Experiences With COVID-19 Stressors and Parents’ Use of Neglectful, Harsh, and Positive Parenting Practices in the Northeastern United States

Abstract

Objectives:

To estimate household exposure to COVID-19 related stress and the association with parent report of neglectful, harsh, and positive discipline practices.

Methods:

Cross sectional survey data was collected from 2,068 parents in the Northeastern US. Parents reported personal and household experiences of COVID-19 stressors, their level of distress, and use of neglectful parenting and discipline practices for a randomly selected child in their home. Analyses estimated rates of COVID-19 related stress and parenting practices. Logistic regression was used to assess the relation of COVID-19 stress to parenting behaviors.

Results:

Individual and household stressor level, as well as distress were each positively associated with likelihood of neglect. Personal exposure to stressors was minimally related to discipline, but household stressor level and parents’ distress were positively associated with harsh and positive discipline.

Discussion:

Indicators of COVID-19 stress (e.g., exposure to stressors and distress) each uniquely predicted parents’ use of neglect, particularly physical and family-based sub-types, and use of harsh and positive discipline practices. Results suggest that parents may require additional support to provide appropriate care for their children while coping with the increased rates of stress associated with the pandemic and the resulting public health response.

The COVID-19 pandemic activated an extensive public health response, including implementation of social distancing measures and mass school and childcare closures across the US (Centers for Disease Control and Prevention, 2020a, 2020b). When implemented in a planful and proactive manner, these measures are critical steps to reducing the transmission of viruses and new incidents of infection (Davis et al., 2015; Markel et al., 2007; Uchida et al., 2012). However, such practices also result in caregivers and children spending more time at home during periods when children would typically be in the care of educational or childcare providers (e.g., Gupta et al., 2020). By March 19, 2020, UNESCO estimated 890.5 million youth, worldwide, had been impacted by school closures related to the COVID-19 pandemic (UNESCO, 2020). By the end of April, 2020, school closures due to COVID-19 had impacted over 124,000 US public and private schools involving over 55 million students; 43 states had ordered or recommended complete or partial school closures through the remainder of the school year (Education Week, April 28, 2020).

Likewise, many caregivers have experienced workplace closures, lay-offs, or shifts to remote work and telecommuting, while also arranging for or providing for their children’s educational and childcare needs. A number of studies have documented adverse parental mental health and wellbeing outcomes associated with the strain of the pandemic (Brown et al., 2020; Cameron et al., 2020; Hessami et al., 2020; Lee et al., 2021; Patrick et al., 2020). Pereda and Díaz-Faes (2020) use general strain theory (Agnew, 1992) and a social-ecological perspective (e.g., Belsky, 1993; Cicchetti & Valentino, 2006) to posit that the cumulative effect of these stressors, especially across multiple contexts, can increase parental anger and stress – increasing risk of intra-familial tension and potential for family violence, including harsh discipline or child maltreatment (e.g., physical, sexual, emotional abuse; physical, supervisory, or other neglect). This risk may be especially great when caregivers experience stressors known to be associated with maltreatment as a result of the COVID-19 response such as job loss, economic stress and social isolation (Bérubé et al., 2020; Lawson et al., 2020; Lee & Ward, 2020; Rodriguez et al., 2020).

It is critical that we understand the impact of the public health response and correspondent stressors experienced by parents on the safety and wellbeing of children. Such information could provide invaluable insight into how to support vulnerable children and their families during the ongoing public health crises. Thus, it is important to know the extent of their adversity during the current COVID-19 pandemic. Acquiring high quality data on this issue is also foundational to building our knowledge base for informing decision-making and resource allocation for future crises or in the event of a protracted response period to the current one.

However, knowledge of these processes is limited. Although state child protection agencies are tasked with monitoring and intervening with maltreatment, there are several reasons why data collected by such agencies are likely to be unreliable during a pandemic. First, new cases of abuse or neglect are more likely to go undetected when social distancing and school closures are in place, as educational and childcare personnel are primary sources of maltreatment report (Fitzpatrick et al., 2020). Second, some agencies have paused in-person monitoring, and surveillance of child safety may be more limited in this context. As a result, child welfare agency data capturing reported incidents of maltreatment likely represent an undercount of maltreatment since it is harder to identify new cases of abuse and neglect and monitored cases will be harder to track. Indeed, data from numerous states document sizable decreases in maltreatment reports to child protective services since the start of school closures (Baron et al., 2020; Bullinger, Boy, et al., 2020; Bullinger, Raissian, et al., 2020; Eldeib, March 24, 2020; Rapoport et al., 2020).

