Pandemic Pets: How The Demand For Dogs Affects In-Home Domestic Violence During COVID-19

doi:10.21203/rs.3.rs-1983761/v1

Download PDF

Research Article

Pandemic Pets: How The Demand For Dogs Affects In-Home Domestic Violence During COVID-19

https://doi.org/10.21203/rs.3.rs-1983761/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

In response to the novel Coronavirus of 2020 the county of Dallas, Texas ordered residents to shelter in place for 52 days. During the pandemic reports of domestic violence increased across the US as victims were forced to remain home with their abusers. Using daily reports of domestic violence and pet adoptions, surrenders, and fosters in Dallas, TX, we find an increase in domestic violence during shelter in place orders and identify the outflow for pets from the home as a trigger for this increase. As animals are surrendered to shelters in-home violence decreases, while violence increases during the shelter in place order– specifically when pets are confiscated from the home. Our results align with family violence literature by identifying a victim’s dependents as a main deterrent to escape. If more women’s shelters accommodated pets, victims would leave their abusers sooner. Any policy intended to decrease family violence ought to include animal shelters, women’s shelters, and police departments. Together, these offices can identify potential abusers, decrease in-home violence, and ensure the safety of families and pets.

JEL Codes: K0, D1, I1

Pets

Shelter in Place Order

Domestic Violence

COVID-19

As of March 2020, Texas is among forty-two states combating the spread of COVID-19 with shelter in place orders (Wu, 2020). In Dallas, TX, local orders required individuals stay indoors and avoid any unnecessary contact with other people. While the order appears to have limited the spread of the virus (Dave et al., 2020), households had to bear other costs due to the policy. Local news articles in Dallas, TX report increased domestic violence during the shelter in place order¹, as victims are isolated with their potential abusers. As a contribution to the criminology literature, we analyze daily reports of in-home domestic violence and find empirical evidence for an increase in violence against women during the COVID-19 shelter in place order in Dallas, TX. Inability to leave the home appears to increase the likelihood of violence in the home.

While the order creates an environment conducive to violence, violence often needs a trigger. Finding triggers like welfare payments (Hsu, 2017) or unexpected sports outcomes (Cardazzi et al., 2020) aids researchers (Dowling et al., 2019) and law enforcement (Johnson and Dai, 2016) in identifying which households are prone to domestic violence. As a contribution to the identification of triggers to domestic violence, we use a novel identification strategy, drawing on recent evidence from the animal sciences literature to identify triggers for in-home domestic violence.

Using daily data from the Dallas Animal Services on pet intake from specific census tracts to local shelters, we investigate how the outflow of pets from the home affects in-home violence. Pets in the household reliably increase domestic violence, so the additional violence from sheltering in place could be fatal to a pet-owning victim (DeGue and DiLillo, 2009). We match pet outflows at the census tract level with reports of family violence, as defined by Card and Dahl (2011); Cardazzi et al. (2020). We find that in-home violence increases as pets are removed from the household via confiscation but decreases when owners surrender their pets to a shelter. Strongest effects exist when we lag in-home violence, which suggests that the removal of a pet from the home affects family dynamics between 3 days to a week later.

Our results contribute to the understanding of triggers of domestic violence and identify in- home pets and their effect on the household as understudied area of research. For a victim in need of refuge the main deterrents to leaving an abusive situation are the victim’s dependents. Almost forty percent of women in shelters report that they would have sought refuge up to two months

earlier if not for their pets, as very few shelters allow animals (Ascione, 2000). For fear of retaliation against those left behind, victims will postpone their escape– sometimes indefinitely. Policymakers should note that animal shelters may play an important role in the identification and prevention of domestic violence (Douglas et al., 2019). In the next section, we present background on domestic violence and pets in the home. In section 3 we describe the setting and data of our analysis. In section 4 we explain our empirical strategy and regression framework, and in section 5 we discuss the results of our analysis.

