Dependent and Independent Variables - Essay Example

Answers to Select Questions Exam 2008 Part C (Q5) Answer on (a The following variables are statistically significant in its relationship with the dependent variable “number of CD albums”: income, interest in music, number of MP3s, and age. The chosen cut-off for significance is p-value=0.10. Those with significant magnitude in its relationship with the dependent variable are variables “income” and “interest in music”. Despite the large magnitude of the coefficient, the variable gender is excluded in the list of variables considered to have a significant magnitude of relationship with the dependent variable because the variable gender is statistically insignificant in its relationship with the dependent variable. The independent variables negatively correlated with the dependent variable: number of MP3s, and age. Answer on (b) The model is a poor model in explaining CD sales because the adjusted R2 is only 0.27. Nevertheless, the F-statistics is significant with a p-value of 0.001. This means that we can reject the null hypothesis that all coefficients of the independent variables of the model are simultaneously equal to zero. Thus, unless a better model is found, the model may be used on a provisional basis. Answer on (c) For inferences to be appropriate, several assumptions are assumed to apply in multiple regressions (Gujarati 2004, p. 203), there are several assumptions involved in using linear regression. The most important of these are as follows: zero mean of residuals, no serial errors among residuals, homocedasticity, zero covariance between independent variable and residuals, absence of specification bias or the model is correctly specified, and no exact collinearity between the independent variables (Gujarati 2004, p. 203 and 204). Answer on (d) No. However, I would conclude that the number of MP3s bought by a person is correlated with the number of CD albums a person possessed. 2. Exam 2009 Part C (Q5) Answer on (a): Except for gender and age, all variables are able to provide a significant explanation on the value of the dependent variable. The variable “task control,” however, is able to explain at a significant level the dependent variable in Netherlands but not in the US. Except for tenure, the background variables are unable to provide a significant marginal contribution in explaining the value of the dependent variable. The variables task control environment control, and tenure consistently manifest a negative relationship and all are significantly related at p-value=0.10 with the dependent variable perceived work stress. Answer on (b): That greater job control can significantly affect the perceived work stress is adequately supported in both regressions. However, I am uncomfortable that the both regression model can only explain not more than 25% of the variation in the dependent variable. This indicates that there can be other variables that explain the dependent variable more significantly. We do not know how the independent variables would express their significance when placed in a regression model that includes the significant variables that may explain the dependent variable better. In the experience of this writer, the value of the coefficients as well as their significance would vary depending on the independent variables covered by a regression function. Further, the assumption in a regression model is that we have the correct model. Thus, all regression functions and their coefficients and significance will be valid only if we have the correct model. Answer on (c): I do not consider the models good in explaining the dependent variable. I have seen better models that can explain their dependent variables better. As mentioned, the adjusted R-squares of the two regression functions indicate that the independent variables included in the regression function can only explain not more then 25% of the variation in the dependent variable. Answer on (d): Yes, the models or results of Table 1 would allow me to say or conclude that the lack of control can cause perception of work stress. However, I will be very careful in my statements and conclusion because I believe we do not have a good model. The models suggest the said relationship (that task control can reduce stress) but we have poor models to lean on in Table 1. Thus, if I am the researcher, I would continue to search for better models. 3. Exam 2005 Part C (Q5) Answer on (a): All independent variables except experience significantly and positively affect the dependent variable. Based on the table, it is variable education that has the largest magnitude in significantly affecting the dependent variable. Answer on (b): Based on the regression model, each year of education contributes 0.575 pound to the annual gross salary of employees. We can say that years of education contribute positively to the magnitude of employees’ gross annual salary. Answer on (c): Based on the adjusted R-square value, the model can explain 49.1% of the variation in the dependent variable. I believe a better model can be constructed but a researcher can use the model to argue that his view that education positively contributes to employee annual gross salary has a good basis. Answer on (d): In my opinion, it is highly possible that education has not been sufficiently operationalized. Perhaps a better model can be constructed if we use average grade at graduation time. Further, the type of education can also be playing a role as well as the school from which the employee graduated. 4. Exam 2006 Part A (Q1) Positivist philosophy and research methodology for work in the social sciences have their origins in the works of Comte de Saint-Simon and August Comte (Routeledge 1998, p. 6709). Positivism has emphasized and supported the quantitative method (Routeledge 1998, p. 6709). Moreover, positivism in the social sciences asserts that the method for the natural and social phenomena research are the same (Routeledge 1998, p. 6709). Thus, positivist research methodology advocates distance between the researcher and researched, arguing that the distance is necessary for scientific objectivity. However, recent developments in social science research have established the importance of qualitative research methodology and the importance as well of ground-theory, immersion with the subject, participatory research, participant observation, and ethnomethodology advocating close contact and even living with the subject of study. Further, while positivism emphasise on the observations gathered from man’s senses, post-positivist research methods equally rely on man’s to use intuition, interpretation, and insights to advance knowledge. Thus, that positivism is inadequate for social science research has been judged de facto by the developments in social science research. Meanwhile, Kuhn’s thesis of a structure in scientific revolution “argued that sociological and psychological factors played an important role in the growth of scientific knowledge, generally mediated through the scientific community” (Narasimhan 1997, p. 4). In particular, Kuhn asserted, “the essential feature of scientific growth is that it is based more on sociological and psychological factors of research rather than any ‘logic of discovery’” (Narasimhan 1997, p. 4). Clearly, Kuhn’s thesis is a clear critique of positivism as the latter argues that pure logic governs scientific inquiry. References Gujarati, D., 2004. Basic econometrics. New York: McGraw Hill. Narasimhan, M., 1997. The Popper-Kuhn Debate on the Nature of Scientific Development. Resonance (Februar6y), 4. Routeledge, 1998. Routeledge Encyclopedia of Philosophy. London and New York: Routeledge. Read More