changed. Towards this, social science
research should examine in detail the
The population control programme has dynamic process of adoption of family
become a part of the national life in India. planning methods and then suggest new
In fact, the first Government operated directions. Such an effort may provide
family planning clinic in the world was policy makers and social planners the
established in India in the year 1930. needed mechanisms by which population
Alarmed at the population growth revealed growth may be reduced.
in the Census of 1961, an all-out effort at
population control was adopted as the
official policy of the Government of India
in 1963 (Schramm, 1971:17). Perhaps, very
The governmental programmes advocat-
few countries in the world have as elaborate ing family planning, by and large, rely on
machinery (both official and voluntary) persuasion and propaganda. The funda-
devoted to population control as India has. mental assumption of such an approach is
In spite of all these, Chandrasekhar that human beings are rational and they
when analysing 1971 census data reports can rationally come to a decision regarding
that, the family planning programme has adoption of family planning methods.
not yet made any considerable dent on the
The family planning techniques pro-
growth of population (Census of India, pagated by the Government of India can
1971:36). It is beyond the scope of this be classified under two headings. They are:
paper to examine the manifold ramifica-
(i) terminal methods (vasectomy and
tions of unchecked/uncontrolled population tubectomy), and (ii) conventional con-
growth on the economic and social traceptives (condom, IUCD, jelly, pills,
development — in short, the "quality of etc.) (The latter methods are usually
life" — in the country. It is sufficient to advocated for spacing of children. But
assert that the present rate of growth of they can be, and are, used by a few who
population should be reduced.
do not want any more children). Between
Perhaps, the policies so far adopted by these two types of methods, the policy of
the Government to popularise family the Government is to' give greater
planning need to be modified or radically emphasis to the former than the latter.
* Dr. Victor Jesudason is Research Associate, Council for Social Development, New Delhi.
**The data for this paper were collected under the auspices of a research grant from the
Godrej Trust. Shalini Bhogle collected the data and prepared the initial report. We are
happy to acknowledge the computational facilities of the Computer Centre of the Delhi
University. Our thanks are due to K. Satyanandam for computational assistance and to B.
R. Patil and Prodipto Roy who critically read earlier drafts of this paper. Grateful
acknowledgement is made to a referee of this journal whose critical reading and suggestions
have greatly improved the paper.

A few studies have shown that some
"irrational" elements are involved in the
adoption of terminal methods. We define
The major objective of this paper is to
"irrational" behaviour as one that will not examine the process by which male workers
be adopted or subscribed to by majority in factories adopt conventional contracep-
of members in a society. In a follow-up tives for family planning purposes.
study of those who underwent vasectomy Specifically, the objectives are:
in Tamil Nadu, it is reported that
18 per cent had fewer than three
(1) to examine the role of background
children, one per cent were over fifty
factors like education, income and
age for adoption of family planning
years of age and three had no children
at all (Repetto, 1968). In another study
(2) to examine the role of mass media,
done in Mysore State, 28.5 per cent of
as indexed by frequency of reading
the men vasectomized had fewer than three
newspapers, in the process of
children, 17.6 per cent were fifty or more
adoption of family planning prac-
years of age, 78 had no children at all and
tices; and
23 were not even married (Quoted in
(3) to examine the role of awareness of
Repetto, 1968). Rationally one would not
small family norm, as indexed by
expect these people to undergo sterilization.
respondents' stated ideal family
Unless social science research explicitly
size, in this process.
takes these "irrational" elements into
account, it will not be able to describe or
predict adoption of terminal family
planning methods. On the other hand,
During May 1969-February 1970, from
adoption of conventional contraceptive among the workers who were married and
methods may not have such irrational living with their wives, a stratified random
elements. So, a generalized description of sample of unskilled, semi-skilled and
the process of adoption common for both skilled workers in two factories (namely,
the methods may mask many interesting Godrej Factory, Bombay and Vazir Sultan
and important differences.
Tobacco Factory, Hyderabad) was drawn.
The adoption of family planning (See: Bhogle and Kaur, 1972:6-7 for
methods, perhaps, is not a simplistic further details regarding sampling). The two
process. Simultaneous influence of many factories may be thought of as at two ends of
a continuum regarding skill required and
variables is involved. So, an univariate labour force commitment.
approach — an approach in which only
1 On the assump-
tion that the process of adoption of family
one variable at a time is related to adop-
planning differs from men to women, only
tion of family planning methods, may be male respondents are included in this study.
hopelessly inadequate. Only a multivariate If for a respondent we do not have in-
approach can capture the dynamic formation on any one of the variables that
process involved.
is of interest to us, that respondent is
On the possibility that the process of adoption of family planning practices may differ
between the workers in the two factories, initial analysis was carried out for these two
factories separately. We found that they do not differ and a few differences we noted
were uninterpretable. So we have decided to ignore the differences between the two
factories and treat the sample as if it is from the same population.

