Taking the stand: defendant statements in court cases of alleged sexual abuse against infants, toddlers and preschoolers

ABSTRACT

Investigating and adjudicating allegations of child sexual abuse are challenging tasks. In the present study, we examined defendant statements concerning charges of sexual abuse against young children in Swedish district court cases (87 defendants, 140 child complainants, tried between January 2010 to December 2015). A main objective was to test predictive factors for admissions of guilt using inferential statistical analyses. Furthermore, using qualitative thematical analysis, we sought to identify common patterns in the defendants’ explanations to the allegation. Approximately one-third of the defendants (31%) pleaded guilty during trial. Admissions of guilt were more likely if the defendant was young, if the child was young at the onset of abuse, if the child and perpetrator had an extrafamilial relationship, and if the defendant possessed child pornography. A conflict with the person who made the report (e.g. a custody dispute), a testimony from the child, a direct eyewitness, or an informal disclosure recipient were significantly more common in cases of denials. In the qualitative analysis, a range of alternative explanations behind the abuse allegations were identified. Legal professionals and investigators may benefit from considering these alternative hypotheses during their investigative and judicial work.

KEYWORDS:

• Admission of guilt
• confession
• legal decision-making
• child sexual abuse
• courtroom

Adjudicating alleged child sexual abuse (CSA) constitutes a challenge for legal systems around the globe. Decision-makers are often faced with the complicated task of assessing likelihood of guilt based largely on oral testimonies and circumstantial evidence (Diesen & Diesen, 2009). A confession can provide vital information regarding the offence and help reduce the burden placed on child complainants. This could be particularly important in cases involving the youngest children, who for developmental reasons can have difficulties providing testimony that meet the standards of the courtroom (Poole, Brubacher, & Dickinson, 2015). A confession might also have a therapeutic function and could contribute to the identification of other victims or perpetrators (Lippert, Cross, Jones, & Walsh, 2010). The scientific understanding of mechanisms and predictive factors associated with suspect confessions has rapidly expanded during the last decades (for overviews, see Gudjonsson, 2003; Lassiter & Meissner, 2010). There is, for example, ample evidence demonstrating the prominence of using information-gathering and rapport-based police interviewing techniques over accusatorial methods to reduce the risk of coercing a false confession (e.g. Carter, 2011; Meissner et al., 2014; Vrij et al., 2017). More recently, academics have also put the searchlight on factors relating to true and false guilty pleas during plea bargaining processes (Redlich, 2010).

In the present study, we sought to examine defendants’ courtroom statements during Swedish legal proceedings concerning sexual abuse against infants, toddlers, and preschoolers. So far, only a handful of studies have systematically investigated factors associated with confessions of CSA (Beauregard & Mieczkowski, 2011; Faller, Birdsall, Henry, Vandervort, & Silverschanz, 2002; Gudjonsson & Sigurdsson, 2000; Lippert et al., 2010) and no large-scale investigation has focused specifically on cases involving the youngest children. However, expanding our knowledge in this context is important considering that alleged CSA against young children has been described as particularly difficult to investigate, prosecute, and adjudicate (e.g. Diesen & Diesen, 2009; Ernberg, Landström, & Tidefors, 2016). The purpose of the current study was twofold. First, we tried to replicate past findings regarding factors associated with CSA confessions in a unique sample of cases involving young children. More specifically, we tested if previously identified suspect characteristics (e.g. the defendant's age) and offence-related factors (e.g. the severity of the abuse) could predict admissions of guilt during trial. Throughout the paper, we will use the term admission of guilt when referring to an admission of having committed the crime (e.g. Yes, I did it) and confession when referring to a more detailed statement of the criminal offence. Second, past research has tended to focus on admission rates rather than statement content. However, insights into the content of defendants’ court statements could be valuable from both a theoretical and applied perspective. Our second aim was therefore to qualitatively explore the defendants’ statements in order to gain a deeper understanding of their admissions and denials of guilt. Focus was placed on identifying common patterns in the defendants’ explanations to the abuse allegation.

In recent decades, there has been an increasing interest in factors associated with the decision to confess to a crime. Extensive research has shown that suspects are more likely to confess if they believe the police hold strong corroborative evidence of their guilt (e.g. Gudjonsson, 2003; Moston, Stephenson, & Williamson, 1992). In child sexual abuse cases, confessions during police interrogations have been found to correlate with the presence of a child disclosure (i.e. the child has told someone about the abuse), a corroborative witness testimony, or multiple victim reports (Lippert et al., 2010). Repeated abuse (Faller et al., 2002) as well as more severe allegations of abuse (Lippert et al., 2010) have also been associated with increased confession rates (but see Beauregard & Mieczkowski, 2011, for an exception). Furthermore, previous research point towards higher confession rates in cases of extra-familial abuse with suspects outside of the child's family (Lippert et al., 2010). However, this association has not been consistent across different studies (Beauregard & Mieczkowski, 2011; Faller et al., 2002).

