Sage Journals: Discover world-class research

Abstract

Arguments about the associations between child maltreatment and empathy remain controversial. This systematic review and meta-analysis aim to estimate the direction and magnitude of the relationships between child abuse and neglect and empathy. Four English databases (Web of Science, PsycInfo, PubMed, and Cochrane Library), three Chinese databases (China National Knowledge Infrastructure, Wanfang, and Weipu), and grey literature were systematically searched. We extracted data related to the associations between child maltreatment and empathy and pooled them using random effects models. A total of 24 eligible studies involving 22,580 participants and 176 estimates were included in the analyses. Overall, child maltreatment was significantly related to reduced empathy ( $\overline{r}$ = −0.15, 95% CI [−0.17, −0.13]). Specifically, the rank-order mean effect size of subtypes of maltreatment on empathy is: emotional neglect ( $\overline{r}$ = −0.18, 95% CI [−0.25, −0.12]) > physical neglect ( $\overline{r}$ = −0.16, 95% CI [−0.23, −0.09]) > emotional abuse ( $\overline{r}$ = −0.15, 95% CI [−0.21, −0.10]) > physical abuse ( $\overline{r}$ = −0.12, 95% CI [−0.17, −0.07]) > sexual abuse ( $\overline{r}$ = −0.07, 95% CI [−0.14, −0.01]). Furthermore, the meta-regression results suggested that the aggregated associations between child maltreatment and empathy were not inflated by publication bias, but they were moderated by the age of respondents, sample size, publication language, empathy measurement, and maltreatment type. The results indicate that general maltreatment and its subtypes are associated with reduced empathy. Parent training and empathy enhancement programs should be designed to help children with a history of childhood maltreatment, respectively, from an early stage.

Keywords

child abuse child neglect empathy moderator meta-analysis

Introduction

Childhood maltreatment is a significant global public health concern causing tremendous damage to the short-term as well as long-term well-being of children. It typically includes five forms of maltreatment: physical abuse, emotional abuse, sexual abuse, physical neglect, and emotional neglect (Bernstein et al., 2003; Gilbert, 2009). A series of meta-analyses have revealed that child abuse and neglect are positively associated with both internalizing problems, such as depression and anxiety (for a review, see Gardner et al., 2019), and externalizing problems, including aggression, violence, and suicidal behaviors (Angelakis et al., 2020; Maas et al., 2008; Miller et al., 2013). Additionally, child abuse and neglect in early life have been found to be positively associated with poor adolescent and adult interpersonal functioning (Davis & Petretic-Jackson, 2000; Haslam & Taylor, 2022).

The main line of studies worldwide has demonstrated that child maltreatment can impair children’s empathy development (Chen et al., 2016; Marta et al., 2018; Ometto et al., 2015). Empathy is the ability to recognize and understand other people’s thoughts, share their feelings, and accordingly respond with an appropriate emotion (Baron-Cohen & Wheelwright, 2004); it involves both cognitive and emotional processes and is other-focused (Davis, 1980; Jolliffe & Farrington, 2004). Cognitive empathy refers to the ability to place oneself in another’s shoes, imagine their emotional states, and predict their behaviors (Ang et al., 2010; Baron-Cohen, 1995). Affective empathy, meanwhile, is the ability to experience another’s mental states and respond with an appropriate emotion (Jolliffe & Farrington, 2006). The negative impacts of child maltreatment on empathy have been reported in both the general population and clinical samples (Meidan & Uzefovsky, 2020; Sun et al., 2020; Wang et al., 2021).

Why does child maltreatment reduce empathy? Previous studies have explored several lines of inquiry to explain the impacts of child abuse and neglect on empathy (Bertsch et al., 2013; Locher et al., 2014; Yang et al., 2019). First, neuroscientific evidence suggests that empathy is shaped by biological dispositions and caregiving experiences (Levy et al., 2019). Early traumatic experiences impact the developing brain, which might result in a dysfunctional superior temporal sulcus and gyrus (Roepke et al., 2013), and they also altered brain function that underlie deficits in affective facial expressions and cognitive empathy (Read et al., 2001). Second, attachment theory argues that early child abuse and neglect are the source of insecure attachment (Bifulco et al., 2002; Bowlby, 1977; Bowlby, 1982) and that the ability to identify and understand another person’s emotions is rooted in an individual’s attachment experience (Gobodo-Madikizela, 2008). Individuals with secure attachment to their parents are more likely to care for others and display empathetic behaviors (Mikulincer et al., 2013); in contrast, maltreated children tend to refuse to take their parents’ perspective or understand others’ emotions (Parlar et al., 2014). Third, social learning theory posits that parents are role models for their children’s development of empathy. Negative parenting behaviors, including abuse and neglect, reflect insensitivity and a denial of children’s needs, which can inhibit children’s empathy development (Yu et al., 2012).

However, the associations between the subtypes of child maltreatment and different components of empathy (e.g., perspective-taking, fantasy, empathic concern, and personal distress) remain inconclusive. For instance, using the Interpersonal Reactivity Index (IRI) to measure empathy, Xu et al. (2010) observed that physical abuse and physical neglect were negatively associated with perspective-taking unlike emotional abuse, emotional neglect, and sexual abuse. Furthermore, all five types of abuse and neglect were positively associated with personal distress but not with fantasy. Other studies have found that emotional abuse, emotional neglect, and sexual abuse are negatively associated with perspective-taking and empathetic concern but not with fantasy (Chen et al., 2016; Dittrich et al., 2019). Using the Basic Empathy Scale (BES), some studies have demonstrated that psychological maltreatment is negatively associated with both cognitive and affective empathy (Li, 2016; Sun et al., 2020). More research has shown that childhood trauma, including emotional maltreatment, is significantly and negatively associated with cognitive empathy but not significantly associated with affective empathy (Meidan & Uzefovsky, 2020; Wang, 2021).

In contrast, another line of studies has indicated that early trauma is positively associated with empathy. Specifically, Greenberg et al. (2018) found that the experience of sexual abuse before the age of 17 years was positively associated with affective empathy and global empathy during adulthood but not associated with perspective-taking, empathic concern, fantasy, or personal distress. The authors also found that other types of child abuse were positively associated with global empathy but not with either cognitive or affective empathy. Meanwhile, using a sample of college students, Ozdemir and Sahin (2020) observed that the five types of child maltreatment were all positively associated with global empathy. The different findings regarding the relationship between child maltreatment and empathy may be related to the onset of trauma, measurement inconsistencies, and the various samples used.

Given the mixed results on the associations between child abuse and neglect and empathy, there is a need to synthesize the available evidence for assessing these associations. To the best of our knowledge, no published research has meta-analytically explored the relationship between child maltreatment and empathy to date, although increasing theories and empirical studies have emerged recently. Thus, the present study aims to estimate the associations between childhood maltreatment and empathy using a systematic approach. Identifying the associations between the different types of childhood maltreatment and empathy is important to allow both researchers and practitioners to design and implement tailored child protection and empathy enhancement programs.

Method

Data Sources and Study Selection

A literature search was performed following the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (Moher et al., 2009). This comprehensive literature search was conducted using four English databases (PsycInfo, PubMed, Cochrane Library, and Web of Science), three Chinese databases (Chinese National Knowledge Infrastructure, Wanfang, and Weipu) and grey literature (e.g., unpublished conference papers and dissertations) since their inception until March 1, 2022. The Chinese databases were specifically selected to improve the diversity of included samples, and the authors of this review have rich experience in Chinese literature and thus are well positioned to contribute.

The following search terms and algorithm were used in both the English and Chinese databases: if (child* maltreat* OR child* neglect OR child* abuse OR child* mistreat* OR child* physical abuse OR child* emotional abuse OR child* sexual abuse OR child* physical neglect OR child* emotional neglect) AND if (empathy). This process yielded 1,429 articles.

Inclusion Criteria

Two researchers deleted any duplicate results, independently reviewed the titles and abstracts of the retrieved literature, and read the full text to select eligible studies. The meta-analysis included studies that met the following criteria: (1) they were original quantitative studies, (2) some of or all the samples had experienced one or more of the five types of child maltreatment, (3) child maltreatment and empathy were assessed with reliable and valid measurement tools, and (4) the association between child maltreatment and empathy was presented in any form of data (i.e., correlation coefficient, regression coefficient, or t-test values). Studies were excluded if they: (1) were not quantitative research, (2) were not in English or Chinese, (3) used overlapped samples, (4) did not focus on child maltreatment, (5) focused on the empathy of professionals, (6) did not use any scale to measure child maltreatment or empathy, or (7) did not present enough information on the relationship between child maltreatment and empathy. A total of 24 independent studies were included in the final analysis (see Figure 1).

Figure 1.

PRISMA flow diagram.

