation in median income throughout the

study area.


Need for Such a Study

The concern of some Americans for a more equitable

distribution of this nation's wealth has shown itself

through civil legislation and action programs by state,









local and federal agencies. Programs such as the Area

Redevelopment Act and the Appalachia Act are examples of the

federal government's increased participation. To implement

these programs rapidly and to the best advantage requires

a thorough knowledge of the problems. Thus research is

essential if the complex facets of poverty are to be under-

stood. An example of such a research program in the South

is The Southern Regional Research Project S-44, "Factors in

the Adjustment of Families and Individuals in Low Income

Areas of the South." This is a cooperative study done by

rural sociologists in nine southern states. These studies

were done in selected counties and dealt primarily with the

attitudes of the rural poor.

There is a need, however, for a comprehensive study

of the Southeast which deals vith the facets associated with

poverty on a larger scale. The Southeast has often been

classified as a retarded area. Recent studies have sug-

gested that this blanket statement can not be applied to

the entire region, as it has made some great economic strides

in the past few decades. At present, there is a lack of

geographic research on this region employing multivariate

statistical techniques, which has emphasized the degree of

variation in economic growth through space and time and

classified economic regions accordingly.

In the remainder of this chapter an overview of the

distribution of poverty in the United States and some






5



approaches to its study are discussed. The pertinent lit-

erature is reviewed as a basis for formulating hypotheses.

The analytical model used in this study is developed in

Chapter II and a brief description of the study area is

given. Chapter III deals with the analysis of the data,

while in ChapterlV the conclusions reached are examined

in view of the hypotheses presented in Chapter I.


Poverty in the United States

Amid the highest standard of living ever achieved

by the United States there exists a forgotten minority of

poverty stricken people. The problems of the poor are often

masked by the middle class way of life in America. There is

no room in this society for the economically less fortunate.

The middle class viewpoint is well expressed in The Affluent

Society by John Galbraith.5 Galbraith's major theme was how,

with increasing automation, the good life could be continued

for all, with less work and more leisure. Poverty was per-

ceived only in two forms: case poverty due to physical

and mental handicaps, and insular poverty such as is found

in West Virginia.

More recent articles have shown that the problems of

poverty are not isolated but nationally widespread. There

are, however, large concentrations of the poor in certain

geographic areas such as the South and in the ghettos of

urban centers. Depending upon the criteria used, there are









estimated to be from 36 to 50 million economically deprived

persons within this wealthy nation.7

This staggering figure is made even more realistic

if one considers the economic, social and psychological

costs of maintaining these individuals in their present way

of life. Not only does the Gross National Product suffer

from the loss of such a large number of human resources,

but there are also heavy social costs. These include wel-

fare payments, costs of mental and physical health, crime

and juvenile delinquency. The national cost of welfare pro-

grams has been estimated at 13 to 15 billion dollars.

Poverty was first recognized as a social problem in

the United States approximately 100 years ago. It existed

before but had been regarded as an individual rather than a

social problem. Prior to this period, individuals attempted

to attain economic freedom by leaving the urban ghettos or

the worn out soils of the East and moving westward. This

movement could only be sustained in a highly agrarian soc-

iety. The slaves on the southern plantations however did

not have this escape valve.

Increasing industrialization and urban growth after

the Civil War brought with it many social evils: poor tene-

ment housing, inadequate schools, low wages, bad working

conditions and no job security. Henry George in his Pro-

gress and Poverty referred to the United States of 1869 as









a land where "amid the greatest accumulation of wealth, men

die of starvation and puny infants suckle dry breasts."10

The plight of the working classes, many of whom were immi-

grants, eventually led to some social reforms. Community-

minded citizens led the battle for public housing regula-

tions, public health, and abolition of child labour. Labour

unions played a major role in obtaining higher wages and

better working conditions.

Major Federal government action on poverty was ini-

tiated by President Franklin D. Roosevelt in the 1930's.

In the State of the Union message in 1933 the former Presi-

dent referred to the "one third of the nation that is ill-

clothed, ill-housed and ill-nourished."1l It was during

this period that the familiar antipoverty tools: social

security, old age assistance, aid to dependent children,

and unemployment compensation were conceived. Furthermore

the Tennessee Valley Authority Act was a recognition that

there were regional inequalities in economic growth, which

resulted in pockets of distressed areas that were resistant

to change and progress.

The next major Federal attack on poverty commenced

in the 1960's. The emphasis was on legislation. The Area

Redevelopment Act of 1962, the Economic Opportunity Act and

the Appalachian Regional Development Act of 1964 are a few

examples. The passage of the acts emphasized rehabilitation

and not monetary relief. They also recognized the areal









distributions of poverty and the need to focus upon the

problems of regional or subregional nature which can not

be solved on the national level. The new progress in

legislation was initiated by a small educated white minor-

ity and a group of dedicated Negro leaders. Civil Rights

marches, sit-ins, and riots have forced the nation's atten-

tion to the problems of her minority poor when other less

peaceful efforts failed.12

The geographic distribution of poverty can readily

be seen if the per cent of families earning less than

$3,000 a year is examined by the major divisions of the

country (Table 1).


Table 1

Per cent of Families Earning Less Than
$3,000 a Year by Regions (1960)

Region Total White Nonwhite
Northeast 9.8 8.4 26.6
North Central 11.5 10.3 29.7
South 24.6 17.9 58.3
West 11.7 11.0 20.7

Source: Louis A. Ferman, Joyce L. Kornbluh and Alan Haber
(eds.), Poverty in America (Ann Arbor: University
of Michigan Press, 1965), p. 72.


This table shows the enormous variation in income across

America. It further points out the plight of the non-

whites. Although the urban centers harbour the majority

of the poor, the rural areas have a higher percentage of

poor families (Tables 2 and 3).








Table 2

Distribution of Poverty by Residence (1960)

Residence % of population % of poor (families
earning less than
$3,000 per year)

Urban 71 54
Rural nonfarm 22 30
Farm 7 16


Source: Louis A. Ferman, Joyce L. Kornbluh and Alan Haber
(eds.), Poverty in America (Ann Arbor: University
of Michigan Press, 1965),p. 104.


Table 3

Median Income by Residence (1960)

Residence Total Population White Nonwhite

Urban $6,580 $6,678 $4,469
Rural Nonfarm $5,486 $5,549 $2,645
Rural Farm $3,779 $3,863 $1,323

Source: Louis A. Ferman, Joyce L. Kornbluh and Alan Haber
(eds.), Poverty in America (Ann Arbor: University
of Michigan Press, 19657, p. 107.

In comparing the income between rural and urban

residents, one should consider the difference in cost of

living. Nevertheless the differences in income are greater

than the variation in the cost of living. From these

figures a picture arises that reveals small concentrated

areas of poverty within cities and a more dispersed form of
rural poor. Within the rural sections of the country, there

are extreme variations in the concentrations of the poor.
Examples of areas with a high rate of unemployment and a

concentrated number of low income groups are: Appalachia,









the cutover regions of the Great Lakes, northern New
13
England, the northern Rockies and the Indian reservations.

Other regions containing large but less concentrated numbers

of poor families are parts of the northern Great Plains,

scattered areas in the West and Midwest and extensive sec-

tions of the Piedmont and the Atlantic Coastal Plain.1

This pattern of dispersion not only shows the

national scale of the problem, but also the variations

from the industrial North to the rural West and the South.

The distribution of natural resources can not account for

all of the variation in poverty. Part of this is due to

demographic factors. Minority nonwhite groups are con-

centrated in many of the distressed economic sections.

Some depressed regions have a high percentage of unskilled

labour, while many of the predominantly white rural coun-

ties have high concentrations of older people. The latter

is the result of the out-migration of the young to more

attractive economic areas. Continuous out-migration over

a period of time, of the highly motivated segment of the

population, will tend to economically depress an area. This

was illustrated and substantiated by Spinelli in his study

of the net out-migration in southeastern Ohio, 1950-1960.


Some Annroaches to the Study of Poverty

One of the most difficult problems in studying pov-

erty is its definition. 4What constitutes poverty? Many









popular definitions adopted are based upon incomes. Sta-

tistics on family and individual income are readily avail-

able and accurate. The present $3,000 criterion used for

families and the $1,500 for single individuals was defined

and used in 1964 by the President's Council of Economic

Advisors. The Council estimated that in 1962 there were

between 33 and 35 million Americans living at or below this

income level.16

Economists such as Leon Keyserling believe that this

criteria are too low. Keyserling raises the income criteria

for families living at the poverty level to $4,000 and for

single persons to $2,000 (1960 dollars). In his analysis

he includes a second grouping called the deprived category.

This group includes all families with incomes between

$4,000 and $5,999 and individuals with earnings between

$2,000 and $3,999. This latter group can subsist on their

incomes, but can not afford many of the luxuries that our

present society has to offer. Keyserling estimates that 73

million Americans fall within these two categories.17

One of the weaknesses of these estimates, based upon

the above definitions, is that they do not account for the

differences in the cost of living throughout the country.

They also do not contain flexibility to take into account

inflationary periods of the economy which increase the cost

of living for individuals on fixed incomes. The social

security administration has made some attempt to account









for the variation in cost of living, but the invariant

$3,000 base is still largely used by them. A more flexible

standard which distinguishes the differences in need between

size of families, rural and urban residence, was published

in 1964 by the Department of Health, Education and Welfare.

This minimum budget ranges from $1,100 for a farm family of

two, aged 65, to over $5,000 for a nonfarm family of seven

or more individuals. This more flexible budget estimated
18
the poor at 34.6 million. More realistic income criteria

are those suggested by the Bureau of Labor Statistics.

Their index was developed for an urban family of four liv-

ing in 20 of the major cities. This index varied from

$5,370 in Houston to $6,567 in Chicago.19

The above estimates show the disagreements among

experts on what is a proper income level for the definition

of poverty. These estimates, furthermore, have not taken

into account transfer payments in the form of financial

assistance from welfare agencies or relatives. They also

have not made any allowance for taxation and therefore,

because of these omissions, do not represent true disposable

income. Indeed most students of poverty feel that the use

of a single index is not sufficient. The trend is to use

several indicators such as color, age, income and education.

Some social scientists have approached the problem

from the sociological and psychological viewpoint. One of









the first to use this definition was Oscar Lewis, who states:

Poverty in modern nations is not only a state
of economic deprivation, of disorganization or
of the absence of something. It is also positive
in the sense that it has a structure, a rationale,
and defence mechanisms without which the poor
could hardly carry on. In short it is a way of
life, stable and persistent passed down from gen-
eration to generation along family lines. The
culture of poverty has its own modalities and
distinctive social and psychological consequences
for its members.20

Economically retarded areas have within them distinct

cultural features. The poor having little money, virtually

no savings and no economic security, have evolved unique

consumer habits, family structures and a way of life. This

subculture, if it may be so called, is composed of the

chronically unemployed or those having irregular employment.

Those who are employed work in the non-unionized low-paying

sectors of the economy. Their low salaries can not provide

for the costs of higher education and health services.

Often their meager earnings are spent on entertainment

luxuries,instead of the basics of food, shelter and cloth-

ing, in an attempt to escape their environment for short

periods of time. The end result is often a low resistance

to disease, ill health and a shorter life span.

The everyday problems of this subculture, due to the

constant stress of maintaining life without hope, has created

in turn psychological manifestations. This has lead to a

lack of motivation and imagination among the destitute. Pro-

tective shells against the outside world are erected. Within










these shells provincialism and deep-rooted prejudices are

maintained. The outside world is viewed with hostility.

This hostility is expressed in their distrust of various

government agencies and programs.

The sociological approach in defining poverty,

although accurately describing the subculture of the poor,

has certain drawbacks. The major restriction is the lack

of sociological data for large sections of the country. To

gather the data requires personal interviews, which not only

can be costly, but also time consuming. In addition, this

definition emphasizes to a great extent the end results

rather than the causes.

A more recent concept, economic health, is gaining

in popularity as it attempts to analyze which indices are

responsible for a lower level of living and how these

indices may vary through space and time. The term economic

health is not universally used, but is represented in a

large number of multivariate studies that use numerous

social, economic and demographic indices. These indicators

are then reduced to a smaller number by elimination or

union with other variables into new ones. This new set of

indices, which accounts for the largest possible variance

within a workable matrix, is then used to explain the dif-

ference in economic growth. This matrix can be applied to

inter-, or intra-regional studies and to a certain extent










be used to explain the changes through time. Economic

health thus tries, by the use of a large number of variables,

to explain why and how different regions of a country,

relative to each other, differ in their overall economic

prosperity. The method of analysis is not new, but it was

not until rapid calculating computers became available that

large regions could be examined on the basis of a multitude

of variables, and at the same time remove personal bias in

regard to the importance of the variables used. This form

of analysis is being used successfully by various social

scientists engaged in regional studies.

