Sage Journals: Discover world-class research

Abstract

Legal professionals and researchers have expressed concern that the cross-examination of complainants in Child Sexual Abuse (CSA) cases is unfair and can leave complainants confused, humiliated, and re-victimised. This study provided a multijurisdictional analysis of cross-examinations of 120 child, adolescent, and adult complainants in 91 Australian CSA trials. We examined (1) defence lawyers’ cross-examination practices, (2) how these varied according to complaint characteristics, and (3) whether cross-examination practices were associated with trial outcomes. Lawyers challenged the inconsistency, plausibility, reliability, and credibility of complainants’ accounts using an average of 75 specific lines of questioning per complainant. Older complainants were more intensely cross-examined about inconsistency, plausibility, and with multipurpose lines of questioning. A greater focus on reliability and credibility was associated with a higher likelihood of acquittal. Penetrative abuse claims were around three times more likely to result in convictions than non-penetrative claims, and joint trials about 10 times more likely to result in convictions than trials with a single complainant. This empirical study provides a foundation for improving the cross-examination of vulnerable complainants.

Keywords

Child sexual assault cross-examination jury verdict vulnerable witness witness credibility

A defendant's right to test the allegations against them is fundamental to the adversarial criminal trial. Yet concerns about the cross-examination of Child Sexual Abuse (CSA) complainants are long-standing (see Zajac et al., 2012, for a review). Despite legislative reforms since the late 19th century (see Martschuk et al., 2022), trial transcript analyses have repeatedly shown that cross-examination of CSA complainants are nonetheless complex, unfair, and upsetting (e.g., Hanna et al., 2012; Quilter & McNamara, 2023; St. George et al., 2022a, 2022b; Zajac et al., 2017).

Although legal and regulatory restrictions in all jurisdictions require cross-examination to be conducted in a civil manner without harassing, offending, oppressing, or humiliating a witness (Evidence Act 1995 (Cth), s 41), two main aspects of child cross-examinations remain problematic. First, cross-examination questions inject misconceptions about how sexually abused children behave into the trial, raising a range of credibility and plausibility issues (Davies et al., 1997; Horan & Goodman-Delahunty, 2020; St. George et al., 2022a, 2022b). For example, defence lawyers might suggest that the complainant's allegations are false because the alleged offending was not reported immediately, and/or the complainant maintained contact with the defendant (Cossins & Goodman-Delahunty, 2017; Davies et al., 1997). Similarly, cross-examining lawyers argue that children are inclined to tell lies and have devious motives (Davies et al., 1997; Stolzenberg & Lyon, 2014). Many lawyers cross-examine children by adapting adult rape myths to the CSA context (Denne et al., 2023b; St. George et al., 2022a).

Second, cross-examination content is frequently driven by misconceptions about human memory. For instance, lawyers often contend that children lack the memory capacity to be reliable witnesses or have difficulty distinguishing fantasy from reality (Davies et al., 1997). Lawyers might also imply that the presence of inconsistencies in a witness's testimony means that their evidence is wholly unreliable (Denne et al., 2023b; Pichler et al., 2021). This approach often includes an unwarranted focus on peripheral details (Hanna et al., 2012), which are prone to memory error and inconsistencies—especially in instances of repeated events of abuse (Coburn et al., 2022).

Adding to these difficulties are procedural delays within the criminal justice system, which increase stress on complainants and add to memory decay—particularly for younger complainants. Policy measures—such as the prerecording of evidence prior to trial—have been implemented to address procedural delays (Jay et al., 2022; Royal Commission into Institutional Responses to Child Sexual Abuse [Royal Commission], 2017). However, the association between procedural delay, cross-examination practices and trial outcome has not been empirically tested.

A comprehensive understanding of how CSA complainants are cross-examined is an important source of information for police, prosecutors, and policymakers. While studies on cross-examination have been undertaken since the late 1980s, they have been somewhat piecemeal in nature, making important comparisons difficult.

There are three main gaps in the research literature on cross-examination of CSA complainants. First, recent years have seen an increase in historical claims of CSA (Alaggia et al., 2019; Australian Bureau of Statistics, 2021; Easton, 2013), with complainants alleging events that happened years—or even decades—earlier. This increase comes as no surprise given rising public awareness of CSA (Mathews, 2019) and the striking prevalence of reported experiences. In one Australian study, nearly one third of adults (28.5%) said they had experienced CSA, with rates ranging from 25.7% among young adults to almost 30% among middle-aged and older adults (Mathews et al., 2023). Yet studies of CSA cross-examinations are typically limited to child or adolescent samples (e.g., Davies et al., 1997; Denne et al., 2023a; Hanna et al., 2012; St. George et al., 2022a, 2022b; Stolzenberg & Lyon, 2014; Zajac et al., 2017). Studies on adults’ allegations of contemporary sexual offences (e.g., Quilter & McNamara, 2023; Zydervelt et al., 2017) provide some clues as to how lawyers cross-examine adults, but a direct comparison of CSA complainants is crucial.

Second, how cross-examination strategies relate to other CSA complaint characteristics is poorly understood. Some research has documented common characteristics of CSA cases (Cashmore et al., 2020). We also have some information about how complaint characteristics relate to each other (Cashmore et al., 2017; Wallis & Woodworth, 2020). What we don’t know is how these factors drive cross-examination practices.

An important case characteristic that is absent from the cross-examination literature is the presence of multiple complainants against the same defendant, which may comprise up to one third of prosecutorial cases (Martschuk et al., 2024). Abuse allegations by more than one complainant are often tried separately due to concerns about unfair prejudice to the defendant. While the likelihood of conviction may increase in joint trials for the same type of offence (Tanford & Penrod, 1984, 1985), that effect appears to be due to evidentiary strength rather than unfair prejudice (Goodman-Delahunty & Martschuk, 2020). Little is known about how defence lawyers approach cross-examination of complainants in trials with one versus multiple complainants.

Finally, we know nothing about how cross-examination strategies and complaint characteristics combine to influence trial verdict. Verdict is the ultimate measure of the effectiveness of cross-examination strategies, as it provides us with some indication of what jurors see—and don’t see—as valid defences to CSA charges.