Current Study

The current study used data from a multi-state web-based survey of households in the northeastern US (Pennsylvania, New York, New Jersey, Connecticut, Rhode Island, and Massachusetts). These states represent a multi-state geographic region that experienced relatively high incidence of COVID-19 (all were in the top 10 for cumulative incidence by April 7, 2020; CDC COVID-19 Response Team, 2020) and (with the exception of Pennsylvania) were early adopters of lockdown and social distancing measures (Allcott et al., 2020). Survey responses, weighted to reflect household characteristics on key dimensions for the region, were used to addresses the following pre-registered hypotheses (https://osf.io/2mscj/?view_only=60a0d589e20a4672a0e2c64fb07a93e9):

(H1) rates of harsh and neglectful parenting would be higher than estimates derived from studies published prior to the current pandemic response; and

(H2) parents who experienced elevated rates of COVID-19 stressors, including higher rates of exposure or greater subjective levels of distress, would report higher rates of engaging in harsh discipline or neglectful parenting.

Parents of children ages birth to 18 in the northeastern US were recruited to participate in the web-based survey study on one of two separate periods, 1–3 months after the onset of public health concerns related to COVID-19, respectively (i.e., April–June 2020).

Method

Survey Sample and Data Collection

Data were collected from an internet panel of parents and caregivers of children ages 0 to 18 years from six states in the northeastern region of the US using a survey developed to assess exposure to COVID-19-related stressors as well as household and parenting experiences. The survey was fielded by a university-based survey research and polling institute, Siena College Research Institute, using Luc.ID, a third-party marketplace research platform that manages an online exchange for survey respondents drawn from more than 250 different sample suppliers. Luc.ID recruits panel participants to complete internet surveys in exchange for points that may be used for online transactions. The panel is monitored to ensure the quality and reliability of its participants, and an independent quality review is conducted quarterly (Lucid Holdings, LLC, 2017). Prior research indicates that demographic and experimental data collected by Luc.ID align with US national benchmarks (Coppock & McClellan, 2019).

Panel participants were asked a series of questions to confirm that they met study inclusion criteria: over 18 years of age, resident of a state in the target region, and parent of a child ages birth to 17 years that resided with the caregiver at least 50% of the time in the prior 30 days. Participants also were required to reference some survey items to a randomly selected child from their household roster. For households with more than one child 15 or younger, parents were instructed to select the child who had most recently had a birthday; if no children in the household were in that age range the same process was applied for children ages 16 or 17.

The survey was fielded on two separate occasions, approximately 40 days apart in April and June 2020. Sample weights were created for each wave through raking, an iterative procedure common in public opinion surveys that adjusts case-level weights to replicate population-level characteristics with a known distribution (Mercer et al., 2018). State-level population indicators from US census data were used to inform raked weight calculation including state-level household count, household income groups, individual race/ethnicity, and community-type (i.e., urban, suburban, or rural designation; GreatData, 2020) based on zip code designation. When applied, these sample weights replicated region-wide population characteristics on these sociodemographic indicators for survey participants.

Study methods were reviewed by the Pennsylvania State University and Yale School of Medicine Institutional Review Boards and classified as exempt. Participants completed an implied informed consent process prior to accessing survey content.

Participants

A total of 2,068 parents completed the survey (Wave 1: 1,019, Wave 2: 1,049). Table 1 summarizes sample characteristics, overall and by wave, with raked weights applied. Nearly two-thirds of respondents were female; three-quarters were married or cohabitating with a partner, and a majority were under 45 years of age. Weighted frequencies of respondent racial/ethnic backgrounds and household income were comparable to census data for the region. Average household composition included 2.14 adults (SD = 0.86) and 1.73 children (SD = 0.84). Average age of randomly selected children was 8.2 years, and 49% were female. Minimal statistically significant differences were observed across wave; wave 1 respondents were more likely to indicate they had been divorced, and their randomly selected child was approximately 7 months older, on average.

Table 1.

Sample Characteristics.

Caregiver Characteristics	Overall Sample (n = 2,068) Weighted %	Wave 1 (n = 1,019) Weighted %	Wave 2 (n = 1,049) Weighted %	χ2(df), p-value
Age				3.99(4), p = 0.41
18–24 years	6.5	7.1	5.9
25–34 years	28.9	27.1	30.6
35–44 years	41.9	42.4	41.4
45–54 years	18.2	18.9	17.5
55 years or older	4.6	4.5	4.6
Gender				1.95(2), p = 0.38
Male	36.7	36.7	36.8
Female	63.2	63.3	63.0
Race
White/Caucasian	75.3	75.2	75.3	0.01 (1), p = 0.93
Black/African American	13.6	13.8	13.4	0.09 (1), p = 0.77
Native American/American Indian	1.5	1.9	1.1	1.74 (1), p = 0.19
Asian/Pacific Islander	6.3	6.3	6.4	0.01 (1), p = 0.94
Other	5.5	5.5	5.6	0.01 (1), p = 0.93
Hispanic Ethnicity	13.2	13.2	13.2	<0.01 (2), p = 1.00
Education Level				0.06 (2), p = 0.97
High School or Lower	19.9	20.1	19.9
Associate or Bachelors	51.7	51.4	52.0
Graduate Degree	27.8	28.0	27.7
Income				<0.01 (5), p = 1.00
Below $10,000	6.2	6.2	6.2
$10,000-$49,000	32.3	32.3	32.2
$50,000-$99,999	27.7	27.7	27.7
$100,000-$149,999	15.2	15.2	15.2
$150,000 or more	15.7	15.7	15.7
Prefer not to answer	2.9	2.9	2.9
Marital Status				14.83(5), p = 0.01
Single (Never Married)	18.7	19.4	18.1
Married	64.4	62.4	66.3
Cohabitating	9.1	8.6	9.5
Divorced	6.2	8.2a	4.3b
Widowed	1.3	1.1	1.4
Selected Child Characteristics	Mean (sd)	Mean (sd)	Mean (sd)	t (df), p-value
	8.2 (5.2)	8.5 (5.3)	7.9 (5.0)	2.89 (2,066), p = 0.004
Age	Weighted %	Weighted %	Weighted %	χ2(df), p-value
Gender				2.81 (2), p = 0.25
Male	51.0	52.5	49.5
Female	49.0	47.5	50.4