A past criminal history has been found to decrease the likelihood of confessing to a variety of crimes (e.g. burglary, drug crimes; Pearse, Gudjonsson, Clare, & Rutter, 1998). Moreover, younger suspects are more likely to confess to crimes than older suspects (e.g. Gudjonsson, 2003; Pearse et al., 1998) and this finding appear to extend to child sexual abuse offences (e.g. Beauregard & Mieczkowski, 2011; Faller et al., 2002; Lippert et al., 2010). Several explanations behind this tendency have been proposed, such as developmental aspects (e.g. higher impulsiveness and risk seeking behaviour, less developed coping strategies, and a more limited understanding of the legal system among younger suspects) and an increased risk for harsher sentencing and loss of employment among older suspects (e.g. Gudjonsson, 2003; Kassin et al., 2010). A conflict between the suspect and the person who reported the abuse has been found to complicate CSA investigations (Diesen & Diesen, 2009). However, the potential association between conflicts beyond the CSA allegation and admission rates has, to our knowledge, not been systematically tested in past research.

The research reviewed above was primarily carried out in North America and the United Kingdom. As the court cases analysed in the current study were tried in Sweden, a few remarks on the Swedish legal system are in order. Swedish criminal law is built around the principles of immediateness and oral proceedings (Swedish Code of Judicial Procedure, chap. 30 par. 2, and chap. 46 par. 5). This means that judicial decisions should be based on the statements and evidence presented in court. The defendant, complainant (children under 15 give testimony via their video-taped police interview) and potential witnesses, are all expected to provide oral and unscripted testimonies during trial. If a defendant has confessed during the preliminary investigation, the prosecutors tend to present this fact to the court.

Furthermore, documentations of police interrogations can be submitted as evidence (a principle called free production of evidence, Swedish Code of Judicial Procedure, chap. 35 par. 1) but this is a procedure rarely used in practice. Instead, court evaluations of police interrogations are typically only carried out in cases where there is reason to suspect police-induced coerced confessions. Importantly, Sweden does not use plea bargaining procedures. Furthermore, defendants are not provided any legally regulated benefits for admitting guilt during trial, but may receive a more lenient sentence for their cooperation during the police investigation. Only 5% of suspects are estimated to confess during Swedish preliminary investigations concerning alleged CSA (Diesen & Diesen, 2009). However, confessions seem to play an important role in the prosecutor's decision to press charges and are present in 30–40% of prosecuted CSA cases (Lainpelto, 2012).

For the present study, we examined defendants’ courtroom statements concerning charges of sexual abuse against infants, toddlers, and preschoolers. The cases analyzed were tried in Swedish district courts between January 2010 to December 2015. We hypothesized that defendants would be more likely to admit guilt if they were young (e.g. Beauregard & Mieczkowski, 2011), accused of sexual abuse against several victims (e.g. Lippert et al., 2010), charged with penetrative abuse (e.g. Lippert et al., 2010), or repeated abuse (e.g. Faller et al., 2002). Furthermore, we predicted that defendants would be less likely to admit guilt if they had a previous criminal history (e.g. Gudjonsson, 2003) or a conflict with the person who made the allegation (e.g. Diesen & Diesen, 2009). Based on research on the perceived strength of evidence (e.g. Moston et al., 1992), we hypothesized that admissions of guilt would be more likely in cases containing DNA evidence, injuries, photo or video documentation of the abuse, child pornography (of other children), a testimony from the child, a direct eyewitness (testifying about observing the abuse take place) or an informal disclosure recipient (testifying about receiving a disclosure from the child). For exploratory purposes, we also examined confession rates in relation to child gender and age during i) the onset of abuse and ii) the forensic interview.

In the qualitative analysis, we searched for shared patterns in the defendants’ courtroom statements concerning their admissions or denials of guilt. The analysis was guided by two central questions; i) What explanations were offered by the defendants who admitted guilt? and ii) What explanations were offered by the defendants who denied guilt? Importantly, as we lacked objective truths of the alleged events, we did not draw any inferences regarding the accuracy of the defendants’ statements.

