Unpacking Sequential Actions Within Use of Force Incidents

Introduction

Use of force by the police has generated research activity dating back several decades (e.g., Garner et al., 1995; Henriquez, 1999), but in recent years, interest in these actions—including threats, attempts, and use as a means to gain civilian compliance—have intensified (e.g., Hollis, 2018; Terrill et al., 2018; Willits & Makin, 2018). Critical aspects of studying and understanding force actions include the scope of actions that constitute force and its measurement, and the complexities associated with measuring police-civilian encounters that provide the context for these actions.

Research has evolved from measuring force actions as a simple dichotomy to measurement of force on a severity continuum and now frequently includes analyzing a force factor ¹ (Alpert & Dunham, 1997, 1999; Crawford & Burns, 1998; Garner et al., 2002; Garner et al., 1995; Terrill & Paoline, 2012; Terrill et al., 2018). Related, there is variability in the actions that are considered force. For example, while use of weapons (e.g., a firearm or TASER) and physical actions are generally considered force, there is less consistency in consideration of verbal commands and handcuffing as force actions (Fridell, 2017; Hollis, 2018; Klahm et al., 2014; Klinger, 1995; Terrill, 2003; Willits & Makin, 2018; Wolf et al., 2009). ² Selecting the scope of actions is critical to conclusions drawn as lower levels of force (e.g., hands on) are more commonly used by police and generally are understudied relative to more serious actions (Bayley & Garofalo, 1989; Garner et al., 1995, 2018; Klinger, 1995; Terrill, 2003; Torres, 2018).

Measurement decisions also have direct implications for prevalence estimates (Terrill, 2003). Use of force is a rare occurrence, but estimates suggest approximately one to five percent ³ of police-civilian encounters result in force depending on the definition and measurement of force actions (Davis et al., 2018; Friedrich, 1980; Garner et al., 2018). Force prevalence is also influenced by the type of police-civilian encounter, with force more likely to occur in more serious situations such as arrest (Davis et al., 2018; Hickman et al., 2008).

Decisions regarding force measurement and scope of actions also underscore the challenges with representing interactional events in a manner that supports greater understanding. Studies generally focus on the most severe type of force (and resistance) applied in the encounter, but frequently they do not measure the sequential nature of these actions to fully unpack the complexities of these events. Recent exceptions include the use of body-worn camera (BWC) footage to measure time to force use and duration of force (Willits & Makin, 2018; also see Hickman et al., 2015; Hine et al., 2018; Kahn et al., 2017; Wolf et al., 2009).

The current study ⁴ contributes to these efforts by examining police-civilian encounters as transactional events in an effort to build knowledge about how events change over time. This is accomplished by applying a detailed coding instrument to narrative descriptions of force encounters that occurred in two police agencies across a 30-month period. This effort focuses on the total number of actions within the encounter and how the starting level of police force and civilian resistance may be instructive in understanding the overall level of force and resistance severity throughout the encounter. The measurement of each sequential action also offers an ability to measure the difference between the initial action in the encounter and the starting level of force and resistance. Disentangling these micro-level interactions and measuring these characteristics is a valuable next step to better understand the complexity of police-civilian encounters that involves use of force by the police.

Examining Use and Severity of Force

Exploration of force use by the police has become increasingly common, and a broad body of research has developed on this topic. Comprehensive reviews of use of force studies underscore the complexities of police-civilian encounters that involve force while simultaneously highlighting important, and at times contradictory, themes stemming from the findings (e.g., the impact of civilian race on the use of force) (Garner et al., 2002; Hollis, 2018; Hollis & Jennings, 2018; Klahm et al., 2014; Klahm & Tillyer, 2010). The intricacies of these encounters are exemplified in efforts undertaken to measure force.

The Current Study

The existing use of force literature consistently demonstrates the effect of civilian resistance on force severity (Fridell & Lim, 2016; Gau et al., 2010; Lawton, 2007; Stroshine & Brandl, 2019; Terrill & Mastrofski, 2002) while also revealing variations in methods, data sources, and measurement (Garner et al., 2002; Hollis, 2018; Hollis & Jennings, 2018; Klahm et al., 2014; Klahm & Tillyer, 2010). More broadly, the quality of data collected and manner in which it is analyzed are critical elements to maximizing our understanding of force actions.

In furtherance of these efforts, the current study engaged a methodological effort to unpack the sequential nature of force encounters by building knowledge about how events change over time from the perspective of a transactional encounter between the officer(s) and civilian(s). A more complete picture of the inherent dynamic process becomes evident when each action and reaction is considered.

