Imputation Methods Of Missing Data For Estimating Entrepreneurship Activity

The research describes entrepreneurial intentions of the Russians. According to “Global Entrepreneurship Monitor (GEM)” Russia has the lowest rates of entrepreneurial intentions among 70 countries participating in the project. Therefore influence of factors of perception of business activity by the Russians is analyzed. GEM database has missing data. This problem leads to overestimation or underestimation of parameters and to mistakes in statistical tests. The purpose of the master thesis is the assessment of entrepreneurial intentions on data which missing values have been imputed. In research several methods of dealing with incomplete data are considered. There are traditional methods, including deletion and single-value imputation, and modern methods. On the sample of Russia’s individual data for 2013 the deletion method, a method of regression imputation, stochastic regression imputation and multiple imputation technique are compared. As a result multiple imputation is revealed as the most effective method. The study provides the assessment of entrepreneurial intentions of Russians using data which missing values have been imputed. Regression estimation reveals positive effect of the knowledge and skills sufficient for entrepreneurial activity and of acquaintance with nascent businessmen on entrepreneurial intentions. Moreover the men who have agreed that the businessman in Russia has the high status, will more possibly choose business as labor activity. Keywords: entrepreneurial intentions, missing data, missing at random, listwise deletion, multiple imputation.