-
International Journal of Applied Research & Studies ISSN
2278 – 9480
1
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
Research Article
Determinants of Household Decision Making among Women in Kolkata
slum Areas: An Application of Multinomial Logistic
Regression
Authors:
Gitanjali Hajra
Address for Correspondence:
Assistant Professor, Economics & Statistics, Brainware Group
of Institutions, Kolkata
Abstract: In today’s world women’s autonomy, which is one of the
most important indicators of women empowerment, is very well-known
concern regarding the gender issue for inclusive
growth of an economy. In most of the cases the role of women in
family decision making is
negligible, where the male member of the family is the final
decision maker, which is more
visible in the poor families of the developing countries. There
are different dimensions of
women’s autonomy. The present paper identifies four such
dimensions like decision in family
planning, Decision in savings, Decision in expenditure and
decision in healthcare. This paper
attempts to find how far education, age and income have an
impact on these four dimensions. In
our study the target respondents were only the women segment
from Kolkata Municipal
Corporation’s slum areas and the primary data has been collected
by applying multistage
sampling technique. Multinomial Logistic Regression was run
through the statistical software
SPSS 15 to test the relationship of these variables to all four
dimensions of Decision making.
Keywords: Decision making, developing countries, Multinomial
Logistic Regression, Multi-Stage Sampling, Women’s Autonomy, Women
Empowerment.
JEL Classification: C01, C12, I25.
1. Introduction: In a less developed country like India, women
played an important role for achieving inclusive growth of the
economy. In the approach paper of 12
th Five year plan (2012-
17) Planning Commission of India addressed economy’s transition
to a higher and more
inclusive growth Path implying removal of Social Exclusion.
Inclusiveness is a
Multidimensional concept; one of such dimensions is women’s
autonomy. Dyson and Moore in
‘on kinship structures and Female Autonomy’ (1983) defined
autonomy as the capacity to
manipulate one’s personal environment and the ability –
technical, Social and Psychology to
obtain information and to use it as the basis of making
decisions about one’s private concerns
and those of one’s intimates’. The National Family Health Survey
(NFHS-3) Report in India of
1998-99 also measured female autonomy in terms of various
indicators on household Decision
Making. Role of women in household decision making Process
depends on various Social and
demographic factors like Income, Age, Occupation, Level of
education and so on. In the present
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
2
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
study finds out the different factors that can influence the
household decision making progress of
women.
2. Review of Literature: There exist several studies conducted
by the different researchers, which examined the determinants of
Women’s Autonomy in household Decision Making.
Dev Raj Acharya, Jacqueline S Bell, Padam Simkhada, Edwin R van
Teijlingen and Pramod Raj
Regmi in the study of ‘ Determinants of Women's Autonomy in
Decision Making’(2010) aimed
to explore the links between women’s household position and
their autonomy in decision
making.
In another research study by Nava Ashraf (Spousal Control and
Intra-Household Decision
Making: An Experimental Study in the Philippines Harvard
Business School) found that
household savings and investments typically depend on how
decision making power distributed
between men and women. It also analyzed the fact that, financial
decisions of the household are
greatly affected by the fact that the income is known to spouses
or not.
Marie Furuta and Sarah Salway in their study of ‘Women’s
Position Within the Household as a
Determinant Of Maternal Health Care Use in Nepal’ (2006)
examined women participation
within their household—decision making, employment and influence
over earnings, and spousal
discussion of family planning by running a logistic Regression
Model and showed that there is a
strong association of women’s education with health.
M. Hemanta Meitei in the study titled ‘Education or Earning and
Access to Resources
Determining Women’s Autonomy: An Experience Among Women of
Manipur’ investigated
how far education or earning and access to resources have a
significant impact on women’s
decision making power. He concluded that, most of the decisions
are taken jointly (both husband
and wife) while working women take more of independent decisions
than the non-working
women. Controlling effect of the other background variables work
status of women turn out a
significant explanatory variable rather education.
Mosiur Rahman, Uzzal K. Karmaker and Abdur R. Mia in the study
of ‘ Determinants of
Women Empowerment at Domestic and Non-domestic Issues: Evidence
from Chapai Nawabganj
District in Bangladesh’ examined the determinants of women
empowerment at domestic and
non-domestic issues in Bangladesh. From the logistic regression
model considering decision-
making power for household affairs as the dependent variable, it
had shown that as the level of
education of the respondents increases their decision making
power also increases.
