1. Dataset apivmdve.dta Download apivmdve.dtacontains information from the Minneapolis Domestic Violence Experiment (MDVE; Sherman and Berk, ASR 1984Links to an external site.).
   • The City of Minneapolis wanted to find out the value of arresting batterers in domestic violence. The research incorporated three treatments: Arrest, ordering the suspected offender off the premises for 8 hours (separation), and a counseling intervention that might include mediation by the officers called to the scene (advice). Participating Minneapolis police officers were randomly instructed to take one of the three actions when encountering a situation that meets experimental criteria. The primary outcome variable was the reoccurrence of a domestic assault at the same address within 6 months of the original random assignment.
     • Experimental criteria are met (i) when there is probable cause to believe that a cohabitant or spouse had committed misdemeanor assault against a partner in the past 4 hours and (ii) when both suspect and victim are present at the time officers arrived.
     • Cases of life-threatening or severe injury (that is, felony assault) were excluded.
   • As you may see already, strict adherence to the randomization protocol is unrealistic and often inappropriate. During the confrontation, officers may likely deviate from the responses instructed. What if a suspect attempts to assault an officer? What if a victim persistently demanded an arrest? What if there are unexpected, injured parties? What if the officer forgot the assigned action? All these render the treatment delivered non-random.
   • Your job is to find the first stage effect of this research (i.e., the effect of action assignment on action taken). In the dataset:
     • The variable "T_RANDOM" indicates action assignments (i.e., what are police assigned to do): 1 Arrest, 2 Advise, 3 Separate
     • The variable "T_FINAL" indicates action outcomes (i.e., what police ended up doing): 1 Arrest, 2 Advise, 3 Separate, 4 Other
     • The chain reaction in this case would be (action assigned) (action outcome) (reoccurrence of a domestic assault)
     • INSTRUCTIONS:
       • Exclude observations with T_FINAL = 4 from your analysis
       • Define a new variable CODDLE for non-arrests (i.e., advise or separate)
       • show how to create a table that looks like the following:

		OUTCOME
		Arrest	Coddle
ASSIGNMENT	Arrest	?	?
Coddle	?	?

• and find the first stage effect following what we did in the class notes. solve using stata