The concept of economic health will be used in this

study to explain the differences in prosperity, if any,

within the five southeastern states in the study area.

Twenty-six variables which are indicators of prosperity

will be utilized to ascertain what role they play within

the study area, and if their importance changes through

space and time.



Review of the Literature

The dependence of the Southeast on agriculture has

been documented by geographers (Hart, Vance).21 This

historical reliance upon a few major crops such as cotton

and tobacco had its drawbacks. These crops were export

orientated and consequently fluctuated greatly in price.









Soil mismanagement often led to the abandonment of farms or

a loss in fertility and yields. The plantation form of

agriculture was not conducive to the modernization of agri-

cultural practices, as cheap labour after slavery made it

economically unfeasible to adopt some of the new techniques

until relative prices of the inputs changed. Rural values

predominated and eventually evolved into a way of life.

Political power came under the control of a small, land

based minority to the detriment of education and industrial-

ization.

This theme, that the South's historical development

acted as a barrier to progress, pervades the literature on

the South. Nichols in a paper "Southern Tradition and

Regional Economic Progress" suggests that five social, polit-

ical, psychological and philosophical factors were respon-
22
sible for the lag in southern economic growth.22 These were:

the dominance of agrarian values, rigidity of a social class

structure which prevented the establishment of a strong mid-

dle class, the control of the political parties by a few which

led to an undemocratic political structure. Furthermore

a lack of social responsibility hindered the development

of a first class public school system. This, tied with the

conformity of thought and behavior which was expected of a

good southerner, left little or no sympathy for dissenters,

many of whom migrated North. The loss of this liberal ele-

ment hindered change and helped maintain the status quo.









In spite of these handicaps the South in the last

few decades has made tremendous strides forward. MacDonald

"On South's Recent Economic Development" points out that

personal income in this region, has increased since 1940,

at a faster rate than in the nonsouth.23 Part of this

increase is attributed by MacDonald to a shift in the Amer-

ican economy from coal and steel to plastics and petroleum,

and the increased exploitation of other southern resources

especially forestry and agriculture. The other main factor

responsible for the rapid increase in personal income was

a slow population growth rate, due to the heavy out-migra-

tion of low income groups. This recent progress of the

South was further verified by Lassiter. In his study

"Education for Males by Region, Race and Age "Lassiter

pointed out that at present the investment in education is

providing better financial returns for white southern males

than white northern males.24

This rapid rate of growth however has not been con-

sistent throughout this region. Moon and McCann, who looked

at the levels of adjustment in 30 rural counties in the

states of Kentucky, Tennessee, North Carolina, Texas,

Louisiana, Mississippi and Alabama, found statistically sig-

nificant variation.25 Similar conclusions were arrived at

by Tarver and Beale in "Population Trends of Southern Non-

metropolitan Towns."26 They concluded that the significant

factors explaining the differences in population growth









among 801 southern towns with populations between 2,500 and

9,999 were: the size of the town and its regional location.

Fuguitt on the other hand found that county seats in rural

nonmetropolitan areas grew at the expense of neighboring
27
centers.2

The importance of urbanization as a significant fac-

tor in explaining regional growth in the United States has

received considerable attention. Schlesinger has argued

that since the Civil War, cities dominated the American
28
growth process. Priedman and Alonso depict a spatial

economy in which the processes of economic development are

leading to greater specialization and increased polarization

of growth in a few urban industrial complexes, each of which

is functionally integrated with a surrounding economic area29

These centers not only grow so rapidly as to create
problems of an entirely new order, but they also act
as suction pumps, pulling in the more dynamic ele-
ments from the more static regions. The remainder
of the country is thus relegated to a second class,
peripheral position. It is placed in a quasi-
colonial relationship to the center, experiencing
net outflows of people, capital, and resources,
most of which is redound to the advantage of the
center where economic growth will tend to be rapid,
sustained and cumulative. As a result, income dif-
ferences between the center and the periphery tend
to widen.30


This viewpoint is also expressed by Schultz, who

argues that economic development occurs primarily within

an industrial urban matrix where the factor and product

markets work best. Agricultural activities located most









favourably in regard to these centers can utilize the fac-

tor and product markets more efficiently than those located

on the periphery of these urban systems.3 This was

partially substantiated by Bryant who used multiple regres-

sion to measure the causes of inter-county variation in

farmers' income in the United States.32 He showed that

variation in farm income is largely determined by age, land

and capital inputs, education and color. Location in respect

to regional urban areas was only significant in the eastern

United States, implying that farmers west of the Mississippi

were more dependent upon national and international markets.

0. D. Duncan, et al., in Metropolis and Region,emphasize the

domination of hinterland activities by metropolitan centers3

Both agricultural and non-metropolitan manufacturing activ-

ities were shown to be a function of general accessibility

to the urban system of the nation and distance to the local

metropolis. Berry concluded in his study of rural poverty

in Ontario that

areas lacking the dynamic conditions of industrial
urban development experience high rates of out-
migration and commitantly realize differences in
per capital income, even though labour migrations
generally appear sufficient to overcome differences
attributable to differences in original resource
endowments. The factors linking rural poverty to
industrial urban growth polarized in urban centers
and varying in amplitude according to the rank of
these centers in an urban hierarchy, are thus clear.

This urban influence has also been clearly demon-

strated by Thompson, et al.,35 Berry,36 Bell and Stevenson,37









38
and Hodge. All of these authors used factor analysis on

a large matrix of economic indices in order to derive a

smaller number of dimensions or factor loadings which would

explain more easily the variations in income levels. In all

cases an urban dimension was derived from the analysis that

was correlated with the rapidly growing areas. The study

by Thompson, et al., on the economic health of New York

State, is of particular interest here. It not only demon-

strated that the large metropolitan cities were the nodes

of economic growth, but also that the predominantly agri-

cultural counties near these centers fared better than

smaller cities, which,had populations of 50,000 or less.

Berry arrived at the same conclusions in his analysis of

rural poverty in Ontario. Location of the rural townships,

in relation to urban areas, played an important role in

determining the level of economic achievement as measured

by 47 variables. In only a few exceptions, most of which

were due to the presence of mineral resources, did a few

townships deviate from the pattern of hinterlands dominated

by large urban industrial clusters. Using similar procedures

Hodge analyzed 473 trade centers in Saskatchewan. He found

that trade center viability was largely a function of rank

on the urbanism scale; the lower the rank the greater the

probability of decline.

The apparent importance of the urban or industrial

dimension is clearly seen in these studies. That these








complexes do grow at the expense of their hinterlands is

readily verified, if one looks at the variables which are

correlated with the urban industrial dimensions. These are

generally high income groups, youthfulness of the popula-

tion, a high level of education and a high level of capital

expenditures. All are indices of a healthy economy. The

converse is true; the dimensions representing poverty have

weaker associations with the above indices.

Are those indices stable through time? The decline

of certain regions relative to others would seem to indicate
39
the contrary. King found that they were not stable. In

his examination of Canadian cities he found that the per-

centage of variance explained by the dimensions arising from

his analysis varied through time, although the factor load-

ings associated with the dimensions remained fairly stable.


The Problem

The problem involved in this study relates to the

variations in the economic environment among 454 counties

in Alabama, Georgia, Mississippi, North and South Carolina.

Specifically does variation exist? What dimensions account

for this variation? Do they vary over time? What percent-

age of poverty do they explain? Are some counties progres-

sing at the expense of others? The literature reviewed pro-

vides a basis from which to examine several useful spatial
hypotheses of economic health in the selected southeastern
states. The hypotheses to be tested in the study are:









(1) There is variation in economic health among the

selected counties and these can be explained in terms of a

few dimensions arising out of factor analysis.

(2) These dimensions are not stable but change

through time as do the conditions required for economic growth.

(3) Within the study area there will be nodes of

rapidly growing areas which thrive at the expense of their

hinterlands.

(4) These nodal areas are associated with urbaniza-

tion and it is postulated that the degree of growth within

these sectors, relative to each other, is a function of popu-

lation size and location, location in relation to local,

regional and national metropolitan centers.

This study will be successful if it can demonstrate

the underlying characteristics of areal variation in economic

health in the five selected southeastern states for two points

in time, and ascertain if these characteristics have changed

between the two time periods (1950 and 1960).












REFERENCES


Richard Hartshorne, "Geography and Economic Growth,"
in Essays on Geography and Economic Development, Norton
Ginsburg (ed.) (Chicago: The University of Chicago Press,
1960), pp. 24-25.

Edward L. Ullman, "Geographic Theory and Under-
developed Areas," in Essays on Geography and Economic Devel-
opment, Norton Ginsburg ed.) (Chicago: The University of
Chicago Press, 1960), pp. 26-27.

Ibid., p. 29.

4Hartshorne, op. cit., pp. 18-23.

5John Galbraith, The Affluent Society (Boston:
Houghton-Mifflin Co., 1958).

Ibid., pp. 323-333.

7Michael Harrington, The Other America, Poverty in
the United States (Baltimore: Penguin Books, 1966), pp.
176-179.

Lar A. Levitan, "Programs in Aid of the Poor,"
Poverty and Human Resources Abstracts, Vol. I (1966), pp.
11-25.

9Herman P. Miller, Poverty American Style (Belmont:
Wadsworth Publishing Co., Inc., 196b), pp. 2-3.

0Louis A. Ferman, Joyce L. Kornbluh and Alan Haber
(eds.), Poverty in America (Ann Arbor: University of
Michigan Press, 1965), p. 1.

11Ibid.

12 Miller, op. cit., p. 5.









13
Gordon E. Record, "The Geography of Poverty in
the United States," in Problems and Trends in American
Geography, Saul B. Cohen (ed.) (New York: Basic Books,
Inc. 1967), p. 105.

14Ibid.

5Joseph Spinelli, "A Study of Net Out-Iigration in
Southeastern Ohio, 1950-1960" (Unpublished Master's Thesis,
Department of Geography, Ohio State University, 1966).

16Miller, op. cit., p. 2.

17Ibid.

18
H18arrington, loc. cit.

19Ibid.

20
2Elizabeth Herzog, "Some Assumptions About the Poor,"
Social Science Review, XXXVII (December, 1963), 390.

21
21John Fraser Hart, The Southeastern United States
(Princeton: D. Van Nostrand Co., Inc., 1967); and Rupert B.
Vance, Human Geography of the South (Chapel Hill: Univer-
sity of North Carolina Press, 1935).

22
22William H. Nichols, "Southern Tradition and
Regional Economic Progress," Southern Economic Journal, XXVI
(January, 1960), 187-198.

23Stephen L. MacDonald, "On South's Recent Economic
Development," Southern Economic Journal, XXVIII (July, 1961),
30-40.

24Roy L. Lassiter, Jr., "Education for Males by
Region, Race and Age," Southern Economic Journal, XXXII
(July, 1965), 15-22.

25
Seung Gyu Moon and Glenn C. McCann, Subregional
Variability of Adjustment Factors of Rural Families in the
South (Raleigh: North Carolina Agricultural Experiment
Station, Southern Cooperative Series, Bulletin No. 111,
January, 1966).









26James D. Tarver and Calvin L. Beale, "Population
Trends of Southern Nonmetropolitan Towns," Rural Sociology,
XXXIII (March, 1968), 19-29.

27
2Glenn V. Fuguitt, "County Seat Status as a Factor
in Small Town Growth and Decline," Social Forces, XLIII
(December, 1965), 245-251.

2Arthur I'. Schlesinger, A History of American Life,
Vol. XI, The Rise of the City 1878-1l89 (New York: The Mac-
millan Co., 1949).

29John Friedman and William Alonso (eds.), Regional
Development and Planning (Cambridge: Massachusetts Institute
of Technology Press, 19b4), pp. 1-11.

0Ibid., p. 3.

31T. W. Schultz, Economic Organization of Agri-
culture (New York: McGraw-Hill Co., 1953).

32
3Keith W. Bryant, "Inter-County Variation in
Farmers' Income," Journal of Farm Economics, XLVIII
(August, 1966), 557-577.

33Otis Dudly Duncan, et al., Metropolis and Region
(Baltimore: The John Hopkins Press, 19b0).

4Brian J. Berry, Strategies, MIodels and Economic
Theories of Development in Rural Regions. Agricultural
Economic Report No. 127 (Washington: Economic Research
Service, U.S. Department of Agriculture, 1967).