Method

Data collection

In Sweden, judges produce a written statement containing information about the defendant's stance on the charges (i.e. admitting or denying guilt), summaries of the testimonies and evidence presented during trial, and descriptions outlining the rationale behind the court's judicial decision (Swedish Code of Judicial Procedure, chap. 30 par. 5). The present analyses were based on information found in these court documents. The data collection procedure was carried out in four steps (for more information, see Magnusson, Ernberg, & Landström, 2017). Initially, the first and second author conducted separate searches in a database containing district court documents. All legal cases (n = 1231) tried between January 1st 2010 to December 31st 2014 containing charges of child sexual abuse were separately screened by both researchers to identify child complainant age. A total of 85 court cases involving children ages six or younger during the first incident of alleged abuse were identified. Second, the two authors separately screened all documents in a database containing court of appeal verdicts (n = 643) following the same procedure as above. A total of 53 court of appeal cases involving young children were identified. Third, we contacted all district courts (n = 48) and courts of appeal (n = 6) and requested access to court cases not available in the databases. Fifty courts (93%) complied with our request and an additional 39 court cases not uploaded to the databases were identified. Fourth, a complementary search of cases tried from January 1st to December 31st 2015 was conducted by searching a database containing district court cases from 2015. This led to the identification of 30 new cases.

To be included in the present study, the case had to involve children seven years or younger during the court hearing. Applying this inclusion criterion resulted in the extraction of 87 District court cases, involving a total of 140 child complainants. Considering the extensive data collection procedure, these cases are likely to reflect most, if not all, court case of alleged CSA against young children during the specified time period.

Case characteristics

The material involved 87 male defendants aged from 15 to 82 years (M = 38.4, SD = 17.6). Sixty-eight defendants (78.2% conviction rate) were convicted of child sexual abuse and sentenced to prison (67.6%), juvenile treatment (25%) or closed psychiatric care (7.4%). The 140 child complainants were between less than one and six years during the first alleged instance of abuse (M = 4.1, SD = 1.5) and between three and seven years during the court hearing (M = 5.2, SD = 1.3). See Tables 1 and 2 for more information.

Quantitative measures

The court documents were coded according to a coding manual containing a total of 142 variables for each district court case (for more information on the coding, see Magnusson et al., 2017). First, a set of variables used in previous court studies and research on CSA (e.g. Diesen & Diesen, 2009; Kuehnle & Connell, 2009; Lamb, Hershkowitz, Orbach, & Esplin, 2008) were operationalised to a preliminary coding scheme. The first and second author thereafter coded five randomly selected cases following both the coding scheme and a data-driven approach. Information not covered in the pre-specified variables was carefully noted and the coding sheets were cross-compared. The results were discussed between the first three authors and a fourth experienced researcher in the field. During this meeting, the coding scheme was refined and new variables were added. The process of separately coding 4–6 randomly selected cases, cross comparing the coding and refining the coding scheme was repeated four times until reaching a point of saturation. An inter-rater reliability analysis was conducted on 25% randomly selected cases following the final draft of the coding manual. The level of inter-rater agreement was calculated at 92%.

Quantitative variables

In line with previous research (Lippert et al., 2010), admission of guilt was coded as either admitting or denying at least one sexual charge of child abuse using a dichotomous variable (0 = denying, 1 = admitting at least one charge of child sexual abuse). In cases where the defendant admitted that the event occurred but denied sexual intent, the statement was coded as a denial. Defendant age during trial was coded as a continuous variable. Previous criminal history reflected if the court document stated that the defendant had previously been convicted of a crime (1 = yes, 0 = no). Past conflict with the person who made the police report (1 = yes, 0 = no) was included if the document described a potential conflict beyond the abuse allegation (e.g. custody dispute, economic conflict). Relationship to the child was coded as intrafamilial or extrafamilial depending on if the defendant was a close relative (i.e. father, step-parent, sibling, cousin, uncle, and grandparent) or not. Repeated abuse was defined as charges concerning at least two separate dates of abuse, and coded as single occasion of abuse (0) or repeated abuse (1). Type of abuse was coded as either penetrative (1) or not penetrative (0). In cases involving multiple charges, type of abuse was coded for the most severe charge. Dichotomous variables, coded as either absent (0) or present (1), were used to categorize the evidence: child testimony, child pornography (of other children), photo or video documentation (of abuse of the complainant), direct eyewitness testimony, informal disclosure recipient testimony, injuries (consistent with sexual abuse) and DNA evidence. The presence of injuries or DNA evidence was collapsed into one variable due to a scarcity of cases that contained physical evidence.

Qualitative analysis

To gain a broader understanding about admission and denials of CSA, we analysed the information found in the court documents pertaining to the defendants’ statements using a data-driven thematical analysis (Braun & Clarke, 2006). Initially, the first two authors separately read written summaries of the defendants’ statements and created descriptive code labels pertaining to the information in cases of admissions (e.g. lost control, mental illness) and cases of denials (e.g. false allegation due to revenge, misunderstanding of non-abusive act). The code labels were cross-compared and disagreements were resolved through discussion. To assess inter-rater reliability, another independent researcher recoded the material after the thematic structure. A comparative analysis resulted in a 94% level of agreement for all labels in cases involving admissions and 85.7% agreement for all labels in cases involving denials. Again, disagreements were closely examined and solved through discussion. Overarching themes were thereafter created to accommodate related sub-themes (e.g. sexual preference towards children, false allegation). Lastly, quotations to exemplify the qualitative process were translated and edited to facilitate reading.