Data Extraction and Coding

We first extracted the following data from the eligible articles using a standardized form: (1) authors and publication year of the study; (2) sample characteristics, including sample size, study region, age (mean), and percentage of female participants; (3) characteristics of the methods used, including study design (cross-sectional or longitudinal), sampling methods (probability sampling or not), report category (parents or children), and tools used to measure child maltreatment (e.g., childhood trauma questionnaire [CTQ]; adverse childhood experience questionnaire) and empathy (e.g., IRI; BES); (4) data characteristics of the two core variables, including the scores of the scales (mean and standard deviations) and the effect sizes (correlation coefficients, regression coefficients, or t-test values) of the relationship between child maltreatment and empathy.

Finally, the quality of each eligible article was assessed based on a five-item checklist (Chen et al., 2019), as follows: (1) Was the research question clearly stated? (2) Did the study report its sampling procedure? (3) Was the study’s response rate higher than 60%? (4) Did the study define child maltreatment and empathy? (5) Was the association between child maltreatment and empathy presented objectively? The scores of each item ranged from 0 to 1, depending on whether the criterion was met or not. The total quality score ranged from 0 to 5, based on the mean scores given by two researchers who evaluated each item independently. If the score was higher than 3 points, the article was considered to be of high quality.

Data Analysis Procedures

Effect size calculations

Since child maltreatment and empathy are both typically measured continuously, effect sizes were represented as Pearson correlations (r). However, different estimation procedures had been employed in the empirical literature on child maltreatment and empathy (e.g., partial correlation, linear regression, F test, and others), meaning that the interpretation of the estimations varies greatly. This is a common situation in meta-analyses, the common solution to which is to convert the respective estimates into partial correlation coefficients. Therefore, we first transformed all effect sizes for analysis from Pearson’s r to Fisher’s Z to provide an approximately normally distributed metric and then transformed them back into $\overline{r}$ for reporting (Dryburgh et al., 2021; Lipsey & Wilson, 2001; Rosenthal, 1991). This process is demonstrated in equations (1)–(3) below, where r is the correlation coefficient and log_e is the natural logarithm, ESz_r is the corresponding individual or mean z_r-transformed correlation, SEz_r is the standard error of ESz_r, and n is the total sample size:

{ES}_{r} = r

(1)

ES z_{r} = 0.5 \log_{e} [\frac{1 + {ES}_{r}}{1 - {ES}_{r}}]

(2)

S E z_{r} = \frac{1}{\sqrt{n - 3}}

(3)

The number of effect sizes of the associations between the subtypes of child maltreatment and the different components of empathy reported in the literature varied greatly (e.g., the number of effect sizes of physical abuse on general empathy was eight, while the numbers of effect sizes of physical abuse on perspective-taking, fantasy, empathic concern, personal distress, cognitive empathy, and affective empathy were four, three, four, three, two, and two, respectively). As suggested by Card (2012), too few effect sizes (less than five studies) can affect the accuracy of the estimate; thus, we aggregated all the effect sizes of child maltreatment on empathy according to the different types of child abuse and neglect. It should be emphasised that a complication arises because increases in the empathy variable can mean an improvement or decline in empathy, depending on how it is measured; for example, a positive correlation between physical abuse and personal distress indicates that child maltreatment has negative effects on empathy. To achieve a consistent sign of the effect across the studies, we standardized the signs of the estimates so that a negative estimate implied that an increase in child maltreatment was associated with a decrease in empathy. Meanwhile, since the studies varied in design and methodology, random effects (RE) models were employed. Cohen’s (1988) guidelines (i.e., correlation values of 0.10, 0.30, and 0.50 in absolute value should be interpreted as “small,” “medium,” and “large” effects, respectively) were used to interpret the magnitude of the effect size for significant correlations.

Heterogeneity

I² statistics were calculated for each pooled estimate to determine heterogeneity. The I² statistic is the proportion of the observed variance that reflects variance in true effect sizes. I² values <25%, <50%, and ≥75% represent low, moderate, and high levels of heterogeneity, respectively (Higgins et al., 2003).

Publication bias

No consistent methods of identifying and correcting publication bias have been presented in the meta-analysis literature; thus, we took several steps to check for publication bias. First, an informal “funnel plot” test was employed to determine any publication selection bias. A funnel plot graph estimated the effects against the standard error of Fisher’s Z, with the most precise estimates at the top. If estimated effects are all drawn from the same normal distribution, the most precise estimates should be closely clustered around the mean of that distribution, with less precise estimates fanning out, thus producing a funnel formation (Sterne & Egger, 2001). If the funnel plot is “lopsided,” with estimates favoring one side or the other, especially for less precise Fisher’s Z values, it indicates evidence of publication selection bias (Gunby et al., 2017).

Second, the classic fail-safe N, Egger’s tests, and the trim-and-fill method were employed. The fail-safe N calculated the number of potential unpublished studies with nonsignificant findings required to reduce the pooled effect size in the meta-analysis to below the level of significance (Zhang et al., 2021). In accordance with the established principles, selective publication bias exists when the fail-safe N is less than 5 times the number of published studies plus 10 (Rosenthal, 1979). Egger’s tests were also conducted to assist in funnel plot interpretation, which are statistical tests used to detect funnel plot asymmetry (Egger et al., 1997). We also conducted trim-and-fill analyses to impute the missing effect sizes for filling in asymmetrical areas of the funnel plot and then recalculated the overall effect size to test for publication bias (Duval & Tweedie, 2000). If the combined effect size estimate does not change significantly compared with the real outcome, it indicates that the influence of publication bias is not large, and that the result is relatively credible (Steichen, 2001).

Third, based on the above tests, as a more precise check, we aggregated all the effect sizes together. A commonly employed statistical test of publication bias is the Funnel Asymmetry Test (FAT), which involved adding the standard error of the estimated effect to the regression specification of equation (4) and testing for its significance. In this analysis, the estimated effect $r_{i}$ became the dependent variable. An ordinary least squares estimation of ‬α₀ in the below equation produces a value equivalent to the arithmetic average of the M estimates. In other words, α₀represents the mean true effect of child maltreatment on empathy while bias-adjusted, and α₁estimates the publication bias (Stanley & Doucouliagos, 2012; Xue et al., 2020).

r_{i} = α_{0} + α_{1} SE (r_{i}) + ε_{i}, i = 1, 2, ..., M

(4)

A publication bias test is $H_{0} : α_{1} = 0$ . The rejection of this hypothesis indicates that the studies in the meta-analysis sample are affected by publication bias. The inclusion of the SE variable also serves to control for the influence of publication bias on the estimate of the mean true effect α₀. Thus, the estimate of α₀ in the specification of equation (4) represents the bias-adjusted estimate of the mean true effect of child maltreatment on empathy. A test of the significance of α₀ is known as the precision effect test (PET). Rejection of $H_{0} : α_{0} = 0$ indicates that the mean true effect of child maltreatment on empathy is nonzero.

Meta-regression analysis

To determine whether there were heterogeneities and the effect of the different study characteristics on the mean true effect, we needed to investigate factors that affected the size of the mean effect. As abovementioned, the number of effect sizes of the associations between the subtypes of child maltreatment and the different components of empathy in the literature varied widely; therefore, we examined moderators at the aggregated level, adding potential moderator variables into the specification of equation (5):

r_{i} = α_{0} + α_{1} S E_{i} + \sum_{k = 1}^{k} α_{k + 1} X_{k i} + ε_{i}

(5)

Here, $X_{k}$ refers to the potential moderator variables, k = 1, 2, . . .K. The coefficient $α_{k + 1}$ captures the change in the effect of child maltreatment on empathy due to $X_{k}$ , where a negative coefficient indicates that analyzed studies that have characteristic $X_{k}$ estimate a larger negative effect of maltreatment on empathy, and vice versa. This also means that the presence of characteristic $X_{k}$ increases the heterogeneity of the pooled estimate.

Furthermore, when we ran the FAT test and meta-regression analysis, a related issue emerged concerning the weighting of the estimates of each study. If all the effects were estimated with equal precision, the meta-regression estimators would give equal weight to every (respectively standardized) estimate. For example, suppose there are two studies: Study A reports one estimate and Study B reports five estimates. A consequence of giving equal weights to the observations is that Study B will be weighted 5 times as much as Study A. An alternative approach is to provide equal weights to each study. In this case, the standardized estimates are further weighted by the inverse of the number of estimates reported by that study. Thus, each standardized estimate from Study B would receive one-fifth the weight of its counterpart from Study A. In the analysis reported below, these weighting schemes were identified as Weight 1 (equal observations) and Weight 2 (equal studies; Doucouliagos & Paldam, 2013). Our study used both for comparing the results and ensuring the robustness of the analyses.