Siwan Anderson and Mukesh Eswaran in the article titled ‘What
Determines Female Autonomy?
Evidence from Bangladesh’ (2005) examined the determinants of
female autonomy within
households in Bangladesh, a developing country. They
investigated the relative contributions of
earned versus unearned income in enhancing women’s autonomy and
the role of employment
outside of their husband’s farm.
Data: For the present study, Primary data have been collected
from the slum areas of Kolkata Municipal Corporations. In this
survey each woman was asked the questions based on structured
questionnaire to measure their role in family decision making
directly. We have taken a sample
of 100 women from Kolkata Slum areas.
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
3
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
3. Variables under the Study: In general there are two types of
variables: dependent and independent.
Independent Variable: Independent variables under the study was
–
a) Age of an Woman b) Educational Qualification of a woman,
which is a categorical variable. This variable is categorized into
six groups: Illiterate, Primary, Secondary, Higher Secondary,
Graduate and Post
Graduate.
c) Income of a woman is categorized into three: < Rs 5000, Rs
5000-Rs.10000 and >Rs 10000.
Dependent Variable: The dependent variable in this study
consists of four types of household
decision making of women which has three aspects:
a) Decision in family planning b) Decision in family saving c)
Decision in family expenditure d) Decision in healthcare
Expenditure The above four variables are categorized into three
levels: 1=High, 2=Low and 3=No Role.
4. Objective of the study: The specific objective of the present
study is to identify the relationship between:
a) Role of level of education of women and decision in family
planning, family savings, family expenditure and healthcare.
b) Role of Age of Women and decision in family planning, family
savings, family expenditure and healthcare.
c) Role of Income of Women and decision in family planning,
family savings, family expenditure and healthcare.
5. Research Methodology:
Multinomial Logistic Regression (MLR): To understand the role of
women education, age and Income in family decision making
process
we have used several statistical and analytical tools. In this
study we have Multinomial logistic
regression (MLR) and to predict the presence or absence of a
characteristic or outcome based on
values of a set of predictor variables.
When the dependent variable is categorized then we can’t use
Ordinary Least Square Method
(OLS) to predict the dependent variable. In that case Logistic
Regression is applied for
predicting the dependent variable.
Logistic regression is useful for situations in which we want to
be able to predict the presence or
absence of a characteristic or outcome based on values of a set
of predictor variables. It is similar
to a linear regression model but is suited to models where the
dependent variable is dichotomous
(0 and 1).
The multinomial (polytomous) logistic regression model is an
extension of the Binomial logistic
regression model, the dependent variable is not restricted only
to two categories and the
independent variables may be categorized or may be continuous.
It is used when the dependent
variable has more than two nominal or unordered categories, in
which Independent variables can
be factors or covariates. In general, factors should be
categorical variables and covariates should
be continuous variables. In our study each of the four dependent
variables is categorized into
three levels: High (Code1), Low (Code 2) and no role (Code 3).
Among the independent
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
4
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
variables level of education is categorized into six dummies: 0
for illiterate, 1 for primary, 2 for
secondary, 3 for Higher Secondary, 4 for Graduate and 5 for the
others.
The overall test of relationship among the independent variables
and the dependent was based on
the reduction in the likelihood values for a model without any
independent variables and the
model with the independent variable. This difference in
likelihood followed a chi-square
distribution. The Logistic Regression equation by using the
Maximum Likelihood Estimation is given by:
Log-odds=A+B(X)
Or, Odds= Antilog (A+B(X))
Where, the Odds ratio (p/(1 − p) indicates the ratio of the
Proportion of Occurrences to the
Proportion of non-occurrences.
The Wald statistic is used to test the significance of
individual logistic regression coefficients for
each independent variable (that is, to test the null hypothesis
in logistic regression that a
particular logit (effect) coefficient is zero). Wald statistic
is the squared ratio of the
unstandardized logistic coefficient to its standard error.
Results of Multinomial Logistic Regression Model:
6. Factors affecting the level of savings as one dimension of
women decision
making: Multinomial Logistic Regression Analysis:
6.1 Overall Test of the relationship
6.1.1 Model Fitting Information: The first analysis of
Multinomial Logistic Regression is to
describe the overall Test of the relationship i.e. the
relationship between the dependent and the
dependent variable. This relationship is based on the
statistical significance of the final model
Chi-Square. Chi-Square Statistic is the difference in 2
Log-Likelihood between the final model
and the reduced model.
