Date of Award
Master of Science
James G. Snell
Irving Dubov, B. D. Raskopf
This study was designed to: (1) derive a consumption function for quantities and expenditures for total meat using economic and socio-economic variables which have been assumed to determine the con-sumption pattern of a typical household; (2) determine which variables significantly influence meat consumption and expenditure when income is treated as a continuous variable; and (3) estimate and compare income elasticities of total meat consumption for different income groups. Data were obtained from 215 families in Knox County, Tennessee, for the period from September 4 to November 12, 1967. A step-wise regression program was used to estimate the influence of 29 economic and socio-economic factors on total quantities purchased and total expenditures for meat. This study made use of three regression models holding both total quantity and total expenditure dependent. The models varied according to the number of independent variables included. Model I included only income as the independent variable; Model II included income and consumer units; and Model III included all 29 independent variables. Five different equations were used in each model. They were: (1) simple linear, (2) cube root, (3) intercept change, (4) slope change, and (5) intercept and slope change. Equation 5, which permitted both intercept and slope to change, was accepted as the best fitting in each model for both quantity and expenditure. Model I—equation 5 showed an R2of 0.197 for quantity and 0.377 for expenditure. Income was statistically significant in income groups 1, 2, and 3. Negative income coefficients were shown in the fitted model for income groups 2 and 3 for both dependent variables. Model II—equation 5 showed an R2of 0.462 for quantity and 0.50 for expenditure. Income group 3 was statistically significant for both dependent variables. The fitted model showed negative income coefficients for groups 2 and 3 for quantity and group 3 for expenditure. Consumer units were significant for both dependent variables. Model Ill—equation 5 showed an R2of 0.573 with a standard error of 31.64 for quantity and an R2of 0.569 with a standard error of 21.84 for expenditure. Income was statistically significant in the third income group when quantity was dependent; however, the fitted model showed a negative income coefficient for income group 3 for both quantity and expenditure. The number of consumer units, race, religion, and number of meals served to guests, were statistically significant for both quantity and expenditure. Number of meals eaten out and skilled labor were significant in determining quantity, while one or more years of college for the homemaker, one to four years of high school for the husband, and one or more years of college for the husband were significant in determining expenditure. The income elasticity values were positive for equations 1 through 4, but negative in some income groups for equation 5. The income-quantity elasticities for the best equation. Model III—equation 5, were 0.1505, 0.2574, -0.6857, 0.1861, and 0.3514 for income groups 1 through 5, respectively, while income-expenditure elasticities were 0,1139, 0.3063, -0.4748, 0.5928, and 0.5947, respectively. Negative income elasticities appeared for Model III—equation 5 in income group 3 for both quantity and expenditure.
Craun, James Whitfield, "Analysis of meat consumption and expenditure in Knoxville, Tennessee. " Master's Thesis, University of Tennessee, 1970.