Results

Study Characteristics

Based on the inclusion criteria, 24 studies with 22,580 participants were included in the analyses (see Table 1 for details of the included studies). The number of effect sizes ranged from 1 to 30 per article, providing a total of 176 effects coded for the meta-analysis. Table 2 reports the study characteristics that were included in our analysis. Among the studies, 78% used mixed-gender samples, 61% used adult samples, and 56% used samples larger than 300. The study’s participants were from three continents (Asia: 61%; Europe: 11%; North America: 28%). Among the studies, 50% were written in English, and 50% were in Chinese. In terms of measurement tools, 67% of the studies used the CTQ or CTQ–Short Form, and 33% of the studies used the IRI or IRI-Short Form. The analyzed studies reported on the relationships of empathy with general child maltreatment (58%), physical abuse (33%), emotional abuse (29%), sexual abuse (42%), physical neglect (29%), and emotional neglect (38%).

Table 1.

Summary of Study Characteristics.

Study No.	Author (Year)	Region	SampleSize	Age (Mean)	Gender(% Female)	Probability Sampling	Child Maltreatment Measurement	Empathy Measurement	Study Quality
1	Bartlett (2012)	United States	447	18.73	100	Yes	CTQ	AAPI-2	High
2	Bujalski et al. (2017)	United States	2551	18.5*	73.8	No	CCMS-A	IRI	High
3	Chen et al. (2016)	China	194	28.6	54.08	No	CTQ	IRI-C	Medium
4	Dittrich et al. (2019)	Germany	251	38.84	53.4	No	CECA	IRI	High
5	Fang et al. (2020)	China	839	19.96	69.7	No	CPM	BES	High
6	Gander et al. (2020)	Austria and Southern Germany	199	15.6*	66.8	No	CTQ	LoPF-Q	Medium
7	Greenberg et al. (2018)	United Kingdom	442	34.84	62.0	No	CTES	IRI	High
7	Greenberg et al. (2018)	United Kingdom	86	34.94	62.0	No	CTES	EQ	High
8	Han & Shao (2019)	China	450	19.69	57.3	No	CTQ-SF	BES	High
9	Li (2016)	China	555	11.68	52.43	No	CTQ-SF	IRI-C	High
10	Li (2016)	China	540	13.2	48.9	No	CTQ-SF	BES-C	High
11	Locher et al. (2014)	South Africa	49	40.0*	50.0*	No	CTQ-SF	SRS	Medium
12	Meidan and Uzefovsky (2020)	Israel	462	32.5*	48.0	No	B-CAP	IRI	High
13	Narvey et al. (2021)	United States	11000	16.5	14.4	Yes	ACE	R-PACT	High
14	Ometto et al. (2015)	Germany	107	12.85	43.93	No	CTQ	SSIA	High
15	Ozdemir & Sahin (2020)	Turkey	926	22.0	76.9	Yes	CTQ	SDIR	High
16	Parlar et al. (2014)	Canada	49	39.77	100	No	CTQ	TEQ	Medium
17	Ren et al. (2018)	China	574	34.0*	35.89	No	CTQ	EQ-C	High
18	Simons et al. (2002)	United States	188	39.0	0	No	SHD	CE Test	High
18	Simons et al. (2002)	United States	188	39.0	0	No	SHD	EFW Test	High
19	Sun et al. (2020)	China	626	20.39	41.1	Yes	CPMS	BES	High
20	Wang et al. (2021)	China	257	33.41	0	No	CTQ-SF	QCAE	High
21	Xu et al. (2010)	China	182	28.21	0	Yes	CTQ	IRI	Medium
22	Yang et al. (2019)	China	125	20.18	51.2	No	CTQ	IRI	Medium
23	Yin & Yao (2021)	China	584	16.27	60.45	No	CPMS	IRIS	High
24	Yu et al. (2020)	China	897	15.41	54.2	No	CTQ-SF	BES	High

Note. CTQ = childhood trauma questionnaire; CCMS-A = comprehensive child maltreatment scale-adult version; CECA = childhood experience of care and abuse interview; CPM = childhood psychological maltreatment; CTES = childhood traumatic events scale; CTQ-SF = childhood trauma questionnaire-short form; B-CAP = brief child abuse potential inventory; ACE = adverse childhood experiences; SHD = redirecting sexual aggression sexual history questionnaire; CPMS = the children’s psychological maltreatment scale. Empathy measurement tools: AAPI-2 = adult-adolescent parenting inventory; IRI = interpersonal reactivity index; IRI-C = interpersonal reactivity index-Chinese version; IRIS = interpersonal reactivity index-short form; BES = basic empathy scale; LoPF-Q = levels of personality functioning questionnaire; EQ = empathy quotient; BESC = basic empathy scale-C; SRS = self-reported emotion scales; R-PACT = residential positive achievement change tool; SSIA = residential positive achievement change tool; SDIR = scale of dimensions of interpersonal relationships; TEQ = the toronto empathy questionnaire; EQ-C = the toronto empathy questionnaire-C; CE Test = child empathy test; EFW Test = empathy for women test; QCAE = the questionnaire of cognitive and affective empathy.

The values are estimations. Child maltreatment measurement tools.

Table 2.

Description of Study Characteristics.

Variable	Description	Mean	Min	Max
Gender
Mixed gender	=1, if the respondents were mixed gender	0.78	0	1
Age
≥18 years old	=1, if the average age of respondents was 18 years and above	0.61	0	1
Sample size
>300	=1, if the sample size was greater than 300	0.56	0	1
District
Asia	=1, if the country studied is located in Asia	0.61	0	1
Europe	=1, if the country studied is located in Europe	0.11	0	1
North America	=1, if the country studied is located in North America	0.28	0	1
Language of publication
English	=1, if the publication language was English	0.50	0	1
Childhood maltreatment and empathy measures
CTQ	=1, if maltreatment was measured with CTQ (childhood trauma questionnaire) scales	0.67	0	1
IRI	=1, if empathy was measured with IRI (interpersonal reactivity index) scales	0.33	0	1
Type of Associations
General Maltreatment	=1, if the study focused on the associations between general child maltreatment and empathy	0.58	0	1
Physical Abuse	=1, if the study focused on the associations between physical abuse and empathy	0.33	0	1
Emotional Abuse	=1, if the study focused on the associations between emotional abuse and empathy	0.29	0	1
Sexual Abuse	=1, if the study focused on the associations between sexual abuse and empathy	0.42	0	1
Physical Neglect	=1, if the study focused on the associations between physical neglect and empathy	0.29	0	1
Emotional Neglect	=1, if study focused on the associations between emotional neglect and empathy	0.38	0	1

Note. The categories omitted in the table are the benchmark categories.

Associations Between Child Maltreatment and Empathy

The results of the associations between each subtype of child maltreatment and the different domains of empathy, including the number of estimates (n), mean effect size ( $\overline{r}$ ) and 95 % confidence interval (95% CI), and heterogeneity statistics (I²) are reported in Table 3. The I² statistics of the relationships between the different types of child maltreatment and aggregated empathy all exceeded 75%, implying heterogeneity in the meta-analysis. Thus, we chose the RE model to conduct the statistical analysis.

Table 3.

Pooled Effects of Effect Size and Heterogeneity from the Meta-Analysis.