35
5John H. Thompson, et al., "Toward a Geography of
Economic Health: The Case of New York State," in Regional
Development and Planning, John Friedman and William Alonso
(eds.) (Cambridge: Massachusetts Institute of Technology
Press, 1964), p. 187.

3Brian J. Berry, "Identification of Declining
Regions: An Empirical Study of the Dimensions of Rural
Poverty," in Thoman, R. S. (ed.), Areas cf Economic Stress
in Canada (Kingston: Queens University, 19b5).

37W. H. Bell and D. W. Stevenson, "An Index of Eco-
nomic Health for Ontario Counties and Districts," Ontario
Economic Review, II, No. 5 (1964), 1-7.





26



38
3Gerald Hodge, "Do Villages Grow? Some Perspectives
and Predictions," Rural Sociology, XXX, No. 2 (June, 1966),
183-196.

3Leslie J. King, "Cross Sectional Analysis of
Canadian Urban Dimensions 1951 and 1961," Canadian Geographer,
X, No. 4 (1966), 205-224.













CHAPTER II

STUDY AREA AND METHODOLOGY


The study area is composed of five states, Missis-

sippi, Alabama, Georgia, North and South Carolina. This

region covers an area of 241,968 square miles, and in 1968

had a population of 16,326,746. These states are similar

in their socio-economic-political characteristics (Hart,l

Vance,2 Odum ). All five withdrew with the Confederacy and

after Reconstruction the states voted solidly Democratic

until 1964, with the exception of North Carolina which went

Republican once.

Perhaps due to the apparent homogenity of the South,

few authors have attempted to subclassify it. What region-

alization of the South has taken place is largely based

upon physical divisions such as the Delta, Piedmont,

Appalachia and the Coastal Plain (White and Foscue,4

Griffin, Young, and Chatham5). Bogue and Beale however,

using 64 variables, have broken the South down into four

distinct regions.6 One of which, the Southeast Coastal Plain

and Piedmont, coincides very closely with the area under

study (Figure 1). Five distinct characteristics separate

this region from the others. It is the least urbanized, has

the highest percentage of Negro population, the highest
27
















7 4
~22202

2*2224
422 2,


^ 1.'1 ..* ^ / /- -^

;.. / i i i '.j ..' \ i ^ -

V.^^Ip ^ yd 5 l -''
.9P 2 '2I -; I

<* 2 ^ ? i/ ^ 4'-
Y 2, i-vE */ *1 i,



M~~ l->-<^ ^- i i
2s






9 20 'Z 2 2
^~i 11 in \rr^^
i ^ -- ~i i ,
< II i- ,' i -^ ,- -^ -v i'i




9 4- I i -
pT-^isn^

S~~ ~~~ b ^ '^ ^^-^1
C/ i, 5 j r/ i^' i 11^ ^
r i I: ; ; '** B I j --P-- '
0) 0


20o 22 i
.2 '2 2,
h E ^ k--^^'-- *) 11 i4-
^ h u s^ ~- ^- ^1^^ <^









percentage of the labour force in agriculture, and the

industrial base is weak and highly dependent upon textiles.

These four characteristics are the cause of the fifth, low

median income, as the lowest median family income of any

region in the United States is found here. Furthermore

Bogue and Beale found enough variation within this distinct

region to classify it into 19 subregions, implying that

there may be differential growth patterns.

The choice of this study area was dependent to a

great extent on the previous work done by Bogue and Beale.

Although the two areas are not exactly the same, they may

be considered as one in their characteristics. The reason

for this discrepancy is that state boundaries were maintained

so as to reduce the effect of state governments in the anal-

ysis.



The Setting

Three distinct physiographic features fall within

the study area, the Coastal Plain, Piedmont and Appalachia.

The boundary between the Piedmont and the Coastal Plain is

demarcated by a Fall Line and a change in soils. The Fall

Line is characterized by a zone of waterfalls and rapids in

the streams which imposed a limit on navigation. Cities

such as Raleigh, Columbia, I.'acon, Columbus and Montgomery

are found along this break. In the Carolinas and Georgia

the Fall Line is marked by a long, low narrow belt of sand










hills, the remnants of ancient beaches. The Piedmont soils

are residual, the product of weathering of the underlying

rock over eons of time. The Coastal Plain soils on the

other hand were formed of sedimentary materials when the

plain was submerged.

Both physiographic divisions have a variety of local

topography ranging from level to hilly, with the older Pied-

mont having the greater extremes of relief. A small portion

of the study area falls within Appalachia. This section

extends from central Alabama to the Blue Ridge of North-

western Georgia and North Carolina. In the Blue Ridge the

mountains rise 3,000 to 6,000 feet above sea level. In the

Appalachian Valley of Alabama the relief is much less, varying

from 500 to 1,000 feet.

The economy was and to a large extent still is

centered on agriculture. Three major crops: tobacco, cot-

ton and peanuts provide the majority of the farmers' cash

receipts. Industrial growth has been based upon service

functions, textiles, food processing, furniture and lumber-

ing. In many respects the economy is similar to that of

underdeveloped countries, in that it is dependent on a few

staple agricultural products and has a weak industrial base.


Historical Background

The retardation of this area,as in the rest of the

South, has largely been due to the historical development of








the region. This historical retardation has been well

documented by historians and economists (Eaton, Maddox ).

It was the development of an agrarian culture that led to

the exclusion of the rise of a strong middle class, and

industry. These could have become the impetus for economic

growth, when agriculture was diminishing in importance.

Once behind, the South was in a poor position to compete

with the industrial North, and it became subservient to a

centralized economy based upon market potentials.

The Southeast was settled by two streams of settle-

ment. The first group occupied the tidewater between

Charleston and Savannah. Plantations were established and

at first indentured servants, later slaves, were brought in

as a source of labour. Rice became the dominant crop along

the coast, while indigo was grown on the plantations further

inland. As the tidewater was being settled a different

group of immigrants were coming into the western Carolinas

and Georgia. These were the Scotch-Irish and Germans who

had come south from Pennsylvania. Many of these settled

permanently in the backcountry and became successful farmers,

raising all their requirements. The excess livestock, hogs,

corn were sold or converted to other more valuable by-

products such as liquor or hides. The heavy migration into

the western area led to a more rapid rate of growth than in

the tidewater. Eventually this may have led to the political

control of the southern colonies by the small farmers, had









not the importance of cotton began to assert itself. Cotton

planters began to move into the Piedmont and it became the

major crop. The search for more and better cotton soils led

to the opening up and settlement of Alabama and Mississippi.

The relatively rapid rise in the dominance of cotton resulted

from two major events, the industrial revolution in Europe,

requiring cotton for its textile mills, and the invention of

the cotton gin by Whitney in 1793.

The emergence of King Cotton as the dominant crop,

controlled by a small group of planters, utilizing slave

labor, pre-empted the use of land for other crops. This

small aristocratic group gained and maintained political

control throughout the area. Furthermore this group man-

aged to impose their values upon the entire southern soci-

ety. The use of slave labor was a barrier to the establish-

ment of more productive agricultural practices which would

utilize a higher share of capital as a factor of production.

The social stigma associated with trade and manufacturing

hindered the development of a middle class and a more diver-

sified economy. Perhaps the greatest damage done was the

concept that education was not for the masses, but for the

select few. The education provided the select few, although

of high calibre, was heavily oriented toward the liberal

arts and was lacking in the field of applied technology.








Other important sources of income were lumber and

naval stores, which initially were of prime importance in

North Carolina but rapidly spread throughout the South.

Tobacco became the second major crop when, in 1850, it was

discovered that bright tobacco did well on the poor sandy

soils of northwestern North Carolina. The third major crop

was the growing of peanuts along the Georgia and Alabama

border.

Thus in the period preceding the Civil War the

southern economy was oriented around the export of a few

major staples as sources of revenue with which to pay for

the importation of manufactured goods from abroad and the

North. This raw material orientation meant that the econ-

omy was affected by any severe fluctuations in prices or

tariffs. The tariff of abominations established by the

North in 1828 was a blow to southern commerce.0

The Civil War brought ruin upon the South. The

abolition of slavery removed a source of cheap labour from

the plantations and left many of these large land holdings

idle. A system of sharecropping evolved to fill the labour

shortage. The system was not satisfactory for it did not

encourage any long term agricultural practices by either

the landowner or the sharecropper. In addition, develop-

ment of either industry or a more efficient system of

agriculture was strongly hampered by the shortage of invest-

ment capital. Inflation during the War had left the








Confederate states with a worthless currency. Deflation

after the war left the South with little buying power.

Some of the capital problems were attributable to national

monetary policies which hurt the South. The National Bank

Reforms in 1865 levied a severe tax on state banks, which

predominantly were located in the South. Furthermore in

these bank reforms, the South did not receive its fair
11
allotment of Federal currency.

The establishment of certain industries such as

iron and steel was also hampered by the monopolistic action

of northern firms who established "Pittsburgh plus basing

point" and by this means removed the locational advantage

of Birmingham. The industrial growth that did take place

occurred mainly in the field of food processing, textiles

and cigarette manufacturing. The use of bright tobacco as

a major ingredient in cigarettes gave rise to cities like

Durham and Winston Salem. The textile industry moved here

from the North during the later part of the nineteenth

century. Many of the mills were established in towns along

the Fall Line where there was a source of power. Despite

these attempts at industrialization, the South did not keep

pace with the North. As most poor or stagnating areas it

suffered from selective out-migration, lack of capital and

the establishment of new firms. All of these factors

encouraged the location of industry in the heart of the

more lucrative northern market.








Thus the South entered the twentieth century in a

disadvantaged position. It was an area short of skilled

labour, investment capital, and entrepreneurs, technolog-

ically and productively lagging behind the other sections

of the country.


Recent Trends in the Economy

Despite these handicaps, great strides have been

made in the last 30 years throughout the study area and

the entire South. This has come about due to the mechaniza-

tion of agriculture, and better utilization of other natural

resources. The development of better educational facilities,

bi-racial labour force, more internal savings and capital

investments has encouraged the location of industry, espe-

cially those that benefit by decentralization.12

The trend has been to a diversification of both the

agricultural and industrial segments of the economy. Exam-

ples are the growth of the dairy industry in Mississippi,

the establishment of beef cattle on the Black Prairie of

Alabama and the Piedmont of Georgia, the location of the

broiler industry since the war in northern Alabama and

Georgia. Abandoned cropland, formerly in cotton, has been

reverting or restocked to pine woodland. Forest management

has been initiated and encouraged by the rapid expansion of

the pulp and paper industry. Better conservation practices,

increased use of fertilizers and mechanization of farms










have resulted in higher yields. Prices of important cash

crops such as cotton, tobacco and peanuts have been stabi-

lized by a system of acreage allotments.

Industrial growth has been spurred by the establish-

ment of large important industries. Some major examples

are: the atomic energy plant near Aiken, South Carolina;

the Redstone guided missile project and other space facilities

at Huntsville, Alabama, the large Lockheed aircraft plant in

Marietta, Georgia. Less dramatic has been the location of

electronic firms in the Golden Triangle of North Carolina.

The location by northern companies of branch firms in south-

ern cities has aided in the establishment and diversifica-

tion of industrial growth. Atlanta owes a great deal of

its growth to the establishment of these regional branches.

It has become an assembly center for Ford and General Motors.

Many insurance and airline firms have their regional offices

here. In addition the overall economic growth in the South

has been aided by the establishment of large military bases

and the rapid increase in tourism, both of which have been

attracted by the mild climate.


Outline of the Analytical Model

Briefly the steps undertaken in the study are as

follows: (1) transfer raw data into an I x N matrix (M

variables and N observations), (2) perform principal com-

ponents analysis of the M x N intercorrelation matrix of









the M variables and rotation of the resulting eigenvectors

to a normal varimax position to satisfy the criterion of a

simple structure, (3) compute the factor scores of N obser-

vations on the R rotated factors, (4) regress a dependent

variable D on the factor scores.



The Model

The mathematical model used to determine the under-

lying dimensions of economic health is known as a principal

axis factor analysis solution.13 The choice of the model

is based upon several considerations. The study of the

covariance of economic, social and demographic character-

istics of regions is a problem of a multivariate nature.