Measures

Demographic indicators included respondent characteristics (e.g., age, gender, race, ethnicity, education, marital status), household characteristics (e.g., household income, number of adults and children, state and zip code of residence), and focal child characteristics (e.g., age, gender). Households were classified as urban, suburban, or rural based on population density, distance from the nearest city, and size of the nearest city using participant zip code responses (GreatData, 2020).

COVID-19 Stress was assessed using the COVID-19 Questionnaire (Wadsworth, 2020), which measures exposure to COVID-19 stressors and ratings of distress due to COVID-19. COVID-19 stressors were assessed using a 12-item index reflecting personal and household member (i.e., spouse/partner, child, or other household member) experiences of specific stressors due to COVID-19 or the public health response (e.g., COVID-19 symptoms, testing, diagnosis, treatment; work- and school-related stressors; having an underlying or chronic medical condition). An additional set of questions assessed whether the focal child had experienced school or childcare disruption. Total sum scores (ranging from 0 to 12) reflected personal and other household member levels of exposure.

COVID-19 distress was assessed using 14 questions that measured the level of stress experienced due to COVID-19 concerns. Sample items include, “Reading or watching news updates about COVID-19 events,” “Fears about your own or loved ones’ health,” and “Difficulty finding childcare because of school/daycare closures.” Items were rated on a 4-point Likert-type scale (0 = Not at all stressful to 3 = extremely stressful), with total distress reflecting the mean of scale items. Inter-item reliability was assessed using McDonald’s ω based on principal factor analyses as confirmatory models did not converge at a proper solution for Wave 1; reliability was good across survey waves (Wave 1 McDonald’s ω = 0.90, Wave 2 McDonald’s ω = 0.89).

Parental neglect was assessed using items from the following subscales of the Multidimensional Neglect Behavior Scale (MNBS; Holt et al., 2004): cognitive (e.g., assisting with school work), supervisory (e.g., leaving child alone in car), physical (having enough food in the home), and family-based neglect (e.g., failing to prevent a household member from hurting the child). The MNBS subscales are developmentally tailored to child’s age, with a total of 9 to 10 questions, depending on the age of the child. Respondents indicated the frequency of specific parenting behaviors over the previous 30-day period using a 4-point scale (0 = Never happened, 1 = Sometimes happened, 2 = Frequently happened, 3 = Always happened), and items were scored affirmatively based on item-specific thresholds. One item, taking the child to the doctor for a check-up, was eliminated from the physical neglect subscale due to limited frequency of this activity in a 30-day period. The MNBS has been extensively tested and validated as a self-report measure of parental neglect (Kantor et al., 2004; Straus, 2006).

Parental discipline was assessed using 11 items from the following scales of the Parent-Child Conflict Tactics Scale (PC-CTS; Straus et al., 1995): positive (non-violent) discipline (e.g., using time out), psychological punishment (e.g., swearing or cursing at child), and physical corporal punishment (e.g., spanking with a bare hand). Respondents indicated the frequency of specific parenting behaviors over the previous 30-day period using a 4-point scale (0 = Never happened, 1 = Sometimes happened, 2 = Frequently happened, 3 = Always happened). The PC-CTS has been extensively tested and validated as a self-report measure of parental emotional and physical discipline practices (Straus et al., 1998; Tonmyr et al., 2011).

Analytic Approach

An initial set of analyses were conducted to estimate rates of personal and household exposure to COVID-19 stressors, caregiver ratings of COVID-19 distress, and rates of parental neglectful, harsh, and positive parenting practices, overall (with 95% confidence intervals) and separately for each survey wave. Exploratory chi-square and t-test analyses were conducted to compare rates of exposure to stressors, COVID-19 distress, and neglectful, harsh, and positive parenting practices across survey waves. Exploratory chi-square analyses also were conducted to compare parenting practices between child age groups. While our preregistered plan also stated that we would compare rates of harsh and neglectful parenting from this study to other published rates, we were unable to find published rates that were comparable with respect to reference period (i.e., prior 30 days). We reflect on this further in the discussion.