Empathy	General Maltreatment		I ²	Physical Abuse		I ²	Emotional Abuse		I ²	Sexual Abuse		I ²	Physical Neglect		I ²	Emotional Neglect		I ²
Empathy	n	$\overline{r}$ [95% CI]	I ²	n	[95% CI]	I ²	n	$\overline{r}$ [95% CI]	I ²	n	$\overline{r}$ [95% CI]	I ²	n	$\overline{r}$ [95% CI]	I ²	n	$\overline{r}$ [95% CI]	I ²
Aggregated empathy	41	−0.19 [−0.24, −0.15]	89.7	26	−0.12 [−0.17, −0.07]	88.8	25	−0.15 [−0.21, −0.10]	87.6	32	−0.07 [−0.14, −0.01]	93.0	23	−0.16 [−0.23, −0.09]	91.7	29	−0.18 [−0.25, −0.12]	90.0
General empathy	14	−0.17 [−0.26, −0.08]	94.3	8	−0.11 [−0.26,0.05]	94.8	7	−0.12 [−0.25, 0.02]	94.6	10	−0.08 [−0.22, 0.07]	94.9	7	−0.17 [−0.34, 0.02]	95.7	9	−0.21 [−0.34, −0.06]	94.9
Perspective taking	5	−0.22 [−0.29, −0.15]	26.3	4	−0.12 [−0.19, −0.04]	63.8	3	−0.16 [−0.23, −0.10]	0	4	−0.04 [−0.19,0.12]	87.6	3	−0.20 [−0.26, −0.14]	0	4	−0.23 [−0.35, −0.10]	74.6
Fantasy	5	−0.17 [−0.23, −0.12]	0	3	−0.05 [−0.19, 0.09]	74.8	3	−0.07 [−0.16,0.03]	44.4	4	−0.02 [−0.18,0.14]	88.2	3	−0.09 [−0.21,0.03]	65.7	4	−0.15 [−0.23, −0.07]	37.9
Empathic concern	5	−0.28 [−0.45, −0.08]	90.1	4	−0.16 [−0.33,0.03]	94.4	3	−0.26 [−0.36, −0.15]	63.1	4	−0.12 [−0.41,0.18]	96.8	3	−0.23 [−0.48, 0.06]	94.1	4	−0.26 [−0.40, −0.11]	82.9
Personal distress	5	0.13 [−0.01,0.25]	74.1	3	0.10 [0.03,0.16]	9.5	3	0.14 [0.08,0.21]	0	4	0.13 [0.07, 0.20]	29.7	3	0.12 [−0.02,0.25]	72.1	4	0.05 [−0.06,0.16]	65.4
Cognitive empathy	4	−0.30 [−0.35, −0.26]	0	2	−0.21 [−0.27, −0.15]	0	3	−0.22 [−0.27, −0.18]	0	3	−0.12 [−0.31, 0.08]	92.8	2	−0.27 [−0.32, −0.21]	0	2	−0.27 [−0.33, −0.21]	0
Affective empathy	3	−0.11 [−0.23,0.02]	79.1	2	−0.08 [−0.23,0.07]	82.1	3	−0.11 [−0.23,0.02]	85.8	3	0.03 [−0.19, 0.25]	94.3	2	−0.11 [−0.22,0.01]	73.1	2	−0.11 [−0.26,0.05]	83.6

Notes. Aggregated estimates: n = 176, $\overline{r}$ = −0.15 [−0.17, −0.13], I² = 91.2. n = number of estimates; 95% CI = 95% confidence interval.

p < .05. **p < .01. ***p < .001.

This analysis yielded the following results. Among the studies that examined the relationships between general maltreatment and empathy, all but four estimates reported negative correlations, and the pooled estimate was $\overline{r}$ = −0.19 (95% CI [−0.24, −0.15]; see Figure 2A). All but two estimates that focused on physical abuse and empathy reported negative relationships, and the pooled estimate was $\overline{r}$ = −0.12 (95% CI [−0.17, −0.07]; see Figure 2B). All but three estimates of the relationship between emotional abuse and empathy reported negative correlations, and the pooled estimate was $\overline{r}$ = −0.15 (95% CI [−0.21, −0.10]; see Figure 2C). The majority of the estimates (68.75%) on sexual abuse and empathy reported that abused children had lower empathy; the pooled estimate was $\overline{r}$ = −0.07 (95% CI [−0.14, −0.01]; see Figure 2D). All but two estimates of the associations between physical neglect and empathy were negative, and the pooled estimate was $\overline{r}$ = −0.16 (95% CI [−0.23, −0.09]; see Figure 2E). Finally, all but two estimates reported negative correlations between emotional neglect and empathy, and the pooled estimate was $\overline{r}$ = −0.18 (95% CI [−0.25, −0.12]; see Figure 2F). Further analyses showed that among all the studies, the pooled estimate was $\overline{r}$ = −0.15 (95% CI [−0.17, −0.13]. In general, there were significant negative correlations between the different types of child maltreatment and empathy; however, all the effects were small. By contrast (see Figure 3), neglect (emotional neglect, physical neglect) and empathy had larger effects than abuse (emotional abuse, physical abuse, and sexual abuse).

Figure 2.

Forest plots of the associations between child maltreatment and empathy. The square shape and bar line represent the estimate of effect size and 95% CI for each study. The diamond shapes represent the pooled estimates of the effect sizes for different maltreatments. DL represent the total effect sizes of the outcome of the random effect models, IV represent the total effect sizes of the outcome of the fixed effect models: (A) Studies of general child maltreatment and empathy, (B) studies of physical abuse and empathy, (C) studies of emotional abuse and empathy, (D) studies of sexual abuse and empathy, (E) studies of physical neglect and empathy, and (F) studies of emotional neglect and empathy.

Figure 3.

Pooled effects of effect size of child maltreatment and empathy (from Table 3).

In addition to the aggregated effects, the associations between the multi-facets of maltreatment and the separate aspects of empathy also showed differential features. In terms of the four dimensions of empathy, both general child maltreatment and its subtypes had stronger negative effects on empathic concern compared with the other dimensions. When empathy was divided into two dimensions, general maltreatment, child abuse, and child neglect were significantly associated with cognitive empathy but not affective empathy. It must be emphasized that these results should be considered with caution as the number of estimates included in each dimension of empathy was small (n ≤ 5) due to the different measures used in the studies.

Publication Bias

Publication bias occurs for many reasons. The key reason is selection bias, which typically favors estimates that are statistically significant and consistent with researchers’ and journals’ preconceived beliefs (Xue et al., 2020). Publication bias represents a serious challenge to the validity of meta-analyses: if the estimates in the literature are disproportionately large and significant, averaging them will preserve this bias, producing a distorted assessment of the mean true effect.

First, an informal funnel plot test was conducted to identify any publication selection bias. Figure 4 illustrates funnel plots of the estimated effects in our analysis. The dispersion at the top of each funnel, with the most precise estimates, is evidence of differences in the “true” effects across the analyzed studies. Judging intuitively, the funnel plots were generally symmetrical, with no serious publication bias exit.

Figure 4.

Funnel plots of the included studies. The hollow circles represent each study included in meta-analysis, and the dashed lines represent the symmetrical boundary: (A) Studies of general child maltreatment and empathy, (B) studies of physical abuse and empathy, (C) studies of emotional abuse and empathy, (D) studies of sexual abuse and empathy, (E) studies of physical neglect and empathy, (F) studies of emotional neglect and empathy, and (G) aggregated studies of child maltreatment and empathy.

Second, classic fail-safe N, Egger’s tests, and the trim-and-fill method were employed; none of them revealed any evidence of publication bias (see Table 4). The fail-safe N indicated that large numbers of unpublished studies would be required to reduce the mean effect size below the level of significance. This would be applicable for the relationships of empathy with general maltreatment (N = 9,333), physical abuse (N = 1,339), emotional abuse (N = 1,959), sexual abuse (N = 683), physical neglect (N = 1,723), emotional neglect (N = 371), and aggregated maltreatment (N = 92,338). To overcome the shortcomings of the funnel plot method, we used Egger’s linear regression test. The results were not significant, indicating that there was no publication bias in our meta-analysis. Finally, the trim-and-fill method did not identify any missing studies for these relationships.

Table 4.

Tests for Publication Bias.

Maltreatment	Number of Estimates	Classic Fail-Safe N	Egger’s Regression	Trim and Fill 95% CI
General maltreatment	41	9333	0.95	[−0.24, −0.15]
Physical abuse	26	1339	0.12	[−0.18, −0.07]
Emotional abuse	25	1959	0.14	[−0.21, −0.10]
Sexual abuse	32	683	0.35	[−0.14, −0.01]
Physical neglect	23	1723	0.55	[−0.24, −0.01]
Emotional neglect	29	3363	0.61	[−0.25, −0.12]
Aggregated estimates	176	92338	0.53	[−0.17, −0.13]

Note. Beg and Egger’s values were significance of p. CI = confidence interval.

Third, a commonly employed statistical test of publication bias is the FAT. Table 5 reports the FAT/PET results, with the FAT results reported on the first row and the PET results on the second. The first four columns report the various combinations of “FE”/“RE” and weighting by individual estimate (“Weight 1”)/weighting by study (“Weight 2”). In all four cases, we accept $H_{0} : α_{1} = 0$ at the 0.5% level of significance, indicating the nonexistence of publication bias. Across all four columns, the PET results demonstrate that child maltreatment was negatively related to empathy, and in Column (3), we reject $H_{0} : α_{0} = 0$ . Bias-adjusted estimates of the mean true effect of child maltreatment on empathy range from −0.06 to −0.13. Columns (5) and (6) in Table 4 report the weighted average estimates of the mean true effect, uncorrected for publication bias, using the “RE (Weight 1)” and “RE (Weight 2)” estimators. The associated estimates are −0.15 and −0.14, close to the unweighted value of −0.15 reported in Table 3. Thus, publication bias did not significantly inflate the observed effects of child maltreatment on empathy.

Table 5.

FAT and PET.