excluded for this analysis. Percentage posure to mass communication media is
distributions, means and standard deviations measured by the stated per week frequency
were computed for both the original of newspaper reading of the respondent. It
sample (N=500) and the reduced sample is coded as: does not read — 0; reads less
(N = 449). We find that both the samples than once a week = l; reads at least once
are similar and no detectable bias is a week = 2; reads three or four times a
introduced by eliminating from analysis week — 3; and reads daily = 4. In order to
those respondents with missing information. measure the respondents' attitude towards
family size, they were asked: "In your
opinion, ideally how many children a
couple should have?" The responses are
The major concern of this paper is the coded in number of children one ideally
process by which factory workers come to should have.
adopt conventional family planning practi-
We have six variables: two related to
ces. In order to measure the degree of socio-economic status, one indicator of
adoption of family planning practices demographic characteristic, one mass com-
(which is our major dependent variable) a munication variable, one attitude variable
series of questions relating to awareness and and one adoption of family planning ,
use of contraceptives were asked. The variable.
responses are coded as: never heard about
conventional contraceptives = 0; heard,
but never used conventional contracep-
tives = 1; and using/used conventional
The review of literature given elsewhere
contraceptives = 2.
(Krishnamurthy, 1968; Agarwala, 1962;
Any study in family planning will be Rao, n.d.) shows that many studies have
incomplete if it fails to take into con-
been conducted on adoption of family
sideration certain demographic characteris-
planning practices. We assume that adop-
tics of the population that are of interest. tion of family planning practice is an on-
Since our sample, by definition, excludes going process. Hence, a model that treats
those who were unmarried / widowed, and adoption of family planning as an ongoing
those who did not have their wives living process might facilitate interpretation of
with them, only age of the respondent is the relationship among variables. For this
included in this paper. It is coded in purpose, the technique of path model is
completed years. Among many variables used here. This technique was developed
which measure the location of an individual by the genetist Sewell Wright (1934) and
in the stratification hierarchy, we have introduced into sociology by Duncan (1966).
selected the level of education attained by The technique is explained and discussed
the respondent (coded in years of schooling in several recent papers (Land, 1969;
completed) and average monthly income Heise, 1969) and numerous applications
(coded in rupees). Presumably, the latter have appeared (Sewell and Shah, 1968:
includes salary, wages for overtime work Shah, et al, 1971; Hauser, 1972). So
and income from other sources.
elaborate discussion about the technique in
Since some studies have shown that this paper is unnecessary.
mass communication media have an im-
The model assumes that the relationship
portant part to play in the adoption of among the variables are linear and addi-
family planning practices, the degree of ex-
tive. The model is written as a set of

