4. Select one of the measures of delinquency, and recode it into three categories representing no delinquency (a score of 0), one delinquent act, and two or more delinquent acts. Compute an ordinal logistic regression model using this recoded measure of delinquency as the dependent variable. From the remaining variables in the data file, select at least five variables that you think might have some relationship with this measure of delinquency. (a) Calculate the odds ratio for each coefficient, and explain what each odds ratio means in plain English. (b) Test the parallel slopes assumption. What does this result mean? (Use the Brant test, if available.) (c) If you have access to Stata or use R, estimate a partial proportional odds model using the same dependent and independent variables that takes into account your results from part (b). If you only have access to SPSS, estimate a multinomial logistic regression model using the same dependent and independent variables. Calculate the odds ratio for each coefficient, and explain what each odds ratio means in plain English. (d) How does the substance of the results from the initial ordinal logistic regression model compare to the substance of the results from either the partial proportional odds model or the multinomial logistic regression model? Which gives a more accurate representation of the relationships in the data? Explain why.