Interpreting the Results of a Regression Analysis - Assignment Example

In this case, the cumulative probability is 0.7049. This implies that the coefficient is at least -20 (100 - 70.5) = 29.5% of the time. We now test the probability that the obtained t-score with 11982 degrees of freedom is greater than 0.5385. Using the t-tables, we obtain the one-tailed p-value as 0.2951. Since this value is greater than the default level of significance (0.05) for the test, we fail to reject the null hypothesis. We establish that the value of the coefficient is not greater than -20 at the 5% level of significance.

From the equations, we now determine whether students whose home language is not English are more or less likely to read fiction compared to those who read fiction at home. Firstly, we realize that the introduction of the variable “fiction” in the second model improves the value of R2 from 0.02 to 0.204. This indicates that the variable alone accounts for (0.204 – 0.02) = 18.4% of the variability in the reading scores. We take the “odds ratio” approach. We apply the formula: Again, nearly everyone in the sample who does not speak English at home reads fiction.

The ratio of fiction readers from the two categorizations is 1:1, implying neither of the groups commands higher fiction reading likelihoods than the other. In the variable “fiction” were to be split into constituent components based on the Likert scale rankings earlier used, such that “fict5” indicates “I like reading fiction very much” and so on, we would have to change the variable responses from scale to dummies. We assume that if an individual who says responding to “fict5” mentions that they “do not like reading fiction very much” will be accorded 0 for the response, and 1 for “I like reading fiction very much”.

The variable “fict3” is the transition between responses that affirm or disaffirm the proposal/ question. This is highlighted by the fact that the scale used has 5 points.