To evaluate our second hypothesis, our pre-registered analytic plan indicated that we would utilize regression-based models if latent variable models failed to converge. Preliminary modeling indicated issues with convergence for confirmatory factor models of key constructs (e.g., COVID-19 distress), so logistic regression models with observed variables were conducted. A series of pre-registered models were run to estimate the association of COVID-19 stress experiences with neglect (overall and by sub-type), harsh discipline (overall and by subtype), and positive discipline practices to assess whether these relations were specific to negative parenting practices or indicative of increased likelihood of a range of both positive and negative discipline practices. For each outcome, model 1 examined the association of personal and household exposure to COVID-19 stressors with parenting practices, model 2 focused on COVID-19 distress, and model 3 focused on the combination of both COVID-19 stressors and distress. All models incorporated sample weights and pre-registered controls for child age and gender, caregiver race and ethnicity, and household SES. Sample wave was also included as an unregistered covariate to account for differences in sample characteristics across waves. The pre-registered plan called for a formative indicator approach to modeling SES; reported household income was used in the current analyses. This process was replicated for planned analyses to assess the association between stressor exposure and distress with subtypes of neglect and harsh discipline.

Results

Rates of Covid-19 Stress Exposure

Rates of exposure to stressors related to COVID-19 or the pandemic response among respondents or household members are summarized in Table 2. Respondents experienced an average of 2.75 stressors (SD = 1.96), and household members 2.83 stressors (SD = 2.41), with no significant difference across waves. The most frequently reported stressors experienced by respondents included canceling a trip or vacation, financial loss, and being required to work from home. These categories were also among the most frequently reported for other household members, though being required to work outside the home was also elevated. Not surprisingly, wave 2 participants reported greater frequency of being tested for and diagnosed with COVID-19. In addition, wave 2 participants were more likely to report being required to work outside of their home and were less likely to report having canceled a trip or vacation due to the pandemic response; they were also less likely to report that a household member had been laid off or fired due to the pandemic. Ratings of COVID-19 distress did not differ by survey wave (Overall M = 1.79, SD = 0.68). A significant majority of participants (82.3%) indicated that their selected child had experienced a school or childcare closure due to the pandemic, with the rate slightly higher among Wave 1 participants (84.8%) than among wave 2 participants (79.9%; χ²₍₁₎ = 8.38, p = .004).

Table 2.

Exposure to COVID-19 related stressors.

	Personal				Other Household Member
	Overall	Wave 1	Wave 2		Overall	Wave 1	Wave 2
Stressor Items	% Yes [95% CI]	% Yes	% Yes	χ2 p-value	% Yes [95% CI]	% Yes	% Yes	χ2 p-value
Experienced COVID-19-related symptoms	10.7 [9.4, 12.0]	10.4	10.9	0.704	11.8 [10.4, 13.2]	10.8	12.8	0.154
Tested for COVID-19	15.7 [14.1, 17.2]	13.7	17.6	0.015	17.8 [16.2, 19.5]	16.7	19.0	0.174
Diagnosed with COVID-19	3.6 [2.8, 4.4]	2.4	4.8	0.003	6.8 [5.7, 7.9]	6.1	7.5	0.206
Treated at a health care facility due to symptoms	3.7 [2.9, 4.6]	3.4	4.0	0.516	7.7 [6.5, 8.8]	6.7	8.6	0.095
Quarantined (e.g., hospitalization, self-quarantine)	23.7 [21.9, 25.6]	23.0	24.4	0.429	25.5 [23.7, 27.4]	24.4	26.6	0.268
Have an underlying or chronic medical condition	21.0 [19.2, 22.8]	21.5	20.5	0.589	25.7 [23.8, 27.6]	25.5	25.9	0.836
Required to work from home	42.2 [40.1, 44.4]	43.8	40.8	0.173	31.7 [29.6, 33.7]	31.5	31.8	0.884
Required to work outside of home despite risks	22.4 [20.6, 24.2]	20.6	24.3	0.048	32.6 [30.5, 34.6]	31.4	33.7	0.276
Increased risk for exposure due to essential job position	21.6 [19.8, 23.4]	21.1	22.1	0.568	28.6 [26.7, 30.6]	30.0	27.2	0.161
Laid off or fired as result of pandemic	18.0 [16.3, 19.6]	17.0	18.9	0.266	18.7 [17.0, 20.3]	20.4	16.9	0.041
Financial loss from pandemic	42.4 [40.2, 44.5]	43.4	41.3	0.351	33.8 [31.7, 35.8]	35.7	31.8	0.062
Canceled a trip/vacation because of pandemic	50.9 [48.7, 53.1]	53.4	48.5	0.027	43.9 [41.7, 46.0]	44.6	43.1	0.516
	Mean (SD)	Mean (SD)	Mean (SD)	t-test p-value	Mean (SD)	Mean (SD)	Mean (SD)	t-test p-value
Total Stressors (Count)	2.75 (1.96)	2.74 (1.82)	2.76 (2.09)	0.783	2.83 (2.41)	2.84 (2.42)	2.83 (2.41)	0.892
COVID-19 Distress	1.79 (0.68)	1.80 (0.69)	1.77 (0.68)	0.369	–	–	–	–

The number of personal and household COVID-19 related stressors were each moderately correlated with caregiver reports of distress (r = 0.27, p < .001; r = 0.28, p < .001, respectively). In addition, caregivers whose child had experienced a school or childcare closure reported higher levels of pandemic-related stress (M = 1.82 vs. M = 1.63; t₍₂₀₅₇₎ = 4.69, p < .001).

