SOC242 Methods of Social Research

SOC242 Methods of Social Research Prof. Younts Fall 2014 Homework Exercise #3: Experimental Designs  You may work in pairs for this exercise, or alone if you prefer. Bring ONE complete copy of all materials to hand in at the beginning of the lab.

As well, at least two examples will be chosen and performed during class on 10/28, so make sure you are prepared to conduct the experiment – bring 22 copies of the vignettes (half- or quarter-sheets are fine) and a method for randomly assigning the 22 students into two or three groups (depending on how many conditions you have). 1. Choose ONE of the following theoretical claims (the variables are in italics): a. The greater an individual’s identification with a target of injustice, the greater her/his negative reactions to that injustice. b. Individuals are more likely to agree with others who are higher status than they are than with others who are lower status than they are. c. The greater the similarity between self and other, the more favorable self’s evaluation of that other. d. The more attractive an individual, the more positive sentiment individuals will express towards them. 2. Design an experiment to test the argument you have chosen that can be conducted during the class period on 10/28. Specifically complete and describe each of the following steps in your typed report: a. Identify and provide an abstract, connotative definition for the independent and dependent variables in the conclusion you have chosen. b. Describe how you will operationalize (i.e., manipulate or measure) each variable. i. For the independent variable, you should plan to have only 2 (i.e., or three at most, including an experimental group(s) and perhaps one control group) each of which receive a different value of the independent variable. 1. The different “conditions” should be identical except for the manipulation of the independent variable they receive. 2. Construct the manipulations of the independent variable so that they can be implemented in the class within a maximum of 10 minutes (vignettes are easiest, but not the only way to go). ii. For the dependent variable, clearly define how you will measure the outcome so that you can clearly observe the outcome at the conclusion of the experiment (HINT: 1-3 survey questions is most efficient, asking participants to react to the scenario in terms of their anticipated behaviors, actual attitudes, or actual emotions on a numerical scale). c. Propose a hypothesis based on the conclusion you chose by substituting your operational definitions into the conlcusion. This should be a statement that clearly predicts the expected differences between the groups in terms of your operationalization of the dependent variable. d. If, after you conduct your experiment, you find support for your hypothesis, can you conclude that the theory is correct/sound? Why or why not? (HINT: Address all three criteria of causality).

1. U.S. Department of Justice Office of Justice Programs National Institute of Justice DEPARTMENT OF JUSTICE OFF ICE OF JUSTICE PROGRAMS BJA NI J OJJDP BJS OVC The KansasCity Gun Experiment by Lawrence W. Sherman, James W. Shaw, and Dennis P. Rogan R e s e a r c h i n B r i e f National Institute of Justice Jeremy Travis, Director January 1995 continued p. 2 risk times. 6 This argument suggests the hypothesis that greater enforcement of existing laws against carrying concealed weapons could reduce gun crime. But this hypothesis had never been tested until the Kansas City gun experiment. The experiment developed out of the first Federal grant awarded under the Bureau of Justice Assistance (BJA) “Weed and Seed” program in 1991. The Kansas City (Missouri) Police Department (KCPD) was given wide latitude in planning its Weed and Seed strategy. Shortly after the BJA Handgun crime is increasing rapidly throughout the Nation, 1 especially in inner-city areas where youth homicide rates have skyrocketed. 2 While some scholars argue that more gun carrying by law-abiding citizens may be the best deterrent to gun violence, 3 others find little evidence to support that view4 but much more evidence that increases in gun availability produce increases in gun homicides. 