Disentangling the interactional nature of these events allow the exploration of whether encounters with more actions are related to the maximum level of force, the maximum level of resistance, or the force factor within the encounter. If empirical evidence verifies these relationships, there are implications for emphasizing quicker resolutions to these situations while balancing force with resistance. Related, understanding the influence of the starting point of force and resistance will inform the broader conversation about de-escalation by revealing information about the transformation of these events over time. To that end, the following research questions are posed:

1. Do the total number of actions predict the maximum force level, the maximum resistance level, or the force factor, while controlling for other relevant factors?

These research questions were examined by coding the time-ordered actions of officers and civilians within force encounters. The specific data coding processes, unit of analysis decisions, and variable measurements are described below.

Data

The Tulsa Police Department and the Cincinnati Police Department provided data on all use of force incidents that occurred between January 1, 2016 and June 30, 2018. Use of force narrative descriptions were the primary data source for the current study. These narratives were recorded by the officer using force in Tulsa and by the supervising officer in Cincinnati.

All available narratives were analyzed using a coding instrument loosely based on prior research by Hickman and colleagues (2015). The coding instrument was pilot tested on several narratives prior to its finalization. During both the pilot test and final coding, all narratives were read by two independent coders to ensure consistency in categorization of all actions. The coding instrument consisted of a six-point force scale and a six-point resistance continuum (see Table 1). Force actions ranged from issuing lawful commands (Level 1) through physical display tactics (Level 3) to use of deadly force (Level 6), while resistance actions included non-compliance (Level 1) through verbal or physical threats (Level 3) to an apparent attempt to cause death or serious bodily harm (Level 6).

OPEN IN VIEWER

Table 1.

Force and Resistance Coding.

Civilian resistance	Officer force	Force factor
		Resist	Force
Non-compliance: Verbal resistance without threats; subject ignores officer or refuses to comply	Issuance of lawful announcements, warnings, orders, or commands	1	1
Passive physical resistance (e.g., “dead weight”)	Physical touch not exceeding a firm grip	2	2
Moved away from officer; fleeing or attempting to flee; defensive resistance	Physical control tactics; pain compliance techniques; hair pulling; joint locks and come-alongs; open-handed strikes; take-downs	3	3
Verbal or physical threats (e.g., fighting stance, reaching for possible weapon, other furtive movements) from officer’s perspective	Display/threat of less lethal weapon (pepper spray/ball, TASER, baton, canine, firearm)	4	4
Unarmed assaultive physical resistance; subject strikes or attempts to strike officer with hands, feet, elbows, knees, or other body parts; includes kicking at officer to avoid control or handcuffing; no apparent attempt to kill or seriously injure officers	Hard hand control; use of pepper spray/ball, TASER, baton, canine, LVNR	5	5
Apparent attempt to cause death or serious bodily injury to officer by using hands, fists, feet; display or threat of weapon; or use of weapon	Deadly force; use of firearm	6	6

The application of the coding instrument to the force narratives was guided by a set of rules designed to identify each unique action (i.e., force or resistance) by any participant (i.e., officer or civilian). The primary goal was to identify every action that fell within either the force or resistance continuums and ascribe that action to a participant in the order described. In situations where multiple actions appeared to occur simultaneously, actions were coded sequentially in the order they were described. If a single participant engaged in multiple actions simultaneously, the highest level of action was recorded. ⁷

Unit of analysis & cases

The process of quantifying complex police-civilian encounters required several methodological decisions including selecting the appropriate unit of analysis. These encounters are frequently coded to reflect the maximum force and resistance levels within the entire incident regardless of how many officers and civilians are involved. In cases involving a single officer and a single civilian, this is a reasonable decision that provides a snap shot of the most serious action; however, it does not allow consideration of the escalating and de-escalating actions taken by an officer or civilian within such an encounter. In situations involving more than a single officer or civilian, it becomes more complicated and challenging to effectively represent all force and resistance actions with a single measure using the incident as the unit of analysis. Given the interest in the influence of total and starting actions within an encounter, the methodology for coding actions adopted here was necessary to more accurately reflect the degree of complexity of these situations and identify the appropriate unit of analysis.

