In this post, we zoom in to the Leeward and central parts of the island to look at violent crime between 2022 and 2024.
In a previous post, we sought to compare the violent crime counts from the Leeward side of O‘ahu (also known as District 8) to a somewhat comparable district–with a similar distribution of urban, suburban, and rural populations–the Windward side (also known as District 4).
In this post, we are going to replicate some of those analyses while comparing the Leeward side to its neighboring District 3, the greater Pearl City area and center of the island.
As we covered in our earlier post, place-based comparisons can be tricky. D3 is a somewhat larger population, estimated to be about 160,000 (larger than D8’s estimated 130,000). It also covers a mix of urban, suburban, and rural populations, although it has more sizeable population centers than the Leeward side. One big difference is a greater concentration of military personnel and centers in the Pearl City area. But one advantage of the comparison is the two districts are neighbors, so there will be some similarities we can’t otherwise measure. Again, the comparison won’t be perfect, which is why we’re interested in doing more than one set of comparisons. If we see similar trends across the Windward and Pearl City comparisons, that might tell us something; if the trends are dissimilar, then we might have too much noise in the data to be able to learn something meaningful.
The data analyzed in this blog post were collected by Emma Burnkrant and Jesus Romero.
Crime Counts in District 3 v. District 8
As we saw with the District 4-District 8 comparisons, there are more instances of violent crime in District 8 (the Leeward side) than in its neighbor District 3 (the island center and Pearl City area), even though District 8 has a somewhat smaller population. This finding holds across the four violent crime types on the HPD dashboard. The difference is also statistically significant for all four except homicide, whether or not we include year-based controls (fixed effects). As is clear from the graphs below, the biggest differences are in sex offenses and aggravated assault, while the robbery and homicide counts are more similar. The difference in homicide counts is just barely not significant, though.
Can We See Evidence of Police-Led Crime Reductions on the West Side?
The biggest question for us in our last comparison post was whether the increase in police presence since August led to a decline in violent crime, at least one large enough to be visible in the data. (Be sure to reread that post to catch all of the caveats about this analysis and some of the technical details.) In theory, if this is a good analysis with good data, we should see similar results from the last analysis. But one limitation to this analysis is it strongly depends on changes in the comparison district. If anything happened in District 4 at the same time as changes in District 8, it might dampen our ability to measure any real effects, if they existed. The same will be true when we add in District 3 in place of District 4. By comparing to another district, we’re hopefully limiting the impact of that possibility.
In the last analysis, we did not see strong evidence for a crime-reduction effect of the greater police presence at the end of last year. There was a hint that robberies might have been affected, but the data were just not quite significant. If we see that trend again, it might indicate something real. If we don’t, it might indicate it was just random chance or not a strong enough difference.
So what do we see?
For aggravated assault, we see no significant difference in either our regular regression or our fixed-effects regression suggesting the added police (or any other changes that were co-occuring) had no (measurable) effect on aggravated assaults.
For homicide, we also see no significant difference in either analysis.
For robbery, we also see no significant difference in either analysis.
For sex offenses, we also see no significant difference in either analysis.
Well, that’s disappointing. We now have two sets of comparisons (Windward and Pearl City/Center Island) and no evidence that the additional policing helped bring down the number of crimes in any of the four categories.
Next steps: When we learn whether policing was scaled back in January as some reporting suggested it might, we will be able to do follow-up analyses since we can see any trends associated with a descaling of police. The crime lab team has also been discussing adding analyses of property crimes so we can also repeat the analyses to investigate what, if any, effect policing practices had on property crime during the period of increased police presence.
The Hawaii Crime Lab 