5 Still others argue that it is not the total number of guns in circulation that increases gun violence, but the carrying of guns in high-risk places at highIssues and Findings Discussed in this Brief: An evaluation of a police patrol project to reduce gun violence, driveby shootings, and homicides in a patrol beat where the homicide rate was 20 times higher than the national average. Key issues: Gun crime is rising rapidly nationwide, while other types of crime are falling. The need for strategies to control gun crime is critical. If police could get more guns off the street, would there be fewer gun crimes? This was the question posed by the Kansas City program. Key findings: The results of the evaluation indicate that directed police patrols in gun crime “hot spots” can reduce gun crimes by increasing the seizures of illegally carried guns. Specific findings include: ● Gun seizures by police in the target area increased by more than 65 percent, while gun crimes declined in the target area by 49 percent (see exhibit 1). ● Neither gun crimes nor guns seized changed significantly in a similar beat several miles away, where the directed patrol was not used. ● There was no measurable displacement of gun crimes to patrol beats surrounding the target area. Exhibit 1: Fir e a rm Of f e nses/Gu ns Se iz e d Pe r 1,000 Pe rso ns Targe t Be a t 37.0 9.9 18.9 16.8 22.6 23.6 10.4 8.8 Gun Crimes Gun Crimes Guns Seiz ed Guns Seiz ed Be fore Pa trols 0 10 20 30 40 During Pa trols Be fore Pa trols During Pa trols Comparison Be a t2 R e s e a r c h i n B r i e f sumed that if enough potential gun criminals in the area had their guns seized, they would be unable to commit gun crimes—at least for as long as it took them to acquire a new gun. Neither of these theories could be directly examined within the limits of the study. Rather, the evaluation study focused on the basic hypothesis that gun seizures and gun crime would be inversely related. From the outset, the project team recognized that confirmation of the hypothesis would not prove that more gun seizures result in reduced gun crime. The design could not eliminate all competing explanations that could be suggested for the results. But if an inverse correlation between gun seizures and gun crime were found, it could suggest the value of further research and development. It could also support a policy of extending the patrols, regardless of the exact reason for their effectiveness. award to the KCPD, the National Institute of Justice (NIJ) awarded the University of Maryland a grant to evaluate the Kansas City effort. This timing allowed the police and researchers to collaborate in planning a focused program with a strong research design. This Research in Brief explains the study’s methodology and key findings, analyzes the reasons for the findings, and concludes with a discussion of policy implications. Study design The program was based on the theory that additional patrols would increase gun seizures, which, in turn, would reduce gun crime. Two possible mechanisms were suggested: deterrence and incapacitation. The deterrence theory assumed that if police took guns away, illegal gun carriers would become less likely to carry them in the area. The incapacitation theory asIssues and Findings continued . . . ● Driveby shootings dropped from 7 to 1 in the target area, doubled from 6 to 12 in the comparison area, and showed no displacement to adjoining beats. ● Homicides showed a statistically significant reduction in the target area but not in the comparison area. ● Before and after surveys of citizens showed that respondents in the target area became less fearful of crime and more positive about their neighborhood than respondents in the comparison area. ● An investment of 4,512 police officer-hours was associated with 29 more guns seized and 83 fewer gun crimes, or 54 patrol hours per gun crime and more than 2 gun crimes prevented per gun seized. ● Traffic stops were the most productive method of finding guns, with an average of 1 gun found in every 28 traffic stops. ● Two-thirds of the persons arrested for gun carrying in the target area resided outside the area. ● Only gun crimes were affected by the directed patrols, with no changes in the number of calls for service or in the total number of violent or nonviolent crimes reported. Target audience: Mayors, law enforcement officials, public health officials, policymakers, community leaders, and researchers. R e s e a r c h i n B r i e f Exhibit 2: 1991 Ch a r act e ristics o f Ta rg e t a n d Comp a riso n Be a ts Charact eristic Targe t Be a t (144) Comparison Be a t (242) Popula tion 4,528 8,142 % Female 53% 56% % Under 25 38% 41% Median Age 32 31 % Nonwhite 92% 85% % Age 25+ High School Graduates 53% 73% Residential Square Blocks 80 150 Popula tion Densit y Per Mile 7,075 4,308 % Single Family Housing 84% 93% % Land Parcels Vacant 34% 14% % Houses Owner-Occupied 63% 71% Median Ye ars O wned 12 10 Median Parcel Value $14,181 $23,953 1991 Fire arms-Rela t ed Crimes 183 252 (Rate Per 1,000) 40 31 1991 Shots Fired Incidents 86 120 (Rate Per 1,000) 19 15 1991 Driv eby Shootings 24 25 (Rate Per 1,000) 5 3 1991 Homicides 8 11 (Rate Per 1,000) 1.77 1.353 R e s e a r c h i n B r i e f stantial volumes of violent crime, which provided reliable statistics for assessing trends over time. Patrol operations. For 29 weeks, from July 7, 1992, to January 27, 1993, the Kansas City Police Department focused extra patrol attention on gun crime “hot Since the target area, patrol beat 144, already selected for the “Weed and Seed” grant had the second highest number of driveby shootings of any patrol beat in 1991, the police and academic team designing the experiment chose the reduction of gun crime as the principal objective of the program. The program budget for police overtime and extra patrol cars was then dedicated to getting guns off the street as costeffectively as possible. While the evaluation concentrated primarily on this first phase of the Weed and Seed grant, additional findings from the evaluation show what happened when the initial funding of patrols stopped (first half of 1993) and continuation funding allowed resumption of the patrols (second half of 1993). 