Table 6.1
From the log-likelihood table the log likelihood function (-2Log
likelihood) will decrease if the
independent variables have a relationship with the dependent
variable. In our analysis (Table 6.1)
the initial Log Likelihood value (with no independent variable)
is 171.513 and the final Log
Likelihood (including the independent variable) is 138.805. The
difference between the two is a
Chi-Square and its significance value is 0.001 (which is less
than 0.05). So, there is significant
relationship between the dependent variable (level of savings)
and the independent variables
(Income, age and level of education). So, the final model is
outperforming the null model.
6.1.2 Goodness of Fit:
Model Fitting Information
171.513
138.805 32.707 12 .001
Model
Intercept Only
Final
-2 Log
Likelihood
Model
Fitting
Criteria
Chi-Square df Sig.
Likelihood Ratio Tests
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
5
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
Pearson and Deviance Goodness of Fit measure how the model
adequately fits the data. The Null
Hypothesis in this case is given by-
H0: There is no Difference between Observed and Expected
Frequencies, i.e. the model is good
fit.
If the significance value is >0.05, it is not statistically
Significant and the model is good fit. In
table 6.2 the significant value of Chi-Square is greater than
0.05. So, we can say that, the data
Table 6.2
Are consistent with the model assumptions i.e. the model
adequately fits with the data.
6.2 Strength of the Multinomial Logistic Regression
Relationship:
6.2.1 Classification Matrix as a measure of Model accuracy:
The classification Table shows the practical results of using
MLR Model. This cross tabulation
of the observed response categories with the predicted response
categories helps to assess the
predictive performance of the model. Cells along the diagonal
represent numbers of correct
predictions. Cells off the diagonal represent numbers of
incorrect predictions.
Table 6.3
The Table 6.3 classifies correctly 62.2 out of 100 respondents
who state that, they have a high
role in family household savings. The model gives better
accuracies for ‘High’ role group and
‘Low’ role group but not for the ‘No Role’ group.
6.2.2 Relationship between Dependent Variable and Independent
variables (likelihood
Ratio Test):
Likelihood Ratio Test Checks the contribution of each
independent variables to the dependent
variable. If the significance of the test is small (i.e., less
than 0.05) then the effect contributes
significantly to the model.
Goodness-of-Fit
112.984 130 .856
114.368 130 .834
Pearson
Deviance
Chi-Square df Sig.
Classification
23 13 1 62.2%
12 37 0 75.5%
1 12 1 7.1%
36.0% 62.0% 2.0% 61.0%
Observed
High
Low
No Role
Overall Percentage
High Low No Role
Percent
Correct
Predicted
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
6
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
Table 6.4
In table 6.4 only level of education is significant, because the
significance value of chi-Square
for this independent variable is 0.005(
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
7
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
Savings(a) B Std. Error Wald df Sig. Exp(B)
95% Confidence Interval for Exp(B)
Lower Bound Upper Bound
High Intercept 20.901 2.318 81.281 1 .000
Age -.069 .053 1.700 1 .192 .933 .841 1.035
[Income=1.00] -.425 .982 .187 1 .666 .654 .095 4.484
[Income=2.00] 0(b) . . 0 . . . .
[Leveledu=.00] -18.305 1.886 94.254 1 .000 1.12E-008 2.79E-010
4.52E-007
[Leveledu=1.00] -16.200 2.123 58.239 1 .000 9.21E-008 1.44E-009
5.91E-006
[Leveledu=2.00] .488 2852.015 .000 1 1.000 1.628 .000 .(c)
[Leveledu=3.00] -16.845 1.505 125.311 1 .000 4.83E-008 2.53E-009
9.23E-007
[Leveledu=4.00] 0(b) . . 0 . . . .
Low Intercept 19.728 1.849 113.887 1 .000
Age -.072 .046 2.407 1 .121 .931 .850 1.019
[Income=1.00] .361 .958 .142 1 .706 1.435 .220 9.377
[Income=2.00] 0(b) . . 0 . . . .
