3. Quantitative Data on Neighbourhood Watch in the Netherlands

According to data from the Dutch Association of Insurers, residents of North Brabant are at the highest risk of home burglaries (6.3 burglary claims per 1000 households per year), followed by Limburg, Utrecht, Flevoland and Gelderland. North Holland has 3.9 burglary claims per 1000 residents per year (SOURCE: Risicomonitor Woninginbraken).

The police are categorized as ‘professionals’ here.

Source: Veiligheidsmonitor Rijk (VMR); Integrale Veiligheidsmonitor (IVM); Veiligheidsmonitor (VM). Crime against property is defined as (attempted) burglary, bicycle theft, car theft, theft from cars, pickpocketing and (attempted) robbery. The figures for 2005–2007 are based on the VMR, the figures for 2008–2011 are based on the IVM, and the figures for 2012, 2013 and 2014 are based on the VM. The index figures (index 2005 = 100) are based on corrected results, making the figures from the VMR and IVM comparable with those of the VM.

Source: Veiligheidsmonitor Rijk (VMR); Integrale Veiligheidsmonitor (IVM); Veiligheidsmonitor (VM). Variable ‘feelings of unsafety’ refer to the share of citizens aged 15 and older that sometimes feels unsafe. The figures for 2005–2007 are based on the VMR, the figures for 2008–2011 are based on the IVM, and the figures for 2012, 2013 and 2014 are based on the VM. The index figures (index 2005 = 100) are based on corrected results, making the figures of the VMR and IVM comparable to those of the VM.

An increasing numbers of citizens in Western countries feel safer and opine that crime has diminished, a development identified by Eysink Smeets and Vollaard (2015) as the fear drop.

The ‘residents’ initiative’ variable was recoded, removing the ‘combination of residents/professionals’, so that only two categories remain: professionals’ initiative (0) and residents’ initiative (1).

The risk of collinearity for the model is negligible here, given that the degree of overlap between the predictive variables is not substantial. The threshold value generally used is that the correlation coefficient between two predictors cannot exceed 0.8. In this case, income and educational level correlate with each other at a level of 0.37. The correlation coefficient of population size and degree of urbanisation is 0.51.

Unfortunately, even the international urban scientific literature offers little insight into this. Although many articles have been published about the concept of collective efficacy (residents’ degree of social cohesion in combination with their willingness to intervene in the public space), in empirical research collective efficacy is usually selected as an independent variable to measure its effect on crime levels (in combination with other neighbourhood characteristics such as income or mobility, see, e.g. Sampson et al. 1997; Morenoff et al. 2001; Browning et al. 2004; and in the Netherlands Kleinhans and Bolt 2010). But in this case, one would have to find out whether the income level of the neighbourhood has any influence on its collective efficacy (hence collective self-efficacy as dependent variable). Only Duncan et al. (2003) and Kleinhans and Bolt (2013) use collective efficacy as a dependent variable. Duncan et al., however, found no significant influence of income level (albeit measured at the household level, not at the neighbourhood level). Kleinhans and Bolt base themselves on qualitative interview material, and state that aspects like public familiarity, communicative skills and fear of negative counter actions were influential in this respect.