7 Target area. The target beat is an 80-by-10 block area with a 1991 homicide rate of 177 per 100,000 persons, or about 20 times the national average. 8 In addition to its 8 homicides in 1991, there were 14 rapes, 72 armed robberies, 222 aggravated assaults (142 with firearms), and a total of 349 violent felonies—close to one a day. Exhibit 2 shows that the beat’s population is almost entirely nonwhite, with very low property values for the predominantly single-family detached homes. Home ownership rates are very high; more than two-thirds of all occupants own their homes. Because the program was restric
   t
   ed to one target patrol beat—see exhibit 3— the planning team selected a beforeafter comparison design. The primary basis for selecting patrol beat 242 in the Metro Patrol District was its almost identical number of driveby shootings9 in 1991; 25 driveby shootings in the control beat compared to 24 in beat 144. Exhibit 2 also shows that the comparison beat, beat 242, is similar to the target beat in many ways. The major difference is that beat 242 has almost twice the population and three times the land area, including a park. The comparison beat also has slightly higher housing prices. Both beats have subExhibit 3: K a nsas Cit y, Misso uri, Police Re p ortin g A r e as 531 232 234 231 233 223 221 222 224 134 132 133 131 141 142 114 123 124 321 322 323 112 122 111 121 312 313 311 NORTH KAN. CITY, MO INDEPENDENCE, MO N RAYTOWN, MO 87TH ST BLUE PKWY South Patrol Division 85TH ST East Patrol Division Central Patrol Division BLUE RIDGE CUTOFF STATE LINE RD SNI-A-BAR RD MISSOURI RIVER 342 241 243 333 331 332 343 341 113 213 214 212 211 143 Metro Patrol Division 242 144 Targe t Pa trol Be a t Control Be a t PROSPECT AVE 39TH ST4 R e s e a r c h i n B r i e f spots”10 in the target area. The hot spot locations were identified by a University of Maryland computer analysis of all gun crimes in the area. The extra patrol was provided in rotation by officers from Central Patrol in a pair of two-officer cars working on overtime under the BJA-funded Weed and Seed program. Four officers thus worked 6 hours of overtime each night from 7 p.m. to 1 a.m., 7 days a week, for a total of 176 nights, with two officers working an additional 24 nights, for a total of 200 nights, 4,512 officerhours, and 2,256 patrol car-hours. They focused exclusively on gun detection through proactive patrol and did not respond to calls for service. While no special efforts were made to limit police activities in the comparison area, beat 242, there were no funds available for extra patrol time in that area. Several different I n early 1992, the success of directed patrols in Kansas City gun crime hot spots was preceded by two apparently unsuccessful attempts to detect guns. These programs are described below: Door-to-door gun patrol. The first attempt was a comprehensive program of door-to-door visits to all 1,259 residences in the 80-block target beat 144, informing residents about a new crackdown on gun carrying and asking them to call an anonymous gun tips “hotline” if they knew of anyone carrying a gun illegally. The officers knocked on 1,410 doors in 173 hours of regular patrol time from March to May 1992, speaking with an adult at 72 percent of the occupied residences11 —one of the highest success rates in any door-to-door policing program. 12 Of the 786 adult residents to whom the police explained the “gun tips” program, 96 percent (756) said they would be willing to call the hotline, and many were extremely enthusiastic. Unfortunately, only two calls were received. The door-to-door gun tip results reveal an important limitation on the police-citizen “partnership” concept of community-based policing. The fact that the officers were white and the area was predominantly black may have made a difference. But the fact that two-thirds of the persons later found carrying guns in the beat resided outside the area may have been more important. 