structural equations that represent the relationship. It is possible that as people
causal process assumed to operate among become more mature they will take time
the variables that are taken up for consi-
to read newspapers. If so, age will be
deration. From the structural equations, positively related to newspaper reading.
the population parameters can be esti-
On the other hand, perhaps, most people
mated, and if found necessary, the model of the older generation did not have an
may be reformulated.
opportunity to develop the habit of read-
ing newspapers in their formative years due
A CAUSAL MODEL OF THE PROCESS OF to the low level of development of com-
ADOPTION OF CONVENTIONAL METHODS OF munication media at that time. If so, age
will be negatively related to newspaper
reading. It is possible that both may be
The variables that are of interest to us operating and one may cancel the influence
(listed above) can be divided into two of the other.
categories, viz., exogenous and endogenous.
We hypothesize that among those who
Exogenous variables are those which are have the same level of ("holding constant",
determined by variables outside the model. or "net of") income, age and newspaper
The exogenous variables may or may not reading, higher the level of education,
be related among themselves. The associ-
their concept of ideal family size will be
ation among them is taken for granted, and smaller. Holding constant education (E),
their influence on endogenous variables is age (A) and newspaper reading (P), higher
of interest to us. Variables which the model the income (I), smaller the family size
is designed to explain are endogenous norm (S) endorsed. Higher the income level
of the people, greater will be the concern
We take that the educational level at-
for consumer goods and comforts. In order
tained by the factory worker (E), average to maintain a high standard of living, many
monthly income (I) and age (A) are exo-
people would opt for a small family. Fur-
genous variables, and that they are associated
ther, higher the level of income, greater
among themselves. We hypothesize that will be their assurance that they will be
these three demographic and social back-
able to provide necessary, nutritional and
ground factors influence the frequency of health care to their children. This will give
newspaper reading (P) of the respondent. It them the assurance that all their children
is reasonable to assume that among those will be alive when the parents reach old
who have the same level of income and age. So, they may not feel the necessity to
age, higher level of education will posi-
wish for more number of children. On the
tively determine the frequency of newspaper other hand, the older respondents were
reading. It is also reasonable to assume that forming families at a time when the con-
among those who have the same level of cept of small family was not actively popu-
education and age, higher the income, larized by the government. So, perhaps
greater the frequency of newspaper reading, they produced large families. Having pro-
as the structural barrier (lack of finance) duced large families, perhaps, they ad-
will be absent. We also hypothesize that justed their ideal to the real situation. So,
given the same level of education and it is possible that higher the age, larger
income, age influences the frequency of their ideal family size. Net of education,
newspaper reading. Unfortunately, we are income and age, we hypothesize that higher
not able to predict the direction of the the frequency of newspaper reading, smaller

FIGURE 1 : A hypothesized model of the process of adoption of conventional contraceptives for
family planning purposes: Male industrial workers in two factories.
NOTE : Variables are: E-education; I-income; A-age; S-ideal family size; P-newspaper reading;
F-adoption of family planning. Figures within parentheses are not significant at .05 level.
the ideal family size. In view of the fact their families. So we assume, older the
that the small family ideal is popularized respondents, higher will be their adoption
through mass media by the government rate. We also assume that frequency of
and other agencies, this hypothesis is newspaper reading and ideal family size
intervene between background factors and
We hypothesize that the location of an adoption of family planning techniques.
individual in the stratification hierarchy, as The former will be positively related and
indexed by education and income, influ-
the latter negatively related.
ences adoption of conventional methods of
In our model we conceptualize news-
family planning. We expect this influence to paper reading (P) and ideal family size (S)
be positive. The assumption underlying this as causally prior to adoption of family
assertion is as follows: higher the location planning practices, and among P and S, P
in the stratification, greater the control over as prior to S. It should be pointed out that
environment, and adoption of family plan-
all these three variables were measured at
ning practices is one indicator of such the same time and so one cannot attribute
temporal priority of one over the other.
It is possible that older respondents One could argue that the habit of reading
would have had a chance to "complete" newspapers and ideal family size are rela-