The present study

The present study explored a wide range of strategies that defence lawyers used when questioning CSA complainants: credibility; the reliability of memory; inconsistency within and across different sources of evidence; plausibility, including misconceptions about CSA; and multipurpose challenges that tap into more than one strategy. The study sought to answer the following research questions:
How does cross-examination strategy use vary according to complaint characteristics, namely gender (male vs. female), relationship to the defendant (extra- vs. intra-familial), complainant age at cross-examination (an indirect measure of reporting delay), abuse type and offending timespan (indicative of offence severity), and procedural delay.

To what extent are these cross-examination practices and complaint characteristics associated with trial verdict?

Method

Transcripts

We analysed 120 transcripts of complainant testimony from CSA trials heard in three Australian states—New South Wales, Victoria, and Western Australia—between 2011 and 2015. All Australian jurisdictions follow a common law adversarial criminal justice system based on the U.K. model. Complainants under 18 typically give direct evidence through prerecorded police interviews and are cross-examined via live closed-circuit television (CCTV) or prerecording before the trial. Adults usually undergo both direct and cross-examination in person or live via CCTV. We analysed the cross-examination of these CSA complainants.

The transcripts were a subset of those obtained from the prosecuting agency in each state in response to a notice or summons issued by the Royal Commission. Each state was required to provide prosecution files for CSA cases that had recently gone to trial, including both historical and contemporary charges. The states provided trial transcripts from 156 prosecution files. Incomplete transcripts of the complainant's evidence were excluded, reducing the sample to 127 cases involving 169 complainants. From these, we randomly selected 120 complainants (across 91 trials) to achieve an even representation of 40 complainants from each of three age groups (≤12 years, 13–17 years, ≥18 years), from three states (NSW: n = 38; Victoria: n = 34; WA: n = 48).

Of the final sample, 76.7% (n = 92) of the complainants were female. Most complainants (59.2%, n = 71) were the sole complainant; 40.8% (n = 49) were co-complainants (between two and five complainants). However, not every multiple complainant case was tried jointly: 61.7% (n = 74) were in separate trials and 38.3% (n = 46) were in joint trials. The complainants were aged between 3 and 17 years at the time of the first alleged offence (M = 9.94, SD = 3.45) and between 6 and 54 years at the time of cross-examination (M = 17.98, SD = 16.24). All 93 defendants were adults at the time of the offending and almost all (97.8%, n = 91) were male. Defendant’s age at trial (available in 67% of allegations, n = 54 defendants) ranged from 22 to 76 years (M = 46.88, SD = 13.49). The relationship between complainant and defendant varied. The accused was either an acquaintance, family friend, or professional (45%, n = 54); an immediate family member, living with the complainant (30%, n = 36); an extended family member (22.5%, n = 27); or a stranger (2.5%, n = 3). Most allegations (58%, n = 70) were penetrative offences. Complainant cross-examinations contained on average 480.91 questions (SD = 360.48, Mdn = 394.5), ranging between 51 and 2,499 questions.

Coding scheme

We applied a coding scheme based on Zydervelt et al. (2017), modifying this scheme to reflect strategies and tactics used with CSA complainants in our sample. Coding distinguished between broad strategies to discredit complainants and specific tactics used within each strategy. Four broad strategies were distinguished: inconsistency (the extent to which a witness recalls an event in same way over time, or diverges from other evidence), plausibility (the apparent reasonable and probable content), reliability (memory accuracy, including susceptibility to errors or biases), and credibility (the veracity of the witness statements and motive to lie or believability). A fifth strategy labelled multipurpose challenges was added to lines of questioning that seemed to incorporate multiple strategies. For example, questions about substance abuse could be aimed at both reliability (e.g., memory problems due to substance use) and credibility (e.g., previous bad character). Questions that were unrelated to any of the five major strategies (e.g., building rapport with the witness, checking understanding) were coded as Miscellaneous (or No Code). For a detailed description of the strategies and respective tactics coded, see Supplemental Materials 1.

Two coders trained in law and psychology first read each transcript to understand the cases context. Tactics were then coded for each line of questioning rather than for individual questions. For instance, if a lawyer asked 10 consecutive questions about why the complainant didn’t report the abuse immediately, it was coded as one tactic. Each line could be coded as multiple tactics when applicable; for example, the question, “You’re lying or confused about this, aren’t you?” was coded as both suggesting the complainant was lying and indicating unreliable memory. The same tactic could be counted multiple times in a cross-examination if separated by at least one different line of questioning. Finally, we coded the inferred intention of the lawyers’ questions, disregarding complainants’ responses to them. For example, if a lawyer suggested that the complainant still hugged the defendant after the alleged offending, a continued relationship (plausibility) code was assigned—even if the complainant denied that hugging took place.

The verdict was coded as “not guilty” or “guilty” based on the trial outcome for one or more charges relating to a particular complainant. Where a complainant's case was tried multiple times, only the final verdict was coded. Where the trial resulted in a hung jury followed by withdrawal of the charges, the outcome was recorded as not guilty.

To calculate inter-rater reliability, 20% of the transcripts were dually coded. Cohen's kappa coefficients, calculated for each tactic, were acceptable (Cohen's κ = .85; SD = 0.15, 95% confidence interval [CI] [0.79, 0.91]). Disagreements were resolved by discussion.

Analytic strategy

Almost every cross-examination employed each of the five broad cross-examination strategies. Analyses focused on the frequency with which strategy was employed. Five negative binomial regressions modelled the number of times that each of the five strategies appeared as a line of questioning, using seven independent variables derived from the key features of a complaint: complainant gender (male vs. female); relationship between the complainant and defendant (intra- vs. extra-familial); offence type (penetrative vs. non-penetrative); number of complainants per trial (one vs. multiple/joint trial); age of the complainant at cross-examination; timespan of the alleged offending (years between the first and last offence); and procedural delay (months between initial police report and cross-examination). Reporting delay (years between the first offence and initial report) was not used in regression models due to high correlation with the age of the complainant at cross-examination (r = .83, r_w = .83), indicating multicollinearity. Offending timespan and procedural delay information was missing for three complainants; these participants were excluded from the regression models. For descriptive statistics and correlations between the complaint characteristics, see Supplemental Material 2.