	Including Publication Bias Term				Excluding Publication Bias Term
	Fixed Effects (Weight 1)	Fixed Effects (Weight 2)	Random Effects (Weight 1)	Random Effects (Weight 2)	Random Effects (Weight 1)	Random Effects (Weight 2)
	(1)	(2)	(3)	(4)	(5)	(6)
FAT	−0.86 (0.87)	−1.71 (0.96)	−0.39 (0.81)	−0.85 (0.79)	—	—
PET	−0.09 (0.05)	−0.06 (0.05)	−0.13^* (0.05)	−0.10 (0.05)	−0.15^*** (0.02)	−0.14^*** (0.02)
Observations	176	176	176	176	176	176

Note. All the estimation procedures calculated the cluster robust standard errors. FAT = funnel asymmetry; PET = precision effect test.

p < .05. **p < .01. ***p < .001.

In summary, the results shown in Table 5 suggest that the “small” effects of child maltreatment found in the analyzed literature were not biased by the publication selection process. Given the large number of studies, data, and estimation characteristics included in our dataset and the substantial heterogeneities among studies included in the meta-analysis (see Table 3), we proceeded with multivariate analysis.

Meta-Regression Analysis

In order to ensure the robustness of the results, a variety of specification tests through the following strategies were performed: First, we examined both fixed effects (FE) and RE models. Second, under each model specification, we explored the results weighted by different study weights (Weight 1 and Weight 2). Third, in each weighted model, we compared the results after including the publication bias term (SE) and omitting it. Thus, eight different estimation procedures were yielded. The meta-regression results are reported in Table 6; the included moderators are listed in Table 2. Due to missing values for different variables, the total number of observations from the meta-regression was 147.

Table 6.

Meta-Regression Analysis.

	Fixed Effects				Random Effects
	Weight 1		Weight 2		Weight 1		Weight 2
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
SE	5.04 (3.99)	–	5.14 (2.81)	–	4.07 (4.33)	–	7.13 (4.54)	–
Intercept	−0.66^* (0.25)	−0.36^*** (0.07)	−0.71^*** (0.17)	−0.43^*** (0.06)	−0.77⁺ (0.43)	−0.37^*** (0.07)	−1.11^* (0.43)	−0.43^*** (0.06)
Gender
Single gender	ref.	ref.	ref.	ref.	ref.	ref.	ref.	ref.
Mixed gender	−0.05 (0.05)	−0.02 (0.05)	−0.05 (0.04)	−0.03 (0.04)	−0.05 (0.05)	−0.03 (0.05)	−0.07 (0.04)	−0.03 (0.03)
Age
<18 years old	ref.	ref.	ref.	ref.	ref.	ref.	ref.	ref.
≥18 years old	0.06^* (0.03)	0.08^*** (0.02)	0.07^** (0.02)	0.09^*** (0.02)	0.08^** (0.02)	0.09^*** (0.02)	0.09^** (0.02)	0.09^*** (0.02)
Sample Size
<300	ref.	ref.	ref.	ref.	ref.	ref.	ref.	ref.
≥300	0.25^* (0.11)	0.10^** (0.03)	0.31^** (0.09)	0.16^*** (0.03)	0.19^* (0.08)	0.11^** (0.03)	0.29^** (0.09)	0.16^*** (0.03)
Region
Asia	ref.	ref.	ref.	ref.	ref.	ref.	ref.	ref.
Europe	−0.03 (0.04)	−0.01 (0.04)	−0.02 (0.04)	−0.03 (0.05)	−0.00 (0.02)	0.00 (0.02)	0.01 (0.03)	−0.01 (0.05)
North America	−0.04 (0.08)	0.02 (0.06)	−0.10 (0.05)	−0.05 (0.06)	−0.00 (0.06)	0.04 (0.04)	−0.07 (0.06)	−0.01 (0.06)
Language
Chinese	ref.	ref.	ref.	ref.	ref.	ref.	ref.	ref.
English	0.14^** (0.04)	0.11^** (0.03)	0.18^*** (0.04)	0.19^** (0.05)	0.12^** (0.04)	0.10^** (0.03)	0.15^** (0.04)	0.16^** (0.05)
Maltreatment
Other Measures	ref.	ref.	ref.	ref.	ref.	ref.	ref.	ref.
CTQ	0.01 (0.04)	0.04 (0.03)	0.04 (0.03)	0.07^** (0.02)	0.01 (0.03)	0.03 (0.03)	0.04 (0.03)	0.07^** (0.02)
Empathy
Other Measures	ref.	ref.	ref.	ref.	ref.	ref.	ref.	ref.
IRI	0.08^*** (0.01)	0.08^*** (0.01)	0.07^** (0.02)	0.10^*** (0.02)	0.09^*** (0.02)	0.09^*** (0.02)	0.09^** (0.03)	0.11^*** (0.02)
Type of association
General maltreatment	ref.	ref.	ref.	ref.	ref.	ref.	ref.	ref.
EA	0.02 (0.02)	0.01 (0.03)	−0.05 (0.03)	−0.06 (0.04)	0.03 (0.02)	0.02 (0.02)	−0.03 (0.04)	−0.05 (0.04)
EN	0.02 (0.02)	0.02 (0.02)	−0.03 (0.02)	−0.00 (0.01)	0.03 (0.02)	0.03 (0.02)	−0.01 (0.02)	−0.00 (0.01)
PA	0.05^* (0.02)	0.05^** (0.01)	0.01 (0.02)	0.03^* (0.01)	0.06^** (0.02)	0.06^** (0.02)	0.03 (0.02)	0.04^* (0.02)
PN	0.03 (0.02)	0.03 (0.02)	−0.01 (0.02)	0.00 (0.01)	0.04^* (0.02)	0.04 (0.02)	0.02 (0.02)	0.01 (0.02)
SA	0.06^* (0.03)	0.07^* (0.03)	0.07 (0.04)	0.08^* (0.04)	0.07^* (0.03)	0.07^* (0.03)	0.08 (0.04)	0.08 (0.04)
Test
EN = EA = 0	F = 0.38	F = 0.84	F = 1.13	F = 0.90	F = 1.51	F = 1.35	F = 0.43	F = 0.66
EN = EA = 0	p = .69	p = 0.45	p = .35	p = 0.42	p = .25	p = .28	p = 0.65	p = .53
EN = PA = 0	F = 3.48	F = 5.67	F = 3.10	F = 2.70	F = 6.71	F = 5.91	F = 3.18	F = 2.60
EN = PA = 0	p = .05	p = .01	p = .07	p = .09	p = .01	p = .01	p = .07	p = .10
EN = PN = 0	F = 1.07	F = 1.55	F = 1.21	F = 0.09	F = 2.51	F = 1.89	F = 1.42	F = 0.36
EN = PN = 0	p = .37	p = .24	p = .32	p = .92	p = .11	p = .18	p = .27	p = .70
EN = SA = 0	F = 2.98	F = 4.70	F = 3.78	F = 2.32	F = 4.41	F = 4.05	F = 3.56	F = 1.89
EN = SA = 0	p = .08	p = .02	p = .04	p = .13	p = .03	p = .04	p = .05	p = .18
PN = EA = 0	F = 1.08	F = 1.47	F = 0.95	F = 1.11	F = 2.95	F = 1.85	F = 0.94	F = 1.18
PN = EA = 0	p = .36	p = .26	p = .41	p = .35	p = .08	p = .19	p = .41	p = .33
PN = PA = 0	F = 3.80	F = 5.81	F = 1.09	F = 2.93	F = 7.34	F = 6.85	F = 1.28	F = 2.46
PN = PA = 0	p = .04	p = .01	p = .36	p = .08	p = .01	p = .01	p = .30	p = .12
PN = SA = 0	F = 2.72	F = 3.52	F = 2.87	F = 2.44	F = 4.35	F = 3.64	F = 2.12	F = 1.62
PN = SA = 0	p = .09	p = .05	p = .08	p = .12	p = .03	p = .05	p = .15	p = .23
Observations	147	147	147	147	147	147	147	147
R ²	0.40	0.76	0.54	0.79	0.29	0.75	0.40	0.75

Note. The top value in each cell is the coefficient estimate, and the bottom value in parentheses is the cluster robust standard error. The test reports the results of testing whether there is no difference in mean r values for the five different maltreatment measures after controlling for the effects of other variables. PA = physical abuse; EA = emotional abuse; SA = sexual abuse; EN = emotional neglect; PN = physical neglect; CTQ = childhood trauma questionnaire.

p < .1, *p < .05, **p < .01. ***p < .001.

The first four columns report FE estimates, and the next four columns report RE estimates. It is to be noted that the SE terms (in Columns (1), (3), (5), and (7)) were positive and statistically not significant once the explanatory variables were included. In other words, the estimated coefficients suggest that the size of the publication bias was small, supporting the previous conclusion drawn from the results shown in Table 5. In comparisons between Columns (1)–(4) and Columns (5)−(8), the conditional mean effects (intercept) were all positive and statistically significant regardless of the weighting scheme and whether SE was included or not, indicating that there is a robust negative correlation between child maltreatment and empathy after strictly controlling the influence of publication bias and study characteristics.

