<p> It is the purpose of this study to determine empirically the contribution of a set of variables to the observed variation in income per hectare on 100 peasant farms in Ecuador. Specific consideration will be given to the relationship between farm size and farm productivity.</p> <p> Twenty-four characteristics were chosen to describe the structure of peasant agriculture. An examination of the distribution of each characteristic over the 100 farms illustrated the heterogeneous nature of the farm sample. Regression analysis applied to the sample at the aggregate level yielded poor results. Large amounts of variability in the dependent variable remained unexplained, and the standard error of the estimate was relatively large in comparison to the observed mean value of the dependent variable. The model was improved when the sample was disaggregated by region, although the standard error remained high which greatly weakened the potential of the model as a predictor equation.</p> <p> To increase the power of the regression model, and to more effectively
analyse the significance of the set of variables to variation in farm income, it was decided to type the farms using factor analysis.</p> <p> Principal axes factor analysis was performed on the matrix of correlation coefficients for the twenty-four standardized characteristics. The factors were then rotated using varimax rotation to obtain a more simplified loading matrix. Eight primary factors were produced which together accounted for 62.25 percent of the variance in the original matrix. Ward's hierarchical grouping algorithm was then applied to the matrix of factor scores for the principal eight factors, and a classification containing fourteen types of farming activity was produced.</p> <p> The relationship between income and farm size was then reconsidered by farming type. There was a slight improvement in the power of the model applied to farm type although the amount of explained variability remained small. Simple regression of income per hectare on farm size, then, failed to explain a large proportion of the variance in the dependent variable even when the sample was considered by farming type.</p> <p> In order to reduce the measure of 'non-explained' variability in
the dependent variable, and to increase the potential of the regression model as a predictor equation, income per hectare was regressed on the rotated factors. Multiple step-wise regression was performed on (a) the complete sample, (b) the sample disaggregated by region, and (c) two major farming types. The multiple step-wise regression model greatly
increased the amount of explained variability in the dependent variable and indicated the significance of the contribution of each independent variable to variation in income per hectare on the farms.
The study is presented in five parts:
Chapter 1 introduces the problem to be analysed.
Chapter II presents the data base, and a simple linear regression analysis examines the relationship between income and farm size on the 100 farms. The results of the regression performed on the aggregate level are poor. The analysis is then repeated on the sample disaggregated by region. The power of the model is greatly increased when the sample is divided into regional subsets, but large amounts of variability are left unexplained and the standard error of the estimate remains high.</p> <p> Chapter III groups the 100 farms according to a typology based on an analysis of the structural and economic organization of the farming unit. A correlation analysis is performed on the twenty-four characteristics and simple correlations between the data are considered. Factor analysis is then performed on the matrix of correlations, and the major dimensions of variation in the data are enumerated. Finally, a grouping algorithm is applied to the matrix of factor scores on the principal factors, and a classification of the farms is produced.</p> <p> Chapter IV reconsiders the relationship between income and farm size by farming type. Multiple step-wise regression then examines the contribution of a set of variables to variation in income per hectare.</p> <p> Chapter V summarizes the merits and weakness in the methodological approach of the study and enumerates the major findings of the analysis.</p> / Thesis / Master of Arts (MA)
Identifer | oai:union.ndltd.org:mcmaster.ca/oai:macsphere.mcmaster.ca:11375/19426 |
Date | 08 1900 |
Creators | McGregor, Elizabeth Ann |
Contributors | Wood, H. A., Geography |
Source Sets | McMaster University |
Language | en_US |
Detected Language | English |
Type | Thesis |
Page generated in 0.0018 seconds