Removing outliers present at the higher end of the distribution led to considerable changes in the results, suggesting that it was important to retain all data including outliers. Further, some data were not independent of each other (e.g., multiple complainants testifying in the same trial). Multilevel analyses were not possible due to low power (21 clusters).

The negative binomial distribution exhibited the best fit for each of these models due to the nature of the data frequencies (distributions were positively skewed, values were positive, and data were overdispersed, i.e., the variance exceeding the mean). The negative binomial distributions exhibited the best fit in all five models, based on goodness-of-fit statistics (e.g., Pearson chi-square and deviance statistics, Akaike’s information criterion) and examination of the residuals after fitting each model. Negative binomial regression analyses were conducted with a 1,000-replication bootstrap—the most robust method to counter data that are skewed, contain outliers, and violate independency assumptions (Tabachnick & Fidell, 2018). Robust estimates of standard error were used to assess the significance of model parameters, in recognition of the potential non-independence of complainants in joint trials.

The impact of each strategy on verdict was explored in a sequential binary logistic regression with a 1,000-replication bootstrap. In the first step, the five strategies were entered as predictor variables. In the second step, the seven complaint characteristics were entered, with reference values as for previous analyses. Improvement in model fit was examined in each successive model as well as the contribution of each variable to the models.

In addition to bootstrapping, the False-Discovery Rate (FDR) method (Benjamini & Hochberg, 1995) was used to assess significant results that might be false positives due to more than 50 individual tests and associated p-values. For comparative purposes, the 35 p-values obtained for each parameter across the five negative binomial models were assessed with reference to FDR = 0.05, FDR = 0.10, and FDR = 0.20, which indicate that a proportion between 5 and 20% of false significant results is acceptable (McDonald, 2014, pp. 254–260). Results of the FDR tests are reported for each strategy variable in tables accompanying the negative binomial regression analyses. The FDR was assessed separately for the binary logistic regression and is reported in the respective table.

Results

Table 1 shows descriptive statistics for each of the five cross-examination strategies, and the specific tactics used to implement each strategy. Defence lawyers employed an average of 74.88 distinct lines of questioning, ranging from 10 to 421 (SD = 55.52). Furthermore, lawyers used every strategy with almost every complainant, ranging from M = 9.29 lines of questioning for credibility (SD = 12.11; 86.6% of transcripts) to M = 27.54 for inconsistency (SD = 24.41; 99.2% of transcripts). Use of each strategy was positively correlated with all others (r = .23–.48, p < .05), and with the overall number of cross-examination questions (r = .54–.71, p < .01), indicating that cross-examination was longer when more strategies were employed to challenge the complainant's account. However, defence lawyers asked an average of 20 lines of questioning that were unrelated to the five major strategies to challenge a complainant (M = 20.51, SD = 15.88, range: 1–121), regardless of complainant age (r = −.09, p = .321). Negative binomial regression analyses revealed that none of the complaint characteristics were associated with cross-examination length (likelihood ratio χ²(7) = 3.02, p = .883, all p _b s > .10).

Table 1.
Lines of questioning employed during cross-examination.

M (SD) Mdn Max % (SD) % Cases

Inconsistency 27.54 (24.41) 19.5 152 27.15 (13.17) 99.2

With own report 12.08 (14.54) 7 90 11.11 (8.71) 92.5

With another witness 9.70 (10.24) 7 59 10.61 (9.25) 86.7

With other evidence 1.78 (3.70) 0 21 1.76 (4.00) 33.3

With accused 3.98 (6.43) 0 31 3.68 (5.46) 49.2

Plausibility 13.13 (12.28) 9 71 14.42 (10.44) 96.7

Delayed report 3.41 (4.07) 2 24 4.40 (6.45) 70.8

Continued relationship 2.33 (4.69) 0 36 2.21 (3.86) 44.2

Lack of resistance 1.83 (3.16) 0 20 1.91 (2.952) 47.2

Lack of distress 0.63 (1.32) 0 10 0.72 (1.38) 32.5

General plausibility 4.95 (6.55) 3 43 5.18 (5.14) 85.8

Reliability 11.53 (15.95) 7 119 11.26 (09.10) 90.8

Memory 10.83 (15.42) 6 117 10.48 (8.78) 90.8

Situational factors 0.71 (2.27) 0 16 0.78 (2.36) 17.5

Credibility 9.29 (12.11) 5 78 8.86 (8.12) 86.7

Lying 3.35 (5.22) 1 33 2.98 (3.66) 62.5

Ulterior motive 3.64 (5.16) 2 42 3.88 (4.45) 72.5

Previous bad character 2.30 (5.09) 0 26 1.99 (3.89) 38.3

Multipurpose 13.38 (11.80) 10 68 14.57 (8.98) 98.3

Substance use 0.83 (2.77) 0 21 0.94 (2.97) 16.7

Mental health 0.30 (1.3) 0 10 0.31 (1.20) 7.5

Custody dispute 0.24 (1.11) 0 11 0.24 (0.95) 10.0

Collusion or contamination 1.48 (3.40) 0 21 1.90 (4.27) 31.7

Complainant is wrong 10.19 (10.04) 8 68 10.87 (9.25) 95.8

History of sexual abuse 0.33 (1.59) 0 16 0.31 (1.20) 12.5

Miscellaneous 20.51 (11.58) 16 121 23.75 (11.66) 100.0

Inconsistency challenges

As mentioned above, inconsistency was the most frequently used cross-examination strategy, applying to 99.2% of the complainants. On average, 27.2% of the cross-examination focused on inconsistency within the complaint's own statement or with other statements (SD = 13.17). Defence lawyers were most likely to question complainants about inconsistencies within their own accounts (92.5% of complainants), pertaining to both major details (e.g., whether the offending had occurred, the nature of the sexual offences) and minor details (e.g., clothing worn, the time of day, the day of the week).

The fitted negative binomial did not significantly improve on the intercept-only model (likelihood ratio χ²(7) = 9.19, p = .239). Model fit was acceptable, although the goodness-of-fit statistics indicated some under-dispersion (e.g., Pearson χ²(106) = 85.26, ratio = .804).