Data of this nature are often not-normally distributed and

are highly interrelated. Regression models using a large

number of variables lacking the assumptions of normality

and independence will give a coefficient of determination

which is biased upwards. In addition, the coefficients and

standard errors of the coefficients may be affected. Fac-

tor analysis, however,can be used on data lacking these a

priori assumptions. This technique can simultaneously

manage a large number of variables, compensate for random

error, and disentangle complex interrelations into their

major and distinct regularities.14 The essence of this

technique is to reduce a large number of variables into a

smaller group of distinct orthogonol factors, components,









or dimensions with a minimum loss of information. On the

basis of these factors the unique variations of a domain can

be discerned. A further step is the obtaining of factor

scores which is the location in space of each observation

on the particular dimensions. These factor scores can be

used as independent variables regressed on any desired
15
dependent variable. This method is suggested by Scott.5

Scott contends that when the assumptions of independence

and normality are not known, this regression technique is

more accurate than the classical least squares technique.16

The essential features of factor analysis are as

follows:17 from a rectangular X matrix of raw data it is

possible to compute a R matrix of simple correlation coef-

ficients.

xll x12 ... xln
x21 x22 ... x2n
xml xm2 ... xmn

XX' = R

rll r12 ... rln
R r21 r22 ... r2n

rn1 rn2 ... rnn

The problem in factor analysis is to find a matrix A such

that R = AA'. This is done by solving for, using an iterative

process, the eigenvalue or characteristic roots L, where L

is the sum of the squared factor loadings of the N variables

on a dimension. The factor loadings are the correlations









between the variables and the dimensions. One can solve

for as many characteristic roots as there are variables;

however, one generally only solves for the most important

roots, that is for the ones that account for the majority

of the variance. The equation for solving L is (R LI) =0

where I is the inverse of R. After finding the eigenvalue

L one can solve for the eigenvector V which is the column

vector of factor loadings, by the following equation (R =

LI) V = 0. Repetition of this procedure yields the factor

matrix A.


all a12 ... aml
a21 a22 ... am2 r is the number of character-
A .. ... istic roots solved for.
aml ... ... amr

Dividing the characteristic roots found by the total com-

munality (communality is the sum of independent common

variances of the components on each variable) yields the

variance accounted for by each dimension.

It is often of interest to find the factor scores,

that is the ranking of each observation on each of the

derived components. The matrix of factor scores is given

by the following equation F- XAL'. In addition the factors

are rotated so as to comply with Thurston's concept of sim-

ple structure, such that each of the original variables

relates highly with one and only one of the new dimensions.l8

This assures that the first factor arising from the analysis









accounts for the maximum variance and each successive factor

accounts for decreased proportions of the explained variance.

The factor scores have further application in a

regression model to explain the per cent of variation of

a dependent variable which is accounted for by the orthogonal

dimensions. This procedure produces an index which has the

largest possible variance of any linear combination. A

lower coefficient of determination is obtained in compari-

son with the least squares method. Factor regression

accounts for errors in the variable high intercorrelations

and gives regression coefficients which are more accurately

representative of economic theory.19



Variables Used

The variables used for the two time periods were

chosen to represent social, economic, demographic and

locational characteristics of the study area. In all, 26

indices were used. They were obtained from the censuses

of population, agriculture, and the city and county data

book for the respective time periods. Three locational

indices were calculated for the study. The following list

is provided to show the nature of the variables.








Table 4

Variables Used


1 Total population
2 Per cent of the population urban
3 Per cent of the population nonwhite
4 Per cent of the population 65 years old or older
5 Per cent of the population 25 years old or over who
completed less than five years of school
6 Per cent of the population 25 years old or over who
completed high school
7 Per cent of housing units sound
8 Median value of owner-occupied dwelling units
9 Median gross rent
10 Value added by manufacturing
11 Total capital expenditures by manufacturing firms
12 Total retail sales
13 Total wholesale sales
14 Per cent of the population under five years of age
15 Per cent of the labour force unemployed
16 Per cent of the farms operated by tenants
17 Average value of land and buildings per farm
18 Total value of farm products produced (sold)
19 Number of commercial farms earning less than $2,500
annually
20 Number of farmers whose off-farm income is greater
than farm income
21 Per cent of families earning between $0 $999 annually
22 Per cent of families earning between $1,000 $1,900
annually
23 Per cent of families earning between $2,000 $2,999
annually
24 Distance in miles from the center of a county to the
nearest city with a population of at least 50,000
25 Distance in miles from the center of a county to the
nearest city with a population of at least 500,000
26 Distance in miles from the center of a county to New
York City


Through these variables an attempt was made to strike

a balance between the rural and urban segments of the study

area. This choice also took into account that the majority

of these indices have been found to be significant by other

researchers, in explaining the variation in economic health.









The major problem with the data is that the areal units

(counties) vary immensely in size. Large counties having

extensive variation within them may not accurately reflect

this in the data. In spite of this inherent difficulty,

county data are at present the best available for such an

analysis.



Summary

This chapter is divided into two sections. The

first describes the study area and the historical basis

for its poverty. This region was selected for analysis

as it is representative of the southeastern United States.

Furthermore, it appears to have had a substantial number

of economic changes within it in the last few decades, and

thus would be an ideal testing area for the analysis of

spatial and temporal patterns of economic health.

Factor analysis was chosen as an appropriate model

because it can handle nonnormal and highly intercorrelated

data. In addition it opens up other avenues of research

beyond the location and identification of the underlying

dimensions.












REFERENCES


John Fraser Hart, The Southeastern United States
(Princeton: D. Van Nostrand Co., Inc., 1967).

Rupert B. Vance, Human Geography of the South
(Chapel Hill: The University of North Carolina Press, 1935).

3Howard Wd. Odum, Southern Regions of the United States
(Chapel Hill: The University of North Carolina Press, 1936).

4G. Langdon White and Edwin J. Foscue, Regional Geog-
raphy of Anglo-America, 2nd edition (Englewood Cliffs:
Prentice-Hall, 1956).

5Paul F. Griffin, Robert N. Young and Ronald L. Chatham,
Anglo-America A Regional Geography of the United States and
Canada (San Francisco: Fearon Publishers Inc., 1962).

Donald J. Bogue and Calvin L. Beale, Economic Areas
of the United States (New York: The Free Press of Glencoe,
Inc., 1961), p. ALII.

7Ibid., pp. 269-270.

Clement Eatoh, A History of the Old South (New York:
The I.acmillan Co., 1965).

James G. MIaddox, et al., The Advancing South: Man-
power Proseccts and Problems (New York: The Twentieth-
Century iFund, 1967).

10
lWilliam E. Laird and James R. Rinehart, "Exogeneous
Check on Southern Economic Development," South Atlantic
.Quarterl, Vol. LXV, No. 4 (1966), pp. 491-5-08

Ibid.

12B. U. Ratchford, "Economic Development in the South,"
South Atlantic Quarterly, Vol. LXIV, No. 4 (1965), pp. 496-
505.
43 "





44



13The type of factor analysis used in this study is
the version known as principal component analysis.

14R. J. Rummel, "Understanding Factor Analysis,"
Conflict Resolution, Vol. XI, No. 4 (1967), p. 444.

15John T. Scott, Jr., "Factor Analysis and Regres-
sion," Econometrica, Vol. 34, No. 3 (July, 1967), pp. 552-
562.

16Ibid.

17For a more detailed explanation of the model see:
H. H. Harmon, Modern Factor Analysis (Chicago: University
of Chicago Press, -TT)- T .; i. G Kndall, A Course in Multi-
variate Analysic (London: Charles Griffin, 957), pp. 10-
36; R. J. Rummel, "Understanding Factor Analysis, Conflict
Resolution, Vol. XI, No. 4 (1967), pp. 444-480; Mary Megee,
"On Economic Growth and the Factor Analysis Method," South-
ern Economic Journal, XXXI (July, 1964), 215-228.

18Qazi Ahmad, Indian Cities: Characteristics and
Correlates (Chicago: University of Chicago Press, 1965),
p. 25.


19Scott, op. cit., p. 552,













CHAPTER III

THE DIMENSIONS OF ECONOMIC HEALTH IN THE
SOUTHEASTERN UNITED STATES (1950-1960)


The analysis as reported in this chapter as well

as in the subsequent chapter, was undertaken at the Univer-

sity of Florida's computing center. The original data, for

the two time periods, were transformed into symmetric 26 by

26 correlation matrices. These matrices were then subjected

to a principal axis solution which yielded six eigenvectors

for the 1950 data and seven eigenvectors for the 1960 data,

based upon the criteria oflAI. The components were then

rotated to a normal varimax position. The underlying dimen-

sions obtained in this manner were used to identify and to

help interpret the areal variation in economic health.

The basic dimensions of economic health are identi-

fied by the analysis of the derived factor loadings and fac-

tor scores. Factor loadings are the correlations of the

original variables with the newly derived dimensions. If

each of the original variables were plotted in n-dimensional

space (n representing the number of derived factors), the

coordinates of the variables would represent the correlations

or loadings on each principal axis (dimension) by the indi-

vidual variables. These factor loadings range from 1.0,

indicating a high positive correlation with a given dimension
45









(which must be interpreted), to a -1.0, indicating a high

negative correlation. Variables with high positive loadings

on a factor are expected to be most characteristic of the

quality represented in the factor, while high negative load-

ings on the same factor should indicate a contrasting qual-

ity. Factor scores on the other hand measure how closely

each observation (county) ranks on each of the individual

dimensions. For example, a high positive factor score for

a county on a given component can be interpreted to mean

that the county is highly representative of that component.

With this in mind, it becomes possible to identify, and

interpret the areal variation of economic health. In order

to facilitate interpretations, of the dimensions only high

loadings on each of the components are shown, as are only

highly negative and positive factor scores in Tables 7-18,

21-34. The factor scores will also be employed later in

this chapter in a stepwise regression as a means of testing

how well the derived dimensions explain the variation in

poverty.


1950 Factor Analysis

Table 5 summarizes the proportion of the total inter-

county variance in 1950 accounted for by each of the six

derived factors. The six rotated factors together explain

70.5 per cent of the total variance of the 26 variables.

Thus these six components are considered representative of

the variables that they replaced.









Table 5

Percentage of Total Variance Explained by Each
Dimension in 1950


Dimension

I
II
III
IV
V
VI


Eigenvalue

8.71
3.09
2.14
1.91
1.45
1.02


Total


Per cent of total variance

33.5
11.9
8.2
7.4
5.6
3.9

70.5


Another indication of the representation of the

original variables by the ncw dimensions is given by the

communalities. These values are the total sum of squares

of the correlations of a particular variable with all six

components. As would be expected, all variables are not

equally represented (Table 6).


Table 6

Percentage of Variance (Communality) of each of the 26
Variables Accounted for by All Six Components in 1950

Variables Communality in Percentage

1 96.6
2 74.5
3 79.1
4 19.3
5 83.0
6 82.2
7 87.1
8 53.4
9 75.5
10 84.2
11 59.6
12 77.2
13 69.4











Variables

14
15
16
17
18
19
20
21
22
23
24
25
26


Table 6--Continued

Communality in Percentage

78.1
10.9
80.2
67.9
84.7
87.9
69.6
93.2
95.5
46.3
39.4
63.6
75.1


Two variables, per cent of the population 65 years of age

and older and per cent unemployed, are poorly accounted for

as only 19.3 and 10.9 per cent of their total sums of squares

are explained. This implies that these variables were not

significant in the analysis. On the other hand the remain-

ing variables are summarized quite adequately by the new

dimensions. Thus the variations of the selected indices

are effectively captured by this statistical technique, a

desirable trait in such an analysis.


The 1950 Dimensions

Factor I (Urban Industrial Counties), the most

significant dimension, contributing 33.5 per cent of the

total variance, isolates the urban ghettos of the Southeast.

This component is associated with poverty found in some of

the urban industrial complexes within this region. Table 7

shows the correlations between this dimension and urban









functions such as retail and wholesale sales. High loadings

are also associated with total population, value added by

manufacturing, capital expenditures, poverty and youthful-

ness of the population. Counties which rank high on this

dimension are those which contain large rapicdy growing

regional centers such as Atlanta, Birmingham, LMobile, and

Charlotte (Table 8 and Figure 2). These centers, due to

their prosperity and rapid rate of economic growth, attract

a large number of immigrants from the hin'terlan i, many of

whom are unskilled and consequently are employed in low

paying industries or seasonal work.