13 Residents of high crime areas may simply not have all the information police need to deal with many crime problems. concealed weapon. These indicators, such as frequent touching of the waist to ensure that a gun stuffed in a belt will not fall down a pants leg, helped Gallagher make more than 1,000 arrests for carrying concealed weapons. But using the same methods, Gallagher was unable to spot any gun carriers during several nights on patrol in the most violent areas of Kansas City in June 1992. This difference may reflect the enormous difference in density between the two cities: New York has 30 times as many people in about the same amount of land. Therefore, in New York most people walk and use public transit; in Kansas City, most people travel by car. The Kansas City officers trained to use these methods did report a few cases in which the techniques led to detection of a concealed weapon, but only 9 percent of guns were found in pedestrian checks. 16 Despite these results in Kansas City, both gun tips hotlines and body language cues may still work well in other cities. The social and physical characteristics of cities vary widely, and these methods may work better in different kinds of communities. Most important, the Kansas City experience demonstrates the importance of trial and error in any city’s efforts to get guns off the streets. Given the complexity of the problem, it is unrealistic to expect the first method tried to be an automatic success. However, the door-to-door program may have produced beneficial results. Exhibit 5 (see page 7) shows that the number of gun crimes in the target beat began to fall sharply in June 1992, the month after the 10-week program of doorto-door visits was completed. Gun crimes continued to decrease up to and after the start of the hot spots patrols. The fact that the number of guns seized in beat 144 declined in the second quarter of 199214 eliminates the high-risk gun seizure theory as an explanation of the June decrease in gun crime. The principal remaining explanations are either the deterrent effect of making all the door-to-door visits (with word of mouth spreading about a police crackdown on gun carrying), or simply random fluctuation that is evident elsewhere in the time series for the target beat—such as in the August to October 1991 period (see exhibit 5). Moreover, the second author found a drop in total serious crimes in the target beat that also began in June 1992, a finding consistent with other door-to-door patrol experiments. 15 Thus, even if door-to-door visits failed to increase gun seizures, they may still have been useful for preventing gun crime and other serious crime. Body language training. A second unsuccessful attempt to detect guns in Kansas City was a method that had worked well in New York City. Detective Robert Gallagher (retired) of the New York City Police Department trained a group of Central Patrol Division police officers in the body language “cues” he used to recognize when someone was carrying a Trial and Error in Gun Detection5 R e s e a r c h i n B r i e f strategies for increasing gun seizures were attempted in beat 144 (see “Trial and Error in Gun Detection”), but Federal funds for extra police patrol were expended entirely upon the overtime patrols. Measures used. Because the extra patrol hours were federally funded, separate bookkeeping was required to document the time. In addition, an onsite University of Maryland evaluator accompanied the officers on 300 hours of hot spots patrol and coded every shift activity narrative for patrol time and enforcement in and out of the area. Property room data on guns seized, computerized crime reports, calls for service data, and arrest records were analyzed for both areas under the study. No attempt was made to conduct victimization surveys, although a before and after survey of the target and comparison beats was conducted to measure citizen perceptions of the program. 17 Data analyses. The data were examined several different ways. The primary analyses compared all 29 weeks of the phase 1 patrol program (July 7, 1992, through January 25, 1993, when the phase 1 funding for the special patrols expired) to the 29 weeks preceding phase 1, using difference of means tests. Other analyses added all of 1991 and 1993. The 1993 data included 6 months with no overtime patrols and phase 2 overtime patrols for 6 months in the second half of 1993. These analyses thus covered six 6-month periods, two of which had the program and four of which did not. The citizen survey analysis compared the amount and direction of before-after differences in attitudes within beats. Both shorter and longer periods around the program were also examTraffic Stops and Reasonable Suspicion any lay people—and even some police—underestimate police powers to search for guns. When a police officer can articulate a reason for believing that a gun crime may be about to occur, the U.S. Supreme Court has ruled that the officer may pat down the outside of the suspect’s clothing to check for guns (Terry v. Ohio, 392 U.S. 1, 1968). This ruling does not give police