As a result, the unit of analysis for this study is an exchange, which was defined as the sequence of actions between one officer and one civilian. As mentioned, many encounters involved multiple officers and/or civilians, resulting in a larger number of cases (i.e., exchanges) than the original number of encounters. For example, a single encounter involving one officer and one civilian represented a single exchange and all actions within that event were coded and assigned to the relevant parties. Encounters involving a single officer and two civilians were measured as two exchanges and actions assigned to the parties in a corresponding fashion, and so forth. The coding structure contained the potential for up to five officers and five civilians to undertake actions within any single encounter. As a result, there was a possibility of up to 25 exchanges within any single encounters (combination of five officers interacting with five civilians within a single encounter; 5 × 5 = 25). After this adjustment, the number of encounter and exchange cases were identical when a single officer and a single civilian were involved; however, encounters involving more than a single officer or civilian resulted in more exchanges than encounters.

Data originally received from the two agencies included 1344 force encounters. An initial data assessment removed duplicate cases and those with missing information resulting in 1180 use of force incidents across the two research sites. Applying the unit of analysis protocol described above resulted in 2084 exchanges across the 1180 incidents. After a data audit to eliminate cases with unusable information, ⁸ 1743 analyzable exchanges remained. Of those, 454 one-to-one exchanges contained full and complete information for analysis. ⁹

Variables

Dependent variables included maximum force, maximum resistance, and a force factor (Alpert & Dunham, 1997, 1999) for each exchange. Maximum force indicated the highest level of the six-category force scale used in the exchange by any officer. Similarly, maximum resistance reflected the highest level of resistance used in the exchange by a civilian on the six-category resistance scale. A force factor measured the difference between the highest level of force used and the highest level of resistance within the exchange and was calculated by subtracting suspect resistance from officer force. A positive value indicates that the officer used a higher level of force than the suspect’s level of resistance, whereas a negative value represents a situation in which the suspect’s resistance was higher than the officer’s level of force.

Independent variables included all actions taken within the exchange by either an officer or suspect (i.e., total actions). Starting force measured initial force within the exchange on the six-point force scale, while starting resistance identified initial resistance within the exchange on the six-point resistance scale. The starting force and resistance measures do not reflect the first action taken in the exchange; they reflect the initial level of action by the officer and civilian, respectively.

Additional independent variables were also created for the one-to-one exchanges including situational variables to indicate whether the exchange occurred on a weekday (1 = Yes; 0 = No) or between 7AM and 7PM (i.e., daytime; 1 = Yes; 0 = No). Officer and civilian characteristics ¹⁰ are also frequently examined to identify whether the use of force and resistance vary across specific officer or civilian profiles. The body of evidence for these characteristics is generally mixed, with some officer and civilian characteristics showing consistent relationships with use of force, but most showing inconsistent findings across studies (Crawford & Burns, 1998; Klahm & Tillyer, 2010; McElvain & Kposowa, 2008; Schuck & Rabe-Hemp, 2007).

Officer characteristics included officer age measured as a continuous variable reflecting age in years. Officer male was coded as a simple dichotomy indicating male officers (1 = Yes; 0 = No). Officer race/ethnicity was also dichotomized (1 = Yes; 0 = No) to identify individuals as White, Black, Hispanic, or Other (Asian, Native American, Pacific Islander, or other). Officer length of service was a continuous measure of the number of years an individual has been a sworn officer with the department. Rank: officer (1 = Yes; 0 = No) was created to identify non-supervisory officers compared to all other sworn officers at any other higher rank.

Civilian characteristics were measured using a similar approach to officer characteristics. Civilian age represented the individual’s age in years. Civilian male was coded as a dichotomous variable indicating whether the civilian was male (1 = Yes; 0 = No). Civilian race/ethnicity was measured in a series of dichotomous variables for White, Black, Hispanic, and persons of Other (Asian, Native American, Pacific Islander, and other) races/ethnicities.

The analytic strategy to examine the exchanges began with descriptive statistics including the range, mean/percent, and standard deviation for study variables. Thereafter, multivariate models were estimated to identify the influence of a single variable on the outcome while simultaneously considering the effect of all other variables simultaneously (Hanushek & Jackson, 1977). Given the limited continuous nature of the dependent variables for force, resistance, and the force factor, ordinary least square regression models were estimated for all exchanges and also for the one-to-one exchanges only. All results included the coefficient, standard error, and beta weight in addition to the level of statistical significance. Supplemental analyses were also undertaken to explore the highest starting point for force and resistance in greater detail.