2.1 Domestic Violence and Animal Abuse

Though quarantine is effective for limiting the spread of COVID-19, the accompanying stressors from illness, unemployment, and uncertainty can be deadly for victims of domestic violence (Bettinger- Lopez and Bro, 2020). Experts have characterized the increase in domestic violence during COVID- 19 as an “invisible pandemic,” as public health officials across the world work to protect the safety and rights of potential victims (Chandan et al., 2020). As reports of domestic violence increase all over the world, evidence suggests that domestic violence increased Dallas, TX during the shelter in place order. Piquero et al. (2020), examining trends in domestic violence during the shelter in place order, find that violence in the home increases initially but are unable to find a lasting effect, perhaps because the analysis ends in April and the Dallas’ shelter in place order lasted until May 20, 2020.

Animal abuse is indicative of other domestic violence, such as child or spousal abuse (Gullone, 2011). An abuser of children is likely to abuse animals and vice versa. An abusive spouse may see an animal as a threat or use it for leverage against a victim (Randour and Davidson, 2008), so a victim with a pet is less likely to leave her abuser.

The abusive cycle consists of three stages: tension-building, an abusive incident, and the honeymoon phase (Focht and Chu, 2020). Abusive households will experience this cycle over and over. After an abusive incident, the victim often finds herself in a “honeymoon phase,” in which her abuser may apologize, show extra affection, or buy gifts in an attempt to “make up for” abuse. To the chronic abuser, a new pet makes a perfect apology gift as well as a convenient obstacle to escape.

Domestic violence increases during stay at home orders. When a pet is removed from the home, in-home violence shifts significantly depending on the reason for the pet’s removal. We suspect that the surrender of a pet from the home is indicative of a victim’s escape. Without an animal, battered women may find shelter more easily thus decreasing reported domestic violence. While the removal or confiscation of a pet from the home increases violence against women during shelter in place orders, we suspect abusers have changed the target of their abuse from the pet to someone in the household.

3.1 The Shelter In Place Order and Family Violence

The Dallas County shelter-in-place order issued by Judge Clay Jenkins “allows for people to leave their homes for outdoor walks, getting necessary items like medicine and groceries, and to care for sick relatives, but otherwise must stay in their households.” The shelter in place order remained in effect for 52 days, beginning on March 24 and expiring on May 15.

Column (1) in Table 1 confirms Piquero et al. (2020) and media reports that in-home violence

increased in Dallas during the shelter in place order.

Table 1: Effect of the Dallas SIPO on In-Home Assaults

Dependent Variables: assault x lag female assault x lag

Model:	(1)	(2)	(3)	(4)	(5)	(6)
Variables SIPO	0.2010^∗∗∗	0.2263^∗	0.0505	0.1275^∗	0.3273^∗∗	0.0634
	(0.0590)	(0.1312)	(0.1817)	(0.0774)	(0.1593)	(0.2064)
Fixed-effects
Week	Yes	Yes	Yes	Yes	Yes	Yes
Month x Police Beat	Yes	Yes	Yes	Yes	Yes	Yes
Day of the Week	Yes	Yes	Yes	Yes	Yes	Yes
Fit statistics
Observations	116,336	25,688	24,797	101,111	18,230	18,230
Squared Correlation	0.05523	0.10303	0.11139	0.03566	0.09576	0.09399
Pseudo R²	0.06967	0.09510	0.10191	0.06417	0.10107	0.10007

One-way (police beat) standard-errors in parentheses Signif. Codes: ***: 0.01, **: 0.05, *: 0.1

Cities across the US have implemented various public health orders in an effort to prevent the spread of the novel Coronavirus. Our analysis focuses on the effects of shelter in place orders on domestic violence in Dallas, Texas. All animal data is geo-coded at the census-tract level. Dallas Animal Shelter uses an intake system called Chameleon. Because we study the county of Dallas, we are unable to aggregate data above the county-level. We match census-tract Chameleon data with assaults, which are reported at the police beat level. Every beat belongs to a sector, which is part of a division (Figure 6). Each census tract is fully encased by a division and sector, and there is reasonable overlap between tracts and beats, as seen in Figure 7. We compare census tracts and beats because these are the most granular geographic measures we can use.

4.1 Dallas Animal Shelter Services

We collect daily animal intake and outcome variables from the Dallas Animal Shelter from January 2018 to June 2020. Staff at the shelter document the characteristics of pets, the owner’s stated reason for surrender, and the animal’s eventual outcome at the shelter (adopted, euthanized, fostered, or transferred). Intake information includes the census tract from which the animal came. We plot the trends of each intake over time in Figures 8 through 12 in the Appendix.