[Leveledu=.00] -16.518 1.364 146.572 1 .000 6.70E-008 4.62E-009
9.72E-007
[Leveledu=1.00] -14.920 1.672 79.603 1 .000 3.31E-007 1.25E-008
8.78E-006
[Leveledu=2.00] .698 2852.015 .000 1 1.000 2.010 .000 .(c)
[Leveledu=3.00] -17.073 .000 . 1 . 3.85E-008 3.85E-008
3.85E-008
[Leveledu=4.00] 0(b) . . 0 . . . .
a The reference category is: No Role. b This parameter is set to
zero because it is redundant. c Floating point overflow occurred
while computing this statistic. Its value is therefore set to
system missing.
From table 6.5 the EXP(B) is 0.933 for ‘High’ group i.e. for
each year difference in age, the
women is 0.933 (7%) times less likely to have ‘high role’ in
family savings compared to ‘No
Role’ Group , similarly, the respondent is 0.931 times less
likely to have low role in family
savings compared to ‘No Role’ Group.
For Income, the odds ratio (EXP(B)) is 0.654 for group 1( <
5000 Rs per month), i.e. the women
within the group 2 ( Rs. 5000- Rs.10,000) are 1.52 times (Odds
Ratio is 1.52, reciprocal of
0.654) are more likely to have high role compared to no role and
. The women within the group 1
(
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
8
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
The Final Log Likelihood value (-2Log Likelihood) decreases and
the significant value of chi-
square is 0.002 i.e. less than 0.05. So, the Null Hypothesis
that there is no significant difference
between the model without independent variables and the model
with independent variables is
rejected. The existence of a relationship between the dependent
and independent variables is
supported.
7.1.2 Goodness of Fit:
Table 7.2
The significance values of Chi-Square of Table 7.2 for both
Pearson and Deviance are >0.05,
indicating the acceptance of the Null Hypothesis that the model
adequately fits the data.
7.2 Strength of the Multinomial Logistic Regression
Relationship:
7.2.1 Classification Matrix as a measure of Model accuracy:
Table 7.3
The Classification table 7.3 indicates that, the overall
percentage of accurate predictions is 70%.
The model gives the better accuracies for the ‘Low’ Role group
and ‘No Role’ Group.
7.2.2 Likelihood Ratio Test:
Table 7.4
Model Fitting Information
139.720
108.288 31.433 12 .002
Model
Intercept Only
Final
-2 Log
Likelihood
Model
Fitting
Criteria
Chi-Square df Sig.
Likelihood Ratio Tests
Goodness-of-Fit
123.248 130 .650
92.169 130 .995
Pearson
Deviance
Chi-Square df Sig.
Classification
0 2 1 .0%
0 23 17 57.5%
0 10 47 82.5%
.0% 35.0% 65.0% 70.0%
Observed
High
Low
No Role
Overall Percentage
High Low No Role
Percent
Correct
Predicted
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
9
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
The Significance value of Chi-Square for Age and Level of
Education are less than 0.05, so these
two explanatory variables have significant effect on the level
of family planning. But, Income is
not statistically significant at 5% level (though significant at
10% level), since the significance
value of Chi-Square for Income is greater than 0.05 (less that
0.1).
7.2.3 Identifying the Statistically Significant Predictors:
Table 7.5
Parameter Estimates
Planning(a) B Std. Error Wald df Sig. Exp(B)
95% Confidence Interval for Exp(B)
Lower Bound Upper Bound
High Intercept -38.805 3108.506 .000 1 .990
Age .111 .107 1.064 1 .302 1.117 .905 1.378
[Income=1.00] 17.315 3108.504 .000 1 .996 33109894.429 .000
.(b)
[Income=2.00] 0(c) . . 0 . . . .
[Leveledu=.00] 13.769 2.047 45.267 1 .000 954757.843 17293.269
52711983.952
[Leveledu=1.00] -1.740 6196.687 .000 1 1.000 .175 .000 .(b)
[Leveledu=2.00] 16.180 1.769 83.674 1 .000 10634781.576
332001.098 340657244.878
[Leveledu=3.00] 17.823 .000 . 1 . 55030877.101 55030877.101
55030877.101
[Leveledu=4.00] 0(c) . . 0 . . . .
Low Intercept -2.851 1.635 3.041 1 .081
Age .109 .039 7.695 1 .006 1.116 1.033 1.205
[Income=1.00] 1.001 .615 2.644 1 .104 2.720 .814 9.085
[Income=2.00] 0(c) . . 0 . . . .