Results

Force and resistance actions were situated on a six-category continuum. The average maximum force was 4.29 with 68.0% of all exchanges involving a Level 5 force action (i.e., hard hand control, pepper spray/ball, TASER, canine) as the maximum (see Table 2). Maximum resistance averaged 3.37 with Level 3 resistance (i.e., defensive resistance, attempting to flee) displayed in 57.2% of all exchanges. The force factor averaged 0.93 across all exchanges which demonstrates higher rates of force on average relative to resistance levels and aligns with the maximum force and resistance descriptives. This pattern was also revealed in the one-to-one exchanges with Level 5 force (88.3%) and Level 3 resistance (64.1%) as the most common categories.

OPEN IN VIEWER

Table 2.

Descriptives.

	All exchanges (N = 1743)				One-to-one exchanges (N = 454)
	Min	Max	Mean/%	S.D.	Min	Max	Mean/%	S.D.
Dependent Variables
Maximum force	1.00	6.00	4.29	1.23	2.00	6.00	4.83	0.59
Level 1	—	—	5.3%	—	—	—	0.0%	—
Level 2	—	—	5.3%	—	—	—	0.2%	—
Level 3	—	—	15.4%	—	—	—	7.9%	—
Level 4	—	—	4.2%	—	—	—	2.0%	—
Level 5	—	—	68.0%	—	—	—	88.3%	—
Level 6	—	—	1.7%	—	—	—	1.5%	—
Maximum resistance	1.00	6.00	3.37	1.30	1.00	6.00	3.26	1.13
Level 1	—	—	11.2%	—	—	—	9.5%	—
Level 2	—	—	0.3%	—	—	—	0.2%	—
Level 3	—	—	57.2%	—	—	—	64.1%	—
Level 4	—	—	13.1%	—	—	—	13.9%	—
Level 5	—	—	7.9%	—	—	—	5.7%	—
Level 6	—	—	10.2%	—	—	—	6.6%	—
Force factor	−5.00	5.00	0.93	1.80	−3.00	4.00	1.57	1.26
Independent Variables
Total actions	2.00	25.00	8.41	4.35	2.00	25.00	9.51	4.59
Starting force	1.00	6.00	2.25	1.58	1.00	6.00	2.17	1.57
Level 1	—	—	55.0%	—	—	—	58.4%	—
Level 2	—	—	9.0%	—	—	—	7.9%	—
Level 3	—	—	8.7%	—	—	—	7.0%	—
Level 4	—	—	11.6%	—	—	—	12.8%	—
Level 5	—	—	14.8%	—	—	—	12.8%	—
Level 6	—	—	0.9%	—	—	—	1.1%	—
Starting resistance	1.00	6.00	2.62	1.27	1.00	6.00	2.44	1.18
Level 1	—	—	30.6%	—	—	—	35.2%	—
Level 2	—	—	0.8%	—	—	—	0.7%	—
Level 3	—	—	54.7%	—	—	—	54.4%	—
Level 4	—	—	7.3%	—	—	—	5.7%	—
Level 5	—	—	2.7%	—	—	—	2.4%	—
Level 6	—	—	3.8%	—	—	—	1.5%	—
Weekday	—	—	—	—	0.00	1.00	70.3%	—
Daytime	—	—	—	—	0.00	1.00	47.6%	—
Officer variables
Age	—	—	—	—	24.00	67.00	40.38	8.59
Male	—	—	—	—	0.00	1.00	93.0%	—
White	—	—	—	—	0.00	1.00	76.0%	—
Black	—	—	—	—	0.00	1.00	14.5%	—
Hispanic	—	—	—	—	0.00	1.00	2.6%	—
Other	—	—	—	—	0.00	1.00	7.3%	—
Length of service	—	—	—	—	0.00	43.70	11.79	8.26
Rank: Police officer	—	—	—	—	0.00	1.00	81.1%	—
Civilian variables
Age	—	—	—	—	12.00	72.00	31.49	11.21
Male	—	—	—	—	0.00	1.00	91.2%	—
White	—	—	—	—	0.00	1.00	36.1%	—
Black	—	—	—	—	0.00	1.00	58.1%	—
Hispanic	—	—	—	—	0.00	1.00	3.7%	—
Other	—	—	—	—	0.00	1.00	1.0%	—

Key independent variables included total actions which averaged 8.4 across all exchanges and 9.5 in the one-to-one exchanges. Starting force actions were most commonly at Level 1 (55.0%), while starting resistance actions were most commonly at Level 3 (54.7%). This pattern was also displayed in the one-to-one exchanges with Level 1 (i.e., verbal commands) force actions (58.4%) and Level 3 resistance actions (54.4%) most common. Additional variables associated with the one-to-one exchanges revealed that 70.3% of these exchanges transpired on weekdays and roughly half of those situations occurred between 7AM and 7PM (47.6%). The one-to-one exchanges involved predominately male (93.0%), White (76.0%), and 40-year-old officers, on average. Officers averaged slightly more than 11 years of service and the large majority were the rank of police officer (81.1%). Civilian characteristics revealed an average age of 31.49 years, with male civilians representing the majority (91.2%), and Black civilians the most common racial/ethnic group to be involved in these exchanges (58.1%).