As we proceed to our discussion of the other variables included in the meta-regressions shown in Table 6, we limit this discussion to those that were significant at the 5% level in at least four of the eight regressions. The respondents’ age and the studies’ sample sizes, publication language, empathy measurements as well as the type of relationships reported were identified as potential moderators that might have contributed to the heterogeneity.

First, the distribution of respondents’ age had a significant effect (β-values ranged from .06 to .09; all p-values were less than .05); that is, compared to adults, the adverse effects of child maltreatment on empathy were more obvious in children. Second, the negative effects of child maltreatment on empathy were relatively weaker in large samples (n ≥ 300) (β-values ranged from .10 to .31; all p-values were less than .05). Third, interestingly, articles published in English reported relatively weaker negative effects of child maltreatment on empathy (β-values ranged from .10 to .19; all p-values were less than .01). Fourth, compared with the results of other measures of empathy, the negative effects of child maltreatment on empathy were significantly weaker when using the IRI scales (β-values ranged from .07 to .11; all p-values were less than .01). Fifth, compared with general child maltreatment, physical abuse (β-values ranged from .01 to .06; six of the eight p-values were less than .05) and sexual abuse (β-values ranged from .06 to .08; five of the eight p-values were less than .05) had more negligible negative influences on empathy.

To verify the differences in magnitude of the relationships between subtypes of child maltreatment and empathy (see Table 3 and Figure 3), a series of F-test were conducted. After controlling for the effects of other variables, the findings demonstrate that compared with emotional neglect, physical abuse (F-values ranged from 2.60 to 6.71; four of the eight p-values were less than .05) and sexual abuse (F-values ranged from 1.89 to 4.70; five of the eight p-values were less than .05) have lesser negative influences; compared with physical neglect, physical abuse (F-values ranged from 1.28 to 7.34; four of the eight p-values were less than .05) and sexual abuse (F-values ranged from 1.62 to 4.35; three of the eight p-values were less than .05) also have more negligible negative influences. These results provide further evidence that emotional neglect and physical neglect generate more negative impacts on empathy than other types of maltreatment. In addition to the above variables, there was weak evidence that other variables included in the regressions had different estimated effects.

Discussion

Accumulating evidence has recently explored the associations between child maltreatment and empathy. Most studies have argued that child abuse and neglect are negatively associated with empathy. However, several contrasting studies have found that child maltreatment is associated with elevated empathy; therefore, these ambiguous findings require further investigation. In the present systematic review and meta-analysis, we synthesized 24 studies to assess the associations between childhood maltreatment and empathy. In general, the findings support those experiences of abuse and neglect during childhood are positively associated with reduced empathy in later life. Although no publication bias was determined, the heterogeneity among the analyzed studies cannot be ignored.

The findings of this meta-analysis suggest that, overall, child maltreatment is negatively associated with empathy, although the estimated effect size was small, which is consistent with most of the previous studies (Chen et al., 2016; Dittrich et al., 2019; Narvey et al., 2021; Yang et al., 2019). Social neuroscience research has revealed that child abuse and neglect can alter empathy-related brain structure (Decety & Michalska, 2010; Levy et al., 2019). Furthermore, attachment theory postulates that an empathic response is formed by the security provided by sensitive caregivers (Mikulincer et al., 2001; Schore, 2001), while abused and neglected children fail to gain psychological safety (Obsuth et al., 2014). As such, abused and neglected children find it considerably difficult to understand others’ mental states or resonate with other’s pain and distress (Fonagy & Luyten, 2009), which results in severe and long-lasting consequences for children’s cognitive and affective development (Currie & Widom, 2010).

Our findings also demonstrate that the estimated effect sizes between the subtypes of child maltreatment and cognitive empathy range from small to moderate. Specifically, emotional neglect and physical neglect are more likely to have destructive impacts on cognitive empathy than physical abuse and sexual abuse, which is consistent with previous research (Locher et al., 2014). Compared with those exposed to physical abuse and sexual abuse, neglected children may display impaired mentalization and unresponsive emotion regulation pattern during their daily interactions with primary caregivers, and thus fail to engage with the thoughts and feelings of others, which has been linked to poor cognitive empathy (Cummings, 1987; Fonagy, 2000). However, physical and emotional neglect are usually overlooked by governments and societies due to their invisible nature and have lagged behind physical and sexual abuse in terms of paediatric practice and training, resource allocation, and political priority (de Braal, 2010). The adverse effects of neglect on cognitive empathy deserve more academic and government’s close attention.

It was further found that neither general child maltreatment nor its subtypes are associated with affective empathy, regardless of the measures of empathy used, which is also consistent with previous findings (Davidov et al., 2013; Wang et al., 2021). Neuroscience and developmental psychology have suggested that affective empathy appears earlier, is involuntary, and relies on mimicry and somato-sensorimotor resonance between others and the self (Decety & Michalska, 2010; Uzefovsky & Knafo-Noam, 2016). Another plausible explanation may be that self-report measures might fail to identify child maltreatment during very early childhood due to the timing of the measurement or recall bias in previous studies.

However, cognitive empathy integrates higher-order cortical regions to understand others and can be trained through psychological interventions (Levy et al., 2019; Pfeifer et al., 2008). Previous studies have shown that the deficits in empathy caused by parental abuse and neglect can eventually lead to the formation of antisocial personalities in adulthood (Jovev et al., 2013) and even criminal behaviors (Docherty et al., 2018). Therefore, cognitive empathy intervention programs should also be implemented for children with a history of maltreatment.

It should be noted that a wide heterogeneity existed across the studies included in our review. First, the negative impacts of child maltreatment on empathy in children’s samples were more significant than those observed in adult samples. This confirms the attachment theory argument that early insecure attachment to primary caregivers can generate more negative influences on younger children (Bowlby, 1982; Levy et al., 2019). Second, measurement inconsistencies were a factor that influenced the association between child maltreatment and empathy. Specifically, compared with the use of IRI scales to measure empathy, the negative association between child abuse and empathy was stronger when adopting the BES and other scales. This may be because the BES is more inclusive than the IRI scales (Davis, 1983). Third, the sample size is another factor influencing the results on the impacts of child maltreatment on empathy, which may be associated with the low robustness of smaller sample sizes. Finally, the association between these two variables was more significant among the Chinese studies than the English studies after excluding publication bias. One plausible explanation is that most of the Chinese studies were based on clinical samples, and these patients are more likely to report more severe child maltreatment and empathy problems (Chen et al., 2016; Xu et al., 2010).

Study Limitations

The present study has several limitations that should be strengthened in future research. First, all the studies included in the analysis adopted a cross-sectional design; therefore, a causal relationship between child abuse and neglect and empathy cannot be inferred. Longitudinal designs should be adopted in future surveys of the impacts of child maltreatment on empathy. Second, there is no consistent measurement of empathy as well as child maltreatment, which may result in an overestimation or underestimation of the relationships between child maltreatment and empathy. Consequently, there is a great need for more reliable measurements of child maltreatment and empathy. Third, only a small number of studies were included under each dimension of empathy due to the different measures used; thus, the true relationship between child maltreatment and empathy may not have been revealed. Therefore, more studies are called for in this field.

Conclusion

This study contributes to our understanding of the negative impacts of child abuse and neglect on empathy, offering relatively reliable evidence to contradict the opposing arguments regarding this relationship. The results regarding the negative influence of child maltreatment on empathy vary depending on the age, sample size, publication language, empathy measurement, and maltreatment type. One important implication is that parents and caregivers should be equipped with scientific knowledge about the negative consequences of child abuse and neglect on empathy to prevent emotional abuse and neglect, allowing them to care for their children more sensitively, especially in the Chinese context. Furthermore, empathy enhancement programs should be designed and implemented as early as possible for those children who have suffered from child abuse and neglect to alleviate the long-term adverse impacts of child maltreatment.

Critical Findings of this Review

Childhood maltreatment is significantly and negatively associated with empathy.

Emotional neglect and physical neglect generate more negative impacts on empathy than other types of maltreatment.

The effects of child maltreatment on empathy vary depending on the age of respondents, sample size, publication language, empathy measurement, and maltreatment type.

Implications for Policy, Practice, and Research

Child protection policy should be given top priority by the government. Meanwhile, parenting programs that aim at child maltreatment prevention and intervention are urgently needed, especially in the Chinese context.

Empathy enhancement programs should be designed and implemented for individuals with a history of childhood maltreatment as early as possible.