[Leveledu=.00] -3.465 1.405 6.087 1 .014 .031 .002 .490
[Leveledu=1.00] -1.678 1.386 1.464 1 .226 .187 .012 2.828
[Leveledu=2.00] -1.062 1.346 .623 1 .430 .346 .025 4.833
[Leveledu=3.00] -1.316 1.465 .807 1 .369 .268 .015 4.736
[Leveledu=4.00] 0(c) . . 0 . . . .
a The reference category is: No Role.
Likel ihood Ratio Tests
108.288a .000 0 .
117.680 9.392 2 .009
113.736 5.448 2 .066
132.433 24.145 8 .002
Ef fect
Intercept
Age
Income
Leveledu
-2 Log
Likelihood of
Reduced
Model
Model Fitting
Criteria
Chi-Square df Sig.
Likelihood Ratio Tests
The chi-square statistic is the dif f erence in -2
log-likelihoods
between the f inal model and a reduced model. The reduced
model is formed by omitting an ef f ect f rom the f inal model.
The
null hypothesis is that all parameters of that ef f ect are
0.
This reduced model is equivalent to the f inal model
because omitting the ef f ect does not increase the
degrees of f reedom.
a.
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
10
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
b Floating point overflow occurred while computing this
statistic. Its value is therefore set to system missing. c This
parameter is set to zero because it is redundant.
Log Odds Ratio for Age i.e. EXP(B) for age is 1.117, i.e. for
each difference in age, women are
1.117 times more likely for ‘High’ Role Group compared to ‘No
Role’ group and for each
difference in age women are 1.116 times more likely for ‘Low’
Role Group compared to ‘No
Role’ Group.
For educational qualification for ‘High’ Income Group, Group
0(Illiterate) and Group 2(
Secondary Education) are statistically significant and for ‘Low’
Income Group only Group
0(Illiterate) is highly significant when we take our reference
category as ‘No Role’ Group.
Having no education (Illiterate: Group 0) make a woman 97%
(100%-3%) less likely to choose
‘Low’ Role group over ‘No Role’ Group.
8. Factors affecting the level of Family Expenditure as one
dimension of
women decision making: Multinomial Logistic Regression
Analysis
8.1 Overall Test of the Relationship:
8.1.1 Model Fitting Information:
Table 8.1
Table 8.1 of Model Fitting Information indicates that, the value
of -2 Log Likelihood Ratio (LR)
of Chi-Square decreases for final model against the initial
model, when we only consider the
intercept. Ie. There is a relationship between the dependent
variable and independent variables,
but since the significant value of Chi-Square (Chi-Square is the
difference between the -2Log
Likelihood of the initial Model and the Final Model) is greater
than 0.05, so we can say that,
there is no significant relationship between the dependent
variable and independent variables.
8.1.2 Goodness of Fit:
Table 8.2
Table 8.2 indicates that, the model adequately fits the data,
since both the values of Chi-Square
in the table are greater than 0.05.
8.2 Strength of the Multinomial Logistic Regression
Relationship:
Model Fitting Information
173.484
157.552 15.932 12 .194
Model
Intercept Only
Final
-2 Log
Likelihood
Model
Fitting
Criteria
Chi-Square df Sig.
Likelihood Ratio Tests
Goodness-of-Fit
128.118 130 .530
133.926 130 .389
Pearson
Deviance
Chi-Square df Sig.
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
11
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
8.2.1 Likelihood Ratio Test:
Table 8.3
The Table 8.3 of the Likelihood Ratio Test shows that, none of
the independent variables are significant, since all the
significance value of Chi-Square are greater than 0.05. It
indicates that,
there is no significant relationship between any of the
independent variables (Age, Income and
Level of Education) and the dependent variable.
8.2.2 Estimating the individual Parameter:
Table 8.4 Parameter Estimates
Expenditure(a) B Std. Error Wald df Sig. Exp(B) 95% Confidence
Interval for Exp(B)
Lower Bound Upper Bound
High Intercept -.419 1.856 .051 1 .821
Age -.008 .043 .035 1 .853 .992 .911 1.080
[Income=1.00] 1.015 .764 1.766 1 .184 2.760 .618 12.334
[Income=2.00] 0(b) . . 0 . . . .