Considering the average levels of maximum force and resistance in relation to the starting levels of force and resistance revealed that force actions frequently began at lower levels and escalated to higher maximum levels, whereas starting and maximum resistance levels were comparable (i.e., Level 3). The relationship between maximum and starting levels of force and resistance are further explored in the multivariate models.

A total of six multivariate models were estimated to explore the maximum force level, the maximum resistance level, and the force factor across all exchanges (Table 3) and those involving one officer and one civilian (Table 4). The maximum force model for all exchanges explained 23.4% of the variation with several independent variables maintaining a statistically significant relationship with maximum force. Exchanges with a higher number of total actions experienced higher force severity, while exchanges with higher levels of starting force (relative to a mid-level of force, Level 3) were also positively related to the maximum force level. The starting levels of resistance were not associated with maximum force levels. Finally, the maximum level of resistance was weakly and negatively related to the maximum level of force. This suggests that as the maximum level of resistance increased in an exchange, the corresponding level of maximum force decreased. Collectively, the most impactful variables were starting force at Level 5 (β = .4) and Level 4 (β = .3) and the total number of force actions (β = .3).

OPEN IN VIEWER

Table 3.

Linear Regression Models: Maximum Force, Maximum Resistance, and Force Factor—All Cases (N = 1743).

	Maximum force			Maximum resistance			Force factor
	Coeff	S.E.	Beta	Coeff	S.E.	Beta	Coeff	S.E.	Beta
Intercept	3.44***	0.13	—	3.01***	0.12	—	0.43**	0.15	—
Total actions	0.09***	0.01	0.31	0.09***	0.01	0.30	−0.00	0.01	−0.01
Starting force
Level 1	0.21*	0.10	0.09	−0.03	0.09	−0.02	0.26	0.14	0.07
Level 2	−0.20	0.12	−0.05	−0.10	0.11	−0.02	−0.11	0.18	−0.02
Level 4	0.99***	0.12	0.26	−0.28*	0.11	−0.07	1.38***	0.17	0.25
Level 5	1.23***	0.11	0.35	0.08	0.11	0.02	1.24***	0.16	0.25
Level 6	2.49***	0.30	0.19	0.77**	0.28	0.06	1.85***	0.42	0.10
Starting resistance
Level 1	−0.02	0.06	−0.01	−0.79***	0.06	−0.28	0.85***	0.08	0.22
Level 2	−0.43	0.29	−0.03	−0.16	0.27	−0.01	−0.28	0.41	−0.01
Level 4	−0.18	0.10	−0.04	0.80***	0.09	0.16	−1.07***	0.15	−0.16
Level 5	0.21	0.17	0.03	1.58***	0.15	0.20	−1.50***	0.23	−0.14
Level 6	0.25	0.16	0.04	2.70***	0.13	0.40	−2.69***	0.20	−0.29
Maximum force	—	—	—	−0.08**	0.02	−0.07	—	—	—
Maximum resistance	−0.09**	0.03	−0.09	—	—	—	—	—	—
R squared	.234			.419			.282

OPEN IN VIEWER

Table 4.

Linear Regression Models: Maximum Force, Maximum Resistance, and Force Factor—1:1 Exchanges (N = 454).