tively stable indicators, at least in the short be mutually uncorrelated and uncorrelated
run, of already established norms. As such, with regressors in their own and preced-
we explicitly assume that they are causally ing equations. We assume that the rela-
prior to adoption of family planning prac-
tionships among the variables are asymme-
tices. Between the frequency of newspaper trical, linear and additive and estimate the
reading and the concept of ideal family model by least squares.
size, we assert that the former requires more
time and energy to develop than the latter.
Further, the mass media campaign to popu-
larize small family is based on the assump-
tion that mass media can influence the
The interpretation of the equations is
formation of the concept about the size of fairly straightforward. In the first equation
the family. So, for the following analysis in Table 2 (for example), for a unit (stan-
it is assumed that newspaper reading is dard deviation) change in education (E)
causally prior to forming a stable concept there is on the average, a change of PPE
about ideal family size.
in newspaper reading (P), net of the effects
Figure 1 is a heuristic device to depict of income (I) and age (A). The numerical
the causal relationships among various calculations are presented in Table 2, line
variables under consideration, which are 1. It shows that one standard deviation
verbally expressed above. The curved two-
(4.027 years of education. Table 1) shift
headed arrows indicate the unanalysed re-
in education produces 0.630 standard de-
lationships among the exogenous variables, viation [or (.630) (1.602) = 1.009] change
E, I and A. The straight unidirectional in newspaper reading, net of influences of
arrows symbolize direct causal influences. I and A. Similarly, a one standard devia-
Since path analytic technique requires that tion change in I (or Rs. 131.423, Table 1)
the system be completely determined, the produces 0.104 standard deviation [or (.104)
influences of unanalysed variables, depar-
(1.602) = 0.167] change in newspaper
tures from linearity and additivity and reading, net of I and A. Although E and
random disturbances are indicated by "e".
I are in different metrics — one in years
The model as formulated verbally and of education and the other in rupees —
represented diagrammatically can also be since they are standardized, their effects
rendered algebraically as a set of linear equa-
on P can be compared. Table 2, line 1
tions. Using the notation of path analysis, shows that on newspaper reading, the effect
we denote the direct effect of variable k of education is about six times larger than
on variable j by Pjk and all the variables the effect of income, when other variables
are expressed in standard form. The equa-
are taken into account.
tions for the model in Figure 1 can be
written as:
In the second panel of Table 2, the re-
gression coefficients express the net effect
on the dependent variable of a unit change
in the predetermined variable. Thus, look-
ing at the corresponding line in the second
panel for each additional year of educa-
tion, the frequency of newspaper reading
averaged 0,25 points higher, net of income
and age.
where the disturbances, ej are taken to

Education (E)
Income (I)
Age (A)
Newspaper reading (P)
Ideal family size (S)
Adoption of Family
Planning (F)
Standard Deviation
Further, for a typical respondent — by 1. Influence of background factors on
"typical" we mean that the educational
adoption of family planning practices.
attainment, income and age are at the mean
of the sample,— the predicted frequency of
Bivariate associations (or gross associ-
newspaper reading is: 0.579 + (6.706) ations) between any two variables are in-
(0.251) + (303.229) (0.001) + (35.795) dexed by zero-order correlation coefficients.
(0.002) or 2.722 which by definition is the The three background variables, namely,
mean frequency of newspaper reading education (E), income (I) and age (A) are
(Table 1). For a respondent who had a positively associated with adoption of family
planning (F). The correlation coefficients
higher level of educational attainment than are: 0.143, 0.424, and 0.223 respectively
the average, say 12 years, the predicted fre-
(Table 1). Except for the gross association
qusncy of newspaper reading is: 0.579 + of I and F (r FI) the other two associa-
(12.0) (0.251) + (303.229) (0.001) + tions are modest. The total effects of each
(35.795) (0.002) or 3.967, which is 1.245 of the three background factors on F are
higher than that of the "typical" given by reduced form coefficients given
respondent. Similarly for a respon-
in Table 2, line 4. The reduced form co-
dent with "typical" education and age but efficients are the coefficients in an equation
with higher income, say Rs. 500.00 per when each of the endogenous variables is
month, the predicted score is 2.834. This related to the exogenous variables in the
is almost same as reading newspapers three model. When a gross association is com-
or four times per week.
pared with its corresponding total effect,
the latter is smaller than the former. For
THE PROCESS OF ADOPTION OF FAMILY example, gross association between E and
P is 0.143, and the total effect of E on F,
net of I and A is 0.115 — a reduction of
Having set the stage, let us examine the 0.028 or about 20 per cent (.028/.143 x 100
process of adoption of family planning = 19.6). Similarly, gross association of I
practices under three headings.
and A with F is also reduced by about