[Leveledu=.00] .268 1.573 .029 1 .865 1.308 .060 28.536
[Leveledu=1.00] 19.320 1.534 158.715 1 .000 245699683.079
12163814.092 4962944501.598
[Leveledu=2.00] 2.135 1.678 1.619 1 .203 8.457 .315 226.839
[Leveledu=3.00] 2.093 1.838 1.296 1 .255 8.105 .221 297.556
[Leveledu=4.00] 0(b) . . 0 . . . .
Low Intercept .038 1.884 .000 1 .984
Age -.026 .045 .344 1 .558 .974 .892 1.063
[Income=1.00] 1.032 .796 1.679 1 .195 2.806 .589 13.359
[Income=2.00] 0(b) . . 0 . . . .
[Leveledu=.00] .355 1.580 .051 1 .822 1.427 .064 31.597
[Leveledu=1.00] 19.290 .000 . 1 . 238610630.073 238610630.073
238610630.073
[Leveledu=2.00] 1.531 1.708 .804 1 .370 4.624 .163 131.361
[Leveledu=3.00] 1.064 1.932 .303 1 .582 2.898 .066 127.749
[Leveledu=4.00] 0(b) . . 0 . . . .
Likel ihood Ratio Tests
157.552a .000 0 .
158.008 .457 2 .796
159.538 1.986 2 .370
172.417 14.865 8 .062
Ef fect
Intercept
Age
Income
Leveledu
-2 Log
Likelihood of
Reduced
Model
Model Fitting
Criteria
Chi-Square df Sig.
Likelihood Ratio Tests
The chi-square statistic is the dif f erence in -2
log-likelihoods
between the f inal model and a reduced model. The reduced
model is formed by omitting an ef f ect f rom the f inal model.
The
null hypothesis is that all parameters of that ef f ect are
0.
This reduced model is equivalent to the f inal model
because omitting the ef f ect does not increase the
degrees of f reedom.
a.
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
12
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
a The reference category is: No Role. b This parameter is set to
zero because it is redundant.
The interpretation of the table 8.4 is that, having a Primary
Educational Degree makes a woman
245699683 times more likely to choose ‘High’ Role group over ‘No
Role’ Group.
9. Factors affecting the level of Family Expenditure as one
dimension of
women decision making: Multinomial Logistic Regression Analysis
9.1 Overall Test of the Relationship:
9.1.1 Model Fitting Information:
Table 9.1
The Model Fitting Information Table (Table 9.1) indicates that,
though there is a relationship
between the dependent variable and the independent variables
since the -2Log Likelihood Ratio
decreases in case of final Model from the model with intercept
only. But this relationship is not
significant, since the significance value of Chi-Square is
greater than 0.05.
9.1.2 Model Fitting Information:
Table 9.2
The significance value of Pearson Chi-Square Test Statistic is
0.026, which is less than 0.05, so
the model does not adequately fit with the given data.
9.2 Strength of the Multinomial Logistic Regression
Relationship:
9.2.1 Likelihood Ratio Test:
Table 9.3
Model Fitting Information
155.313
148.607 6.705 12 .876
Model
Intercept Only
Final
-2 Log
Likelihood
Model
Fitting
Criteria
Chi-Square df Sig.
Likelihood Ratio Tests
Goodness-of-Fit
163.204 130 .026
133.639 130 .396
Pearson
Deviance
Chi-Square df Sig.
Likel ihood Ratio Tests
148.607a .000 0 .
148.712 .105 2 .949
148.793 .186 2 .911
155.221 6.614 8 .579
Ef fect
Intercept
Age
Income
Leveledu
-2 Log
Likelihood of
Reduced
Model
Model Fitting
Criteria
Chi-Square df Sig.
Likelihood Ratio Tests
The chi-square statistic is the dif f erence in -2
log-likelihoods
between the f inal model and a reduced model. The reduced
model is formed by omitting an ef f ect f rom the f inal model.
The
null hypothesis is that all parameters of that ef f ect are
0.
This reduced model is equivalent to the f inal model
because omitting the ef f ect does not increase the
degrees of f reedom.
a.
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
13
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
Likelihood Ratio Tests indicate that, none of the independent
variables are statistically
significant at 5% level of significance.
