class.
X3 = Third independent variable or Family size in number.
X4 = Fourth independent variable or Earning member in
number.
X5 = Fifth independent variable or House rent in taka.
Rationale:
The theoretical relationship between the dependent variable (Y) and each of the independent variables (X1, X2, X3, X4, and X5) is easy to understand using simple theory and logical deduction.
The Expense for Consumption of food by the garment workers is dependent on the total family income. They are positively related because as the total family income increases workers will be able to spend more for their food. Again if the income decreases they will cut down their induced food consumption. Thus we get
dY/dX1 > 0
Education changes human nature and life style. Higher educated are aware about their health and willing to buy rich food than the others. So they spend more. Thus the relationship between food expenditure and education is positive. This relationship can be summarized in the following equation:
dY/dX2 > 0
Food expense are also affected by the number of family member. If the family member increases the food expenditure per member will decrease, ceteris paribus. The more family member will reduce the food expenditure. Thus the relationship is negative. We can write this relationship by following equation:
dY/dX3 < 0
On the other hand if the earning member of entire family increases, total food consumption will also change. In general this relationship is positive because the additional income will increase the average food expenditure of that family. Thus we get
dY/dX4 > 0
The relationship between food expense per head and house rent is difficult to understand at first, but in the end make sense. Workers have to pay a large amount of their income for the accommodations. Due to this they have to cut down their other expenditure. Sometime they reduce their food budget to live vicinity of their working place. The relationship between these two is expected to be negative. Thus we get
dY/dX5 < 0
So the regression equation would be –
Y = b1 + b2*X1 + b3*X2 - b4*X3 + b5*X4 - b6*X5 + ei
Where,
Ŷi = Expenditure on food per month by the female worker.
b0 = Y intercept .
b1 = Net change in Y for each unit change in X1,
holding other independent variables constant.
b2 = Net change in Y for each unit change in X2,
holding other independent variables constant.
b3 = Net change in Y for each unit change in X3,
holding other independent variables constant.
b4 = Net change in Y for each unit change in X4,
holding other independent variables constant
b5 = Net change in Y for each unit change in X5,
holding other independent variables constant
Ei = Error term
Data Sources and Collection Procedures
Sources of Data:
We administrated a structured questionnaire (Table-, Page ) to the garments female workers. Successful completion of the survey crucially depended on the cooperation of the owners of the garment industries. We made appointment with the managing director who could be conducted after several phone calls and some times after 5 p.m. only or on Friday. In some industries it took lot of persuasion to convince them.
Methods:
The data has been conducted by directly questioning the female worker randomly at the lunch lime. This also helped us to eliminate possible bias that could have been there, if they were questioned on the production floor. As the interview was conducted within the factory premises and adequate care was taken to ensure the interview was confidential and none of the management personal was around during the interview.
Problem Area
In the initial collection of data we included a variable i.e. family member. But, in this variable we only included members who are living in Dhaka. But, later we felt that we should actually include members in the workers’ village. But, we could not arrange any other appointment so we had to stick with the earlier data. By not including the village members we are actually overlooking a very important variable so our regression results could have a very high residual, in other way the R2 might be small.
Regression Table
Table:1
Table: 2 White Heteroskedasticity Test:
Analysis of the Regression Equation
By including the variables for our statistical model, we can create a regression equation for the study. This is stated as follows: (figures within the bracket are Std. Error)
Ŷi = 504.4228 + 0.5915 X1 + 27.48818 X2 – 114.584 X3 + 50.268 X4 + 0.099 X5
(27.089) (0.01734) (5.4124) (17.4545) (25.8696) (0.05017)
The coefficient of determination (R-square) is a measure of the goodness of fit of the estimated regression equation. In the table 1, the R-square value 0.82722 shows the 82.72% variation in the dependent variable Y (food expenditure per head) is explained by the estimated regression equation.
The reliability of the data set is shown by the value in the standard error column. This value shows how well the values in the model relate to each other.
In this equation the Y intercept (bO) is 504.4728 .It shows if all variables are zero there will be an autonomous expenditure on food about 504.47 taka.
Positive slope coefficient for variable X1 is 0.05916. It shows if the total family income increases by 1 Taka the partial expenditure on food per head will increase by 0.05915 taka & vice versa.
There is a great effect on food consumption by the Education. Usually higher educated and experienced labors get higher payment. Positive slope coefficient for education level 27.49 represents the partial expenditure on food per labor will increase by 27.49 taka if the education level increase by 1 class (grade) & vice versa.
The slope coefficient for the variable family member is negative .It refers if the number of family member in a family increase by one the partial expenditure on food per head will decrease by 114.58 taka & vice versa.
On the other hand if the earning member of a family increase by one the partial expenditure on food will increase by 50.279 taka & vice versa which is represented by positive slope coefficient 50.279.
The slope coefficient for house rent is positive. The value 0.0999 shows if the house render expense increase by one the expenditure on food per head also increase by 0.0999 taka & vice versa. This result contradicts our earlier assumption. This might be because the workers decide on the housing depending on the income level. So, the housing is actually representing the income level of the worker. So, there is a positive relationship.
Heteroscedasticity
The white Heteroscedasticity test shows that our data do not have heteroscedasticity problem as p>.05, which is logical as we are dealing with people with the same level of income, their expenditure pattern should not vary much.
Durbin Watson Test
The DW test statistic is 1.6327 and DL = 1.230 and DU = 1.786. So, the test is inconclusive, but we can see from the residual graph in Appendix that there is no serial-correlation in our data.
Correlation Coefficient Matrix
Table 2: Correlation Matrix of Variables.
So, we can see from the correlation matrix that there is a high correlation between total family income and family member, total family income and house rent, family member and house rent. But, these variables are important so, we would not exclude these variables.
Conclusion
This study proposed to develop a statistical model on Money Spent on Food by the Female Garment Worker. Finally we decided that all the five variables that started with are important. In our ultimate model we decided that expenditure on food of garments workers depend on Family income, Education Level, family size, Earning member, House rent.
Bibliography
-
Afsar, R .1998. “ A case Study of the Gender Dimension of Labor Migration in the formal Manufacturing Sector Of Dhaka City”. Paper prepared for Center for Policy Dialogue and UNRISD Project on Women’s Employment in Bangladesh.
-
B U P 1993: ‘A Study on Female Garment Worker In Bangladesh’, Bangladesh Unnayan Parishad, Dhaka.
-
Rausan Jahan .2001. ‘ Women Workers in the Garment Industry ‘, Bangladesh journal of Political Economy, Vol.9,no.3 B E A.
Appendix
Table A: Data For Regression Analysis
Table C: Questionnaire for the Study