Table 1.
Reported crimes in Santa Monica, CA, 2006–2019, by crime type.
Fig 1.
Reported crimes in Santa Monica, CA, 2006–2019, by crime type.
Reclassified crimes are as follows: receiving/possessing stolen goods, forgery, narcotics possession, fraud, and larceny. All others do not fall under the provisions of Prop. 47. Data is from Table 1.
Fig 2.
Reported crimes in Santa Monica, CA, 2006–2019, by year.
The gray area indicates the years during which Prop. 47 has been enforced. Prop. 47 went into effect during November, 2014, the calendar year which had the fewest reported crimes. The Santa Monica Police Department launched new community policing initiatives in September, 2018. We note a sharp decline in reported crime for 2019.
Fig 3.
Density of average annual reported larceny incidents in Santa Monica, CA, before Prop. 47 (left) and after Prop. 47 (right).
The right-hand panel displays more intense coloring in the central contours, reflecting higher incidence of larceny after implementation of Prop. 47. Though these maps are restricted to larceny, all crime types are more prevalent Downtown compared to residential areas. Created using OpenStreetMaps.
Fig 4.
Reported crimes per month in Santa Monica, CA, before and after implementation of Prop. 47 in November, 2014.
For Prop. 47 crimes (left panel) the monthly average before implementation of the new law was 281.4 crimes, whereas after November, 2014 the average was 322.9 crimes per month. A Welch’s t-test shows that this 14.7% increase is statistically significant (p < 0.05). The same computations applied to non-Prop. 47 crimes (right panel) show that, to the contrary, non-reclassified crimes decrease by 9.2%, from 443.2 to 402.4 per month, after passage of the initiative. A Welch’s t-test shows that this 9.2% decrease is statistically significant (p < 0.05).
Fig 5.
Additive Seasonal and Trend decomposition using Loess (STL) for reported monthly crimes in Santa Monica, CA, 2006–2019.
The time series of reported crimes per month Y(t) is separated into trend T(t), seasonality S(t), and remainder R(t) components. We perform this decomposition for Prop. 47 crimes (left) and non-Prop. 47 crimes (right). The colored regions in the plot of the trend T(t) indicate the time period following the enactment of Prop. 47. An increase in the trend emerges for the Prop. 47 crimes towards the end of 2014, but not for non-Prop. 47 crimes. For these results, the STL smoothing window is wtrend = 19.
Fig 6.
Change-point analysis of reported crime in Santa Monica, CA, 2006–2019.
We perform our change-point analysis on the slope M(t) of the crime trend T(t) for Prop. 47 crimes (left) and non-Prop. 47 crimes (right). The top panels are the trends T(t) derived from raw monthly data Y(t) using STL decomposition with smoothing window wtrend = 19 months. The lower panels visualize the slopes M(t), the resulting change-points, and their 95% confidence intervals. Base parameters used in the mosum change-point detection algorithm appear in the main text. The lower left panel for reclassified crimes shows a change-point in June 2014 (dotted line) with a 95% confidence interval between September 2013 and April 2015, which includes November, 2014, when Prop. 47 went into effect. This result is fairly robust to changes in the algorithm’s parameters. For non-reclassified crimes in the right panel, using our base algorithm parameters, the change-point is October 2006 (dotted line) with a 95% confidence interval between September 2006 and November 2006. However, this non-Prop. 47 change-point is strongly dependent on the algorithm parameters. No time-frame emerges that is robust to parameter changes.
Fig 7.
Segmented regression and breakpoint analysis of reported crime in Santa Monica, CA, 2006–2019.
According to Davies’ test, the preferred model has two breakpoints. Top and bottom panels show segmented regression for, respectively, the original time series Y(t) and the smoothed trend T(t) with wtrend = 19 months. Left panels are for Prop. 47 crimes and right panels are for non-Prop. 47 crimes. Breakpoints appear as dots and the respective 95% confidence intervals are marked by bars. For Prop. 47 crimes, the first breakpoint (red) is November, 2014 for Y(t) and October, 2014 for T(t). Values of 1 < wtrend < 19 months preserve the T(t) breakpoint. The second breakpoint (black) is September, 2018 for Y(t), and October, 2018 for T(t). The first breakpoint signals a transition to higher crime for both Y(t) and T(t), the second one a decrease. For the non-Prop. 47 crimes, breakpoint dates are highly unstable and sensitive to changes in wtrend.