9.2.2 Parameter estimation:
Table 9.3 Parameter Estimates
Healthcare(a) B Std. Error Wald df Sig. Exp(B)
95% Confidence Interval for Exp(B)
Lower Bound Upper Bound
High Intercept -.236 1.859 .016 1 .899
Age .014 .045 .105 1 .746 1.015 .929 1.108
[Income=1.00] -.213 .794 .072 1 .788 .808 .170 3.829
[Income=2.00] 0(b) . . 0 . . . .
[Leveledu=.00] -1.302 1.561 .696 1 .404 .272 .013 5.793
[Leveledu=1.00] -2.770 1.815 2.328 1 .127 .063 .002 2.199
[Leveledu=2.00] -2.325 1.649 1.989 1 .158 .098 .004 2.476
[Leveledu=3.00] -2.047 1.796 1.298 1 .255 .129 .004 4.365
[Leveledu=4.00] 0(b) . . 0 . . . .
Low Intercept .087 1.723 .003 1 .960
Age .003 .037 .007 1 .932 1.003 .933 1.078
[Income=1.00] -.251 .640 .153 1 .695 .778 .222 2.730
[Income=2.00] 0(b) . . 0 . . . .
[Leveledu=.00] -.761 1.513 .253 1 .615 .467 .024 9.061
[Leveledu=1.00] -1.551 1.596 .944 1 .331 .212 .009 4.841
[Leveledu=2.00] -1.537 1.552 .980 1 .322 .215 .010 4.506
[Leveledu=3.00] -2.020 1.789 1.274 1 .259 .133 .004 4.427
[Leveledu=4.00] 0(b) . . 0 . . . .
a. The reference category is: No Role.
The interpretation of the table 9.3 is that, increases in age of
women make 1.015 times more
likely to have high role in Healthcare over No Role Group.
Similarly, increases in age of women
make 1.003 times more likely to have Low Role in Decision making
of Health care over ‘No
Role’ Group.
Women within the income Group 1 (Monthly Income Rs.
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
14
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
determinants of Women Empowerment and also have tried to find
out the various factors
responsible for taking a role in household Decision making among
women.
We have applied the Multinomial Logistic Regression (MLR) Model
that best explains the
Decision making Progress in household based on the data
collected from the women of Kolkata
Municipal Corporation Slum areas.
In this study we have shown that, level of education of women
have significant effect on
decision in level of savings and family planning (The
significance value of Chi-Square is less
than 0.05), whereas, level of education has no significant
effect on decision in family
expenditure and Healthcare (the significance value of Chi-Square
is greater than 0.05). Age is a
significant determinant in decision of family planning but age
of a woman does not play any
significant role in taking decision in family savings, family
expenditure and Healthcare
expenditure. But income does not play any significant role in
any one of the four dimensions of
family decision making progress within the household among
woman.
References:
1) Siwan Anderson and Mukesh Eswaran. What Determines Female
Autonomy? Evidence from
Bangladesh. Bureau for Research and Economic Analysis of
Development Working Paper No.
101, September 2005.
2) M. Hemanta Meitei. Education or Earning and Access to
Resources Determining Women’s
Autonomy: An Experience Among Women of Manipur.
3) Acharya et al. Determinants of Women's Autonomy in Decision
Making Reproductive Health
2010, 7:15,
http://www.reproductive-health-journal.com/content/7/1/15.
4) Marie Furuta and Sarah Salway. Women’s Position Within the
Household as a Determinant Of
Maternal Health Care Use in Nepal. International Family Planning
Perspectives, 2006, 32(1):17–27.
5) Anu Rammohan and Meliyanni Johar. The Determinants of Married
Women's Autonomy in
Indonesia. Feminist Economics. 16 Oct 2009.
http://www.tandfonline.com/loi/rfec20.
6) Agarwala, Rina. Lynch, Scott M. Refining the Measurement of
Women’s Autonomy: An
International Application of a Multi-dimensional Construct.
Social Forces, Volume 84, Number
4, June 2006, pp. 2077-2098 (Article).
7) Pavalavalli Govindasamy and Anju Malhotra. Women’s Position
and Family Planning in
Egypt .Studies in Family Planning by Population Council. Volume
27, Issue 6 (Nov-Dec 1996),
328-340.