Dime vision I


Table 7

Urban Industrial Counties in 1950


Prima y V riables F

Total population
Per cent of population urban
Value added by manufacturing
Total Capital expenditures
Retail sales
Wholesale sales
Number of persons under five year:, of age
Per cent of families earning between
0--999
Per cent of families earning between
$1,000- i,999
Per cent of families earning between
$2,000-$2,999


actor Loadings

.92
.44
,88
.73
.85
.82
.83

.76

.90


.65









Table 8

Counties with Extreme Factor Scores on Dimension I
in 1950


High Positive

Jefferson, Ala.
Fulton, Ga.
Mobile, Ala.
Mecklenburg, N. C.
Forsythe, N. C.
Guilford, N. C.
Chatham, Ga.
Greenville, S. C.
Charlestovwn, S. C.
Spartanburg, S. C.
Gaston, N. C.
Fayette, Ga..
Bibb, Ga.
Richland, S. C.
Montgomery, Ala.
Buncombe, N. C.
Hinds, Miss.
Wake, N. C.
Durham, N. C.
Richmond, Ga.


12.32
11.84
3.94
3.66
3.51
3.10
2.78
2.52
2.33
2.29
1.97
1.59
1.59
1.50
1.48
1.48
1.40
1.38
1.35
1.30


High Negative

Chattahooche, Ga.
Fannin, Ga.
Houston, Ga.
Carteret, N. C.
Dare, N. C.
Onslow, N. C.
Lafayette, Miss.
Ohowan, N. C.
Stone, Miss.
Orange, N. C.


The second dimension which explains 13.3 per cent

of the total variance is a rural poverty indicator. This

component is related mainly to types of farming and sources

of farm income. High positive loadings are found on total

value of farm products produced and number of commercial

farms earning less than $2,500 annually, the tenancy rate,

and number of farmers whose off-farm income is greater than

farm income (Table 9). This dimension isolates the poor

tenant and the small individual farmer who, because of lack

of finances, education, or the small scale of his holding,

has not been able to make a financial success of his


-2.18
-1.34
-1.03
- .82
- .81
- .78
- .76
- .71
- .71
- .70













+ +


+
0


cn) CM

ED


'
':'''
'''''
''''
'''


C


C

o

L.
0


0) 0


I iv









enterprise. Factor scores fail to reveal any easily defined

grouping of counties. Instead they suggest that this type

of poverty is quite widespread, especially in areas outside

the dominance of any large metropolitan center (Table 10).

There does appear, however, to be a concentration of poor

counties in the old cotton belt of the Southeast, suggesting

that these rural counties have not fully recovered from the

negative effects of the sharecropping system of agriculture

(Figure 3).


Dimension II


Table 9

Tenant and Small Farm Poverty in 1950


Primary Variables Factor Loadings

Per cent of farmers that are tenants .49
Total value of farm products produced (sold) .83
Number of commercial farms earning less than
$2,500 annually .93
Number of farmers \h'ose off-farm income is
greater than their farm income .39
Per cent of families earning between
$0-$999 .55


Table 10

Counties with Extreme Factor Scores on Dimension II
in 1950


High Positive

Bolivar, Miss.
Sunflower, MIiss.
Robeson, N. C.
Leflore, Miss.
Washington, Miss.
Coahoma, Miss.
Cullman, Ala.
De Kalb, Ala.
Johnston, N. C.


5.38
5.30
3.61
3.38
3.28
3.15
3.09
3.08
3.00


High Negative

Evans, Ga.
Glynn, Ga.
Camden, Ga.
I:uscogee, Ga.
McIntosh, Ga,
Chatham, Ga.
Liberty, Ga.
Fulton, Ga.
Charlton, Gao


-2.43
-1.93
-1.89
-1.64
-1.60
-1.59
-1.52
-1.51
-1.43








Table l0--Continued

High Positive High Negative

Sampson, N. C. 2.76 Long, Ga. -1.41
Horry, S. C. 2.74 Bryan, Ga. -1.39
Orangeburg, S. C. 2.40 Camden, N. C. -1.37
Tallahatchie, Miss. 2.36 Currituck, N. C. -1.37
Columbus, N. C. 2.36 Doughert,, Ga. -1.33
Florence, S. C. 2.34 Clink, Ga -1.32
Williamsburg, S. C. 2.22 Turner, Ga. -1.32
Marshall, Ala. 2.15 Peach, Ga. -1.31
Pitt, N. C. 2.15 Dare, N. C. -1.28
Quitman, Miss. 2.09 Hancock, Miss. -1.28
Cleveland, N. C. 2.05 Jones, Ga. -1.26
Tunica, Mirs. 2.02 Wayne, Ga. -1.26


The third most important factor (Wealthy Urbanized

Counties) identifies a pattern of urban wealth. It is

associated with variables, such as per cent of the popula-

tion urban, median value of owner-occupied homes, and per

cent of the population that completed high school (Table 11).

The economic influence of Atlanta is clearly seen in De Kalb,

Clayton and Cobb which rank high on this dimension. These

counties are to an extent dormitory suburbs of Atlanta and

as such contain a high proportion of residents with above

average incomes. Many of the other counties listed in Table

12 owe their prosperity to the location of state institutions,

an example is Orangeburg, South Carolina. Other counties

that fall into this category are Hinds, Missi'sippi; Mont-

gomery, Alabama and Richland, South Carolira. The influence

of state spcdii g is felf-evident in these counties as the

respective state capitals are located there. Federal pay-

rolls are another important source of prosperity, for example










CM m +
+ + A



- I,
0
o 0
*- O


L~jiwI I
0n


0


44 z


^


O
d Ir,
6 cs\
L4









Houston, Chattahooche and Muscogee counties in Georgia, the

sites of large military bases. Prosperity in these urban-

ized counties can be traced to concentrations of highly edu-

cated people working in professions that pay above average

salaries.

Table 11

Dimension III Wealthy Urbanized Counties in 1950


Primary Variables


Factor Loadings


Per cent of the population urban
Per cent of the population that completed
less than five years of school
Median value of owner-occupied homes
Average value of farms (land and buildings)
Per cent of the population that completed
high school


.73

-.43
.70
.55

.82


Table 12
Counties with Extreme Factor Scores on Dimension III
in 1950


High Positive

Chattahooche, Ga.
De Kalb, Ga.
Harrison, Miss.
Glynn, Ga.
Houston, Ga.
New Hanover, N. C.
Dougherty, Ga.
Muscogee, Ga.
Forrest, Miss.
Clarke, Ga.
Adams, Miss.
Hinds, Miss.
Montgomery, Ala.
Richland, S. C.
Jackson, Miss.
Orange, N. C.
Pasquotank, N. C.
Chatham, Ga.


5.54
4.39
3.72
3.71
3.33
3.08
3.06
2.98
2.73
2.71
2.70
2.62
2.44
2.35
2,31
2.24
2.16
2.11


High Negative

Fayette, Ga.
Jefferson, Ala.
Dawson, Ga.
Jasper, S. C.
Heard, Ga.
Union, Ga.
Stokes, N. C.
Forsythe, Ga.
Yancey, N. C.
Berkeley, S. C.
Calhoun, S. C.
Banks, Ga.
Echols, Ga.
Lowndes, Ala.
Fulton, Ga.
Gilmer, Ga.
Greene, Ala.
Clebourne, Ala.


-2.25
-2.16
-1.85
-1.53
-1.42
-1.40
-1.35
-1.35
-1.31
-1.31
-1.31
-1.30
-1.30
-1.29
-1.28
-1.27
-1.26
-1.26










Dimension IV (Rural White Counties) shows the con-

trast in levels of living between primarily white and black

counties. Variables that correlate highly negative with

this component are: per cent of the population nonwhite,

per cent completed less than five years of school and per

cent of farmers that are tenants (Table 13). A positive

correlation is found between this dimension and number of

farmers whose income from off-farm employment is greater

than faii income. High positive factor scores are asso-

ciated with the Appalachian counties of Alabama, Georgia,

North and South Carolina, with the heaviest concentration

in western North Carolina (Table 14). The predominantly

Negro counties rank low on this factor (Table 14). These

Negro counties are located in a belt which runs from east-

ern North Carolina to the Mississippi Delta. The largest

concentrations are found in the Delta, southwestern Georgia

and the tobacco area of eastern North Carolina. The major-

ity of the Negroes in this belt are the descendants of the

former slaves who worked on the cotton and tobacco planta-

tions. Thus this area of colored poverty has its roots in

the plantation system of land tenure, whose remnants still

remain. Although both groups of counties are poor, the

Negro counties are more so, indicating the traditional

disparity between the black and the white segments of our

society.









Table 13


Dimension IV


Rural White Counties in 1950


Primary Variables


Factor Loadings


Per cent of population nonwhite
Per cent of farmers that are tenants
Per cent of population that completed less
than five years of school
Number of farmers whose income from off-farm
employment is greater than farm income


Table 14


Counties with Extreme Factor
in 1950


High Positive


Wilkes, N. C.
Cullman, Ala.
Walker, Ala.
Ashe, N. C.
Fannin, Ga.
Madison, K. C.
Caldwell, N. C.
Randolph, N. C.
Greenvill S. C.
De Kalb, Ala.
Watauga, N. ,.
Buncombe, N. C.
Catawaba, N. C.
Jackson, N. C.
Cherokee, N. C.
Spartanburg, S. C.
Davidson, N. C.
Burke, N. C.
Mitchell, N. C.


2.30
2.25
2.13
2.06
2.01
1.99
1.89
1.89
1.82
1.80
1.77
1.66
1.66
1.65
1.63
1.59
1.58
1.55
1.55


The last two components


Scores on Dimension IV


High Negative


Lee, Ga.
Union, Miss.
Calhoun, Ga.
Bur: e, Ga.
Turner, Ga.
Coahoma, Miss.
Terrel, Ga.
Peach, Ga.
Lowndes, Ala.
Stewart, Miss.
Issaquena, Miss.
Greene, Ala.
Baker, Ga.
Washington, Miss.
Bullock, Ala.
Edgecombc, N. C.
Quitman, Ga.
Webster, Ga.
Adams, Miss.


-2.56
-2.54
-2.20
-2.02
-2.01
-2.01
-1.92
-1.84
-1.81
-1.79
-1.78
-1.77
-1.73
-1.713
-1.71
-1.70
-1.66
-1.63
-1.62


together contribute 10.1 per


cent to the total explained variation. These two dimensions

are difficult to identify. They appear to be indicators of

urban influence upon their hinterlands (Tables 15 through

18). Factor V indicates that a close proximity to a large


-.85
-.72

-.78

.60









city may not always be beneficial. For a small urban center

generally is at a disadvantage competing with a large

regional center for a regional market, especially if the

regional center contains a large proportion of the regional

market. A large city in this instance may cause services

and industries to gravitate to it. This may be true of the

southwestern Mississippi counties that rank highly negative

on Dimension V (Table 16), These counties are depressed

economically not only because of their poor agricultural

base but also due to the lack of thriving urban centers.

In this case the presence of a large urban center is not

beneficial, rather it appears that New Orleans is growing

at the expense of this area. This in turn suggests that New

Orleans is losing its national significance to other more

rapidly growing centers such as Atlanta. This is in con-

trast to the more prosperous counties of North Carolina,

whose produce is related more to national markets. Thus

this dimension appears to reflect the importance of national

markets to economic growth.

The last dimension is another locational index as it

has negative loadings with all the distance variables (Table

17). The positive loading with median value of owner-occupied

dwellings is an indicator of wealth. This factor reflects

a locational advantage in regard to regional and local mar-

kets, A glance at the factor scores will verify this loca-

tional advantage. For example. counties with high positive








scores arelfound to be centered near large metropolitan cen-

ters (Table 18). The economic effect of Atlanta is clearly

seen on the Georgia counties. The counties with high nega-

tive factor scores tend to be more isolated and economically

depressed.

Table 15

Dimension V Locational Advantage in Relation to
National Market (New York) in 1950


Primary Variables


Factor Loadings


Per cent of population 65 years of
age or older -.31
Median gross rent .31
Avel-ge value of farm (land and buildingr) .48
Total value of farm products produced .27
Number of farmers hose incon from off-
farm employment is greater than farm income -.22
Distance in miles fron- the center of a county
to a city of at lear b 500,000 .66
Distance in miles from the center of a county
to New York -.81

Table 16

Counties with Extreme Factor Scores on Dimension V
in 1950


High Positive

Horry, S. C.
Yadkin, N. C.
Glynn, Ga.
Alleghany, N. C.
Robeson, N. C.
Surry, N. C.
Columbus, N. C.
Stokes, N. C.
Davie, N. C.
Harnett, N.C.
Forsythe, N. C.
Pitt, N. C.
Florence, S. C.
Cabarrus, N. C.
Dillon, S. C.
Camden, Ga.
Nash, N. C.