Rates of Neglectful, Harsh, and Positive Discipline

Rates of caregiver neglectful, harsh, or positive disciplinary practices in the prior 30-day period are summarized in Table 3. Overall, more than half of caregivers (55.6%) reported engaging in neglectful parenting behavior in the previous month, with physical neglect and supervisory neglect being most common (33.9% and 31.8%, respectively). Similarly, more than half of caregivers (59.0%) reported using harsh discipline in the previous month, with psychological aggression occurring more frequently than physical discipline (57.2% and 26.9%, respectively). Finally, two-thirds (67.1%) of caregivers reported using positive disciplinary practices in the previous month. Minimal differences were observed for neglectful, harsh, or positive parenting practices across survey waves, except that more wave 2 caregivers reported engaging in family-based neglect (e.g., witnessing violence or failing to prevent family violence against the child).

Table 3.

Parental Use of Neglectful and Harsh or Positive Discipline Behaviors.

	Overall(% Yes) [95% CI]	Wave Comparison			Age Comparison
		Wave 1 (% Yes)	Wave 2 (% Yes)	χ2 p-value	Age 0–4	Age 5–9	Age 10–15	Age 16–17	χ2 p Value
Any neglectful behavior	55.6 [53.5, 57.7]	56.4	54.8	0.460	53.0	58.9	54.5	59.7	0.138
Emotional	4.5 [3.6, 5.4]	4.5	4.4	0.887	3.7a	3.0a	4.6a	12.3b	< 0.001
Cognitive	3.6 [2.8, 4.5]	3.9	3.4	0.551	3.0a	2.6a	3.6a	10.3b	< 0.001
Supervisory	31.8 [29.8, 33.9]	31.3	32.3	0.622	39.1a	42.0a	20.9b	21.3b	< 0.001
Physical	33.9 [31.8, 35.9]	33.7	34.0	0.858	28.3a	37.7b	35.8b	34.4a, b	0.004
Family-based	12.6 [11.2, 14.0]	11.0	14.1	0.033	8.6a	14.1b	14.9b	11.7a, b	0.003
Any harsh discipline	59.0 [56.8, 61.1]	58.7	59.2	0.812	57.1a	67.0b	57.1a	47.4a	< 0.001
Psychological punishment	57.2 [55.1, 59.4]	57.1	57.3	0.935	54.2a	65.3b	56.1a	47.1a	< 0.001
Physical corporal punishment	26.9 [25.0, 28.8]	27.1	26.8	0.879	30.0a	30.3a	23.9a, b	17.5b	0.001
Any positive discipline	67.1 [65.1, 69.2]	65.6	68.6	0.136	64.2a	80.8b	64.6a	44.5c	< 0.001

Note. For age comparisons of individual parent behaviors, each subscript letter denotes a subset of age categories whose proportions do not differ significantly from each other at the .05 level.

Child age was significantly related to all subtypes of neglectful parenting and to both harsh and positive discipline practices. Caregivers reported higher rates of emotional or cognitive neglectful behavior for older adolescents, largely due to decreased assistance with academic work. Rates of supervisory neglect were greatest for children ages 0 to 9, including elevated rates of being able to monitor young children or check-in on school-age children while working, and of leaving school-age children in the car for short periods of time. Physical neglect and family-based neglect were most prevalent for children ages five to nine and 10 to 15. Harsh disciplinary practices, as well as positive disciplinary practices were most common for caregivers of children ages five to nine, and least common for caregivers of older adolescents.

COVID-19 Stress and Rates of Neglectful, Harsh, and Positive Discipline Practices

Logistic regression model results are presented in Table 4 (neglect, overall and by subtype) and Table 5 (harsh discipline, overall and by subtype; positive discipline). Focusing first on neglect, higher levels of personal and household exposure to COVID-19 stressors, as well as caregiver distress related to COVID-19, each were associated with increased risk of overall neglect occurrence in the preceding month when treated individually (models 1 and 2) and when modeled jointly (model 3). Subtype analyses revealed that this overall pattern primarily was due to associations with physical and family-based neglect sub-types. Higher rates of personal and household exposure to stressors, as well as higher ratings of distress were individually and jointly associated with significant increases in risk of occurrence for each. In the joint models, each additional personal stressor increased the odds of physical neglect by 13 percent and family-based neglect by nearly 14 percent, and each additional household member stressor increased odds by approximately six percent and 18 percent, respectively. Odds ratios associated with caregiver distress ratings were even greater. On the other hand, levels of exposure to stressors and COVID-19 distress were not statistically related to cognitive or supervisory neglect in any models, and distress was associated with decreased likelihood of emotional neglect separately and when accounting for household levels of stressor exposure.

Table 4.