The top ten breeds of dogs confiscated and surrendered are in Figures 2 and 3. The top ten reasons owners gave for surrendering their dogs are in Figure 4. This graph does not include the 80% of owners who gave “no reason” or “other reason.” Note that fewer than one percent of owners cited financial reasons for their surrender. During lockdown these reasons changed slightly, but financial reasons were still insignificantly low on the list of surrenders. These reasons align with (Jr. et al., 2000), who found that relinquishment was associated with physical and behavioral characteristics of the animals and owner characteristics and knowledge.

4.2 Dallas Police Public Data

We collect daily crime incidents data from the Dallas Police Department from January 2018 to Octo- ber 2021. The data available provides information on the victim (age, gender, relationship to abuser) and details on the incident (time and date, beat where the incident occurred, and injuries suffered in the incident). The Dallas Police Department disseminates this information to be transparent regarding policing practices and criminal activity across Dallas.

Table 2: 2020 Monthly Reported In-Home Violence By Victim Type

Month (2020)	Average Monthly Reported Violence	Against Women	Against Non-White Victims
January	0.083	0.044	0.068
February	0.082	0.036	0.067
March	0.091	0.052	0.078
April	0.092	0.045	0.071
May	0.102	0.053	0.082
June	0.109	0.054	0.087
July	0.107	0.054	0.091
August	0.102	0.054	0.088
September	0.094	0.051	0.081
October	0.099	0.051	0.077
November	0.087	0.047	0.067
December	0.090	0.046	0.077

In order to measure how demand for pets affects in-home domestic violence, we need to link violence across Dallas with the reporting of Dallas Animal Services. The Dallas Police Department collects data regarding police incidents according to beat maps, while Dallas Animal Services collects data regarding animal intake by census tract. We overlay these two maps in Figure 7. Because all data is from the county of Dallas, this matching strategy helps us specify connections between assaults and animals. Dallas county is one single jurisdiction and our data cannot be aggregated any higher.

We define domestic violence as simple assault, aggravated assault, or intimidation between men and women in the household (single family, condominium/town home and apartment resi- dences). We use this definition as it closely follows the criminology and economics literature on how to measure the causes of domestic violence (Card and Dahl, 2011; Cardazzi et al., 2020).

Our study is subject to some limitations inherent to crime data. We assume laws and enforce- ment are constant across precincts, as the county of Dallas is a single jurisdiction. Criminologists use police reports as well as victim surveys to approximate crime. Though we rely ultimately on reports of assault recorded daily, our data aligns with the CDC’s National Intimate Partner and Sexual Violence Survey, which is reported annually. We assume that a decrease in reports is indicative of a decrease in assaults. In an extension of this paper, we will incorporate data from Dallas Women’s Shelters for more accurate measures of assault. This may be wishful thinking, as data on women’s shelters is limited to protect the anonymity of victims.

5.1 Regression Framework

We begin our analysis by measuring how the shelter in place order has affected in-home domestic violence. y_ct represents a count variable of daily reported in-home assaults per beat. First, we investigate changes to in-home vs. outside the home assaults, and then we look specifically at assaults against women and children. We estimate the following equation in Table 3.

y_ct = β₀ + β₁SIPO_t + σ_wy + µ_c + ϵ_ct.

To measure violence against women, we interact the number of female assault victims with our shelter in place order variable. We estimate how the demand for Dogs during the shelter in place order affects in-home domestic violence using a difference-in-differences framework. In the following equation, y_ct represents a count variable of daily reported in-home assaults per council districts.

y_ct = β₀ + β₁SIPO_t + β₂DogConfiscated_it + β₃DogFostered_it

+β₄DogStray_t + β₅DogSurrendered_it

+β₆DogTreatment_it + +β₇CatConfiscated_it + β₈CatFostered_it

+β₉CatStray_t + β₁₀CatSurrendered_it

+β₁₁CatTreatment_it + +β₁₂DogConfiscated_it × SIPO_t+ β₁₃DogFostered_it × SIPO_t

+β₁₄DogStray_t × SIPO_t+ β₁₅DogSurrendered_it × SIPO_t

+β₁₆DogTreatment_it × SIPO_t+ +β₁₇CatConfiscated_it × SIPO_t+ β₁₈CatFostered_it × SIPO_t

+β₁₉CatStray_t × SIPO_t+ β₂₀CatSurrendered_it × SIPO_t

+β₂₁CatTreatment_it × SIPO_t+ +σ_wy + µ_c + ϵ_ct.