N O T E : Figures within parentheses are not significant at 0.05 level. Variables a r e :
E—Education; I—Income; A—Age; P—Newspaper reading; S—Ideal family size;
F—adoption of family planning practices.
17 per cent and 21 per cent respectively. conclude from the reduced form coefficients
In other words, the total association of that an important precondition for higher
each of the three background variables rate of adoption is that the level of living
with F is reduced by about a fifth when the of the people should rise.
other two variables are taken into account.
This shows that although the three back-
2. Influence of background factors on
ground factors are correlated among them-
mediating factors.
selves, their total effect on adoption of
family planning are mostly (at least four-
The model identifies two mediating vari-
fifths) independent of each other.
ables, namely, frequency of newspaper
reading (P) and ideal family size (S). Be-
Among the three factors, income has fore examining the role of mediating fac-
the largest effect on adoption of family tors on adoption of family planning, it
planning, and it is three times larger than may be instructive to examine the influence
that of education and twice as large as of background factors on mediating fac-
age. If it can be assumed that income is tors.
an indicator of level of living, and if it
(a) Newspaper reading: The gross asso-
is the policy of the government to encourage ciation of the three background factors
more and more people to adopt family with newspaper reading (P) is given in
planning practices, then we can tentatively Table 1. Education (E) and income (I) are

positively associated with P whereas age 0.630 — a negligible reduction (Table 2,
(A) is negatively associated. When we re-
line 2). On the other hand, the reduction
gress P on the three background factors, for I is 60 per cent and for A it is 107
the effects of all the three factors, including per cent. This shows that the zero-order
that of A are positive (Table 2, line 1). relationships of A and I with P are due to
When we take into account the other two their association with E. The gross asso-
variables (E and I), the bivariate negative ciation between two variables can be de-
relationship between A and P revealed in composed into its components according to
Table 1 becomes positive. For these res-
the basic theorem of path analysis, which is:
pondents we find that when education and
income are taken into account, the relation-
ship between age and newspaper reading is
where i and j are any two variables in
the system and k runs over all variables in
The gross association of E with P is the system from which direct paths lead
0.651 (Table 1). When I and A are taken to variable j. (For proof of this theorem,
into account, the effect of E on P becomes see: Duncan, 1966).
This shows that almost all (about 97 per very little is due to association of E with
cent) of the effect of E on P is direct and I or A. On the other hand,
This shows about 2/5th of the effect of I is due to their association with education
on P is direct and about 3/5th is due to only.
association of I with E. Similarly, only 7
per cent of the effect of A on P is direct.
(b) Ideal family size: The gross asso-
All these show that education has large ciations of education (E) and income (I)
effect on newspaper reading and that effect with ideal family size (S) are negative
is not due to association of education with (Table 1). The reduced form coefficients
income or age. Further, most of the effect show that the total effects are also nega-
of income and age on newspaper reading tive (Table 2, line 2). This shows that