8) Dr.Thresiamma Varghese, Women Empowerment in Oman: A study
based on Women
Empowerment Index. Far East Journal of Psychology and Business
Vol. 2 No 2, February 2011.
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
15
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
9) Rosalinda Coppoletta. Women in Palanpur: An Empirical Study
of the Determinants of
Autonomy in a North Indian Village.20th Jan, 2010.
10)Acharya, M. and L. Bennett (1982) “Women and the Subsistence
Sector: Economic Participation and
Household Decision Making in Nepal”, World Bank Working Paper
no. 526.
11)Eswaran, M. (2002) “The Empowerment of Women, Fertility, and
Child Mortality: Towards a
Theoretical Analysis”, Journal of Population Economics, 15, pp.
433-454.
12)Manser, M. and M. Brown (1980) “Marriage and Household
Decision-Making: A Bargaining
Analysis”, International Economic Review, 21(1), 31-44.
13) Woolley, F.R. (1988), “A Non-Cooperative Model of Family
Decision Making”, TIDI Working
Paper 125, London School of Economics.
14)Thomas, D. (1990) “Intrahousehold Resource Allocation: An
Inferential Approach”, Journal of
Human Resources, 25(4), 635-664.
15)Acharya, M. and Lynn Bennett. 1981. The Status ofWomen in
Nepal Vol II: Field Studies
RuralWomen of Nepal. Kathmandu: Centre for Economic Development
and Administration.
16)Jejeebhoy, S. 1995. Women’s Education, Autonomy and
Reproductive Behaviour. Oxford:
Clarendon Press.
17)Dyson T, Moore M: On kinship structure, female autonomy, and
demographic behaviour in India.
Population and Development Review 1983, 9:35-60.
18)Sathar ZA, Shahnaz K: Women's autonomy in the context of
rural Pakistan. The Pakistan
Development Review 2000, 39:89-110.
19)Kritz MM, Adebusoye PM: Determinants of women's
decision-making authority in Nigeria: the
ethnic dimension. Sociological Forum 1999,Volume 44
:399-424.
20)Jejeebhoy S, Sathar ZA: Women's autonomy in India and
Pakistan: influence of religion and region.
Population and Development Review 2001, 27:687-712.
21)Chanana K: Education attainment, status production and
women'sautonomy: a study of two
generations of Punjabi women in New Delhi. In Girls' Schooling,
Women's Autonomy and Fertility
Change in South Asia, Sage Publications; 1996:107-132.
22)Jayaweera S: Higher education and the economic and social
empowerment of women-the
Asian experience. A Journal of Comparative and International
Education 1997, 27:245-261.
23) Mosiur Rahman, Uzzal K. Karmaker and Abdur R. Mia.
Determinants of Women
Empowerment at Domestic and Non-domestic Issues: Evidence from
Chapai Nawabganj District
in Bangladesh. Dhaulagiri Journal of Sociology and Anthropology
Vol. 3: 143-161.
-
International Journal of Applied Research & Studies ISSN
2278 – 9480
16
iJARS/ Vol. I/ Issue I/Jun-Aug, 2012/158
24)Mercy Tembon and Lucia Fort: Gender Equality, Empowerment and
Economic Growth.
Chapter 11, Page Number 183 to 197.
25)Jin H: A study of rural women's decision making power on
reproduction and fertility. Chinese
Journal of Population Science 1995, 7:241-257.
26)Bloom, Shelah S., David Wypij, and Monica Das Gupta. 2001.
Dimensions of Women’s
Autonomy and the Influence on Maternal Health Care Utilization
in a North Indian City.
Demography 38(1): 67–78.
27)Heaton, Tim B., Tina J. Huntsman, and Dallan F. Flake. 2005.
The Effects of Status on
Women’s Autonomy in Bolivia, Peru, and Nicaragua. Population
Research and Policy
Review 24(3): 283–300.
28)Williams, Linda B. 1990. Development, Demography, and Family
Decision-Making: The
Status of Women in Rural Java. Boulder, CO: Westview Press.
29)UNDP In India: Results from 2010. Empowered Lives. Resilient
Nations.
30)UNDP Human Development Report
2010.www.undp.org.in/sites/default/files/reports_publication/UNDP_Annual_Report_2011.pdf
31)National Family Health Survey (NFHS-3). Ministry of Health
and Family Welfare, Govt. Of
India. Gender Inequality and Women Empowerment in India