More solid studies on the measurement of empathy are needed, and longitudinal designs in the causal relationship between child maltreatment and empathy should be adopted in future research.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research has been supported by Supported by National Social Science Fund of China (20ASH018).

ORCID iD

Huiping Zhang

Author Biographies

Huiping Zhang is a professor in Center for Studies of Sociological Theory and Method, and Department of Social Work and Social Policy, Renmin University of China. Her research interests cover vulnerable children and family, and child protection in rural China.

Xinyi Gao is an MSW student at Department of Social Work and Social policy, Renmin University of China. Her research focuses on mental health and prevention of vulnerable children.

Yuzhu Liang is an MSW student at Department of Social Work and Social policy, Renmin University of China. Her research focuses on child welfare and child policy assessment.

Qinyun Yao is an MSW student at Department of Social Work and Social policy, Renmin University of China. Her research focuses on family services for hospitalized children.

Qingong Wei is an associate professor at National Academy of Development and Strategy, Renmin University of China. His research interests include community building and social policy evaluation.

References

Ang

R. P.

Goh

D. H.

(2010). Cyberbullying among adolescents: The role of affective and cognitive empathy, and gender. Child Psychiatry & Human Development, 41(4), 387–397.

Angelakis

Jennifer

L. A.

Gooding

(2020). Association of childhood maltreatment with suicide behaviors among young people: A systematic review and meta-analysis. JAMA Network Open, 3(8), e2012563.

Baron-Cohen

Gillberg

(1995). Mind blindness: An essay on autism and theory of mind. Developmental Medicine and Child Neurology, 37(12), 1124–1124.

Baron-Cohen

Wheelwright

(2004). The empathy quotient: An investigation of adults with asperger syndrome or high functioning autism, and normal sex differences. Journal of Autism and Developmental Disorders, 34(2), 163–175.

*Bartlett

J. D.

(2012). Young mothers, infant neglect, and discontinuities in intergenerational cycles of maltreatment. Tufts University.

Bernstein

D. P.

Stein

J. A.

Newcomb

M. D.

Walker

Pogge

Ahluvalia

Stokes

Handelsman

Medrano

Desmond

Zule

(2003). Development and validation of a brief screening version of the childhood trauma questionnaire. Child Abuse & Neglect, 27(2), 169–190.

Bertsch

Grothe

Prehn

Vohs

Berger

Hauenstein

Keiper

Domes

Teipel

Herpertz

S. C.

(2013). Brain volumes differ between diagnostic groups of violent criminal offenders. European Archives of Psychiatry Clinlical Neuroscience, 263(7), 593–606.

Bifulco

Moran

P. M.

Ball

Bernazzani

(2002). Adult attachment style: I. Its relationship to clinical depression. Social Psychiatry and Psychiatric Epidemiology, 37(2), 50–59.

Bowlby

(1977). The making and breaking of affectional bonds: I. Aetiology and psychopathology in the light of attachment theory. The British Journal of Psychiatry, 130(3), 201–210.

10.

Bowlby

(1982). Attachment and loss. Basic Books.

11.

*Bujalski

A. S.

Chesin

M. S.

Jeglic

E. L.

(2017). Cognitive empathy partially mediates the relationship between childhood physical abuse and borderline personality disorder features in college students. Current Psychology, 38(1), 121–127.

12.

Card

N. A.

(2012). Applied meta-analysis for social science research. Guilford.

13.

*Chen

Kong

Zhang

Hong

Zhang

(2016). Study on relationship between childhood trauma experience and empathy in patients with schizophrenia. Journal of Qiqihar Medical University, 28, 3493–3496.

14.

Chen

Sun

Xie

Chan

K. L.

(2019). The impacts of internal migration on child victimization in China: A meta-analysis. Trauma, Violence, & Abuse, 20(1), 40–50.

15.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Erlbaum.

16.

Cummings

E. M.

(1987). Coping with background anger in early childhood. Child Development, 58, 976–984.

17.

Currie

Widom

C. S.

(2010). Long-term consequences of child abuse and neglect on adult economic well-being. Child Maltreatment, 15(2), 111–120.

18.

Davidov

Zahn-Waxler

Roth-Hanania

Knafo

(2013). Concern for others in the first year of life: Theory, evidence, and avenues for research. Child Development Perspectives, 7(2), 126–131.

19.

Davis

J. L.

Petretic-Jackson

P. A.

(2000). The impact of child sexual abuse on adult interpersonal functioning: A review and synthesis of the empirical literature. Aggression and Violent Behavior, 5(3), 291–328.

20.

Davis

M. H.

(1980). A multidimensional approach to individual differences in empathy. Journal of Personality and Social Psychology, 44, 113–126.

21.

Davis

M. H.

(1983). Measuring individual differences in empathy: Evidence for a multidimension approach. Journal of Personality and Social Psychology, 44(1), 113–126.

22.

de Braal

. (2010). Understanding emotional abuse. Journal of Family Health Care, 20(3), 82–84.

23.

Decety

Michalska

K. J.

(2010). Neurodevelopmental changes in the circuits underlying empathy and sympathy from childhood to adulthood. Developmental Science, 13(6), 886–899.

24.

*Dittrich

Bermpohl

Kluczniok

Attar

C. H.

Jaite

Fuchs

Neukel

Herpertz

S. C.

Brunner

Winter

S. M.

Lehmkuhl

Roepke

Kaess

Heim

Boedeker

(2019). Alterations of empathy in mothers with a history of early life maltreatment, depression, and borderline personality disorder and their effects on child psychopathology. Psychological Medicine, 50(7), 1182–1190.

25.

Docherty

Kubik

Herrera

C. M.

Boxer

(2018). Early maltreatment is associated with greater risk of conduct problems and lack of guilt in adolescence. Child Abuse & Neglect, 79, 173–182.

26.

Doucouliagos

Paldam

(2013). The robust result in meta-analysis of aid effectiveness: A response to Mekasha and Tarp. The Journal of Development Studies, 49(4), 584–587.

27.

Dryburgh

N. S. J.

Ponath

Bukowski

W. M.

Dirks

M. A.

(2021). Associations between interpersonal behavior and friendship quality in childhood and adolescence: A meta-analysis. Child Development, 93(3), e332–e347.

28.

Duval

Tweedie

(2000). Trim and fill: A simple funnel-plot based method of testing and adjusting for publication bias in meta-analysis. Biometrics, 56(2), 455–463.

29.

Egger

Smith

G. D.

Schneider

Minder

(1997). Bias in meta-analysis detected by a simple, graphical test. British Medical Journal, 315, 629–634.

30.

*Fang

Wang

Yuan

K. H.

Wen

(2020). Childhood psychological maltreatment and moral disengagement: A moderated mediation model of callous-unemotional traits and empathy. Personality and Individual Differences, 157, 109814.

31.

Fonagy

(2000). Attachment and borderline personality disorder. Journal of the American Psychoanalytical Association, 48, 1129–1146.

32.

Fonagy

Luyten

(2009). A developmental, mentalization-based approach to the understanding and treatment of borderline personality disorder. Development and Psychopathology, 21(4), 1355–1381.

33.

*Gander

Buchheim

Bock

Steppan

Sevecke

Goth

(2020). Unresolved attachment mediates the relationship between childhood trauma and impaired personality functioning in adolescence. Journal of Personality Disorders, 34(Supplement B), 84–103.

34.

Gardner

M. J.

Thomas

H. J.

Erskine

H. E.

(2019). The association between five forms of child maltreatment and depressive and anxiety disorders: A systematic review and meta-analysis. Child Abuse & Neglect, 96, 104082.

35.

Gilbert

Widom

C. S.

Browne

Fergusson

Webb

Janson

(2009). Burden and consequences of child maltreatment in high-income countries. Lancet, 373(9657), 68–81.

36.

Gobodo-Madikizela

(2008). Trauma, forgiveness and the witnessing dance: Making public spaces intimate. Journal of Analytical Psychology, 53(2), 169–188.

37.

*Greenberg

D. M.

Baron-Cohen

Rosenberg

Fonagy

Rentfrow

P. J.

(2018). Elevated empathy in adults following childhood trauma. PLoS One, 13(10), e0203886.

38.

Gunby

Jin

Reed

W. R.

(2017). Did FDI really cause Chinese economic growth? A meta-analysis. World Development, 90, 242–255.

39.

*Han

Shao

(2019). A study on childhood trauma and empathy ability of college students. The Guide of Science & Education, 15, 177–178.

40.

Haslam

Taylor

E. P.

(2022). The relationship between child neglect and adolescent interpersonal functioning: A systematic review. Child Abuse & Neglect, 125, 105510.

41.

Higgins

Thompson

S. G.

Deeks

J. J.

Altman

D. G.

(2003). Measuring inconsistency in meta-analyses. British Medical Journal, 327(7414), 557–560.