higher the location of individuals in the income and age are not mediated by news-
stratification hierarchy, smaller is their paper reading. But, when newspaper
ideal family size. In this study we find, reading is taken into account, the total
that among those who have the same level effect of E is reduced from 0.115 to 0.019
of education and age, those who have — a reduction of almost 92 per cent. This
higher income opt for smaller families. It shows that almost all the effect of educa-
is reasonable to argue that higher the in-
tion on adoption of family planning is via
come, higher the ability to support a newspaper reading, or in broader terms,
large family. So one may expect higher participation in mass media. Many studies
the income, larger will be their ideal family on family planning have identified educa-
size. But we find that this is not so. To tion as a very important variable for
put it differently, the argument that in-
adoption of family planning practices. For
creased financial resources of the family this sample of factory workers we have
will result in a large number of children is identified a mechanism, namely newspaper
not supported by our data. Further, if we reading, through which the influence of
interpret income as an indicator of standard education operates. From the point of view
of living, if more and more people are to of social engineering, those studies hypo-
develop the concept of small family, their thesize education as a precondition for
standard of living should rise.
adoption of family planning practices. Our
study suggests that instead of waiting for
As pointed out earlier, it is possible that the population to be educated, manipula-
when the older factory workers were form-
tion of mass media may result in desirable
ing their families, the concept of small level of adoption of conventional methods
family was not popularized by the govern-
of family planning.
ment and other voluntary agencies. So, per-
haps many of them produced many children.
When we include ideal family size (S) in
(The zero-order correlation coefficient bet-
the model as an intervening variable, it
ween age and total number of children is mediates still further the influence of
0.492). This is reflected in our data by the education on S (compare Table 2, lines 5
positive total effect of age, net of education and 6). But the effects of all other
and income, on ideal family size (Table variables in the model are not modified.
1, line 4).
The effect of S, net of all the variables in
the model, on adoption of family planning
3. The role of mediating factors in the is negative. In other words, larger the ideal
process of adoption of family planning family size of the respondent, lower the
adoption of family planning practices. Net
of all variables in the model, one addi-
In addition to the three background tional child in the ideal family size, on the
factors, namely E, I and A, when we take average, results in about 0.05 less adoption
into account newspaper reading (P), the of family planning practices (Table 2,
total effect of income and age is hardly panel B, line 6).
reduced (compare Table 2, lines 4 and 5).
The total effect of I on F, net of E and A
One advantage of path analytic technique
is 0.356. When P is also added, the effect is that the total effect of one variable on
becomes 0.340 — a reduction of about 4 the other can be decomposed into its con-
per cent only. This shows that the effects of stituent parts. For example:

This shows that over 90 per cent of the 6 years as it is now, then the predicted
total effect of income on family planning score for newspaper reading will be 3.97
is direct and very little is via other (or almost reading newspapers daily).
mediating variables in the model. Earlier, Given the typical score for income and age,
we pointed out that very little of the the policy manipulated score for education
gross association is reduced when we (12 years) and the predicted score for
compare it with total effect. All these show newspaper reading (3.97), the predicted
that income has an important direct effect score for ideal family size will be 3.45
on adoption of family planning. On the children. Given the typical score for income
other hand only 9.4 per cent of the total and age, the manipulated score for educa-
effect of education on adoption of family tion (12 years), the predicted scores for
planning is direct and another 82 per cent newspaper reading (3.97) and ideal family
is via newspaper reading.
size (3.45 children), the predicted score for
adoption of family planning is 1.41 or an
increase by 0.07.
Let us assume that the respondents
A model like that of ours, which is remain "typical" in two background
written in structural equations, facilitates us variables (viz. education and age), but the
to predict the outcome if certain variables income level is about twice as that of the
are manipulated or changed. If such a present level (i.e., Rs. 600.00). Then the
variable is capable of being manipulated, predicted score for newspaper reading will
then the direction in which policy measures be 3.95. The predicted number of children
may be undertaken can be specified.
in the ideal family size will be 3.30. For
The predicted score for "typical" res-
adoption of family planning, the predicted
pondent with regard to adoption of family score will be 1.79. This is just 0.21 less
planning is 1.34 (see Appendix A for the than 2.00, the latter indicating adoption by
relevant calculations for this and following all respondents.
equations). It may be recalled that the score
Similarly, through social engineering if
for adoption of family planning ranges respondents are "made" to read news-
from 0 to 2. So, the predicted score for papers every day (score of 4) all other
the "typical" respondent implies that the conditions being the same, the predicted
"typical" predicted score is little more ideal family size will be 3.62 and the
than having knowledge about conventional predicted score for adoption of family
planning will be 1.39.
Let us assume that by policy interven-
Among the three possibilities for mani-
tions the educational level of the workers pulation of variables, increasing the level
is increased,. say to 12 years rather than of income seems to be more promising, if