Ethical considerations

The data collection was reviewed by the Regional Ethical Regulation Board. All identifiable markers have been omitted and the case information in the quotations have been slightly changed to protect the identities of the involved parties.

Results

Quantitative analyses

Twenty-seven of the eighty-seven defendants (31%) admitted guilt to at least one charge of child sexual abuse against a preschooler, toddler, or an infant. Examining the defendants’ admissions of guilt per child complainant, the defendants admitted sexual abuse against 63 (45%) of the 140 child complainants. Among the 18 defendants tried for sexual abuse against multiple victims, 6 defendants (33.3%) admitted guilt on all charges, 9 defendants (50%) denied guilt on all charges and 3 defendants (16.7%) admitted guilt on at least one, but not all, of the charges. Notably, all charges involving an admission of guilt resulted in a conviction (n = 63, 100% conviction rate). In cases involving a denial, the charges resulted in a conviction in 56 of 77 cases (72.7% conviction rate).

Suspect factors

A binary logistic regression indicated that defendant age during trial was significantly associated with admissions of guilt, χ²(1) = 17.20, p < .001. The mean age of defendants who admitted guilt was 27.7 years (SD = 13.1) and the mean age of defendants who denied was 43.2 years (SD = 17.34). Furthermore, a past conflict with the person who made the police report was significantly associated with denials, χ²(1) = 5.63, p < .018. The statistical tests for previous criminal history and the number of child complainants did not reach the specified significance threshold (p > .05). See Table 1 for more information and odds ratios.

Table 1. Case characteristics per defendant, admission rates, inferential statistics, and odds ratios.

Characteristics	Total n (%)	Admission rate	Statistic	p-value	Odds ratio	Odds ratio 95% CI
Number of child complainants			χ^2 = 1.91	.167	0.48	0.16–1.38
One child	69 (79.3%)	27.5%
Multiple children^a	18 (20.7%)	44.4%
Defendant age during trial^b			χ^2 = 17.20	<.001***	0.93	0.90–0.97
15–17 years (juvenile)	13 (14.9%)	69.2%
18–30 years	19 (21.8%)	47.4%
31–40 years	22 (25.3%)	22.7%
41–50 years	12 (13.8%)	16.7%
>50 years	21 (24.1%)	9.5%
Conflict with the person who made the report			Fisher's exact test	.018*	8.67	1.08–69.45
No	71 (81.6%)	36.6%
Yes	16 (18.4%)	0.06%
Previous criminal history			χ^2 = 1.97	.161	2.58	0.66–10.01
Past criminal history	17 (19.5%)	17.6%
No criminal history	59 (67.8%)	35.6%
No information	11 (12.6%)	27.3%

Note: *p < .05, ***p < .001.

^aIn cases involving multiple children seven years or younger, the frequencies were as follows: charges involving 2 children (9 defendants, 10.3%), 3 children (5 defendants, 5.7%), 4 children (2 defendants, 2.3%), 13 children (1 defendant, 1.1%) and 19 children (1 defendant, 1.1%).

^bThe expected association between defendant age during trial and confession rate was examined in a binary logistic regression with age as a continuous variable. R² = .18 (Cox & Snell), .25 (Nagelkerke).

Offence-related factors

Admissions of guilt were more likely if the child was young at the onset of abuse, if the child and perpetrator had an extrafamilial relationship, and if the defendant possessed photo documentation of the child complainant or child pornography of other children. At odds with our expectations, admissions were less likely in cases containing a child testimony, a direct eyewitness testimony or a testimony from a disclosure recipient. The presence of DNA evidence or injuries, the type and duration of abuse, and the child's gender were not significantly associated with admissions. See Table 2 for an overview.

Table 2. Case characteristics per child complainant, admission rates and inferential statistics.