Although the overall model was not significant, the results from the bootstrap analyses indicated a significant main effect of trial type and complainant age at the time of the trial (see Supplemental Materials 3, Table S3.1). Complainants in joint trials were almost 1.5 times more frequently challenged on inconsistencies than complainants in separate trials (Wald χ²(1) = 4.77, p_b = .050, odds ratio (OR) = 1.48, 95% CI [1.04, 2.10]; joint trials: M_EM = 31.55, 95% CI [24.46, 40.71]; separate trials: M_EM = 21.32, [16.69, 27.24]). Figure 1 shows the different tactics that challenged inconsistency as a function of trial type. Complainants in joint trials were more intensely questioned on inconsistencies with other witnesses (Cohen's d = 0.53, [0.16, 0.90]), and with the accused's version of the events (d = 0.66, [0.28, 1.03]); there were no differences for inconsistency with own report (d = 0.05, [−0.31, 0.42]), or with other independent evidence (d = −0.07, [−0.43, 0.29]).

Figure 1.
Tactics used to challenge the inconsistency of a complainant's account: separate versus joint trials. Error bars are 95% CIs.

Older complainants were more intensely cross-examined about inconsistencies, holding all other variables in the model constant (Wald χ²(1) = 9.84, p_b = .007, significant at FDR = 0.10), such that with each additional year of age, the number of questioning lines on inconsistency increased by 2.1% (OR = 1.02, [1.01, 1.03]).

Challenges to plausibility

On average, 14.4% of the cross-examination focused on the plausibility of the allegation(s). The most frequently used plausibility tactic was general plausibility (85.8% of complainants); lawyers contended that the complainant's version of events did not align with “common sense” (e.g., other people's reactions, the defendant’s character, or the physical likelihood of an event). In one trial, the defence lawyer suggested that thin walls in the house meant someone would have overheard the offending had it taken place. Over two-thirds of complainants (70.8%) were challenged about delayed reporting. These delays ranged from minutes to decades. Almost half of the complainants were challenged on their failure to resist the alleged offending (47.2%) and on maintaining a relationship with the defendant (44.2%). Around one-third of complainants (32.5%) were challenged on their perceived lack of distress following the alleged offence (e.g., noting that the complainant went out to play afterwards) or when recounting it.

The fitted negative binomial model significantly improved on the intercept-only model (likelihood ratio χ²(7) = 15.28, p = .033) (see Supplemental Materials 3, Table S3.2). Again, model fit was acceptable, although the goodness-of-fit statistics indicated some under-dispersion (e.g., Pearson χ²(106) = 71.40, ratio = .674).

Offence type and complainant age at trial emerged as significant main effects in the model. Holding all other variables constant, lawyers used 1.5 more lines of questioning about plausibility when the charge was for a penetrative versus a non-penetrative offence (Wald χ²(1) = 6.63, p_b = .015, significant at FDR = 0.20; penetrative offences: M_EM = 13.34, 95% CI [11.03, 16.14]; non-penetrative offences: M_EM = 8.81, [6.81, 11.42]). Further, with each additional year of age, complainants were subjected to 2.9% more lines of questioning about the plausibility of their account (Wald χ²(1) = 9.53, p_b = .014, significant at FDR = 0.20, OR = 1.03, [1.01, 1.05]).

Given that the model was significant, we examined the number of discrete plausibility tactics as a function of offence type. Compared to non-penetrative abuse, complainants who alleged penetrative offences were subjected to more lines of cross-examination about delayed reporting (Cohen's d = 0.52, 95% CI [0.15, 0.88]) and their lack of resistance (d = 0.45, [0.09, 0.82]), as shown in Figure 2. No other significant differences were attributable to plausibility tactic use (continued relationship: d = 0.16, [−0.21, 0.52]; lack of distress: d = −0.20, [−0.20, 0.52]; general: d = 0.18, [−0.18, 0.54]).

Figure 2.
Tactics used to challenge the plausibility of a complainant's account: non-penetrative versus penetrative abuse. Error bars are 95% CIs.

Challenges to reliability

On average, 11% of cross-examination focused on the reliability of complainants and their evidence (SD = 9.10). Lawyers predominantly questioned complainants about internal memory errors (90.8%) as opposed to external situational influences (17.5%). When addressing potential memory errors, lawyers often focused on details central to the alleged offence, such as incremental disclosures or a lack of key information. At other times, lawyers challenged complainants on their (lack of) memory for minor details—including the colour of their bedsheets; the type of pyjamas they wore; whether they had a bath on the night in question; whether the alleged offence lasted 2 or 3 min; and the frequency, timing, and details of conversations with their parents about the abuse. Tactics that challenge the reliability of a complainant's account overlapped considerably with tactics that highlight inconsistencies within their statement.

The fitted negative binomial model did not significantly improve on the intercept-only model (likelihood ratio χ²(7) = 9.69, p = .207). Model fit was acceptable, but the goodness-of-fit statistics indicated some over-dispersion (e.g., Pearson χ²(106) = 172.39, ratio = 1.63).

Although the overall model was not significant, the main effect of offending timespan was significant (Wald χ²(1) = 4.24, p_b = .043) (see Supplemental Materials 3, Table S3.3). Holding all other variables constant, the number of lines of questioning challenging the reliability of the complainant's account was 1.10 times greater for each additional year of the duration of the offending (OR = 1.10, 95% CI [1.01, 1.21]).

Challenges to credibility

On average, 8.9% of the cross-examination focused on credibility (SD = 8.12). The most common tactic was suggesting an ulterior motive for the allegation against the defendant (72.5% of complainants), closely followed by direct accusations of lying (62.5%). Suggested ulterior motives were highly case-specific, but consistent with stereotypes of complainants of different ages. Children were accused of lying due to pressure from others, adolescents to gain independence and adults for financial gain.

Lawyers questioned 38.3% of the complainants about their prior dishonesty or bad character. Sometimes these accusations were related to the offence (e.g., suggesting previous false allegations of abuse), while other times they were not (e.g., highlighting an adult complainant's disruptive behaviour at school). In one trial, after a 10-year-old complainant answered “yes” to the judge's competency question about whether she had ever told “a little lie”, defence counsel inferred that she was also lying about the alleged offending.