We estimate the differences in assaults given the intake of dogs and cats to the Dallas Animal Services during the shelter in place order using an indicator variable, SIPO_t, for the 52 days that the order is in place. Next, we include indicators for the daily number of dog and cat brought into a shelter because it was confiscated from owner, surrendered by owner, returned after fostering, received medical treatment, or is a stray animal. We then interact these variables with our indicator

for the shelter in place order to estimate differences in assaults given the outflow of pets from the household during the SIPO. σ_w and µ_d represent week and day of the week fixed effects, respectively. We also include month by beat fixed effects, λ_it. The day of the week fixed effects account for any differences in weekday vs. weekend in-home domestic violence and beat by month fixed effects account for any time-invariant factors that may affect in-home domestic violence at the beat level. ϵ_idt is our error term.

Table 3 holds the results of regressing domestic violence on pet intake during shelter in place orders. In Column (2) of Table 3 we show that a confiscated dog significantly increases in-home violence three days after removal. In Table ?? Column (2) we show that in-home violence against women decreases significantly three days after a dog is voluntarily surrendered by the owner. This supports our theory that battered women will find shelter for their dogs before escaping an abusive home. Column (2) in Table 4 shows a similarly significant effect for non-white victims who surrender their dogs to a shelter.

6.1 Heterogeneous Effects

Domestic violence is a social issue and can be mitigated with carefully applied policy, keeping in mind that individuals will include pets in their optimization framework. People are so unwilling to leave their animals in a dangerous situation that they will forego their own safety. In the aftermath of Hurricane Katrina, so many survivors refused to leave their pets behind (Haeffelle et al., 2015) that Congress passed the Pets Evacuation and Transportation Standards Act of 2006, requiring pets to be evacuated with people. ² In any emergency–natural or man-made–good pet owners will find sanctuary for their animals first and then ensure their own safety.

We show that domestic violence decreases– meaning victims of domestic violence escape their abusers– after surrendering their pets to shelters. Many victims of domestic violence will stay with their abusers just to be with their pets. If animal shelters informed women’s shelters of increases in ²H.R. 3858 (109th): Pets Evacuation and Transportation Standards Act of 2006

Table 3: Effect of Pet Intake on In-Home Violence During the SIPO

	In-Home Assault	Lagged 3 Days In-Home Assault	Lagged 7 Days In-Home Assault
	(1)	(2)	(3)
SIPO	-0.4448	-1.576^∗	0.7215
	(0.3538)	(0.9508)	(0.8085)
Dog Confiscated	-0.0246	-0.2278	0.3089
	(0.0800)	(0.1668)	(0.2244)
Dog Stray	0.0243	-0.0457	0.0834
	(0.0163)	(0.0440)	(0.0577)
Dog Owner Surrender	-0.0732^∗	0.0082	-0.1266
	(0.0392)	(0.0255)	(0.1126)
SIPO × Dog Confiscated	-0.1818	3.478^∗∗	-0.5205
	(0.8368)	(1.649)	(1.302)
SIPO × Dog Stray	0.2892	-0.5946	-0.3796
	(0.1958)	(0.8041)	(0.6568)
SIPO × Dog Owner Surrender	0.1942	-3.782	-515.5
	(0.4064)	(8.611)	(10,333.8)
Controls	Yes	Yes	Yes
Week	Yes	Yes	Yes
Month by Beat	Yes	Yes	Yes
Day of the Week	Yes	Yes	Yes
Observations	9,190	1,535	1,364
Squared Correlation	0.09949	0.31552	0.30403
Pseudo R²	0.08827	0.17626	0.17232
BIC	12,022.4	3,548.3	3,142.8

One-way (Beat) standard-errors in parentheses Signif. Codes: ***: 0.01, **: 0.05, *: 0.1

pet surrenders, women’s shelters could predict demand for beds and resources, resulting in a faster, smoother transition for victims. Very few women’s shelters accept pets, and the uncertainty of an animal’s fate if left with her abuser is enough to keep a victim from escaping.

