If you feel it was random variation then you have to solve for at least 3 things.
1. A 20% increase in murder over 2 years are reasonably within the range of deviations from a fitted trend. That sounds like an outlier to me. An underlying reason is much more likely.
By 20%, you are a referring to change in the homicide rate from 4.5/100,000 (as seen in the graph Zombie gave us in post #67 in this thread) to 5.3/100,000 (for 2014-2016 (which is not quite 18%)? So, a total change of 0.8/100,000. Looking at Zombie's first chart, there's a larger total change from 1980-1982 and from 1989-1991, even though statistically, the whole period of 1972-1996 looks like random variations.
2. The questioned Ferguson Effect does in fact not exist. I find the arguments against both forms of manifestation to be flimsy at best.
My understanding of the Ferguson Effect is the claim that police are reluctant to leave their vehicles. To my understanding, most murders are done when police are not in the vincinity at all. Could you please explain how the Ferguson Effect could have any relation to the crime rate? Also, what did Trump do to end the Ferguson Effect? Did he pardon all the police officers who shot black people, or something?
Also, if it is the Ferguson Effect, why is it stronger in smaller cities than in medium cities?
https://ucr.fbi.gov/crime-in-the-u.s/2016/crime-in-the-u.s.-2016/tables/table-10
Citiy size and percent change from 2015 to 2016:
1,000,000 and over (Group I subset) 20.3
500,000 to 999,999 (Group I subset) 6.5
250,000 to 499,999 (Group I subset) 11.1
GROUP II (100,000 to 249,999) 6.3
GROUP III (50,000 to 99,999) 0.7
GROUP IV (25,000 to 49,999) 2.4
GROUP V (10,000 to 24,999) 3.4
GROUP VI (under 10,000) 8.4
Does the Ferguson Effect lose potency in cities between 50K and 100K, and then regain strength in cities under 10K? Same question for between 500K and 1000K.
3. That Obama's speech and his policies had no effect. The latter here is kind of dumb argument to make to begin with. If Obama's policies have no effect then why was he making them?
Which policies by Obama do you think had an effect on the murder rate? The ACA? Net neutrality?
Somehow confirm all three and I could see believing the 20% increase truly was random variation within a trend. That seems like quite the challenge, one criminologists havent been able to square up.
How would I confirm a lack of causation between the Ferguson Effect (to the degree it exists at all) and the homicide rate? How would I confirm a lack of causation between Obama policies (assuming you can name them) and the homicide rate? What would such hypothetical data look like?