The fitted negative binomial model significantly improved on the intercept-only model (likelihood ratio χ²(7) = 19.79, p = .006) (see Supplemental Materials 3, Table S3.4). Model fit was acceptable, although the goodness-of-fit statistics indicated some over-dispersion (e.g., Pearson χ²(106) = 164.98, ratio = 1.56).

Procedural delay emerged as a significant main effect in the model (Wald χ²(1) = 35.07, p_b = .001, significant at FDR = 0.05). Holding all other variables constant, an additional month of procedural delay was associated with 1.1% fewer lines of questioning about credibility-related issues (OR = 0.99, 95% CI [0.99, 0.99]).

There was also marginal effect of the nature of the relationship between the offender and the complainant (Wald χ²(1) = 3.87, p_b = .079), such that complainants in intra-familial cases were asked 1.6 times more lines of questions about their credibility than their counterparts in extra-familial cases (intra-familial: M_EM = 9.29, 95% CI [6.16, 14.02]; extra-familial: M_EM = 5.64, [3.59, 8.85]).

Multipurpose challenges

Multipurpose lines of cross-examination (accounting for an average of 14.6% of cross-examination, SD = 8.98) did not suggest any obvious strategy other than to fulfil obligations pursuant to the rule in Browne v. Dunn (1893), according to which a witness must be given the opportunity to respond to the opposing party's argument and defend their character. By far the most common tactic was simply to suggest the complainant was wrong (95.8% of complainants). One lawyer used this tactic 68 times during a single cross-examination. Some lawyers explained that it was part of their job to make suggestions of this nature, and that the complainant did not have to agree.

Lawyers suggested to 31.7% of complainants that their evidence was contaminated by, or the result of collusion with, another person. Even brief discussions between co-complainants were used to imply that the evidence was either deliberately or unintentionally incorrect. Additionally, 16.7% of complainants were questioned about their use of substances, and 7.5% about their mental health. These tactics were aimed at undermining both credibility (e.g., by arguing that self-harm is attention-seeking behaviour) and reliability (e.g., by arguing that anti-depressants could disrupt memory).

Lawyers questioned 12.5% of complainants about their sexual history or family abuse, alleging they might be confusing memories for sexual experiences or fabricating accounts. This tactic was employed even when complainants denied previous sexual experience. For example, an 11-year-old was repeatedly asked if a “similar thing” had happened to her; she denied it each time. A 26-year-old was questioned about her mother's history of sexual abuse, despite reporting no knowledge of it. Additionally, 10% of complainants were asked about custody disputes, regardless of their awareness of such disputes or the defendant's involvement.

The fitted negative binomial model did not significantly improve on the intercept-only model (likelihood ratio χ²(7) = 10.00, p = .189) (see Supplemental Materials 3, Table S3.5). Model fit was moderate, with the goodness-of-fit statistics indicating some under-dispersion (e.g., Pearson χ²(106) = 60.55, ratio = .571).

Although the overall model was not significant, a significant main effect of complainant age emerged. Holding all other variables constant, older complainants were asked 2.5% more multipurpose lines of questioning for each additional year of age (Wald χ²(1) = 11.66, p_b = .005, significant at FDR = 0.10; OR = 1.03, 95% CI [1.01, 1.04]). We therefore examined the number of specific multipurpose tactics as a function of complainant age. As shown in Figure 3, adult complainants were more often challenged on substance use (M_EM = 1.75, SD = 4.37) than child complainants (M_EM = 0.03, SD = 0.16, Cohen's d = 0.56, 95% CI [0.11, 1.00]), while means for adolescents fell between the two other age groups (M_EM = 0.73, SD = 1.65). In other words, the older the complainants, the more likely they were to be challenged about substance abuse (r_s(120) = .27, p = .002). Further, adolescents were more likely to be challenged about mental health issues (M_EM = 0.60, SD = 1.84) than children under the age of 13 years (M_EM = 0.00, SD = 0.00, d = 0.46, [0.02, 0.91]).

Figure 3.
Multipurpose tactics used to challenge a complainant's account: mean counts by age of the complainant. Error bars are 95% CIs.

Predictors of verdict

Overall, 55.6% of the complainant claims resulted in a conviction on at least one count (18.8% guilty on some counts; 36.8% guilty on all counts); the remaining claims resulted in acquittals on all counts. A binary logistic regression assessed whether verdict could be predicted based on proportion of cross-examination strategy use and complaint characteristics, using the 1,000-replication bootstrap. The five strategies were entered in the first step, followed by the seven complaint characteristics in the second step.

The first logistic regression model did not significantly improve on the intercept-only model (χ²(5) = 9.26, p = .099) (see Table 2, Model 1). However, the model improved verdict prediction slightly, from 56.8% to 63.1% (43.8% correct for acquittals, 77.8% correct for convictions). Greater focus on inconsistency challenges (Wald χ²(1) = 3.98, p_b = .045, OR = 0.96, 95% CI [0.92, 0.99]), and credibility challenges (Wald χ²(1) = 5.04, p_b = .023, OR = 0.94, [0.89, 0.99]) during cross-examination was associated with a higher likelihood of an acquittal.

Table 2.
Sequential logistic regression results predicting verdict.