higher rate of adoption of family planning similar. So, our interpretation of the causal
is the desired ideal. We would like to relationships, presented earlier, does not
point out that we use level of income as
an indicator of standard of living. Mere
increase in wages and salaries does not lead EFFECTS OF SELECTED VARIABLES ON VARIOUS
to an increase in the standard of living, if THE PROCESS OF ADOPTOIN OF CONVENTIONAL
the increase in wages is eaten up by rising CONTRACEPTIVES FOR FAMILY PLANNING PURPOSES:
inflation. So, if it is socially desirable to MALE INDUSTRIAL WORKERS IN TWO FACTORIES.
increase the rate of adoption of family
planning practices and thereby to reduce

population growth, then increase in stan-
dard of living alone can achieve it.
We also
assert that the standard of living should
increase for all people and not only for a
few. Only then there will be increased
adoption of conventional methods of family
Table 2 contains many coefficients which
are not statistically different from zero. In
other words, the null hypothesis that the
slope equals zero (b*=0) is not rejected at
.05 probability level.
If we fail to reject a hypothesis when it
should have been rejected, we will commit
Type II error. In order to examine this NOTE : Variables are: E — education; I — in-
possibility, we erased the paths that are
come; A — age; P — newspaper
not statistically significant and recomputed
reading; S — ideal family size; F —
adoption of family planning practices.
the coefficients. This procedure implies
that the omitted variables make a negligible differ whether the hypothesized model or
contribution to explanation of variation in modified model is presented. In the case
the dependent variable. The results are of S, the coefficients in the modified model
presented in Table 3 (and also as Figure are smaller than the ones in the hypothe-
la). When we compare the relevant sized model. This suggests, that for ideal
coefficients of determination in Table 2 and family size if we erase the paths that are
3, we find that they are similar for news-
now not significant at .05 level, we lose
paper reading (P) and adoption of family some information and E and P also have
planning (F). For ideal family size (S) and an influence on S.
R2 of the modified model is smaller than
the R2 of the hypothesized model (com-
pare Table 2, line 3 and Table 3, col. 2).
On the assumption that adoption of
Further, when the relevant regression family planning is an ongoing and multi-
coefficients (both standardized and
faceted process, we constructed a multi-
unstandardized) are compared, we
variable model to explain and interpret
find that for both P and F, they are the inter-relationships among various vari-

FIGURE la : A modified model of the process of adoption of conventional contraceptives for
family planning purposes: Male industrial workers in two factories.
N O T E : Variables are: E-education; I-income; A-age; P-newspaper reading; S-ideal family size;
F-adoption of family planning.
ables. In constructing the model, we made would like to point out that these variables
certain simplifying assumptions and limited are identified as some of the more impor-
our attention only to a few selected tant variables in prior research in this area.
variables (Blalock, 1961: 7-17). Two
A discerning reader would have noted
simplifying assumptions we made use of large paths from residually defined
are: first, that the variables are linear, and variables. This shows that more variables
second, that the inter-relationships among than that are now included in the model
variables are additive. We also limited are involved in the process of adoption of
our attention to a total of six variables family planning. Perhaps variables like
only. These variables are selected on the type of family, number of children, upward
basis of prior research and theoretical mobility orientation are some of those
variables. Further, adoption of family
A critic may charge that the choice of planning is a joint decision between
variables included in the model is not husband and wife. So, an enquiry into the
adequate and variables other than the ones manner in which women adopt family
now included should have been included. planning is very essential. Above all, more
Without dogmatically asserting that our accurate measurement of variables, better
choice of the variables is the best, we scaling techniques and in-depth multivariate