Characteristics	Total n (%)	Admission rate	Statistic	p-value	Odds ratio	Odds ratio 95% CI
Child gender			χ^2 = 0.60	.44
Girl	117 (83.6%)	45.3%
Boy	17 (12.1%)	35.3%
No information	6 (4.3%)	66.7%
Child age at onset of abuse^a			χ^2 = 24.21	<.001***	0.54	0.41–0.71
< 2 year	9 (7.0%)	100%
2 years	13 (5.7%)	84.6%
3 years	27 (19.3%)	37.0%
4 years	33 (23.6%)	51.5%
5 years	28 (20.0%)	39.3%
6 years	30 (21.4%)	16.7%
Child age during child interview^b			χ^2 = 0.78	.38
3 years	15 (10.7%)	53.3%
4 years	18 (12.9%)	16.7%
5 years	25 (17.9%)	36.0%
6 years	34 (24.3%)	26.5%
7 years	19 (13.6%)	31.6%
No information	29 (20.7%)	96.6%
Relationship to perpetrator			χ^2 = 12.59	<.001***	4.74	1.93–11.59
Intrafamilial	42 (30%)	19%
Extrafamilial	74 (52.9%)	52.7%
No information	24 (17.1%)	66.7%
Type of abuse			χ^2 = 0.21	.65
Penetration or oral act	63 (45.0%)	42.9%
No penetration	77 (55.0%)	46.8%
Duration of abuse			χ^2 = 0.78	.38
Repeated abuse	68 (48.6%)	41.2%
Single occasion	72 (51.4%)	48.6%
DNA or injuries			Fisher's exact test	.19
Present	10 (7.1%)	20%
Absent	130 (92.9%)	46.9%
Photo or video of abuse			χ^2 = 37.12	<.001***	0.08	0.03–0.20
Present	45 (32.1%)	82.2%
Absent	95 (67.9%)	27.4%
Child pornography			χ^2 = 22.86	<.001***	0.17	0.08–0.36
Present	54 (38.6%)	70.4%
Absent	86 (61.4%)	29.1%
Child testimony			χ^2 = 43.77	<.001***	38.71	8.74–171.52
Present	106 (75.7%)	29.2%
Absent	34 (24.3%)	94.1%
Direct eyewitness			χ^2 = 3.98	.046*	2.69	0.99–7.31
Present	23 (16.4%)	26.1%
Absent	117 (83.6%)	73.9%
Testimony from child disclosure recipient			χ^2 = 42.25	<.001***	16.84	6.37–44.54
Present	43 (30.7%)	26.8%
Absent	97 (69.3%)	86%

Note: *p < .05, ***p < .001.

^aMean child age at onset in cases of denials (M = 4.60, SD = 1.28) and admissions (M = 3.38, SD = 1.55).

^bMean child age at child interview (M = 5.29, SD = 1.23) and admissions (M = 5.06, SD = 1.41).

Qualitative analysis

Admissions of guilt

Twenty-seven defendants admitted to at least one charge of sexual abuse. Out of these defendants, seventeen provided an explanation for their actions that could be identified in the court document. Two main themes emerged from the cases containing descriptions of the defendants’ admissions of guilt: Explanations concerning sexual preference towards children and Explanations concerning mental state (see Table 3 for an overview of themes and sub-themes).

Table 3. Themes for admissions of guilt.

Themes	Sub-themes	n	%
Explanation concerning sexual preference towards children	Sexual drive	10	58.8%
	Lost control	5	29.4%
	Initiative of the child	3	17.6%
Explanation concerning mental states	Thrill of getting caught	3	17.6%
	Mental illness	3	17.6%

Explanations concerning sexual preference towards children. Most of the defendants described the abuse in relation to their sexual preference towards children. Three different sub-themes were identified within these statements: sexual drive, lost control of impulses and initiative of the child. The most common sub-theme, discussed by ten of the seventeen defendants, was to bring up their sexual drive and paedophilic urges:

Even before puberty, he was aware of his feelings towards children. These thoughts have remained since then. He has been in control over his emotions, but this time, he could not control them.

Several of these defendants stated that the abuse was a next step after a prolonged consumption of child pornography. Perceived expectations and requests from their paedophilic networks could push them towards acting, documenting, and distributing the abuse online. Another reoccurring sub-theme concerned explanations about experiencing a loss of control over their impulses. Being in a situation where they were left alone with a child was frequently discussed as a trigger. The last sub-theme concerned claims about the child's actions. Three defendants stated that the child had initiated the abuse by asking the defendant to touch them or show the defendant's genitals:

He was fast-forwarding adult films on his computer when the complainant came up behind him and asked what it was. The complainant then asked if he could do the same to her, which he did.

Explanations concerning mental state. The second theme found in the statements concerned explanations surrounding mental states, including two sub-themes; thrill of getting caught and mental illness. Three defendants stated that a risk seeking behaviour and the thrill of potentially getting caught was a contributing factor to the abusive acts:

What made him do it, beyond the temptation, was that he had not done something to her before. He thought that he could add one more child to his ‘checklist’. It was some form of ‘madness’, to go through as many children as possible before getting caught.

In three other cases, the defendants described the abuse in relation to mental illness. This sub-theme included statements concerning self-destructive thoughts and behaviour, depressive symptoms, mental retardation, and psychosis. One defendant for example explained to the court:

He cannot say with certainty if the abuse actually took place or if it is something he has imagined. // He was intoxicated and had not taken his anti-psychosis medicine at the time.

In this case, the defendant was sentenced to psychiatric care partially based on his detailed confessions throughout the police investigation and court hearing. His statement could, however, also be corroborated by hearsay testimony and behavioural changes in the child.