2.22
2.18
2.05
2.05
2.00
1.97
1.95
1.94
1.81
1.79
1.79
1.78
1.74
1.72
1.68
1.68
1.63


High Negative

Evans, Ga.
Pike, Miss.
Hinds, Miss.
Harrison, Miss.
Lincoln, Miss.
Copiah, Miss.
Amite, Miss.
Wilkinson, Miss.
Walthall, Miss.
Marion, Miss.
Forrest, Miss.
Jefferson, Miss.
Lafayette, Miss.
Marshall, Miss.
Tuscaloosa, Ala.
Jefferson Davis,
Dallas, Ala.


-2.55
--2.08
-2.07
-1.99
-1.97
-1.89
-1.80
-1.74
-1.73
-1.73
-1.71
-1.68
-1.68
-1.61
-1.61
Misse-l.56
-1.56










Table 17

Dimension VI Locational Advantage in Relation to
Local Regional and National Markets in 1950


Primary Variables


Factor Loadings


Median value of owner-occupied
dwelling units
Median gross rent
Distance in miles from the eel'ter of a
county to a city of at least 50,000
Distance in miles from the center of a
county to a city of at least 500,000
Distance in miles from the center of a
county to New York


Table 18


Counties with Extreme Factor
in 1950


High Positive

Fannin, Ga.
Fayette, Ga.
De Kalb, Ga.
Edgecombe, N. C.
Clayton, Ga.
Wilson, N. C.
Turner, Ga.
Hertford, N. C.
Evans, Ga.
Tunica, Miss.
Nash, N. C.
Wake, N. C.
Greene, N. C.
De Soto, Miss.
Henry, Ga.


12.76
2.90
2.15
2.02
1.95
1.86
1.86
1.81
1.76
1.75
1.69
1.66
1.63
1.55
1.52


Scores on Dimension VI


High Negative


Lowndes, Ga.
Georgetown, S. C.
Issaquena, Miss.
Wayne, Ga.
Choctow, Ala.
Washington, Ala.
Clarke, MIiss.
New Hanover, N. C
Columbus, N. C.
Marion, S. C.
Louderdale, Miss.
McIntosh, Ga.
Choctow, Miss.
Jasper, Miss.
Geneva, Ala.


-2.51
-1.84
-1.79
-1.76
-1.76
-1.64
-1.62
-1.55
-1.52
-1.46
-1.44
-1.44
-1.43
-1.42
-1.41


1960 Ana ysis

In 31950 seven 'imen3ions (,nrrg d from the factor

analysis solution. T1 so seven components explain 77.9 per

cent of the total variance (Table 19). The higher percentage


.69
.22

-.47

-.43

-.26









of explanation over the 1950 period is due to the additional

dimension and a higher percentage accounted for by the other

dimensions. A check of the communalities reveals that the

higher percentage of explained variation is due to a better

factoring of the variables, including old age and unemploy-

ment, which were poorly accounted for in 1950 (Table 20).


Dimen s

I
I]
II
IV
V
VI
VI

Total


Table 19

Percentage of Total Variance Explained by Each
Dimension in 1960

pion Eigenvalue Per cent of tot,

8.80 33.8
[ 3.44 13.3
I 2.40 9.2
J 1.86 7.2
1.47 5.7
1.13 4.4
I 1.12 4.3

77.9


al variance


Table 20

Percentage of Variance (Communality) of Each of the 26
Variables Accounted for by All Seven Components in 1960

Variables Communality in percentage

1 97.1
2 65.9
3 77.6
4 92.4
5 85.4
6 78.0
7 91.3
8 89.5
9 83.5
10 87.1
11 58.9
12 81.5
13 67.2
14 67.2









Table 20--Continued

Variables Communality in percentage

15 73.1
16 74.8
17 68.6
18 65.1
19 83.9
20 75.5
21 91.6
22 95.4
23 94.2
24 57.2
25 54.2
26 69.2


The 1960 Dimensions

The largest dimension, as in 1950, is labelled

urban industrial counties. It contributes 33.8 per cent of

the total variance. High positive correlations are found

between this dimension and population, value added, capital

expenditures, retail sales, wholesale sales, and poverty

(Table 21). The concentration of the poor in the industrial

area is not unexpected. As employment diminishes in the

rural areas due to increased mechanization and larger land

holding the only promise of jobs lies in the growing indus--

trial and service sectors of the economy. Consequently

there is a large migration from the country to the cities.

At times the flow is too great to be absorbed at once into

the economy, resulting in temporary unemployment. On the

other hand, due to their lack of skills or color, many of

those who are absorbed by the labour market are engaged in

low paying occupations, especially in some branches of the

service sector of the economy. Consequently the faster a









city grows and the larger it becomes the larger the concen-

tration of urban poverty. For as a city achieves a higher

rank in the hierarchy of urban systems, it assumes more

service functions, many of which are low paying. In addi-

tion the expansion of urban functions often makes the city

more attractive to future immigrants; this then compounds

the problem of absorbing these migrants into the existing

labour force rapidly. Furthermore, if the supply of labour

greatly exceeds the demand, wages may remain low unless the

workers are unionized or covered by wagc laws. This notion

is suggested by the county factor scores in Table 22. It

is observed that the cities ranking highest on the urban

industrial poverty dimension are also the largest in popula-

tion (Figure 4).

Table 21

Dimension I Urban Industrial Counties in 1960

Primary Variables Factor Loadings

Total population .93
Per cent of the population urban .43
Value added by manufacturing .84
Total capital expenditures .73
Retail sales .80
Wholesale sales .80
Per cent of families earning between
$0 $999 .80
Per cent of families earning between
$1,000 $1,999 .90
Per cent of families earning between
$2,000 $2,999 .91









Table 2

Counties with Extreme Factor
in 1960


High Positive

Fultan, Ga.
Jefferson, Ala.
hecklenburg, N. C.
Mobile, Ala.
Guilford, N. C.
Forsythe, N. C.
Greenville, S. C.
Chatham, Ga.
Charleston, S. C.
Spartanburg, S. C.
Gaston, N. C.
Richland, S. C
Hinds, Iliss.
De Kalb, Ga.
Etowah, Ala.
Jackson, liss.
Buncombe, N. C.
Wake, N. C.


11.99
11.33
4.89
4.13
3.63
3.12
2.90
2.58
2.53
2.48
2.08
1.90
1.82
1.62
1.61
1.57
1.51
1.46


2

Scores on Dimension I


High Negative

Houston, Ga.
Chattahoochee, Ga.
Clayton, Ga.
Catoosa, Ga.
Lafayette, Miss.
Hancock, Miss.
Lumpkin, Ga..
Stone, 1iss.
Columbia, Ca.
Walthall, ILiss.
Candler, Ga.
Liberty, Ga.
Watauga, N. C.
Lamar, Ga.
George, Miss.
Oconee, Ga.
Wayne, Ga.
Polk, N. C.


The second most important dimension contributes 13.3

per cent to the total variance. This component is called

Rural Negro Counties as it relates to the poverty found in

black rural counties. It has high factor loadings on per

cent of the population nonwhite, per cent of the population

who have completed less than five years of school, per cent

of the population under five years of age and per cent of

farmers who are tenants (Table 23). The other factor load-

ings also emphasize the rural nature of this dimension. The

ranking of factor scores clearly indicates the differential

in levels of living between the white and nonwhite (Table

24). High positive scores are associated with predominantly


-1.17
- .99
- .97
- .92
- .86
- .75
- .71
- .70
- .70
- .67
- .67
- .67
- .67
- .66
- .62
- .61
- .60
- .60












*: o

0 4.

00













<0







cZ -

< Z

L,
QO ^^
Z c^_____ W









Negro counties. These counties are located in an arc that

runs from eastern North Carolina through the old cotton belt

to the Mississippi Delta where the heaviest concentration

is found. White counties with low factor scores are gen-

erally found in the Appalachian region of the study area.

The distribution of the counties is similar to that found

in 1950.

The third dimension is also an indicator of rural

poverty but it is not related to color. This component

reflects the residual poverty left over from the sharecrop-

ping era. This group of rural poor is made up of tenant

and small commercial farmers who have not been able to cope

with the changing trends in agricultural production (Table

25). In many instances their deprivation is attributed to

the lack of education, capital and land resources. Although

this form of poverty is widespread (Table 26), Figure 5

indicates that it is most prevalent in the areas where share-

cropping prevailed.


Table 23

Dimension II Rural Negro Counties in 1960

Primary Variables Factor Loadings

Per cent of the population nonwhite .82
Per cent of population that completed
less than five years of school .72
Per cent of population under five years
of age .77
Per cent of farmers who are tenants .74
Average value of farm (land and buildings) .47









Table 23--Continued


Primary Variables


Factor Loadings


Total value of farm products produced
Number of farmers whose off-farm income
is greater than farm income
Per cent of dwelling units sound


.35

-.48
-.45


Table 24

Counties with Extreme Factor Scores on Dimension II
in 1960


Hi gh Positive

Tunica, Miss.
Coahoma, Miss.
Bolivar, Miss.
Washington, Miss.
Sunflower, Miss.
Leflore, Miss.
Sharkey, Miss.
Quitman, Miss.
Humphreys, Miss.
Tallahatchie, Miss.
Lee, Ga.
De Soto, Miss.
Lowndeo, Ala.
Burke, Ga.
Pitt, N. C.


High Negative


3.55
3.15
3.10
3.09
3.02
2.91
2.66
2.44
2.23
2.20
2.04
2.03
2.01
1.94
1.91


Ashe, N. C.
Wautauga, N. C.
Jackson, N. C.
Dare, N. C.
Chattahoochee, Ga.
Avery, N. C.
Macon, N. C.
Tishomingo, Miss.
Buncombe, N. C.
Towns, Ga.
Fannin, Ga.
Marion, Ala.
Alleghany, N. C.
iitchell, N. C.
Darlington, S. C.


Table 25


Dimension III


Tenant and Small Farm Poverty in 1960


Primary Variables


Factor Loadings


Per cent tenants
Average value of farms
Value of farm products produced
Number of commercial farms earning less
than $2,500 annually
Per cent of families earning between '0-$999
Per cent of families earning between $1,000-
$1,999
Number of farmers whose off-farm income is
greater than farm income


.38
-.38
.65

.89
.45

.26


.65


-1.92
-1.92
-1.87
-1.83
-1.81
-1.79
-1.79
-1.78
-1.77
-1.74
-1.73
-1.71
-1.66
-1.59
-1.58






68







rn+
+ + A















I I_
C









I
















Jo
O '























ZI


1 - :-.! -.. .-v..:..'-..









Table 26

Counties with Extreme Factor Scorer. on Dimension III
in 1960


Higp Positive

Robeson, N. C.
Cullman, Ala.
De Kalb, Ala.
Hinds, Miss.
Panola, Miss.
Dallas, Ala.
Williamsburg, S. C.
Sunflower, Miss.
Orangeburg, S. C.
Horry, S. C.
Columbus, N. C.
Sampson, N. C.
Marshall, Miss.
Madison, Miss.
Marshall, Ala.
Anderson, S. C.
Florence, S. C.


3.37
3.32
3.23
2.84
2.62
2.58
2.56
2.51
2.50
2.49
2,48
2.47
2.25
2.19
2.14
2.12
2.08


High Negative

Fulton, Ga.
Camden, Ga.
McIntosh, Ga.
Clinch, Ga.
De Kalb, Ga.
Jackson, Miss.
Charlton, Ga.
Peach, Ga.
Mecklenburg, N. C.
Muscogee, Ga.
Bryan, Ga.
Chatham, Ga.
Currituck, N. C.
Liberty, Ga.
Dougherty, Ga.
Lee, Ga.
Jones, Ga.


The fourth dimension (Locational Advantage in Rela-

tion to National Market) has high correlations with distance

to New York and distance to a city of at least 500,000 per-

sons (Table 27). Counties with extreme negative scores on

this dimension are generally concentrated in southwestern

Mississippi (Table 28). Other low factor scores are found

in central Alabama and northern Mississippi, These factor

scores, although they do to a certain extent reflect poverty,

reflect largely the proximity of these counties to large

regional centers such as Memphis, Birmingham and New Orleans.

It is of interest to note that no large regional grouping

appears around Atlanta. The absence of such a grouping

around Atlanta may be attributed to the closer proximity

of New York. On the other side of the scale are found the


-2.71
-2.53
-2.40
-2.05
-1.86
-1.80
-1.76
-1.74
-1.63
-1.59
-1.58
-1.55
-1.46
-1.42
-1.39
-1.39
-1.37










tobacco-producing and the furniture manufacturing counties

of North and South Carolina. Although these counties are

not in close proximity to regional markets they are close

to the national market. This favourable location is impor-

tant in regard to their specialization in the growing of

tobacco and in the manufacturing of cigarettes and furniture.