the right to stop cars or persons in an arbitrary manner. But it does imply that when police stop people for other legally sound reasons, they may find further evidence that justifies proceeding to frisk a suspect and to search the passenger compartment of the car. 18 When one considers that traffic stops are the leading cause of police murders in the line of duty, the logic of this policy may be clearer, as the following true case study from Indianapolis implies: An officer stopped a car in a high crime neighborhood for running a stop sign. As the officer approached the driver, he saw the driver reaching into a belt pack. He then directed the driver to get out of the car so he could pat down the belt pack. Feeling hard metal inside, the officer opened the pack and found a small revolver. The evidentiary standard of reasonable suspicion is necessarily lower than the standard of probable cause, which is the level of evidence required to justify an arrest. In the case study, the driver could not have been arrested for reaching into his belt pack. But the behavior did provide a basis for articulating why the officer thought the driver might have had a gun. Only after the gun was actually found was there sufficient evidence to make an arrest for carrying a concealed weapon without a permit. But the articulable suspicion allowed the officer to detect the hidden evidence in a lawful and constitutional manner. Other methods used in Kansas City included looking into the car for guns in plain view on the seat or the floor and looking for body language of pedestrians for telltale signs of a gun stuffed inside a suspect’s clothing. Consent searches of glove compartments or car trunks are also legal, as long as the consent is truly voluntary. M Search Upon Arrest 45% (13) Frisk for Safety 34% (10) Plain View 21% (6) Phase 1 Pa trols 07/07/92–01/25/93 N=29 How Hot Spot Patrols Seized Guns6 R e s e a r c h i n B r i e f Exhibit 4: Gu n Crimes Be f or e a n d Durin g Ph ase 1 b y Be a t Be fore During Be a t 12/17/91–7/6/92 7/7/92–1/25/93 %Change Target (144) 169 86 -49%* Comparison (242) 184 192 +4% Adjoining Target 141 76 57 -25% 142 106 84 -21% 143 39 44 +13% 213 143 158 +10% 214 104 138 +33% 331 143 175 +22% 332 153 160 +5% All Kansas City 4,359 4,287 -2% * Statistically significant t value, P < .05. Before and during weekly gun crime means were tested for significant differences in all areas displayed. Only the target area showed enough change for it to be unlikely a result of chance or random fluctuation. ined for overall impact. Autoregressive moving averages (ARIMA) models were used to compare gun crime in the 52 weeks before and after the patrols in both the target and comparison beats. Standard chi-square tests were used to compare 1991 versus 1992 differences in gun crimes for all four quarters, as well as both half-years, in both target and comparison beats. No matter how the data were examined, the results were similar. The program in action Patrol activity. Officers reported spending 3.27 car-hours of the 12 carhours per night actually patrolling the target area (27 percent), for a total of 1,218 officer-hours of potential gun detection and visible patrol presence in the area. The officers thus spent 70 percent of their time processing arrests and performing other patrol-related duties, as well as some patrol work outside the target area. Despite their limited time in the area, the officers generated a lot of activity. Both in and out of target beat 144, the directed patrols issued 1,090 traffic citations, conducted 948 car checks and 532 pedestrian checks, and made 170 State or Federal arrests and 446 city arrests, for an average of 1 police intervention for every 40 minutes per patrol car. There is some evidence that activity levels declined during October through January, just as street activity usually does at the onset of colder weather. 19 The average number of car checks made per day, for example, began at a high of 6.5 in July, and dropped to a low of 3.2 in November, but time in the target area, miles driven, and traffic citations issued did not change substantially during the first 6-month period. The actual techniques the officers used to find guns varied, from frisks and searches incident to arrest on other charges to safety frisks associated with car stops for traffic violations (see exhibit 3). 