Denials of guilt

Forty-eight court documents contained richer descriptions about the defendants’ denial of guilt during trial that could be analysed for the purpose of this study. See Table 4 for an overview of the themes and sub-themes.

Table 4. Themes for denials of guilt.

Themes	Sub-categories	n	%
Misunderstandings	Misinterpretation of non-abusive action	24	50.0%
	Accidental touching	11	22.9%
	Initiative of the child	8	16.7%
Unreliable child testimony	Source or reality monitoring errors	13	27.1%
	Social influence on the child's testimony	11	22.9%
	Child is intentionally fabricating	4	8.3%
False allegation	False allegation due to revenge	8	16.6%
	False allegation due to other conflict	7	14.6%
	False allegation due to custody dispute	2	4.2%
Unreliable evidence	Misinterpreted forensic evidence	6	12.5%
	Fabricated forensic evidence	3	6.3%
Claiming memory loss	Claiming memory loss due to intoxication, brain injury, or sleep	3	6.3%

Misunderstandings. The most common theme in the material concerned misunderstandings. This theme included three different sub-themes; misinterpreted non-abusive actions, accidental touching, and initiative of the child. Twenty-four defendants discussed misinterpreted non-abusive actions as an explanation for the allegation, such as intimate child caring behaviour (e.g. helping the child on the toilet or getting dressed, putting lotion on a rash on the child's genitalia) or play activities. In a few cases, the defendants claimed that someone had misinterpreted their non-abusive behaviour of showing affection (e.g. hugging, kissing) as sexual due to cultural differences. The second sub-theme contained explanations regarding accidental touching. Twelve defendants described that they had accidentally touched the child in an inappropriate place, but without sexual intent:

When he bent down to kiss her he slipped and the kiss landed further down on her stomach than he had planned, approximately at the pubic bone. It was not on the complainants’ genitals, but it's possible that his beard might have touched her further down and that it felt uncomfortable.

The last sub-theme in this category included explanations about how the child had acted in ways that could be interpreted as sexual, by for example rubbing on the defendant's leg or grabbing the defendant's genitals. However, the defendants claimed that they themselves had no sexual intent and that the situation was a big misunderstanding:

He was lying naked in bed and the complainant was building a fort under the blanket. He was startled when the complainant touched his penis, but he did not want to make a big deal out of it. He told her “that's enough” and got out of bed.

Unreliable child testimony. Another reoccurring theme in the defendants’ statements pertained to the credibility of the child complainant's testimony. Three different sub-themes were identified: source or reality monitoring errors, social influence on the child's testimony and child is intentionally fabricating. In thirteen cases, the defendant speculated that source or reality monitoring errors, where the child incorrectly transferred other information to their testimony, could be an explanation behind the allegation. These statements for example included descriptions about accidental exposure to nudity, sexual acts, or adult pornography. In a few cases, the defendant stated that the child could have been victimized by someone else and inaccurately reported this memory. In other cases, the defendants brought up the possibility of the testimony just being fantasies or a retelling of someone else's memory:

The complainant knows that her own mother was sexually abused as a child, as her mother told her that a man touched her mother's body in a bad way. The defendant claimed that the mother has also given a detailed description of the abuse over the phone when the children were present.

Another sub-theme concerning the credibility of the child testimony was to discuss if the child had been exposed to various forms of social influence from adults or other children that in turn had tainted their testimony. This was often coupled with statements about false allegations, where the defendant argued that the child had been coached by another adult. Lastly, four suspects accused the child of intentionally fabricating their testimony to revenge some other form of injustice. One defendant tried for attempted rape of two children at a preschool for example accused the complainants of lying to avenge a push:

The staff came up to him and told him what the girls said that he had done. He was chocked. // The complainants’ stories are made up. He does not know how they have been able to come up with this. He thinks that one of the complainants was angry about being pushed. // The complainant wanted to see his penis and he got tired of her nagging and left. When she continued, he pushed her so that she fell.

False allegations. False allegations were another commonly discussed theme in the defendant statements. Three sub-themes could be identified: false allegations due to revenge, custody disputes or other conflicts. Eight defendants explained that the allegation was falsely made due to revenge and jealousy. In most of these cases, the police report came from the child's mother. The motive could, for example, be that the defendant had cheated or started seeing someone else. Two defendants also explained the allegation in relation to an ongoing custody dispute. The remaining seven defendants stated that the allegation was due to some other conflict. This sub-theme included motives of gaining some advantage (e.g. claim a house, win a divorce settlement) or from a financial conflict between parties;

The complainant's father was very angry with him during an argument over the church money. // He thinks that the complainant's father has told her to give false information because he and the father do not get along.