COVID-19 Related Stress Experiences and Parents’ Use of Neglectful Parenting (Overall and By Subtype).

	Any Neglect			Any Emotional Neglect			Any Cognitive Neglect
	Model 1a	Model 2a	Model 3	Model 1a	Model 2a	Model 3	Model 1a	Model 2a	Model 3
	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]
Stressor Exposure
Personal	1.075** [1.022, 1.130]	–	1.056** [1.003, 1.111]	1.026 [0.914, 1.152]	–	1.065 [0.946, 1.198]	0.933 [0.814, 1.070]	–	0.949 [0.825, 1.091]
Other Member	1.083*** [1.040, 1.129]	–	1.069** [1.025, 1.114]	0.973 [0.883, 1.071]	–	1.000 [0.905, 1.105]	0.934 [0.834, 1.046]	–	0.944 [0.841, 1.059]
Distress	–	1.416*** [1.241, 1.615]	1.281*** [1.115, 1.472]	–	0.667** [0.500, 0.891]	0.634** [0.464, 0.865]	–	0.747 [0.540, 1.034]	0.815 [0.579, 1.149]
	Any Supervisory Neglect			Any Physical Neglect			Any Family-Based Neglect
Stressor Exposure
Personal	0.986 [0.935, 1.038]	–	0.989 [0.937, 1.043]	1.167*** [1.109, 1.228]	–	1.130*** [1.072, 1.190]	1.172*** [1.097, 1.252]	–	1.137*** [1.063, 1.217]
Other Member	1.018 [0.976, 1.061]	–	1.020 [0.978, 1.065]	1.088*** [1.045, 1.134]	–	1.059** [1.016, 1.104]	1.215*** [1.153, 1.280]	–	1.184*** [1.123, 1.248]
Distress	–	0.960 [0.835, 1.104]	0.950 [0.819, 1.101]	–	1.977*** [1.703, 2.295]	1.716*** [1.468, 2.006]	–	2.318*** [1.846, 2.911]	1.747*** [1.373, 2.224]

Note. ^apre-registered model; All models controlled for sample wave and for child age and gender, parent race/ethnicity, and household income.

^*p < .05.

^**p < .01

^***p < .001.

Table 5.

COVID-19 Related Stress Experiences and Parents’ Use of Harsh and Positive Discipline Practices (Overall and By Subtype).

	Any Harsh Discipline			Any Positive Discipline
	Model 1a	Model 2a	Model 3	Model 1a	Model 2a	Model 3
	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]	OR [95% CI]
Stressor Exposure
Personal	1.049 [0.997, 1.103]	–	1.012 [0.960, 1.065]	1.059* [1.004, 1.118]	–	1.018 [0.963, 1.076]
Other Member	1.099*** [1.054, 1.146]	–	1.070** [1.025, 1.117]	1.111*** [1.062, 1.163]	–	1.080** [1.031, 1.132]
Distress	–	1.760*** [1.538, 2.015]	1.643*** [1.425, 1.895]	–	1.826*** [1.589, 2.099]	1.688*** [1.458, 1.954]
	Any psychological punishment			Any physical corporal punishment
Stressor Exposure
Personal	1.053* [1.002, 1.107]	–	1.017 [0.966, 1.070]	1.055* [1.001, 1.113]	–	1.028 [0.973, 1.085]
Other Member	1.075*** [1.032, 1.120]	–	1.048* [1.005, 1.093]	1.154*** [1.106, 1.203]	–	1.130*** [1.082, 1.179]
Distress	–	1.708*** [1.494, 1.953]	1.617*** [1.404, 1.862]	–	1.748*** [1.491, 2.049]	1.519*** [1.284, 1.796]

Note. ^apre-registered model; All models controlled for sample wave and for child age and gender, parent race/ethnicity, and household income.

p < .05.

^**p < .01.

^***p < .001.

With respect to parental discipline practices (Table 5), our results indicated that higher levels of household exposure to COVID-19 stressors and higher levels of caregiver distress associated with COVID-19, each were associated with increased risk of overall harsh discipline occurrence in the preceding month when treated individually (models 1 and 2) and when modeled jointly (model 3); personal levels of exposure to stressors did not predict occurrence of harsh discipline. A similar pattern was also observed for each subtype of harsh discipline, although personal levels of exposure did significantly predict increased risk of both harsh psychological and physical discipline in models that did not include personal distress. In the joint models, each additional type of stressor experienced by a household member increased the odds of harsh psychological discipline by nearly five percent and harsh physical discipline by 13 percent. As with neglect, the association between caregiver distress and each form of harsh discipline was of even greater magnitude.

Finally, the results for likelihood of positive discipline mirrored those for each subtype of harsh discipline. Specifically, levels of personal and household member exposure to COVID-19 related stressors each were associated with increased likelihood of parents’ use of positive discipline strategies when modeled separate from distress (model 1), but only other household member levels were associated with such behavior when distress was also included (model 3). Caregiver distress was consistently associated with increased likelihood of using positive discipline strategies when modeled separately (model 2) and jointly (model 3). The effect sizes observed for positive discipline practices were comparable to those observed in the harsh discipline models both in terms of magnitude and direction.