Wald (df = 1) p_b B [95% CI] SE B OR [95% CI]

Step 1

Inconsistency strategy 3.98 .045* −0.04 [−0.09, 0.00] 0.02 0.96 [0.92, 0.99]

Plausibility strategy 0.22 .633 −0.12 [−0.08, 0.05] 0.03 0.99 [0.94, 1.04]

Reliability strategy 2.46 .082* −0.05 [−0.11, 0.01] 0.03 0.96 [0.90, 1.01]

Credibility strategy 5.04 .023 −0.07 [−0.14, −0.02] 0.03 0.94 [0.89, 0.99]

Multipurpose strategy 1.23 .291 −0.03 [−0.09, 0.03] 0.03 0.97 [0.92, 1.02]

Constant 4.32 .024 3.11 [0.43, 6.77] 1.49 22.32

Step 2

Inconsistency strategy 9.10 .003* −0.08 [−0.19, −0.04] 2.74 0.92 [0.87, 0.97]

Plausibility strategy 1.56 .253 −0.04 [−0.14, 0.03] 3.14 0.96 [0.90, 1.02]

Reliability strategy 3.94 .047* −0.08 [−0.20, −0.01] 3.83 0.93 [0.86, 0.99]

Credibility strategy 6.28 .008*** −0.09 [−0.22, −0.03] 3.58 0.91 [0.85, 0.98]

Multipurpose strategy 3.60 .071 −0.07 [−0.19, 0.01] 3.60 0.93 [0.87, 1.00]

Gender 0.00 .990 0.00 [−1.76, 1.92] 0.62 1.00 [0.30, 3.37]

Abuse type 3.58 .037* 1.07 [0.03, 2.60] 0.56 2.90 [0.96, 8.75]

Relationship 0.01 .964 −0.04 [−1.45, 1.49] 0.55 0.96 [0.33, 2.82]

Trial type 10.96 <.001*** 2.28 [1.39, 4.64] 0.69 9.76 [2.53, 37.59]

Complainant age at cross-examination 0.35 .500 −0.02 [−0.10, 0.05] 0.03 0.98 [0.93, 1.04]

Offending timespan 2.30 .152 0.21 [−0.05, 0.81] 0.14 1.23 [0.94, 1.60]

Procedural delay 1.14 .084 0.01 [−0.01, 0.03] 0.01 1.01 [0.99, 1.02]

Constant 6.23 .013 4.15 [0.72, 9.43] 1.66 63.42

Note. p_b and 95% CI for B are based on the 1,000-replicated bootstrap.

95% CI: 95% confidence interval; SE: Standard Error; OR: Odds Ratio.

Significant at <20%, significant at <10%, significant at <5% false discovery rate.

The second model, comprising the seven complaint characteristics as well as the five strategies, was significant (χ²(12) = 38.61, p < .001, Nagelkerke R² = .294, Cox and Snell R² = .394) (see Table 2, Model 2). Verdict prediction accuracy improved to 71.2% (64.6% correct for acquittals, 76.1% correct for convictions). Holding all other cross-examination strategies and complaint characteristics constant, greater focus on cross-examination about inconsistency (Wald χ²(1) = 9.10, p_b = .003, OR = 0.92, 95% CI [0.87, 0.97]), reliability (Wald χ²(1) = 3.94, p_b = .047, OR = 0.93, [0.86, 0.99]), and credibility challenges (Wald χ²(1) = 6.28, p_b = .008, OR = 0.91, [0.85, 0.98]), was associated with a higher likelihood of an acquittal. In other words, the odds of acquittal increased by about 8.6% when the cross-examiner asked proportionally more questions on inconsistency, by about 7.9% when more questioning related to reliability, and by about 9.4% when the questioning was on credibility.

Whether the trial was separate involving one complainant or joint involving multiple complainants was the strongest predictor of verdict (Wald χ²(1) = 10.96, p_b < .001, significant at FDR = 0.05). Joint trials involving multiple complainants were 10 times more likely to result in a conviction (OR = 9.76, 95% CI [2.53, 37.59]). Further, cases involving penetrative abuse claims were almost three times more likely to result in a conviction than non-penetrative abuse claims (Wald χ²(1) = 3.58, p_b = .037, OR = 2.90, [0.96, 8.75]).

Complainant age at cross-examination was not associated with verdict. Given that complainants were significantly older in joint trials (M = 2.77, SD = 11.99) than separate trials (M = 16.10, SD = 6.83), we investigated the relationship between complainant age and verdict as a function of trial type. To account for a possible curvilinearity, complainant age was squared. Figure 4 shows a slightly negative association between complainant age and guilty verdicts in separate trials. In joint trials, however, the association was curvilinear; the likelihood of conviction increased until around ages 30–35, then decreased.

Figure 4.
Verdict as a function of complainant age at cross-examination and the number of complainants in the case.

Discussion

Defence lawyers used an average of 75 identifiable lines of questioning with each CSA complainant, employing four main strategies—inconsistency, plausibility, reliability, and credibility—in nearly every case, along with multipurpose strategies (e.g., highlighting substance use or mental health issues) that could act to challenge a complainant's credibility or reliability. Within each strategy, lawyers used a variety of tactics, often applying the same tactic in different ways or in the same way interspersed among other types of challenges. As a result, the average number of lines of questioning was higher than the overall number of tactics used.

Cross-examination practices and complaint characteristics

Many cross-examination challenges were nonspecific, merely accusing the complainant of being “wrong”. However, some specific tactics revealed misconceptions about how CSA victims report and remember their experiences. Complainants’ evidence was often held to unrealistic standards, based on the mistaken belief that memory functions like a recording system (e.g., Akhtar et al., 2018).

The transcripts that we analysed involved the almost ubiquitous presence of several features that are detrimental to recall, and especially the recall of minutiae: long delays between the event and cross-examination and a lack of knowledge of sexual assault at the time of encoding. The expectation that complainants of CSA should recount highly specific and consistent details is out of step with decades of research. Yet our data revealed that this expectation was present during cross-examination, with challenges to inconsistency being the most common strategy, accounting for nearly 30% of all cross-examination, regardless of the offending timespan.

We know that further memory difficulties arise when complainants are required to recall individual instances of repeated abuse. Exposure to repeated instances of similar abusive events leads to difficulties isolating what happened during a particular occurrence (Coburn et al., 2022; Powell & Thomson, 1996). Although repeated events strengthen memories for core actions, peripheral details, and variable actions are less memorable (Cossins & Goodman-Delahunty, 2017). Yet defence lawyers in our study employed more lines of questioning that challenged memory reliability when complainants reported more protracted and therefore more repeated instances of abuse.

Other cross-examination tactics focused on common misconceptions about the ways that victims of CSA behave, both during and after abusive episodes. Defence lawyers often argued that genuine victims would “put up a fight”, report abuse immediately, and cut ties with the perpetrator—particularly in cases involving allegations of penetrative offences. Similar tendencies to cross-examine children about lay misconceptions of sexual offending have been observed in U.S. trials (St. George et al., 2022a, 2022b) and with adult sexual abuse complainants in Australian trials (Quilter & McNamara, 2023).