Unreliable forensic evidence. Another theme found in the defendant's explanations concerned the reliability of the forensic evidence. Two different sub-categories emerged; misinterpreted forensic evidence and fabricated forensic evidence. Misinterpretations of evidence for example included statements of mistakenly downloading child pornography in the belief that it was adult actors or possessing child pornography for research purposes (e.g. to write a book). In two cases, the defendant claimed that the photo evidence of the child complainant had been misinterpreted (i.e. that the real purpose was to document injuries on the child's genitals, or accidentally getting the focus wrong on the camera). This sub-theme also comprised explanations of misinterpreted DNA evidence:

There is a very simple explanation to why his sperm was found inside the complainant's pajama pants. A few weeks after the complainant's visit, he woke up wet. The wetness was clearly sperm. He got up and went to the bathroom, where there was a drawer for dirty laundry. He took the complainant's pajama pants and wiped his genitals. He used the crotch of the pants, since that is the thickest part of the pants and therefore the part best suited for wiping.

The second sub-theme concerned fabricated forensic evidence. Three defendants stated during trial that they were being framed, either by the person who made the police report, or in one case, by the police. This category included explanations of having their computer stolen and returned with child pornography, or having their DNA placed on the child's clothes or bed:

The defendant claimed that the complainant's mother could have planted his sperm from some used condom that he had thrown away in the bathroom waste bin.

Claiming memory loss. Lastly, three defendants denied the allegation and claimed to suffer from memory loss due to severe intoxication, brain injury, or being asleep.

Discussion

The present study examined defendant statements during court proceedings concerning alleged sexual abuse against infants, toddlers, and preschoolers. Using quantitative and qualitative analytical approaches, we sought to address two primary research questions. First, what suspect- and offence-related factors predict admissions of child sexual abuse against young children? Second, what explanations are given during trial by defendants who admit or deny guilt? In line with previous research, approximately one-third (31%) of the defendants admitted guilt during trial (e.g. Lainpelto, 2012). However, it should be noted that the estimated confession rate during Swedish preliminary investigations of CSA is around 5% (Diesen & Diesen, 2009). This is considerably lower than in, for example, studies from the United States (e.g. 32%, Cross, De Vos, & Whitcomb, 1994). Speculatively, this difference might be explained by the fact that Sweden does not employ plea bargaining processes before trial. However, since the standards and legal principles behind crime statistics differ between countries, it is difficult to make international comparisons.

In the quantitative analysis, we found support for a number of our hypotheses. Consistent with past research (e.g. Beauregard & Mieczkowski, 2011; Faller et al., 2002; Lippert et al., 2010), young defendants were more likely to admit guilt. A similar trend was found for defendants who had an extrafamilial relationship with the child (Lippert et al., 2010), which might in part be explained by the added complexity and conflicts commonly found in intrafamilial abuse cases. In general, cases involving a conflict beyond the abuse allegations were more likely to contain a denial from the suspect. At odds to our expectations, the absence of testimony from the child, an eyewitness, or an indirect disclosure recipient was associated with increased probabilities of admissions. This directional trend contradicts the findings by Lippert et al. (2010), who reported that a child or witness testimony increased the probability of confessions during police interrogations. The conflicting result could be due to our unique sample of only prosecuted cases. It is reasonable to assume that prosecutors rely largely on testimonial evidence in cases where other types of corroborative evidence are lacking. Furthermore, the cases involving infants and toddlers frequently contained a confession from the suspect. This finding could reflect the difficulties involved in identifying and prosecuting CSA cases involving children who are too young to provide testimony. A confession may be necessary to bring these cases to court in lack of evidence of strong corroborative value.

Suspects’ beliefs regarding the amount of corroborative evidence held by the police can affect the decision to confess a crime (e.g. Gudjonsson, 2003; Moston et al., 1992). In CSA investigations, the presence of corroborative evidence has often been found to increase the likelihood of confessions (Lippert et al., 2010). Interestingly, in the present study, we only observed significant effects in relation to the presence of photo or video of the abuse, as well as child pornography of other children. Neither testimonial nor physical evidence (DNA and injuries) were associated with increases in admissions. The theoretical framework of evidence elasticity may help us understand these results. According to the assumption of evidence elasticity, assessments of evidence differ depending on the type of evidence (see Ask, Rebelius, & Granhag, 2008). For example, video-recordings of sexual abuse may be more difficult to rationalize with alternative explanations than a child testimony. In other words, more ambiguous evidence is associated with higher levels of elasticity and hence leaves more room for different interpretations. The results in our qualitative analysis seem to support this idea, as defendants who denied guilt frequently questioned the reliability of testimonial and physical evidence. In the next section, we will briefly discuss these results in relation to past research regarding the reliability of child testimony, false allegations and physical evidence.