Furthermore these commodities have been less susceptible to

mechanization than other enterprises such as cotton growing.

This is in contrast to southwestern Mississippi which suf-

fers from being unable to compete effectively in the agri-

culture market. As a result there has been a large rever-

sion to forestland, especially with the coming in of large

pulp and paper companies. The surplus of labour resulting

from the dwindling agricultural base has resulted in a high

unemployment rate and out-migration.


Table 27

Dimension IV Locational Advantage in Relation to
National Market (New York) in 1960

Primary Variables Factor Loadings
Distance in miles to a city of at
least 500,000 persons .72
Distance in miles to New York -.78
Per cent of population 65 years of age
or over -.19
Median gross rent .29
Total value of farm products produced .21
Per cent of population nonwhite -.14
Per cent of population unemployed -.17
Distance in miles from center of county
to nearest city of at least 50,000 .22








Table 28

Counties with Extreme Factor Scores on Dimension IV
in 1960

High Positive Loadings Hijh Negative Loadings

Gilmer, Ga. 4.61 Pearl River, Miss.
Horry, S. C. 2.52 Harrison, Miss.
Florence, S. C. 2.43 Mobile, Ala.
Columbus, S. C. 2.20 Pike, Miss.
Dillon, S. C. 2.19 Hancock, Miss.
Robeson, N. C. 2.13 Copiah, Miss.
Darlington, S. C. 2.10 Tuscaloosa, Ala.
Alleghany, N. C. 2.10 Lincoln, Miss.
Marlsboro, N. C. 2.02 Hinds, Miss.
Marion, S. C. 1.87 Amite, Miss.
Yadkin, N. C. 1.85 Marion, Miss.
Surry, N. C. 1.83 Wilkinson, Miss.
Scotland, N. C. 1.81 Lawrence, Miss.
Alexander, N. C. 1.79 Jefferson Davis, Miss.
Carteret, N. C. 1.71 Walker, Ala.
New Hanover, N. C. 1.71 Adams, Miss.
Wilkes, N. C. 1.71 George, Miss.


-2.42
-2.08
-2.08
-2.04
-2.03
-1.93
-1.92
-1.87
-1.79
-1.78
-1.76
-1.73
-1.71
-1.70
-1.65
-1.62
-1.61


Dimension V labelled wealthy urbanized counties,

reflects urban prosperity. It has high factor loadings on

per cent of the population urban, per cent of population

completed high school, per cent of population with sound

housing, median value of owner-occupied homes and median

gross rent (Table 29). Counties which rank high on this

component are prosperous small cities such as Macon, Albany,

and Columbus in Georgia, Huntsville, Alabama and Wake Forest

in North Carolina (Table 30). Other counties that have high

positive scores are Cobb, Clayton and De Kalb in Georgia.

These counties contain some of the more prosperous suburbs

of Atlanta. Negative factor scores are associated with

counties both rural and urban, that have a high percentage

of low income groups (Table 30).








Table 29

Dimension V Wealthy Urbanized Counties in 1960


Primary Variables


Factor Loadings


Per cent of population urban
Per cent of population completed less
than five years of school
Per cent of population completed high
school
Per cent of population with sound housing
Median value of owner-occupied homes
Median gross rent
Retail sales
Average value of farm (Land and buildings)


.67

-.52

.80
.80
.70
.75
.39
.48


Table 30

Counties with Extreme Factor Scores on Dimension V
in 1960


Hig-h Positive Loadings


High Negative Loadings


De Kalb, Ga.
Houston, Ga.
Clayton, Ga.
Dougherty, Ga.
Harrison, Miss.
Cobb, Ga.
Muscogee, Ga.
Nadison, Ala.
Onslow, N. C.
Choctow, Ala.
Clarke, Ga.
Orange, N. C.
Montgomery, Alao
Wake, N. C.
Cumberland, N. C.
Camden, Ga.


4.12
3.69
3.52
3.31
3.10
3.08
3.05
2.91
2.84
2.69
2.60
2.35
2.31
2.23
2.15
2.01


Webster, Ga.
Jefferson, Ala.
Fulton, Ga.
Glascock, Ga.
Quitman, Ga.
Hancock, Ga.
Talbot, Ga.
Issaquena, Miss.
Stewart, Ga.
Banks, Ga.
Wheeler, Ga.


The last two dimensions like those in 1950 are dif-


ficult to identify.


Dimension VI appears to be another


locational index. This component is related to national, re-

gional and loca] market potentials. Table 32 suggests that


-2.45
-2.43
-2.43
-1.82
-1.74
-1.67
-1.58
-1.58
-1.54
-1.50
-1.50









counties which rank high on this factor are located close

to urban centers. They tend to have lower unemployment

rates and a younger population than counties which have a

less favourable location.

The last component is difficult to name. Table 33

shows that the salient variables correlated with this factor

are old age and median value of owner-occupied dwelling units,

and value added by manufacturing. This dimension may isolate

areas of declining prosperity, suggesting the out-migration

of the younger and better educated labour force. Although

Table 34 is not conclusive, it implies that out-migration

may be the key to the explanation of this dimension.



Table 31

Dimension VI Locational Advantage in Relation to
National, Regional and Local Markets in
1960

Primary Variables Factor Loadings

Distance in miles from the center of the
county to a city of at least 50,000 -.60
Per cent of the labour force unemployed -.83
Distance in miles from the center of the
county to a city of at least 500,000 -.10
Dif.tance in miles from the center of the
county to New York -.19
Per cent of the population 65 years of
age or older -.10








Table 32

Counties with Extreme Factor Scores on Dimension VI
in 1960


High Positive Loadings


Crawford, Ga.
Edgefield, S. C.
Guilford, N. C.
Baker, Ga.
Hertford, N. C.
De Kalb, Ga.
Wake, N. C.
Fayette, Ga.
Glascock, Ga.
Webster, Ga.
Candler, Ga.
Forsythe, N. C.


2.08
2.06
2.04
1.84
1.81
1.76
1.74
1.73
1.71
1.71
1.63
1.63


High Negative Loadings

Bolivar, Miss.
Washington, Miss.
Pamlico, N. C.
Carteret, N. C.
Clay, N. C.
Hyde, N. C.
Avery, N. C.
Brunswick, N. C.
Rabun, Ga.
Pearl River, Miss.
Jackson, Miss.
Adams, Miss.
Jasper, Miss.
New Hanover, N. C.
Pitt, N. C.


-3.85
-3.28
-3.17
-3.10
-2.88
--2.88
-2.72
-2.56
-2.36
-2.31
-2.12
-2.29
-2.02
-2.00
-2.00


Table 33

Dimension VII Areas of Declining Prosperity
in 1960


Primary Variables

Per cent of the population 65 years
of age or older
Median value of owner-occupied
dwelling units
Median gross rent
Value added by manufacturing


Factor Loadings


-.89

-.54
.30
-.31


Table 34


Counties with Extreme Factor Scores on Dimension VII
in 1960


High Positive Loadings

Berkeley, S. C.
Liberty, Ga.
Hinds, Hiss.
Beaufort, S. C.
Cumberland, N. C.
George, Miss.
Camden, Ga,
Chattahoochee, Ga.
Onslow, N. C.
Jackson, Miss.
Charlton, Ga.


High Negative Loadings


1.77
1.73
1.71
1.70
1.58
1.34
1.27
1.25
1.24
1.16
1.16


Choctaw, Ala.
Forsythe, N. C.
Baldwin, Ga.
Fulton, Ga.
Hyde, N. C.
Alleghany, N. C.


-18.77
-1.38
-1.28
-1.21
-1.04
-1.00








Table 34--Continued

High Positive Loadings High negative Loadings

Houston, Ga. 1.16
Mobile, Ala. 1.09
Charleston, S. C. 1.04
Marengo, Miiss. 1.02
Muscogee, Ga. 1.00
Pearl River, Miss. 1.00


Comparison of the Two Time Periods

The purpose of examining two time periods is to

interpret changes through time, in the overall economic

growth of the southeastern United States, to analyze whe-

ther there has been a change in the importance of the dimen-

sions of economic health, and whether new factors have arisen

over the decade 1950-1960. Such a comparative study, however,

is limited to a descriptive analysis as tests of significance

between dimensions over two time periods have not yet been

derived.

The 1950 and 1960 factor analyses revealed essentially

the same patterns of economic health, although in 1960 an

additional dimension emerged. This dimension, however, con-

tributed little to the explanation of the total variation.

In spite of this similarity important changes have occurred

in the rank order of the components, indicating that the

dimensions are not stable through time, and that changes have

occurred in the overall factor pattern. Similar results were

noted by King in his study of Canadian cities and by Hoffman

and Romsa2 in their analysis of urban dimensions in the south-

eastern United States.








Although dimension IV in 1950 (Rural White Counties)

and dimension II in 1960 (Rural Negro Counties) are named

differently, they are considered to be similar as both are

indicators of the color problem. They have similar factor

loadings with the exception that the correlation signs are

reversed. Thus one dimension can be considered a reciprocal

of the other.

The changes in rank order are reflected in dimensions

II through V. Negro poverty has moved from fourth in order

of importance in 1950 to second in 1960 and the per cent of

variation contributed has increased from 7.4 to 13.3 per

cent, an increase of 5.9 per cent. This change in the impor-

tance of the Negro dimension implies that the Negroes have

not shared equitably in the economic progress of the South-

east, during this decade. The decrease of 2.7 per cent in

explained variation of component II in 1950, tenant and small

farm poverty, has reduced it to third in order of importance

in 1960. The decrease in this form of rural poverty is in

accord with the mechanization and the consolidation of farm

units that has occurred in the Southeast. In many instances

this process of consolidation and mechanization has resulted

in a surplus labour supply. Agricultural unemployment is

responsible for the increased concentration of Negro poverty

in the Mississippi Delta. The mechanization of cotton farm-

ing in the Delta has greatly reduced the labour requirements

and has either left the Negro unemployed, or has forced the









excess labour supply to migrate elsewhere, particularly to

the North. This similar process has been at work in south-

ern Georgia, southern Alabama and southwestern Mississippi,

where the pulp and paper companies have become established

and have reverted large areas of crop land to pine forest.

The shift of dimension III in 1950 (wealthy urban-

ized counties) to dimension V in 1960 is an indication that

there has been a narrowing of the range in prosperity

between cities. This is interpreted as demonstrating that

urban centers have shared in economic growth more equitably

than rural areas. This is shown by the increase in rural

poverty in certain sections of the study area as expressed

by components II and IV in 1960. In 1950 these components

accounted for 13 per cent of the total variation while in

1960 they accounted for 20.5 per cent.

The locational factor relating to the national

market has moved in this time period from fifth place in

1950 to fourth place in 1960. On the other hand the regional

and local market locational index has remained stable not

only in rank but it also only registered a slight increase

in total explained variation. This may be the result of an

improved transportation network that has made these markets

more acccssibae. The new dimension that emerged in 1960 is

not only difficult to identify but it appears to have con-

tributed little to the overall analysis.









From these dimensions it is clear that the variance

in economic health throughout the study area in 1950 and

1960 is a function of urbanism, color, location, industrial-

ization, and rural poverty. It is observed that the rate

and form of urban growth is most important, for not only do

the urban areas harbour the majority of the region's poverty,

but also influence to some extent the growth rates of their

hinterlands. It is suggested that the increased rural pov-

erty that is arising in certain sections, due to increasing

technology, will become of greater concern unless an outlet

is found for the surplus supply of labour.



Regression Analysis

The major underlying patterns of economic health

factored from the original 26 variables can be tested for

hypotheses by the use of a regression model. As regression

analysis requires normal and independent data, the testing

of the normality of the factor scores and the independence

of the dimensions is of theoretical importance. In addition

it is of interest to ascertain how well these dimensions

explain the variation in poverty throughout the study area.

To test these and additional hypotheses a dependent variable,

median family income, vias chosen as an index of poverty.

This variable was then regressed on the dimensions in a

stepwise regression.









The Hypotheses

Since the dimensions are theoretically representative

of the original data, it is hypothesized that a high coef-

ficient of determination between the dimensions and median

family income will be obtained. It is hypothesized that the

1960 dimensions will result in a higher coefficient of deter-

mination, as they have higher commonalities than the 1950

components. In addition, hypotheses in regard to the expected

correlations between the dimensions and the dependent variable

can be tested. The expected relationships are shown in Table

35.

