DETERMINATION OF APPROPRIATE STATISTICAL TECHNIQUES FOR COLLECTION AND ANALYSES OF SOCIO-DEMOGRAPHIC AND FOOD CONSUMPTION DATA- A CASE STUDY

Abstract

This study “Determination of Appropriate Statistical Techniques for Collection and Analyses of Socio-demographic and Food Consumption Data- A Case Study” primarily aimed to determine a unique sampling scheme for collecting and analyzing of food consumption data from a diverse population, impact of improper analytical tools on the estimates of population characteristics; socio-demographic determinants influencing food demand and consumption behavior of the rural people in regions similar to study area. The formula for determining the sample size is proposed as a weighted average of sample sizes of individual variables where the weights are proportional to variance given by Sample size=n=\W,n,, where W,=S,?/>\S;?;_S,’= population variance of i® variable and n,= sample size for i" variable obtained using usual formula. Sample size using the proposed formula was estimated as 402 for simple random sampling, 398 for stratified random sampling and 405 for two-stage random sampling with an average sample size of 400 for all three sampling methods. Stratified random sampling appears to be superior over simple random sampling and two-stage random sampling from a diverse population in a compact and small area like a union in terms of closeness of the estimates to the corresponding population value and precision of the estimates. For sampling from an unknown population, before conducting actual sample survey a 1% sample from the population from each stratum or 10 sampling units from each stratum would be good enough to use for determining sample size for conducting sample survey using stratified random sampling technique. Data gathered using a particular sampling technique but analyzed using the analytical tools of other sampling technique is a completely wrong practice and produces distorted results. To study the food demand and consumption behavior a sample of size 400 households was drawn from the enumerated population of 7001 households employing stratified random sampling method. Results of logistic regression analysis identified that household income had a positive impact and household size had a negative impact on food security of the households. The food insecurity index may be defined as 4/5 of the mean per capita food expenditure of all households that yielded about 64.2% of the households as food secure and 35.8% as food insecure in study area. Households with more male members appeared to be more food secured than the households with more female members and household food security increased with the increase of SES. Household budget allocation on different food items differed from food item to food item but for particular food item, the budget share was almost identical among the religious/ ethnic groups. Engel models and Linear Approximate Almost Ideal Demand System (LA/AIDS) model were used to study the demand and consumption behavior of different food items by rural households. Budget share appeared to be highly responsive to the expenditure on food in each form of Engel models and LA/AIDS model for different food items under consideration. Households were responsive to prices in adjusting their consumption of the food items. Socio-demographic variables such as household size, number of wage earners in household, religion, education, socio-economic status and occupation of household head significantly impacted the food budget share. Engel models classified cereals as necessary food items and, roots and pulses, vegetables, rich foods, milk and sugar, oil and spices as luxury food items with one misclassification of oil and spices as luxury food item which was correctly classified as necessity food item by LA/AIDS model. Vegetables is substitute of the food groups rich foods and, milk and sugar. Roots and pulses and, milk and sugar are substitute of rich foods whereas, cereals is complementary to vegetables, roots and pulses, rich foods and, milk and sugar. LA/AIDS model is more informative and capable to explain differential effect of price and sociodemographic factors on the demand and consumption behavior of different food items by the households and classification of food items as necessary and luxury compared to Engel models. The implication of price sensitivity of household on food consumption such as own and cross-price elasticity is that the policy makers and producers will understand precisely how consumer will respond if there is a change in the price of any or some of the food items and help them to formulate the pricing policy of different food items.