Our thematical analysis showed that defendants tended to question the reliability of the child's testimony (due to source monitoring errors, social influence, or intentional fabrications). Decades of research demonstrate that preschoolers can be reliable witnesses if interviewed with scientifically-based and age-appropriate methods (Saywitz, Lyon, & Goodman, 2017). However, young children tend to be more susceptible to social influence and source monitoring errors (Poole et al., 2015). As has been exemplified in a number of high-profile miscarriages of justice (e.g. the McMartin preschool case), suggestive influence can contribute to false testimonies of CSA (e.g. Bruck & Ceci, 1999). Police organisations in many countries, including Sweden, have therefore undergone substantial reforms to provide child interviews based on scientific recommendations (for more information, see Poole et al., 2015). Furthermore, many defendants discussed the issue of false allegations due to revenge, other conflicts or custody disputes, thus questioning the reliability of the prosecutions’ witnesses. While important to note that intentional false allegations do occur, the prevalence is estimated around 5–8% in CSA investigations (see Howitt, 2015, for an overview). Considering that more than one third of the defendants’ in the current sample discussed malicious reports, the police and prosecution may benefit from investigating and presenting evidence aimed at falsifying potential claims of intentional false allegations. Lastly, some defendants questioned the reliability of physical evidence. However, as so few cases included DNA evidence or injuries, it is difficult to draw any conclusions regarding these findings. The clear absence of physical evidence is on the other hand consistent with past research. Researchers have therefore argued that confessions may be particularly important for identifying and prosecuting crimes against children (e.g. Cross & Whitcomb, 2017).

Some methodological concerns need to be addressed before discussing the practical implications of the current results. First, the analyses are based on summaries of the defendants’ statements written by judges after the court hearing. Hence, different biases and memory errors may have influenced the data. However, this procedure enabled us to collect and analyze a larger sample of court cases than would have been possible from transcripts of the defendants’ full-length statements. Second, since we lack objective truths of the alleged events, we can not draw any inferences regarding the accuracy of the defendants’ admissions or denials of guilt. Third, we did not have access to case files from the police investigations conducted prior to the court hearing. This clearly limits the conclusions we can draw, as a prior confession during the police interview is likely to affect the suspect's subsequent statement during trial. As such, we could not capture factors related to the interrogative technique and investigative context (e.g. custody, use of legal aid, number of interviews) that can impact confession rates (e.g. Gudjonsson, 2003; Kassin et al., 2010). In fact, few studies have examined suspect interviews in Sweden (Granhag & Magnusson, 2017) and we encourage researchers to shed light on this important topic.

Although our object of study was not the suspect interviews conducted during the investigative phase, we still believe that our results send a message with relevance also for this phase. It was not rare for defendants to raise alternative explanations to the existing evidence. One empirically supported technique for optimizing the value of the interviewer's available background information is the Strategic Use of Evidence (SUE) technique (Granhag & Hartwig, 2015; Vrij & Fisher, 2016). One key feature of the SUE-technique is to systematically address possible alternative explanations to the evidence during the interview with a suspect. Specifically, to either exhaust alternative explanations or – if an alternative explanation is raised by the suspect – carefully examine this explanation. The outcome of the present paper indicates that there is room for a more extensive use of the SUE-technique (or similar techniques) during the investigative phase.

Intensive training courses in suspect interviewing techniques seem insufficient to maintain long-term improvements, as trained police officers often deviate from the techniques during field interviews (e.g. see Carter, 2011). Scholars therefore emphasize the need for continuous feedback, supervision, and knowledge on the benefits of using empirically supported techniques to facilitate improvements over time (Lassiter & Meissner, 2010; Vrij et al., 2017). Our findings may also be of practical value when planning and preparing questions for child interviews. Child interviewing protocols in, for example, Norway and Finland instruct interviewers to formulate questions regarding alternative hypotheses behind the CSA report prior to the interview (Langballe & Davik, 2017). This procedure could be particularly important in countries were children only testify via their video-recorded police interview (e.g. the Scandinavian countries), considering that child complainants have no opportunity to provide explanations or clarifications to the defendant's statement during trial.

Conclusions

The current study contributes to the growing literature on suspect confessions in child sexual abuse cases. To the authors’ knowledge, this is the first study focusing exclusively on cases involving the youngest children. Some previously identified predicative factors from past research were applicable to admissions of CSA against infants, toddlers and preschoolers. However, our hypotheses regarding associations between corroborative evidence and admission rates were largely unsupported. Police and prosecution may benefit from focusing their investigative resources into falsifying potential alternative explanations to different types of evidence before taking the case to court. We hope that our overview of explanations raised by defendants could assist during their hypothesis generation.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the Swedish Research Council for Health, Working Life, and Welfare under grant number [1533, 2013]. A portion of this research was presented at the annual conference of the European Association of Psychology & Law (EAPL) in May 2017.