Is History a Threat? (Comment)

Erick Hobsbawm (d.2012) was a Marxist historian, a professor of history at Birkbeck College, University of London, where he taught till the end of his life. Birkbeck is an evening college where most of the students belong to the working class and continue with studies to improve their educational qualifications. Conscious of the condition of students who attended his lectures after a hard day of physical labour, he devised lectures on the history of Europe to create emancipatory thought. His four books use history as a tool to comprehend present-day Europe. They are titled The Age of Capital, The Age of Industrial Revolution, The Age of Revolution and The Age of Extreme. Besides these four books, he has published a number of papers and essays that attempt to build a new historical consciousness. There are other progressive historians besides Hobsbawm who believe that history can be liberating if it is not written in the service of the rulers of the time. It is also not a static and depressing account of the past but is dynamic and constantly changing. While history tells the story of unjust rulers and the institutions they created, it also reveals how the rulers and their unjust institutions meet their end. The reign of absolute kingship where the king was considered the Viceregal God, brutal military dictatorships, and exploitative economic systems that lasted for centuries with the aim of torturing and subduing the common working people were all destroyed over time, especially when people joined forces and resisted them to gradually establish a just order.

Determination of Sample Size and Comparison of Estimation Methods in Multilevel Models

The main purpose of this study was to examine the impact of sample size on multilevel model estimates and their standard errors under different methods of estimations. Three different studies were designed to achieve the objectives of the study. In study 1, two level binary logistic random intercept and random slope regression model was used. The performance of two estimation methods was observed under varying conditions of the design factors i.e the number of groups, group sizes and intraclass correlation (ICC). Maximum Likelihood (ML) with adaptive quadrature and Penalized Quasi-likelihood (PQL) methods of estimation were used in study 1. Similarly, three categories and five categories two level ordinal logistic random intercept and random slope regression models were used in study 2. The performance of ML and PQL methods of estimation was observed under varying conditions of the design factors i.e., the number of groups, group sizes, ICC and distribution of category responses. Moreover, a two level random intercept and random slope linear regression model was used in study 3. The performance of Restricted Maximum Likelihood Method (REML) and Bootstrap by means of Minimum Norm Quadratic Unbiased Estimators (MINQUE) was observed under varying conditions of the design factors. In all the three studies relative parameter bias and 95% confidence interval coverage rates were used to assess accuracy and precision of estimates and their standard errors. Further, empirical power rates were also computed in study 1 and study 2.