Discussion

To our knowledge, this study represents the largest regional investigation, to date, of the association between caregiver and household experiences of COVID-19 and the related public health response to parents’ use of neglectful, harsh, and positive discipline practices. In addition, the study incorporated rigorous sample weighting methods to reflect a large region of the US (northeastern US) heavily impacted by COVID-19 and social distancing measures during the period of investigation. Our results provide evidence that caregivers: (a) have experienced numerous stressors due to COVID-19, directly and indirectly via other members of their household; (b) engaged in neglectful and harsh discipline practices, as well as positive discipline practices, in the months following onset of the pandemic and associated public health response; and (c) that greater levels of stressor exposure and of COVID-19 related distress were associated with increased likelihood of engaging in neglectful, harsh, and positive discipline strategies among caregivers.

With respect to caregiver and household experiences of COVID-19 related stressors, these included both health-related concerns as well as disruptions in economic, work, and school or childcare related domains. Although direct experiences of health-related consequences were infrequent (e.g., caregivers reported that fewer than 4% had personally been diagnosed with COVID-19 and only 7% had a member of their household experience a diagnosis), many households had experienced self-quarantine, altered planned activities, or were coping with financial loss and employment-related challenges; and over 80 percent had experienced school or childcare closures. These cumulative experiences, either personally or among members of their households, contributed to elevated ratings of caregivers’ distress related to the pandemic and the public health response.

A majority of caregivers reported engaging in some level of neglect and harsh discipline practices. Physical and supervisory neglect were most common across all child age groups, though age-specific patterns of exposure were observed. Caregivers of younger children, ages 9 and below, were more likely to report engaging in supervisory neglect than caregivers of adolescents, whereas caregivers of school-age children and young adults were most likely to report physical neglect. This later group was also most likely to report family-based neglect behaviors. With respect to harsh discipline practices, harsh psychological behaviors (e.g., threats and insults) were more likely to be reported than physical discipline (e.g., spanking or slapping) for all age groups. School-age children (ages 5 to 9) were most likely to experience these types of behaviors, while older adolescents were least likely.

It is difficult to contextualize these past-month rates of parenting behaviors relative to estimates prior to COVID-19, as most studies that assess these behaviors examine past-year prevalence. Maguire-Jack and Font (2017), for example, reported significantly higher rates of physical discipline and both physical and supervisory neglect, but their estimates were based on past-year behavior and lowered the threshold for identification of these behaviors (i.e., any occurrence rather than item-specific thresholds recommended by measure authors). Conversely, our estimates of neglect were significantly higher than those reported in a Canadian study that assessed past-year prevalence (Clément et al., 2016). The authors also recently published a study of post-COVID neglect behaviors in Canada that did examine past month behaviors (Bérubé et al., 2020), but the authors did not estimate prevalence rates given the non-representative nature of their sample (Bérubé, personal communication, Jan. 20, 2021). Finally, our estimated rates were greater than those of a nation-wide web-based survey conducted post-COVID (Lawson et al., 2020). That study estimated past week harsh psychological and physical discipline rates using the PC-CTS at 47.7 and 17.8 percent, respectively.

Our logistic regression results demonstrate the strong association between household exposure to COVID-19 stressors, caregiver distress related to the pandemic and public health response, and multiple forms of neglect, as well as both harsh and positive discipline practices. Higher rates of personal and household exposure to stressors was associated with overall likelihood of neglect and with physical or family-based neglect, specifically. In addition, higher rates of household exposure to stressors was associated with likelihood of harsh discipline, overall and by subtype, and with positive discipline. Personal exposure was not directly related to harsh or positive discipline practices in models that accounted for COVID-19 distress, but it was positively correlated with that distress, which uniquely predicted elevated rates of overall neglect (and both physical and family-based sub-types), overall harsh discipline (including both psychological and physical forms), and with positive discipline practices. Interestingly, higher rates of distress were associated with lower likelihood of emotional neglect, which may suggest that caregivers compensated for higher rates of distress by engaging in some practices to spend quality time with their children or communicate feelings of warmth and support.

Taken together, our results highlight the relation of household stressors (including health, employment, and school or childcare disruptions) and associated distress with increased likelihood of both neglectful and harsh parenting practices. The association of COVID-19 stressors and distress with physical neglect could suggest that families coping with economic and other disruptions may be facing challenges providing for basic needs (e.g., having sufficient food). Further research is needed to better understand how COVID-19 stressors are affecting issues such as economic strain and food insecurity for parents and families who have experienced job loss or instability—particularly for those households that rely on economic supports (e.g., TANF), or nutritional supports (e.g., SNAP, free-or-reduced lunch, etc.).