Fig 8.
Neighborhoods of Santa Monica, CA, and their average annual reported crimes.
(Top) The eight neighborhoods of the city of Santa Monica. To the north of Santa Monica is Pacific Palisades, to the south is Venice, and to the east is West Los Angeles, which are all part of the city of Los Angeles proper. To the west is the Pacific Ocean. Created using OpenStreetMaps. (Bottom): Average annual reported crimes in each neighborhood for Prop. 47 crimes (top) and non-Prop. 47 crimes (bottom). Gray bars indicate the average prior to implementation of the new law in November, 2014. The color-coded bars represent the averages after Prop. 47 came into effect.
Fig 9.
Neighborhood-level analysis of reported crime in Santa Monica, CA, 2006–2019.
Normalized histograms and densities of Prop. 47 (top) and non-Prop. 47 (bottom) monthly crimes in the eight Santa Monica neighborhoods. The gray bars and curves represent the time period before November, 2014, and colored ones represent the period after. Table 2 provides summary statistics.
Table 2.
Statistical tests on reported crime in neighborhoods of Santa Monica, CA, 2006–2019, before and after the passage of Prop. 47.
Fig 10.
Map of the Metro Expo Line light rail extension to Santa Monica, CA.
Operations began on May 20, 2016 at seven new stations, of which four are located within Santa Monica city borders. They are Expo/Bundy, 26th Street/Bergamot, 17th Street/Santa Monica College, and Downtown Santa Monica. The other three stations (Palms, Westwood/Rancho Park, Expo/Sepulveda) are within neighboring Culver City. Prior to May 20, 2016 the, Expo Line’s terminus was at the Culver City train stop. Picture courtesy of the Los Angeles Metro.
Fig 11.
Reported crime near Expo Line light rail stations in Santa Monica, CA, 2006–2019, before and after opening of the light rail extension.
These normalized histograms (bars) and densities (curves) provide monthly reported crime counts in a 450 meter radius around locations at which new stations opened in 2016. Gray represents the time period through May 2016, prior to this opening. Colored data is from the time period afterwards. Changes to the mean are statistically significant for the Downtown Santa Monica station, the 17th Street/Santa Monica College station, and the Expo/Bundy station. They are not significant for the 26th Street/Bergamot station. See Table 3.
Table 3.
Statistical tests on reported crime in the vicinity of light rail station locations in Santa Monica, CA, 2006–2019, before and after opening of the Expo Line light rail extension.
Fig 12.
Reported crime near Expo Line light rail stations in Santa Monica, CA, during three successive time periods spanning 2006–2019.
These normalized histograms (bars) and densities (curves) provide monthly reported crime counts in a 450 meter radius around locations at which new stations opened in 2016. The time periods are: (1) January, 2006—October, 2014, before implementation of Prop. 47; (2) November, 2014—May, 2016, between the implementation of Prop. 47 and the opening of the Expo Line; and (3) June, 2016—December, 2019, after the opening of the Expo Line. For each location, the left plot incorporates periods (1) and (2), and the right plot incorporates periods (2) and (3). Changes to the mean are not statistically significant prior to opening of the Expo Line (left plots) except for the Downtown station. After opening of the Expo Line (right plots), crimes increased significantly at the Downtown and 17th Street/Santa Monica College stations. See Table 4.
Fig 13.
Reported Prop. 47 crime near Expo Line light rail stations in Santa Monica, CA, during three successive time periods spanning 2006–2019.
This figure is analogous to Fig 12, but restricted to Prop. 47 crimes only. See Table 5 for statistical test results.
Fig 14.
Reported non-Prop. 47 crime near Expo Line light rail stations in Santa Monica, CA, during three successive time periods spanning 2006–2019.
This figure is analogous to Fig 12, but restricted to non-Prop. 47 crimes only. See Table 6 for statistical test results.
Table 4.
Statistical tests on reported crime in the vicinity of light rail station locations in Santa Monica, CA, during three successive time periods spanning 2006–2019.
Table 5.
Statistical tests on reported Prop. 47 crime in the vicinity of light rail station locations in Santa Monica, CA, during three successive time periods spanning 2006–2019.
Table 6.
Statistical tests on reported non-Prop. 47 crime in the vicinity of light rail station locations in Santa Monica, CA, during three successive time periods spanning 2006–2019.