approaches are essential to comprehend the
Making use of the technique of path
process of adoption of family planning so analysis, we tested the model on a sample
that effective policy measures may be of male respondents drawn from two
factories. We found that education has a
large effect on newspaper reading. The
effects of income and age on newspaper
reading, although positive, are mostly due
In order to understand the complicated to> association of these variables with
process of adoption of family planning, a education. Although there is a substantial
multivariate model based on prior research total effect of education on adoption of
is constructed. We made two simplifying family planning practices, when newspaper
assumptions, namely, that the variables reading is taken into account, the total
are linear and the relationship among the effect of education is very much reduced.
variables are additive. Further, we limited Newspaper reading or participation in mass
our attention to only six variables.
media is the mechanism through which the
effect of education operates. Further, income
We hypothesized that three background has the largest effect on adoption of family
factors (education, income and age) in-
planning. The effect of income is not
fluence frequency of newspaper reading. reduced by the mediating variables
The background factors and newspaper included in the model. In terms of policy
reading influence the concept of ideal implications we showed that in order that
family size and all these variables deter-
more people may adopt family planning
mine adoption of family planning in a devices, the level of income and standard
causal sequence.
of living should increase.
Agarwala, S. N.
Attitudes Towards Family Planning in India. Bombay: Asia
Publishing House.
Bhogle, Shalini
Adoption of Family Planning in Two Industrial Settings. New
and Surjit K a u r
D e l h i : Council for Social Development.
Causal Inferences in Nonexperimental Research. Chapel H i l l :
Hubert M., Jr.
University of North Carolina Press.
Census of India
Provisional Population Totals. Paper 1 of 1971-Supplement.
D e l h i : Manager of Publications, 1971.
Duncan, Otis Dudley
"Path Analysis: Sociological examples". American Journal of
Sociology 72 (July): 1-16.
Hauser, Robert M.
Socio-Economic Background and Educational Performance. Rose
Monograph Series. Washington D C : American Sociological
Heise, David R.
"Contingencies in Constructing Causal Models". pp. 38-73 in E.
F. Borgatta and G. W. Bohrnstedt (eds.), Sociological Methodo-
logy, San Francisco: Jossey-Bass.
Krishnamurthy, K. G.
Research in Family Planning. D e l h i : Sterling Publishers.
Land, Kenneth C.
"Principles in Path Analysis", pp. 3-33 in E. F. Borgatta and
G. W. Bohrnstedt (eds.) Sociological Methodology. San
Francisco: Jossey-Bass.

R a o , Kamala Gopal
Studies in Family Planning in India and Review for Programme
Implications. New Delhi: Central Family Planning Institute.
Repetto, Robert
"India: A Case Study of the Vasectomy Programme" Studies
in Family Planning 31 (May): 8-18.
Schramm, Wilbur
"Communication in Family Planning". Reports of Population/
Family Planning N o . 7. April.
Sewell, William H.,
"Socio-Economic Status, Intelligence, and the Attainment of
and Vimal P. Shah
Higher Education". Sociology of Education 40 (Winter): 1-23.
Shah, Vimal P ,
"Social Class and Educational Aspiration in an Indian Metro-
Tara Patel and
pol's". Sociological Bulletin, 20 (Sept): 113-133.
William H. Sewell.
Wright, Sewell
"The Method of Path Coefficients". Annals of Mathematical
Statistics 5 (September): 161-215.

The Indian Journal of Social Work, Vol. XXXV, No. 3, (October 1974)