Ascione (2000) surveys forty-one domestic violence agencies across the US, only three of which can house pets, and eight of which can offer pet carriers or port-o-pets for the short term only. The meager resources these shelters do have are donated by the community, pet stores, or an animal welfare agency. Though women’s shelters may lack the resources to house animals, a relationship between institutions would enable women to board their pets with a local animal shelter for the duration of her refuge. If more women’s shelters allowed animals or provided resources to connect victims to animal shelters for temporary boarding, victims would leave their abusers sooner.

If animal shelters cross-referenced adoption applications with police records, shelters could choose to deny an abuser’s application. Our research shows that this screening process would have a significant impact on reports of domestic violence. Any policy encouraging animal shelters, women’s shelters, and law enforcement agencies to work together can decrease domestic violence.

Table 4: Effect of Shelter In Place Order on In-Home Violence By Non-White Victims

	In-Home Assault	Lagged 3 Days In-Home Assault	Lagged 7 Days In-Home Assault
	(1)	(2)	(3)
SIPO	-0.4342	-0.7838	0.4665
	(0.4569)	(0.4874)	(0.8045)
Dog Confiscated	-0.0080	-0.3324	0.6157^∗∗
	(0.0882)	(0.2412)	(0.2665)
Dog Stray	0.0295^∗∗	-0.0003	0.1366^∗
	(0.0150)	(0.0596)	(0.0709)
Dog Owner Surrender	-0.0751	0.0950	-0.0593
	(0.0541)	(0.0991)	(0.0695)
SIPO × Dog Confiscated	-0.0993	2.373^∗	-14.85
	(1.032)	(1.429)	(28.41)
SIPO × Dog Stray	0.1767	-1.105	-0.3443
	(0.2551)	(1.057)	(0.7219)
SIPO × Dog Owner Surrender	0.2150	-3.749^∗∗∗	-12.07
	(0.4844)	(1.170)	(45.50)
Controls	Yes	Yes	Yes
Week	Yes	Yes	Yes
Month by Beat	Yes	Yes	Yes
Day of the Week	Yes	Yes	Yes
Fit statistics
Observations	8,118	1,142	1,123
Squared Correlation	0.09974	0.33611	0.34626
Pseudo R²	0.09474	0.17633	0.19604
BIC	10,200.3	2,823.4	2,586.4

One-way (Beat) standard-errors in parentheses Signif. Codes: ***: 0.01, **: 0.05, *: 0.1

With access to police records, an animal shelter can keep a new pet out of a potentially abusive home, thus removing both a trigger for violence and an obstacle from a victim’s escape. If more women’s shelters allowed animals or provided resources to connect victims to animal shelters for temporary boarding, victims could leave their abusers sooner. Open communication between these three unlikely institutions–women’s shelters, animal shelters, and local police– could accelerate relief for victims of domestic abuse.

We do believe our identification strategy finds a unique relationship between pet ownership and domestic violence during the shelter in place order. However, limitations to our results include the inability to directly match which households adopt, foster, or surrender pets. Though it would perfect our matching strategy, data on women’s shelters is difficult to obtain for privacy reasons so we are unable to map victims to animals. Finally, we assume that a decrease in domestic violence reports is consistent with a real decrease in domestic violence, which may not be the case.

Domestic violence increased across the US during the 2020 COVID-19 pandemic. To a pet-owning family who are already more likely to have violence in the home, this increased violence from shel- tering in place could be fatal. Violence against women increases as dogs are fostered, but violence against children decreases when dogs are adopted.

We measure the additional violence a pet-owning family may experience while sheltering in place. We collect daily animal intake and outcome variables from the Dallas Animal Shelter during the pandemic and combine this data with daily crime incidents from the Dallas Police Department. During shelter in place orders, we find that the removal of pets from the home affects family violence, as dogs confiscated from a home leads to increased in-home violence, whereas dogs surrendered to shelters leads to a decrease in violence. As victims are able to place their pets in a shelter, it appears they are more free to leave abusive relationships. Alternatively, pets confiscated from the home suggests an preexisting elevated level of violence in the home, which may affect the target of violence once a pet is removed.