	Maximum force			Maximum resistance			Force factor
	Coeff	S.E.	Beta	Coeff	S.E.	Beta	Coeff	S.E.	Beta
Intercept	4.18***	0.23	—	2.85***	0.50	—	1.62***	0.44	—
Total actions	0.02**	0.01	0.14	0.08***	0.01	0.32	−0.06***	0.01	−0.23
Starting force
Level 1	0.36**	0.11	0.30	0.23	0.18	0.10	0.15	0.21	0.06
Level 2	−0.03	0.14	−0.01	0.15	0.23	0.04	−0.19	0.27	−0.04
Level 4	0.46***	0.12	0.26	0.04	0.21	0.01	0.47	0.25	0.12
Level 5	0.58***	0.13	0.33	0.38	0.22	0.11	0.25	0.25	0.07
Level 6	1.71***	0.29	0.31	1.09*	0.51	0.10	0.74	0.58	0.06
Starting resistance
Level 1	−0.07	0.06	−0.06	−0.38***	0.10	−0.16	0.31**	0.12	0.12
Level 2	0.12	0.32	0.02	0.40	0.55	0.03	−0.27	0.65	−0.02
Level 4	−0.21	0.12	−0.08	0.86***	0.20	0.18	−1.12***	0.23	−0.21
Level 5	0.36*	0.18	0.10	1.67***	0.30	0.23	−1.33***	0.35	−0.16
Level 6	−0.17	0.24	−0.04	2.64***	0.39	0.29	−2.91***	0.46	−0.28
Maximum force	—	—	—	−0.09	0.08	−0.05	—	—	—
Maximum resistance	−0.03	0.03	−0.06	—	—	—	—	—	—
Weekday	0.02	0.06	0.02	−0.02	0.10	−0.01	0.05	0.12	0.02
Daytime	0.13*	0.05	0.11	0.09	0.09	0.04	0.05	0.11	0.02
Officer Variables
Male	−0.07	0.10	−0.03	0.21	0.18	0.05	−0.29	0.21	−0.06
White	−0.03	0.06	−0.02	−0.06	0.11	−0.02	0.03	0.13	0.01
Length of service	−0.00	0.00	−0.04	0.00	0.01	0.00	−0.00	0.01	−0.02
Rank: officer	−0.01	0.07	−0.01	−0.03	0.13	−0.01	0.02	0.15	0.01
Civilian variables
Age	0.00	0.00	0.02	−0.01	0.00	−0.06	0.01	0.01	0.06
Male	0.25**	0.09	0.12	−0.02	0.16	−0.01	0.30	0.19	0.07
Black	0.03	0.06	0.02	−0.21*	0.10	−0.09	0.25*	0.11	0.10
Hispanic	0.12	0.14	0.04	0.51*	0.25	0.09	−0.40	0.29	−0.06
Other	0.20	0.19	0.05	0.68*	0.33	0.08	−0.49	0.39	−0.05
R squared	.181			.357			.272

***p < 0.001, ** p < 0.01, * p < 0.05, Excluded categories: Starting Force Level 3, Starting Resistance Level 3.

Officer age was not included due to multicollinearity concerns with officer length of service.

The maximum resistance model for all exchanges explained 41.9% of the variation in maximum resistance based on the cumulative effects of all independent variables. Total actions increased the maximum resistance level while higher levels of starting resistance (e.g., Levels 4, 5, and 6) were associated with increased resistance severity. Of note, exchanges beginning with a starting force Level 6 were also associated with higher maximum levels of resistance. Finally, exchanges with lower maximum force levels also experienced higher maximum resistance levels; this finding mirrors the result in the maximum force model. The most impactful variables were starting resistance at Level 6 (β = 0.4) and total number of actions in the exchange (β = 0.3).

Results from the force factor model revealed several statistically significant results while explaining 28.2% of the variation in the force factor. Higher starting levels of force (e.g., 4–6) resulted in higher force factors, while higher starting points of resistance (e.g., 4–6) were negatively associated with the force factor. In other words, exchanges that began with higher resistance levels experienced lower overall force factors. Starting force at Levels 4 or 5 and starting resistance at Level 6 were the most impactful variables on the force factor (β = 0.3).

Three additional multivariate models were estimated to identify key variables related to the maximum level of force, maximum resistance level, and force factor for exchanges involving only one officer and one civilian (see Table 4). The maximum force model explained 18.1% of the variation in this outcome. Total actions in the exchange, starting levels of force (e.g., 1, 4–6), starting resistance (at Level 5), daytime exchanges, and those involving male civilians all exerted a statistically significant positive impact on maximum force. Notably, maximum resistance and officer characteristics were not related to maximum force. The most impactful variables were starting levels of force (β = 0.3).

The maximum resistance model explained 35.7% of the variation in this outcome based on the measured variables. Total actions in the exchange, starting force level of 6, and higher levels of starting resistance (e.g., 4–6) were positively related to the maximum resistance level. Conversely, starting resistance at Level 1 decreased the maximum resistance level. Several civilian characteristics were also associated with maximum resistance levels including the presence of Black civilians reducing the maximum level of resistance, while the involvement of Hispanic and Other civilians increasing the maximum level of resistance (compared to White civilians). Maximum force, situational characteristics, and officer variables were not statistically related to maximum resistance. Total actions in the exchange (β = 0.3) and higher levels of starting resistance (β = 0.2 and 0.3) were most impactful on maximum resistance.