Defence lawyers in trials with more than one complainant were more likely to cross-examine on inconsistencies with other witnesses and with the accused's version of the events. This greater focus on inconsistencies with other witnesses is expected, given that allegations need to be similar for evidence from multiple complainants to be admitted (Goodman-Delahunty et al., 2024).

There was some evidence that defence lawyers used some lines of questioning more often with older complainants than with younger complainants. Adult complainants, for example, were more likely than adolescents or children to be questioned about substance use (a multipurpose tactic), while adolescents were more likely to be asked about their mental health than younger children. Further, older complainants faced more challenges regarding inconsistencies in their evidence and the plausibility of their accounts.

In contrast, there was no association between credibility challenges and complainant age (an indirect indication of reporting delay). On closer inspection, lawyers used this strategy liberally—even with complainants who made a prompt contemporaneous disclosure. One complainant, for example, was challenged about waiting until the defendant left the house before disclosing to his mother. Other complainants were asked why they didn’t disclose immediately to a stranger (e.g., a service station or pool attendant) rather than waiting a matter of hours or even minutes until they encountered a trusted adult. Similarly, the number of lines of questions on memory reliability did not differ according to complainant age, suggesting that defence lawyers asked older complainants questions related to memory and CSA myths to the same extent as younger complainants.

Other patterns in our data were less intuitive. For example, longer procedural delays were not associated with more lines of questioning about inconsistencies and reliability. In fact, lengthy procedural delays were associated with fewer credibility challenges. This effect could be attributable to unidentified factors. For instance, procedural delay was positively correlated with complainant age (see Supplemental Materials 2). One possible explanation, then, is that courts have started prerecording younger complainants’ evidence in advance of the trial, while older complainants were more likely to be cross-examined during the trial, either via CCTV or in person. Indeed, prerecorded cross-examination was associated with younger complainant age and shorter procedural delay than live cross-examination. Therefore, holding all other factors—including complainant age—constant, lawyers appeared more likely to challenge complainants’ credibility when they cross-examined complainants in advance of the trial, perhaps refining their case theory afterwards.

Trial verdict

Ultimately, the challenges that defence lawyers select are aimed at jurors, who determine the trial outcome. For this reason, defence lawyers are unlikely to use tactics unless they perceive them to align with—or at least not conflict with—jurors’ existing beliefs (Temkin, 2000). Indeed, researchers have noted that potential jurors endorse some beliefs about complainants of CSA that are not grounded in reality (Goodman-Delahunty et al., 2017a, 2017b; Shackel, 2009); even judges demonstrate misconceptions about CSA victims (Connolly et al., 2009). When cross-examination tactics tap into—or generate—misperceptions, defence lawyers can create reasonable doubt where none exists.

Some strategies were indeed associated with verdicts. Controlling for all complaint characteristics, we found that emphasis on inconsistencies, reliability, and credibility of the complainant’s account was associated with conviction likelihood, suggesting that these strategies were successful in establishing reasonable doubt. In contrast, multipurpose strategies and plausibility challenges linked to CSA misconceptions or rape myths were unrelated to verdicts.

Only two of our complaint characteristics were associated with verdict. First, the presence of more than one complainant at trial increased convictions by a factor of 10, despite an increase in questions about memory contagion and collusion. This effect is foreseeable given that lack of corroboration is a major barrier to CSA convictions (Connolly et al., 2009; Mills & Sharman, 2017), and that other research echoes an association between multiple complainants and conviction (Goodman-Delahunty & Martschuk, 2020; Leipold & Abbasi, 2006). The presence of multiple complainants who corroborate an offender's modus operandi strengthens the credibility of each complainant—and therefore the prosecution case—considerably (Saks & Spellman, 2016). This increase can be attributed to more incriminating evidence against the defendant (anticipated prejudice), as opposed to unfair prejudice (Saks & Spellman, 2016). We note, however, that our effect appeared to be driven by joint trials involving 20- to 40-year-old complainants.

Second, charges involving penetrative abuse claims were three times more likely to result in convictions than non-penetrative claims. While we cannot pinpoint which specific charges led to convictions in cases involving both offence types, prior research on mock juries’ decisions in an historical CSA case suggests a pattern: when faced with both non-penetrative (acts of indecency) and penetrative charges (sexual intercourse) against a defendant, mock juries are more likely to convict on less severe charges and acquit on the more serious ones (Goodman-Delahunty & Martschuk, 2020).

Our findings about the impact of certain case-related features on the perceived strength of the evidence—specifically regarding single versus multiple complainants, and penetrative versus non-penetrative sexual conduct—contribute to the case trajectory at all phases of the criminal justice process. These factors independently shape the availability and/or effectiveness of certain cross-examination strategies. For example, a single complainant alleging non-penetrative abuse is not only more likely to face all four cross-examination strategies, but also more vulnerable to them—more so than a single complainant alleging penetrative abuse, or any complainant in a case with multiple accusers, regardless whether the alleged abuse involved penetration.

Limitations and future directions

As with any study using court transcript data, several limitations must be considered. First, we should question whether our sample is representative of trials in Australia and further afield. To counter selection bias, we requested trial transcripts of all CSA trials from 2010 to 2015 in three Australian jurisdictions and randomly selected an equal number of complainants across age groups (child, adolescent, adult). We note that the conviction rate in our sample (55.6%) aligns with other Australian studies on CSA cases (e.g., 55.5% in Cashmore et al., 2020).

Second, research of this kind can never establish the ground truth of the allegations. It is possible that some genuine allegations were unsubstantiated and vice versa. However, it is important to consider that Australian cases do not proceed to trial unless the prosecutor can establish a reasonable prospect of conviction and that the prosecution is in the public interest (Hodgson et al., 2020).