Pets remain an understudied component of the household. We encourage further research on the demand for pets and the effect of in-home pets on family decision-making, financing, and spillover behaviors such as violence. As understanding of pets and their effect on the home increases, animal shelters may be able to play an even more important role in the community. Our results indicate that Dallas Animal Shelters and the Dallas Police Department could combine their resources to identify potentially violent households. If equipped to identify prior violent activity in potential adopters, animal shelters could more reliably ensure safety for both families and animals.

Compliance with Ethical Standards

We have no conflicts of interest with regards to our study. All data used is publicly available.

Ascione, F. R. (2000). Safe havens for pets: guidelines for programs sheltering pets for women who are battered. Utah State University.
Bettinger-Lopez, C. and Bro, A. (2020). A double pandemic: Domestic violence in the age of COVID-19. Gender-based violence, 3.
Card, D. and Dahl, G. B. (2011). Family violence and football: The effect of unexpected emotional cues on violent behavior. The quarterly journal of economics, 126(1):103–143.
Cardazzi, A. J., McCannon, B. C., Humphreys, B. R., and Rodriguez, Z. (2020). Blaming the ref: Interpreting the effect of unexpected emotional cues on family violence. Working Paper 20-11.
Chandan, J. S., Taylor, J., Bradbury-Jones, C., Nirantharakumar, K., Kane, E., and Bandyopad- hyay, S. (2020). COVID-19: a public health approach to manage domestic violence is needed. The Lancet Public Health, 5(6):e309.
Dave, D., Friedson, A. I., Matsuzawa, K., and Sabia, J. J. (2020). When do shelter-in-place orders fight COVID-19 best? policy heterogeneity across states and adoption time. Economic Inquiry.
DeGue, S. and DiLillo, D. (2009). Is animal cruelty a “red flag” for family violence? investigating co-occurring violence toward children, partners, and pets. Journal of Interpersonal Violence, 24(6):1036–1056.
Douglas, H., Harris, B. A., and Dragiewicz, M. (2019). Technology-facilitated domestic and family violence: Women’s experiences. The British Journal of Criminology, 59(3):551–570.
Dowling, C., Morgan, A., et al. (2019). Predicting repeat domestic violence: Improving police risk assessment. Trends and Issues in Crime and Criminal Justice, (581):1.
Focht, J. and Chu, A. (2020). The cycle of domestic violence.
Gullone, E. (2011). Conceptualising animal abuse with an antisocial behaviour framework. Animals, 1(1):144–160.
Haeffelle, S., Storr, V., and Grube, L. (2015). Community Revival in the Wake of Disaster: Lessons in Local Entrepreneurship.
Hsu, L.-C. (2017). The timing of welfare payments and intimate partner violence. Economic inquiry, 55(2):1017–1031.
Johnson, R. R. and Dai, M. (2016). Police enforcement of domestic violence laws: supervisory control or officer prerogatives? Justice quarterly, 33(2):185–208.
Jr., J. C. N., Salman, M. D., King, M., Scarlett, J. M., Kass, P. H., and Hutchison, J. M. (2000). Characteristics of shelter-relinquished animals and their owners compared with animals and their owners in U.S. pet-owning households. Journal of Applied Animal Welfare Science, 3(3):179–201.
Piquero, A. R., Riddell, J. R., Bishopp, S. A., Narvey, C., Reid, J. A., and Piquero, N. L. (2020). Staying home, staying safe? a short-term analysis of COVID-19 on dallas domestic violence. American Journal of Criminal Justice, pages 1–35.
Randour, M. L. and Davidson, H. (2008). A common bond: Maltreated children and animals in the home. American Humane and The Humane Society of the United States.

No competing interests reported.

Appendix.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Pandemic Pets: How The Demand For Dogs Affects In-Home Domestic Violence During COVID-19

Status:

Version 1

Abstract

Figures

1 Introduction

2 Literature Review

2.1 Domestic Violence and Animal Abuse

3 Setting

3.1 The Shelter In Place Order and Family Violence

4 Data

4.1 Dallas Animal Shelter Services

4.2 Dallas Police Public Data

5 Identification Strategy

5.1 Regression Framework

6 Results

7 Policy Implications

8 Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1