Important terms
“The 'Input', also known as 'corrected mean', is a global measure of rate of rule application and can be thought of as an 'overall indication of the strength of the rule' or 'an average frequency of occurrence of the application value of the dependent variable'” (Tagliamonte 2006: 141)
“It measures the goodness of fit of the analysis. Figures closer to 0 represent better models than those further removed from 0” (Tagliamonte 2006: 156)
1) Conduct a binominal one-step analysis; 2) detect files with the highest error and remove them in the cell file 3) re-run the data and observe the manner in which the log likelihood moves closer to 0. The higher the number rises, the better the fit of the model to the data will be. However, keep in mind that this exercise is solely meant to clarify how the log likelihood measurement works in principle, obviously a reliable analysis would not be conducted in this manner.
“The 'iterations' show you 'an account of the program's progress in finding the “maximum likelihood” estimation of the factor weights to a certain degree of accuracy, at which point “convergence” is indicated.” (Tagliamonte 2006: 141)
Created with the Personal Edition of HelpNDoc: Free CHM Help documentation generator