Guns, Cops and Public Opinion

Following the spate of terror attacks in China earlier this year, particularly the violent Kunming attacks on March 1, China has begun arming regular police patrols in several areas. The move is intended to ensure that police have the necessary skills and equipment to combat serious crimes, but the increased visibility of firearms has also raised public concerns about possible police negligence or abuse.

The journal ‘Public Opinion on Politics and Law’ (《政法舆情》) has recently released a report considering public opinion about the new policy and recent incidents of police gun use. Some of this data was released publicly online and is summarized and commented on here with its charts reproduced and translated. The Chinese review can be found here, with links to the full article, which I have not yet seen.

I began writing this summary because I saw what looked like concrete and relevant data, and there is a bit of that, but what stands out more is the report’s lack of focus and methodology—a nod to media research which formally resembles meaningful insight, but has little substance behind it. I continued writing the summary both to share what could be learned in it, but also to point out some of the articles failures along the way.

1. Public Opinion Background

The report first considers the increased prominence of reporting on police use of guns by comparing the number of news stories with relevant key words over the past several years. It is no surprise that the media attention around the new gun policies resulted in a massive increase in the number of stories on police and guns generally, but the increase in stories concerning the actual discharge of police weapons is more telling.

Although before this year, regular police forces did not carry firearms, special units of armed police or special police ( 武警, 特警)could be called in when needed, and ordinary police had legal authority to use weapons, but guns were not standard issue. It isn’t clear whether stories with keywords relating to the armed police and special police were included in the media review.

The increased number of stories on fatalities caused by police firearms might be the result of several factors, such as increased coverage due to public interest in the new policies and greater permission to report such events, but also a genuine uptick in the number of such incidents.

II. Public Opinion Trends`

The report also shows the distribution of news reports on instances of police use of guns over the course of 2014. This breakdown reveals that only 13.3% of such reported incidents occurred before the Kunming terror attacks. Also, it indicates that even before the implementation of the new policy, there was a general trend of a greater number of police gun incidents reported each month. [the percentage on the translated chart below total more than 100%, which I cannot explain, but may be a simple rounding error]

Reports on a a total of 60 incidents of police gun use were carefully studied for this report, all occurring in the first half of 2014, with only fifteen incidents occurring in the first three months. The report provides a breakdown of these incidents by region province as well:

The larger number of reported incidents in those areas where police use of guns has been authorized, such as Yunnan, Xinjiang and Guangdong, is expected, although the prominent number of incidences in Sichuan is unexplained. The absence of Beijing from the chart is somewhat distressing, as at least one incident of police gun discharge was publicly reported as early as May.

The language of the report summary is ambiguous as to whether these 60 events reflect all such incidents or only a sample. It is questionable what value the data has if it is not a complete set of at least the reported cases, particularly the breakdowns by time and geography, without at least explaining the selection criteria for understanding the representativeness of the data.

The inclusion of Taiwan is also confusing. Regardless of political sentiment on cross-strait relations, it is undeniable that Taiwan functions under a local legal system that is quite different than that of the other listed regions and cannot simply be lumped together with them. If Taiwan was also included in all the media reporting data analysis, it would undoubtedly skew the results even further, as Taiwan’s free press would be less guided in its coverage (or non-coverage) of incidents and have an entirely different set of criteria.

The chart above reflects something of whether incidents are more prevelant in urban or rural areas, but really needs to be compared to the total populations of these region types to give an idea of whether there is a bias of any kind in either reporting or actual incidents.

The chart below shows the prevalence of different types of firearm incident included in the report. You can see, that fatal shootings are the highest category, with warning shots close behind.

It appears that each incident was only coded in a single category for this chart. What then of incidents with multiple victims, where some targets were injured while some were killed, or accidental fire that resuleds in a death or injury? Both of these situations are know to have occurred: the Kunming terror events had 5 shot with 4 killed and 1 injured, and a misfire in a kindergarten resulted in five injuries.

The decision to include ‘incidents’ without mentioning the number of persons affected is the most obfuscating part of the analysis. The chart above shows that 38.3% (23) of incidents reviewed resulted in at least one fatality. It also tells us that the total number of suspects killed by police in the Kunming terror attack was four, but that this is still counted as a single incident. While it is too recent to be included in this report, nine suspects were reported shot dead by police in a raid in Xinjiang yesterday. Does it make sense to code a paramilitary raid by a large police force leading to multiple deaths as a single incident of gun use just as a lone officer against a knife weilding suspect?

The answer is that it depends on what data you are trying to get at, and here, the stated goal is to understand public opinion about these incidents. Although no hint is given as to its source, the next chart looks most clearly at that:

Hopefully, the full paper gives more information on how this information was coded- was the tone of a single report on each incident evaluated? Were people polled for opinion upon reading the reports? Should a tightly controlled media system ever be considered as reflecting upon public sentiment, or does it only ‘guide public sentiment’. Later parts of the article refer to web user opinions, but this raises more questions than it answers in terms of how they were polled.

The report summary astutely notes that fatal shootings arouse the most diverse range of opinions, and that people should be prepared for public feedback following fatal shoots by police. It is also encouraging that there were no ambivalent or positive stories of lost guns or accidental firing.

The report continues to mention that public opinion is also concerned with a number of other factors including the identity of the target. When the target is female, for example, the public becomes more concerned about the specifics of how the police fired, where they struck and the degree of harm. The chart below shows that only about 3 cases can be confirmed as including females.

No data is provided to confirm, but the article continues to discuss that when they target is a petitioner, they are more suspicious of the police action, saying that they sympathize with the target, and some said that the incident wreaked of ‘social stability’ measures.

The target having a weapon was also seen as influencing public opinon on the action. Again, no data is provided to confirm this, but the article asserts that when the target uses a highly dangerous weapon such as firearms, there is almost no questioning of police tactics. When they only carried a knife, however, web users are more likely to ask whether they really posed a threat and whether other areas of their body might have been hit to render them harmless.

When coverage fails to mention whether there was a weapon, the internet is likely to be very suspicious. Ultimately, the article concludes that it is very important that media and officials be accurate and forthcoming in their reporting of such details.

The remainder of the report is likely a continuation of data in this vein, rather than an elaboration on what has already been said.

While it is encouraging always to see data of any kind, and charts can make the data look persuasive, this data isn’t terribly useful as a measure of public opinon. Most of the data is descriptive of the cases only, without giving any evidence to explain why it’s important; assertions as to public opinon are only tangentially connected to the data.

Conclusions

We live in an era of unprecedented access to information and ever evolving techniques for analyzing it. In China, where social maintainance and guiding public opinion areviewed as important government functions, the use of data to fully understand public opinion and ground conditions is even more important. The goal of the social scientist and the investigative reporter is not just to gather facts but to explain what can be understood from those facts, and what limitations the data has. This report has assembled facts about certain articles reporting on police use of firearms, but seems lost as to what that data means, or which parts are important. The conclusions drawn seem drawn from something other than the data, without even an explanation.