The final model examined the force factor in one-to-one exchanges and this model explained 27.2% of the variation in the outcome. The total number of actions reduced the force factor (i.e., related to higher levels of resistance compared to force) so that the longer an exchange continued, force and resistance moved toward alignment or greater resistance than force. In other words, exchanges involving more actions (presuming this operates as a loose proxy for length of the exchange) exerted an effect of reducing the force factor toward zero (i.e., similarity between force and resistance levels) or even toward a negative force factor reflecting higher civilian resistance relative to officer force. Other statistically significant relationships include situations that began with resistance Levels 4, 5, or 6, which reduced the force factor. Finally, encounters involving Black civilians increased the force factor. In other words, the gap between force and resistance was greater in situations involving Black civilians.

Across the six models, the statistical and substantive importance of higher starting levels of force and resistance was a consistent result. ¹¹ In particular, starting force and resistance levels at four or above were among the most impactful variables in an expected direction. Additional analyses using the sequential coding structure were possible to further explore these findings.

Exchanges that included starting force and resistance at or above Level 4 were selected and analyzed separately to identify the type of force and resistance applied in the initial action of the exchange (i.e., Action 1). Note that starting force and resistance measured the first action by officers and civilians, respectively, but this behavior did not necessarily appear as the first action of the exchange. ¹² By examining Action 1, three specific patterns emerged within the exchanges.

First, initiating actions occurred when Action 1 aligned with the starting point of force or resistance. Matching actions transpired when the level of resistance displayed in Action 1 aligned with the starting level of force or vice versa (i.e., the level of force displayed in Action 1 matched the starting level of resistance). Finally, escalating situations occurred when Action 1 reflected a lower level of resistance relative to the starting level of force or when Action 1 was a lower level of force compared to the starting point of resistance.

Table 5 reports bivariate analyses between the Action 1 behavior and the starting point of force and resistance for all exchanges. ¹³ For exchanges that involved starting force levels of 4–6, an initiating action (i.e., Force Level 4–6) occurred in 49.2% of all exchanges. In other words, the starting level of force was identical to the initial action in the exchange. Only 8.5% of the exchanges involved Action 1 as a resistance Level 4–6 and matching the starting force level. Finally, 42.2% of these exchanges involved a Level 1–3 resistance behavior as Action 1 with a subsequent response of force at Level 4 or above (i.e., escalating). In other words, these escalating exchanges reflect a “low” level of civilian resistance as the first action within the exchange, but a higher starting point of force applied by the officer in their first action.

OPEN IN VIEWER

Table 5.

Bivariate Analyses of Starting Force and Resistance at Level 4 or Higher.

	All exchanges	All exchanges
	Force 4–6, action 1	Resistance 4–6, action 1
Initiating (force 4–6)	49.2%	—
Matching (resistance 4–6)	8.5%	—
Escalating (resistance 1–3)	42.2%	—
Initiating (resistance 4–6)	—	55.4%
Matching (force 4–6)	—	8.2%
Escalating (force 1–3)	—	36.4%

When considering only exchanges that involve resistance levels of four or above as a starting point, a similar pattern of results emerged. The most common initial action in an exchange was an initiating one (i.e., a match between this level of resistance and the starting level of resistance). Slightly more than 8% of the exchanges involved a similar level of force in Action 1 to the starting point of resistance (i.e., Level 4–6). Finally, slightly more than one-third of exchanges (36.4%) involved a low force level (i.e., 1–3) prior to the starting level of resistance of 4–6 (i.e., escalating).

Discussion

The current study utilized a methodology for unpacking complex police-civilian encounters to understand the importance of total actions and starting force and resistance levels on maximum force and resistance responses by officers and civilians, respectively. The primary data source was detailed and comprehensive narrative accounts recorded by officers and their supervisors of 1743 police-civilian exchanges that occurred across 1180 force incidents during a 30-month period. Force and resistance actions were measured on corresponding six-category scales in all models, with one-to-one exchanges also including contextual, officer, and civilian characteristics.