20 Every arrest for carrying concealed weapons had to be approved for adequate articulable suspicion with a supervisory detective’s signature. Results of increased patrol Gun seizures. The federally funded hot spots patrol officers found 29 guns in addition to the 47 guns seized in the target beat by other police units during phase 1 (second half of 1992), increasing total guns found in the beat by 65 percent over the previous 6-month period and almost tripling the number of guns found during car checks. The ratio of guns seized to directed patrol time in the target area was 1 gun per 156 hours, but the ratio to time actually spent in the area (and not processing arrests) was 1 gun per 84 hours and 1 gun per 28 traffic stops. Overall, there was an increase from 46 guns seized in beat 144 in the first half of 1992 to 76 seized in the last half. Once the guns were seized, most of them were then permanently removed from the streets. Not all of the guns were carried illegally; about one-fifth (14) of the total 76 guns seized in the target area during phase 1, and 4 of the 29 guns seized by the extra hot spots patrols were confiscated by police for “safekeeping,” a practice followed by many police agencies when officers have reason to believe gun violence may otherwise occur. While guns taken for this reason are usually returned to their registered owners upon application at the property room, the process can take several days to several weeks to complete. Illegally carried guns, on the other hand, are destroyed by Kansas City police and not returned to circulation. Gun crime. There were 169 gun crimes in the target area in the 29 weeks prior to the hot spots patrols, but only 86 gun crimes in the 29 weeks during the phase 1 patrols—a 49 percent decrease, with 83 fewer gun crimes (see exhibit 4). This change was statistically significant in both a test of differences of means7 R e s e a r c h i n B r i e f (t-test) for that period, and in an ARIMA model covering an even longer before and after period. 21 The comparison beat 242 showed a slight drop in guns seized, from 85 in the first half to 72 in the second half of 1992. It also showed a slight increase in gun crimes, from 184 in the 29 weeks before the program to 192 gun crimes in the 29 weeks during the program (see exhibit 5). Neither change was statistically significant. 22 In addition, while gun crime dropped in beat 144, none of the seven contiguous beats showed any significant change in gun crime, as exhibit 5 shows for the 29 weeks before and after tests. Both the Homicides. Homicides were also significantly lower in beat 144 during the two 6-month program periods than in other 6-month periods, from 1991 through 1993, while there were no significant differences in homicides across those periods in comparison beat 242. Other crimes. Neither total calls for police service, calls about violence, property or disorder crimes, total offense reports, nor property or violent offenses showed any effect of the increased patrol. There were no changes in these measures in either the target or comparison area. The target area hot spots patrols focused specifically on guns, and their effects were limited to gun crimes. increases and decreases in gun crime found across the contiguous beats were small enough to have occurred by chance. The 52 weeks before and after special tests (ARIMA models) showed significant reductions in g

   un crimes in beats 141 and 143. Driveby shootings. Driveby shootings in beat 144 dropped significantly during both 6-month periods of hot spots patrols (second halves of 1992 and 1993) compared to the 6-month periods without them. The same analysis showed no differences in the beats surrounding 144 and an increase in the comparison beat 242. 23 Exhibit 5: To t a l Of f e nses Wit h Fir e a rms b y Mo n t h in Ta rg e t a n d Comp a riso n Be a ts 1989 Monthly Fire arm Off enses Hot Spot Pa trols Pause in Hot Spot Pa trols Hot Spot Pa trols Reinitia t ed 1 121 121 121 121 12 1990 1991 1992 1993 0 10 20 30 40 50 60 Targe t Be a t Comparison Be a t8 R e s e a r c h i n B r i e f Community perceptions. Community surveys before and after the intensive patrols showed that respondents in the target area, beat 144, became less fearful of crime and more satisfied with their neighborhood than respondents in the comparison beat 242. Target area respondents also perceived less physical and social disorder after phase 1. While target beat respondents were only marginally more likely to say that the shooting problem had gotten better and no more likely to say that overall crime problems had improved, they were significantly more likely than comparison area respondents to say that neighborhood drug problems had gotten better. When the experimental period was over, crimes involving firearms gradually increased again for 5 months in the first half of 1993, consistent with the typical police crackdowns pattern. 24 When the phase 2 patrols began in the second half of 1993, gun crimes dropped again, although not as consistently as in phase 1. Analysis of the gun crime reduction Assuming that there are 100,000 handguns in Kansas City, 25 the seizure of 29 handguns may be considered a drop in the bucket, an implausible reason for any significant reduction in gun crime (Exhibit 6 indicates how gun crime was defined and recorded). But there are at least three plausible theories for