The relation of stressors and distress with family-based neglect (which include potential exposure to partner-related aggression or failure to prevent others in the household from hurting the child) may suggest that this distress is increasing risk of couple-related conflicts. This is also paralleled by the increased risk of physical discipline practices. Numerous researchers have noted that pandemic-related concerns may also affect partner conflict and risk of family violence (Piquero et al., 2020; Sharma & Borah, 2020). It is critical that research investigate the mechanisms that contribute to this increased risk in the face of public health emergencies. The Family Stress Model (Conger et al., 2000), for example, posits that an external stressor (e.g., COVID-19) may affect parenting by contributing to economic strain, marital or couple distress, and parent-related stress challenges that all increase risk of adverse parent-child interactions (e.g., including neglectful or harsh practices). More research is needed to understand whether these multiple pathways are implicated in risks to children during the COVID-19 response.

Lastly, the observed associations between stressors and distress with positive discipline practices merits further investigation. It is worth noting that COVID-19 stressors and distress not only predicted increased risk of potentially negative parenting practices, but also that it increased likelihood of parents’ use of positive practices as well. This suggests that the distress of the pandemic and public health response may be leading to increased likelihood of parent-child conflicts, and that parents are likely to respond with a variety of discipline practices – some negative (e.g., threats, physical discipline) and some positive (e.g., time out, rewards for good behavior). It may be that parents who have positive parenting practices in their repertoire are more likely to rely on these strategies to address conflicts. Future research is needed to understand the factors that contribute to a positive response to distress rather than a harsh response and whether supports can be offered to parents to minimize reliance on more negative practices.

Although this study draws on a large, population-based sample of a US region impacted during the early phase of the COVID-19 pandemic to investigate the association with parenting practices, our findings should be interpreted in the context of several potential limitations. First, parenting behaviors were based on self-report—raising potential concerns about the accuracy of reporting past-month neglectful, harsh, or positive parenting practices. Survey participants were informed that their identities were anonymous and could not be linked with responses to minimize concern about the nature of their responses, but it is possible that parents responded in socially desirable fashion to describe their practices. Second, the sample was drawn from a web-based marketplace research platform that may not reflect the general population with respect to access to a computer for participation and motivation to participate in survey-based research. The third-party platform draws survey respondents from more than 250 different sample suppliers, conducts regular and independent quality reviews to monitor panel characteristics and reliability of its participants, and has been shown to align with US national benchmarks in previous research (Coppock & McClellan, 2019). The organization that fielded the survey also provided raked weights which, when applied, produced demographic estimates consistent with census figures for the survey region.

Another potential limitation is that survey participation was limited to households in the northeastern US region. This region was heavily affected by COVID-19 during the period of investigation, and it is possible that results may not generalize to other regions or to those that were less directly impacted. Though reflective of the region on important dimensions (e.g., socioeconomic, racial and ethnic composition, community type, etc.), these characteristics may differ significantly for other regions of the US. Further investigation of rates of exposure to COVID-19 stressors and the effects on parenting practices in other regions is needed. The cross-sectional design of the survey is another potential limitation. Although data collection was completed in two separate waves to investigate potential changes over time, each wave was independent, and participants were not followed over time. Thus, we are unable to assess the temporal relationships between COVID-19 stressors and subsequent distress or the causal effects on parenting behaviors. Finally, the study was initiated after the onset of COVID-19 and the public health response and limited to a relatively brief time period to assess associations to parenting (i.e., the previous 30 days). Thus, we are unable to assess whether COVID-19 stressors and the public health response contribute to increased rates of neglect or harsh discipline – as we have no similar estimates of rates of these behaviors in preceding periods.

Despite these limitations, our results provide some clear guidance for prevention and intervention strategies to reduce risk of maltreatment among households disproportionately affected by COVID-19 and its repercussions on caregiver and household stressors. First, the significant associations of cumulative household stressors, including economic and employment disruption, as well as school and childcare closures, highlights the need for robust assistance for families in need. Federal and state efforts to expand economic supports, including eligibility for cash and food assistance (TANF, WIC, SNAP, and P-EBT programs) or subsidized and emergency childcare resources may be useful strategies to reduce COVID-19 effects on household stress. However, these programs and policies have suffered from uneven implementation and adoption (Patrick et al., 2020; Shantz et al., 2020) and more should be done to expand and target these resources to at-risk families and households. Second, while official records demonstrate that, coincident with school and childcare closures, rates of CPS reporting have declined, our results indicate that risk of neglect and harsh psychological and physical parental behaviors persists. School and childcare sites are a primary source of child maltreatment identification and reporting, and alternative means of detecting risk of child maltreatment must be enacted when these settings are closed or transition to remote supports. These strategies could include training education personnel to identify signs or risks of maltreatment during remote lessons or virtual check-ins (Thomas et al., 2020), and expanding opportunities for such check-ins among other professionals, particularly those that may be required rather than voluntary (e.g., medical personnel, public benefit reauthorizations; Font, 2021). Finally, availability of virtual supports (e.g., online resources, telehealth) and hybrid programs (e.g., direct support coupled with virtual check-ins) may offer a means to support parents and reduce risk of child maltreatment. Harris et al. (2020) revealed that such programs can be effective at improving parenting behaviors including disciplinary strategies and, when direct contact is involved, also reduce parental stress.