There are several key takeaways from the results for maximum force, maximum resistance, and a force factor models. First, the total number of actions in an exchange was positively correlated with maximum force and maximum resistance in four models and in one of the two force factor models. Moreover, this relationship was consistently one of the most impactful variables of statistical significance in the models. One implication of this result is that exchanges that are able to be resolved in a more expeditious manner may result in lower levels of force and resistance. It is important to caution that quicker encounter resolution should not be used as a justification or reason for force to exceed resistance, rather, that if fewer actions can be utilized, the overall intensity of the encounter may also be lowered. In other words, all things being equal, the fewer actions required to bring the civilian safely under control the better (Willits & Makin, 2018). The application of de-escalation strategies, with an emphasis on verbal approaches, may be particularly relevant and impactful in reducing the number of actions.

Another key finding was the importance of starting force and resistance in relation to the overall level of force and resistance severity. The models consistently demonstrated that higher levels of starting force and resistance exerted an increasing impact on overall force and resistance severity. Supplemental analyses further revealed that the starting level of force and resistance differed from the initial action undertaken in the encounter. Two critical observations are relevant regarding this result. First, the sequential coding of actions within an exchange is crucial to developing a comprehensive understanding of how events develop over time and the action-reaction patterns that occur within these situations. Second, the process by which maximum force and resistance is achieved is not uniform and varies to a considerable extent. For example, in exchanges involving starting force at levels 4–6, less than 10% of those situations began with a matching level of resistance. Conversely, over 40% of exchanges that involved a starting force level between 4–6 were initiated with civilian resistance at a level between 1–3. While additional civilian resistance behaviors above levels 1–3 could have occurred prior to the initial officer force action, it is clear that in a number of situations that began with a relatively low level of civilian resistance (i.e., Levels 1–3) the initial action by an officer was to employ a high level of force (i.e., levels 4–6).

Several ancillary observations are also noteworthy. The force factor models largely performed as expected and their results aligned with the maximum force and resistance models. The one exception is that total number of actions was only related to the force factor in one-to-one exchanges, while the starting force variables were not significantly related to the force factor in these exchanges. Broadly, however, the force factor models were consistent with the results from the other models. This suggests the use of a force factor may continue to offer benefit in understanding the overall force-resistance dynamic in these events. Future research should explore, however, the utility of a force factor when analyzing the sequential actions within an event. Work by Hickman and colleagues (2015) and Kahn and colleagues (2017) underscores the value in creating action dyads as a mechanism to maintain a force factor.

The models also revealed a somewhat surprising result with respect to civilian resistance and its impact on maximum force and maximum resistance. Unlike most previously reported studies (Fridell & Lim, 2016; Garner et al., 2002; Gau et al., 2010; Stroshine & Brandl, 2019; Terrill & Mastrofski, 2002; Terrill & Paoline, 2012; but see Lawton, 2007 for contradictory findings), a weak, but statistically significant, negative relationship was revealed between maximum force used by the police and maximum civilian resistance. In other words, as civilian resistance increased along the continuum, the maximum force used by officers slightly decreased. It is somewhat unclear as to what explains this result. It is possible that this represents a change in the fundamental approach officers use toward civilians in such situations and potentially could be a product of the changing cultural environment within which policing occurs. It may be changes in police training and awareness of and concern toward use of force actions have begun to influence police behavior. Alternatively, it may be that this is an artifact of these specific data and the methodology used to code these narratives. Ultimately, additional research is needed in other jurisdictions to further probe this effect to determine its prevalence and identify whether the categorization of force and resistance actions on a continuum influence this effect.

Also of note, several contextual, officer, and civilian characteristics were statistically significant in the one-to-one models suggesting that situational factors continue to play a role in the severity of force and resistance (Crawford & Burns, 1998; Klahm & Tillyer, 2010; McElvain & Kposowa, 2008; Schuck & Rabe-Hemp, 2007). Related, some differences were revealed between one-to-one exchanges and exchanges that were nested in events involving multiple officers and/or civilians. While the current data were unable to identify and measure the specific characteristics that may explain these differences, the results highlight the importance of considering the context within which force and resistance appear. Future research should mine narrative and other data sources to flesh out the nuances of these situations and probe the potential impact these contextual factors may exert on decisions to use force.

Finally, the current study and its findings demonstrate the value of measuring actions within police-civilian encounters in a sequential manner. Results confirm that aspects of these events (i.e., total actions and starting point of force and resistance) can only be measured through intricate coding of these situations, but are informative and critical to consider. In particular, this methodology allowed an identification of how some exchanges involved an initial high level of force or resistance, while others escalated to these levels after starting with lower levels of force or resistance. While there is much to learn about these processes, these findings confirm the importance of recognizing the complexities inherent in police-civilian encounters and measuring the interactional nature of these events as they develop over time.