42.

Jolliffe

Farrington

D. P.

(2004). Empathy and offending: A systematic review and meta-analysis. Aggression and Violent Behavior, 9(5), 441–476.

43.

Jolliffe

Farrington

D. P.

(2006). Development and validation of the basic empathy scale. Journal of Adolescence, 29(4), 589–611.

44.

Jovev

McKenzie

Whittle

Simmons

J. G.

Allen

N. B.

Chanen

A. M.

(2013). Temperament and maltreatment in the emergence of borderline and antisocial personality pathology during early adolescence. Journal of the Canadian Academy of Child and Adolescent Psychiatry, 22(3), 220.

45.

Levy

Goldstein

Feldman

(2019). The neural development of empathy is sensitive to caregiving and early trauma. Nature Communications, 10(1), 1–10.

46.

*Li

H. R.

(2016). The relationship between child maltreatment and aggressive behavior: The mediating effect of empathy. South-Central University for Nationalities.

47.

Lipsey

M. W.

Wilson

D. B.

(2001). Practical meta-analysis. Sage.

48.

*Locher

S. C.

Barenblatt

Fourie

M. M.

Stein

D. J.

Gobodo-Madikizela

(2014). Empathy and childhood maltreatment: A mixed-methods investigation. Annals of Clinical Psychiatry, 26(2), 97–110.

49.

Maas

Herrenkohl

T. I.

Sousa

(2008). Review of research on child maltreatment and violence in youth. Trauma, Violence & Abuse, 9(1), 56–67.

50.

Marta

G. R.

Doretto

V. F.

Scivoletto

(2018). Maltreatment and emotion recognition among Brazilian adolescents. Frontiers in Psychiatry, 9, 625.

51.

*Meidan

Uzefovsky

(2020). Child maltreatment risk mediates the association between maternal and child empathy. Child Abuse & Neglect, 106, 104523.

52.

Mikulincer

Gillath

Halevy

Avihou

Avidan

Eshkoli

(2001). Attachment theory and reactions to others’ needs: Evidence that activiation of the sense of attachment security promotes empathic responses. Journal of Personality and Social Psychology, 81(6), 1205.

53.

Mikulincer

Shaver

P. R.

Sahdra

B. K.

Bar-On

(2013). Can security-enhancing interventions overcome psychological barriers to responsiveness in couple relationships? Attachment & Human Development, 15(3), 246–260.

54.

Miller

A. B.

Esposito-Smythers

Weismoore

J. T.

Renshaw

K. D.

(2013). The relation between child maltreatment and adolescent suicidal behavior: A systematic review and critical examination of the literature. Clinical Child and Family Psychology Review, 16, 146–172.

55.

Moher

Liberati

Tetzlaff

Altman

D. G.

(2009). Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. Annals of Internal Medicine, 151, 264–269.

56.

*Narvey

Yang

Wolff

K. T.

Baglivio

Piquero

A. R.

(2021). The interrelationship between empathy and adverse childhood experiences and their impact on juvenile recidivism. Youth Violence and Juvenile Justice, 19(1), 45–67.

57.

Obsuth

Hennighausen

Brumariu

L. E.

Lyons-Ruth

(2014). Disorganized behavior in adolescent–parent interaction: Relations to attachment state of mind, partner abuse, and psychopathology. Child Development, 85(1), 370–387.

58.

*Ometto

De Oliveira

P. A.

Milioni

A. L.

Dos Santos

Scivoletto

Busatto

G. F.

Nunes

P. V.

Cunha

P. J.

(2015). Social skills and psychopathic traits in maltreated adolescents. European Child & Adolescent Psychiatry, 25(4), 397–405.

59.

*Ozdemir

Sahin

S. K.

(2020). The impact of childhood traumatic experiences on self-esteem and interpersonal relationships. Psychiatry and Behavioral Sciences, 10(4), 185–191.

60.

*Parlar

Frewen

Nazarov

Oremus

MacQueen

Lanius

McKinnon

M. C.

(2014). Alterations in empathic responding among women with posttraumatic stress disorder associated with childhood trauma. Brain Behavior, 4(3), 381–389.

61.

Pfeifer

J. H.

Iacoboni

Mazziotta

J. C.

Dapretto

(2008). Mirroring others’ emotions relates to empathy and interpersonal competence in children. Neuroimage, 39(4), 2076–2085.

62.

Read

Perry

B. D.

Moskowitz

Connolly

(2001). The contribution of early traumatic events to schizophrenia in some patients: A traumagenic neurodevelopmental model. Psychiatry, 64(4), 319–345.

63.

*Ren

Zhou

Yang

Wang

Guo

Zhou

(2018). Drug abstainers’childhood trauma and antisocial personality symptoms: Mediating of empathy. Chinese Journal of Clinical Psychology, 4, 701–705.

64.

Roepke

Vater

Preißler

Heekeren

H. R.

Dziobek

(2013). Social cognition in borderline personality disorder. Frontiers in Neuroscience, 6, 195.

65.

Rosenthal

(1979). The file drawer problem and tolerance for null results. Psychological Bulletin, 86(3), 638–641.

66.

Rosenthal

(1991). Meta-analytic procedures for social research. Sage.

67.

Schore

A. N.

(2001). Effects of a secure attachment relationship on right brain development, affect regulation, and infant mental health. Infant Mental Health Journal: Official Publication of the World Association for Infant Mental Health, 22(1–2), 7–66.

68.

*Simons

Wurtele

S. K.

Heil

(2002). Childhood victimization and lack of empathy as predictors of sexual offending against women and children. Journal of Interpersonal Violence, 17(12), 1291–1307.

69.

Stanley

Doucouliagos

(2012). Meta-regression analysis in economics and business. Routledge.

70.

Steichen

T. J.

(2001). Nonparametric trim and fill analysis of publication bias in meta-analysis. Stata Technical Bulletin, 10(57), 8–14.

71.

Sterne

J. A.

Egger

(2001). Funnel plots for detecting bias in meta-analysis: Guidelines on choice of axis. Journal of Clinical Epidemiology, 54(10), 1046–1055.

72.

*Sun

Chen

Wang

(2020). The link between childhood psychological maltreatment and cyberbullying perpetration attitudes among undergraduates: Testing the risk and protective factors. PLoS one, 15(9), e0236792.

73.

Uzefovsky

Knafo-Noam

(2016). Empathy development throughout the life span. In Sommerville

J. A.

Decety

(Eds.), Social Cognition (pp. 89–115). Routledge.

74.

*Wang

Guo

Yin

Zhang

Dong

Wang

(2021). The role of empathy in the relationship between childhood trauma and borderline personality tendencies in male offenders. The Journal of Forensic Psychiatry & Psychology, 32(5), 679–696.

75.

*Xu

Wang

Cao

Guan

Liu

Zuo

Miao

Qian

(2010). The psychological trauma, empathy deficits and antisocial personality disorders. Neural Injury and Functional Reconstruction, 4, 253–258.

76.

Xue

Reed

W. R.

Menclova

(2020) Social capital and health: A meta-analysis. Journal of Health Economics, 72, 1–16.

77.

*Yang

Zhang

Zhao

(2019). Childhood emotional neglect predicts college students’ empathy ability. Community Psychology Research, 9(1), 143–160.

78.

*Yin

Yao

(2021). The effect of psychological maltreatment on externalizing problem behavior of high school students: The chain mediating effect of empathy and moral disengagement. Chinese Journal of Special Education, 10, 70–76.

79.

*Yu

Zhao

(2020). Childhood maltreatment and prosocial behavior among Chinese adolescents: Roles of empathy and gratitude. Child Abuse & Neglect, 101, 104319.

80.

Wang

Liang

(2012). A correlation study of medico’s empathy and parental rearing patterns. Chinese Journal of School Health, 33(1), 28–30.

81.

Zhang

Sun

Wei

(2021). Effects of childhood maltreatment on self-compassion: A systematic review and meta-analysis. Trauma, Violence, & Abuse, 1–13, https://doi.org/10.1177/15248380211043825

Does Child Maltreatment Reduce or Increase Empathy? A Systematic Review and Meta-analysis

Abstract

Keywords

Introduction

Method

Data Sources and Study Selection

Inclusion Criteria

Data Extraction and Coding

Data Analysis Procedures

Effect size calculations

Heterogeneity

Publication bias

Meta-regression analysis

Results

Study Characteristics

Associations Between Child Maltreatment and Empathy

Publication Bias

Meta-Regression Analysis

Discussion

Study Limitations

Conclusion

Critical Findings of this Review

Implications for Policy, Practice, and Research

Footnotes

Declaration of Conflicting Interests

Funding

ORCID iD

Author Biographies

References