Table 35

Expected Correlations Between Median Family
Income and the Derived Dimensions
Median Family Income

Urban Industrial Counties
Tenant and Small Farm Poverty
Wealthy Urbanized Counties +*
Rural Negro Counties
Locational Advantage in Regard to
National Market 1
Locational Advantage in Regard to
Regional and Local Markets +
Areas of Declining Prosperity
Rural White Counties .


The analysis verified the normality of the factor

scores and the independence of the dimensions (Tables 36,

37, 39 and 40). The hypothesis regarding the coefficient of

determination was found to be correct (Tables 38 and 41).

However, the correlations between the dependent and the






80



independent variables were found as expected with one

exception, Urban Industrial dimension had a positive cor-

relation instead of a negative correlation as expected

(Tables 37 and 40).

Table 36

Normality of 1950 Factor Scores


Variable


Mean


Median Family Income 1445.38525
Urban Industrial Counties -0.00002
Tenant and Small Farm Poverty 0.00003
Wealthy Urbanized Counties 0.00022
Rural White Counties -0.00013
Locational Advantage in Regard
to.National Market .-0.00000
Locational Advantage in Regard
to National, Local and
Regional Markets -0.0004


Standard Deviation


572.95850
.99995
i99997
.99997
.99997

.99997


.99995


Table 37

1950 Correlation Matrix


Variable


3 4


1 Median Family
Income 1.000 0.365 -0.124 0.623
2 Urban Industrial
Counties 1.000 -0.000 0.000
3 Tenant and Small
Farm Poverty 1.000 0.000
4 Wealthy Urbanized
Counties 1.000
5 Rural White Counties
6 Locational Advantage in Relation
to National Market
7 Locational Advantage in Relation
to National, Regional and Local Markets


0.334 0.301 0.212

-0.000 0.000 0.000

-0.000 -0.000 -0.000


-0.000
1.000


-0.000 -0.000
0,000 0.000

1.000 0.000


1.000







Table 38

1950 Summary Table


Step Variable
Number Entered


Multiple


0.6232
0.7223
0.7959
0.8510
0.8770
0.8857


2 2
R Increase in R


0.3884
0.52].8
0.6334
0.7243
0.7692
0.7845


0.3884
0.1334
0.1116
0.0909
0.0449
0.0153


Table 39

Normality of 1960 Factor Scores


Variable


Mean


Standard Deviaticn


IMedian Family Income 3
Urban Industrial Counties
Tenant and Small Farm Poverty
Wealthy Urbanized Counties
Rural Negro Poverty
Locational Advantage in Relation
to National Market
Locational Advantage in Relation
to National, Regional and
Local Markets
Areas of Declining Prosperity


072,64746
-0.00003
0.00007
0.00018
0.00013

0.00002


-0.00006
-0.00011


986.50854
.99992
.99998
.99995
.99999


.99995


.99996
.99995


Table 40
1960 Correlation Matrix


Variable


4 5 6 7


1 Median Family
Income 1.000 0.349 -0.363 -0.187 0.046 0,695 0.186 -0.042
2 Urban Industrial
Counties 1.000 -0.000 0.000 0.000 -0.000-0.000-0.000
3 Rural Negro
Counties 1.000 0.000 -0.000 -0.000-0.000 -0.000
4 Tenant and Small
Farm Poverty 1.000 -0.000 0.000-0.00 --0.000
5 Locational Advantage in
Relation to National M:arket 1.000 -0.000 0.000 -0.000
6 WealthyUrbanized Counties 1.000 -0,U00 -0.000
7 Locational Advantage in Relation to
National, Regional and Local iMarkets 1.000 0.000
8 Areas of Declining Prosperity 1.000


__ __








Table 41

1960 Summary Table


Step Variable
Number Entered

1 6
2 3
3 2


Multiple


0.6954
0.7844
0.8584
0.8785
0.8979
0.8991
0.9001


0.483
0.615
0.736
0.771
0.806
0.8o8
0.810


2 Increase in R

i6 0.4836
3 0.1316
8 0.1.215
7 0.0349
1 0.0344
3 0.0022
)1 0.0018


Table 42


Comparison of the Beta, Correlation Coefficients
and the Coefficient of Determination Between the
Two Time Periods


Beta


Urban Industrial


1950

R R2


209.26 .365*


Counties
Tenant and Small *
Farm Poverty -70.92 -.124
Wealthy Urbanized .
Counties 357.09 .623
Rural Negro Counties -- --
Locational Advantage
in Relation to the .
National Market 172.72 .301
Locational Advantage
in Relation to National,
Regional and Local *
Markets 121.44 .212
Areas of Declining


Prosperity
Rural White
Counties


191.42 .334


Beta


. 33


343.94


1960

R R2

.349 .121


.015 -184.26 -.187. .034


.388 686.11l
-- -357.91


.090


45.80


.044 183.09

-41.85*


.695 .
-.363


.483
.131


.046 .002


.186

-.042


.034

.001


.111


Significant at .05 level.

The Beta and Correlation Coefficients (Table 42)

indicate that the variance in median family income in both

periods was a function of location, color, industrialization,


_ I~YI~









urbanization and rural poverty. Dimension VII contributed

very little to the analysis. Although the urban industrial

dimension did not correlate negatively as expected, this

was probably due to the large percentage of high incomes

found in these urban industrial counties which tend to inflate

median family incomes, and thus to a degree hide the large

number of poor families in these counties.

There have, however, been changes in the significance

of these dimensions in explaining the differences in median

family income (Table 41). The influence of the locational

dimensions has dropped rapidly from 13.58 per cent of the

explained variation in 1950 to 3.62 per cent in 1960. Almost

all of this loss is accounted for by the dimension represent-

ing location in regard to the national market, as this com-

ponent v.as not significant in the 1960 regression analysis.

This may reflect the overall economic growth that was achieved

in the South during the two time periods, especially in the

cities. In contrast to the decrease in the importance of

location, urban growth played a larger role in 1960. The

dimension representing urban wealth accounted for an addi-

tional 10 per cent in explained variation. On the other hand

rural poverty had a slightly higher negative correlation in

1960 than in 1950, indicating that in general rural areas are

today at a larger disadvantage. This also applies to the

Negro, who has lost ground during the time period studied.

This appears also to be borne out by the beta coefficients, with

the exception of the beta value, associated with location in









relation to the national market, as the beta coefficients

in 1960 are substantially larger than those in 1950. This

is an indication that the income gap between the poor and

the rich has actually widened in this time period,



Summary

The factor and regression analysis shows that the

urban-rural and the Negro-white dichotomy is increasing.

The analysis indicates that economic growth and opportunity

are being centered around urban complexes to the disadvantage

of rural areas. Furthermore, the economic gap between the

Negro and white segments of the population in this region

has not narrowed. These conclusions are in accord with the

theories regarding poverty and regional growth which were

examined in Chapter I.













REFERENCES


Leslie J. King, "Cross Sectional Analysis on
Canadian Urban Dimensions 1951 and 1961." Canadian Geojraapher.
X, No. 4 (1966), 205-224.

2
WV/ayne Hoffman and Gerald Romsa, "Some Temporal Con-
siderations of Basic Urban Dimensions in the Southeast,"
Paper read before the meeting of the Southeastern Division
of the Association of American Geographers, Greenville, North
Carolina, November 26, 1968.














CHAPTER IV

SUMMARY AND CONCLUSIONS


This chapter summarizes the findings of the factor

and regression analysis for two time periods of five south-

eastern states, Alabama, Mississippi, Georgia, South Carolina

and North Carolina. These states were selected as they

constitute a socio-cconomic region and are representative of

the southeastern United States.

The prime objective of the study was to identify the

major dimensions of economic health of the study area for

two time pe-iods 1950 and 1960, and to ascertain if thes-,

dimensions had c' anged throu ;h ti, e. In the analysis, the

area variation in economic )'e~tth was measured by using the

multivariate procedure of factor analysis, as the 26 variables

selected and used as indices of economic health did not meet

the statistical premises of normality and independence.

Furthermore, factor scores derived from factor analysis were

used in interpreting the areal pattern of the dimensions of

economic health., These same factor scores were also used in

a regression equation to explain the variance in median

family income.










1950 Analysis

The 1950 data were grouped into six dimensions which

represented 70.5 per cent of the total variance of the

original 26 variables. Dimension I was interpreted as being

an indicator of industrialization and urbanization. Counties

ranking high on this component contained the largest urban

centers in the study area. The highest factor scores were

associated with the larger and more rapidly growing centers

such as Atlanta, Birmingham and Charlotte.

Factor II was a low income agricultural component.

Variables characterizing this dimension revealed an agricul-

tural structure which was remnant of the sharecropping era.

Many of the farmers, in the counties that were identified by

this component, are not able to compete successfully with

agricultural enterprises that use large inputs of land,

capital and managerial skill as substitutes for labour and

thus are able to operate on a larger scale. Counties fitting

this pattern were generally located in the old cotton belt.

The third dimension revealed a pattern of urban

counties. These counties contained smaller but wealthier

cities tha l the urban nodes represented by dimension I.

These cities contain a high proportion of residents who are

employed in certain high paying segments of the service sec-

tor, such as education and the military. Other counties

which are represented by this dimension contain wealthy

dormitory suburbs.









Factor IV was a component defining the largely

white population of Appalachia and the predominantly Negro

Mississippi Delta. This dimension although reflecting

poverty in both areas, identified the disparity in level of

education attained and land ownership between white and the

nonwhite segments of the population. The last two dimen-

sions were a reflection of the effect of location, in regard

to local, regional and national markets. No definitive pat-

tern arose, although there is some empirical evidence that

poverty in southwestern Mississippi is due to its disadvan-

tageous location.

On the basis of the derived dimensions, areal pat-

terns of economic health were derived. In general a picture

arose of scattered peaks representing counties with a high

proportion of urban residents and or industrial development

that are surrounded by a sea of poverty. In the northern

portion of the study area are the poor whites of Appalachia.

In. the Mississippi delta poverty is largely comprised of

Negro agricultural workers, many of whom are unemployed.

Poverty in southwestern Mississippi appears to be a reflec-

tion of isolation. Throughout the entire area are found the

poor tenant and small farmer.

Factor scores obtained from the analysis were util-

ized as independent variables in a regression model to test

the significance of the dimensions in explaining the varia-

tion in median family income. The regression an:-Jysis did









support the area pattern of economic health as identified

by the six dimensions. The six factor scores obtained from

the factor analysis explained 78.45 per cent of the varia-

tion in median family income. Median income was found to be

positively correlated with urbanization, industrialization,

and the white portion of the population, The three dimensions

representing these indices accounted for 63.34 per cent of the

explained variation. Location and agricultural poverty con-

tributed smaller amounts to the total coefficient of deter-

mination.

The beta coefficients also substantiated the correla-

tion coefficients. Dimension III had a beta value of 357.09

implying that a one unit increase in this dimension would

result in an increase of $357.09 in median family income.

Since this dimension is largely a reflection of education,

it bears out the economic value of public investment in the

schools of this region. A one unit increase in the urban

industrial dimension would result in an increased median

family income of $209.26. Being white or located close to

urban centers resulted also in higher incomes. Agricultural

employment, however, resulted in lower incomes as did a

decrease in the white dimension.


1960 Analysis

The 1960 data were reduced by factor analysis to

seven dimensions, which explained 77.9 per cent of the varia-

tion in economic health. The increase in explained variation









over the 1950 period was due to the derivation of an addi-

tional component and a more complete factoring of the

original variables. Except for dimension VII (areas of

declining prosperity) the components obtained for this time

period were similar to those derived in 1950, although their

order of importance had changed.

Factor I as in 1950 was associated with urban

industrial counties. It had similar factor loadings and

was identified with counties containing the larger urban

centers. The second component pointed out the economic

discrimination against the Negro. This factor differentiates

between Negro poverty, of which a large proportion is con-

centrated in the Mississippi Delta, and the poverty of white

Appalachia. Although the poverty of Appalachia has been

recognized by various levels of government through different

aid programs, little has been done for the Delta area.

Dimension III represents rural poverty found

throughout the Southeast. Rural poverty is found among

both the white and the Negro operators of small land hold-

ings or tenants.

Dimension IV, a locational index, reflects the

effect of proximity to urban markets. The areal pattern

of factor scores although difficult to interpret suggests

that Mississippi and Alabama suffer by not having closer

access to the national market.




University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.7 - mvs