Third, regardless of the complaint characteristics, we found near-ceiling effects for all five cross-examination strategies; every cross-examination strategy was used with almost every complainant. It is possible that defence lawyers employ a scattershot approach—using every available means to challenge a complainant's evidence, rather than an approach closely targeted to the individual complaint (Hampel et al., 2016). However, we are cautious about this interpretation, given that (1) the same cross-examination tactics were used in diverse ways, and (2) we analysed cross-examination strategies without any explicit knowledge of the defence theory. We note that even a well-planned defence might entail multiple challenges to a complainant's account. Future research should explore patterns of cross-examination strategies across different jurisdictions and investigate the frequency with which specific tactics co-occur within a single cross-examination. Similarly, researchers should investigate cross-examination strategies in light of defence theories and their questioning plan. Where scientific evidence does not support application of a defence theory, research on how courts can best consider this information could be useful.

Fourth, the trials analysed in this study took place before Australian states implemented ground rules hearings—in which judges set parameters for questioning children (e.g., manner, duration, or adherence to the rule in Browne v. Dunn (1893)) and intermediary schemes to assist children with communication (e.g., S 389E Criminal Procedure Act 2009 [Vic] ), and S 294M Criminal Procedure Amendment (Child Sexual Offence Evidence) Act 2023 [NSW] , respectively). These reforms may have altered some cross-examination practices. However, both reforms target question structure and language rather than cross-examination strategies aiming to undermine the complainant's evidence. Future research should examine how, if at all, defence lawyers have adjusted their strategies and tactics in cases where ground rules hearings were held and/or intermediaries were present at trial.

Fifth, examining complainants’ responses to cross-examination was beyond the scope of our study. Some cross-examination challenges might backfire by allowing complainants to provide information that counters the challenges. This possibility might explain the observed lack of association between plausibility challenges and verdict.

Finally, jurors reach their verdict based on all the evidence presented at trial, including the child's pre-recorded and in-court responses. Previous research showed some overlap in the way that prosecution and defence lawyers question sexual abuse complainants (Powell et al., 2022; Quilter & McNamara, 2023). Therefore, the evidence presented at trial and the questioning strategies employed by prosecutors—and where applicable, police interviewers—are likely to account further for variance in verdicts.

Practical implications and conclusions

Across the globe, courts are adopting trauma-informed criminal justice policies to safeguard sexual assault complainants from experiencing secondary victimisation in the courtroom (American Bar Association, 2018; Cooper & Mattison, 2017; Goodman-Delahunty & Cowdery, 2023). Our findings provide information for defence lawyers about which strategies are likely to be fruitless in challenging complainants’ accounts. For instance, longer cross-examinations did not decrease the likelihood of conviction; nor did plausibility strategies that play into misconceptions of CSA. Further, these findings can assist prosecutors to decide whether expert evidence is required to counter misconceptions about CSA, about memory or about children in general. Similarly, by knowing which strategies and topics tend to dominate cross-examination in different cases, police investigators can pre-emptively address these in their interviews.

Some academics have been criticised for insensitivity to the presumption of innocence in their push for legislative changes to increase successful prosecution of child abusers (Powell et al., 2007; Yehia, 2010). But legal practice matters, too. For instance, the Australian High Court reviewed questions that the law entitles cross-examiners to ask ( Libke v. The Queen [2007] HCA 30 , Heydon J. [127]). Our work provides a detailed analysis of defence lawyers’ cross-examination practices, offering practical insights into which approaches are more or less effective, along with implications for prosecutors seeking to engage more actively with these practices at trial.

Independent enquiries have suggested reducing procedural delay to trial in CSA cases (Jay et al., 2022; Royal Commission, 2017), but little empirical research has examined the relationship between procedural delay, cross-examination practices and verdict. In our study, reduced procedural delay was associated with more challenges to credibility, but not with other challenges. However, procedural delay was not associated with verdict, suggesting that efforts to reduce delay might expedite closure for complainants without disadvantaging defendants.

Yet regardless of cross-examination content and timing, lengthy and challenging cross-examinations can be detrimental to complainants’ wellbeing and the quality of their evidence (Quilter & McNamara, 2023). Even short laboratory interviews mimicking cross-examination's leading, complex, and confrontational nature reduce children's accuracy (see Zajac et al., 2012); fatigue associated with longer cross-examinations could further reduce performance (Enriques et al., 2023; Gudjonsson, 1992). More broadly, challenges to complainants during cross-examination—for example, repeatedly being called a liar—can result in short- and long-term effects on various facets of wellbeing (Grob et al., 2023; Quas & Goodman, 2012; see Enriques et al., 2023, for a review). Although some legal professionals consider these effects necessary collateral damage that pales in comparison to being wrongfully accused or convicted (Cashmore & Bussey, 1996), distress suffered during (and after) cross-examination has far-reaching consequences for the justice system itself. The preponderance of CSA complainants who say they would not re-engage with the justice system (Field & Katz, 2023) should be a wake-up call to policymakers (Ellison & Munro, 2017; Kezelman & Stavropoulos, 2016; Pemberton & Mulder, 2023).

Implementing meaningful criminal justice reforms is difficult in CSA cases (Dallaston & Mathews, 2022; Goodman-Delahunty & Cowdery, 2023), where an absence of corroborating evidence means that cases often hinge on the complainant's memory. This study provides a deeper understanding of cross-examination strategies and tactics in CSA cases and how these practices affect verdicts—in turn providing a stronger platform for meaningful reform. One key implication of our findings is the need to limit cross-examination strategies based on misconceptions about CSA and memory. This could potentially be achieved through legislative reforms (e.g., Evidence Acts) or other procedural and regulatory changes, ensuring that cross-examination tactics serve the interests of justice.

Supplemental Material

sj-docx-1-anj-10.1177_26338076251335352 - Supplemental material for Lawyers’ strategies for cross-examining complainants of child sexual abuse

Supplemental material, sj-docx-1-anj-10.1177_26338076251335352 for Lawyers’ strategies for cross-examining complainants of child sexual abuse by Natalie Martschuk, Jane Goodman-Delahunty, Martine B Powell, Nina Westera and Rachel Zajac in Journal of Criminology

Supplemental Material

sj-docx-2-anj-10.1177_26338076251335352 - Supplemental material for Lawyers’ strategies for cross-examining complainants of child sexual abuse

Supplemental material, sj-docx-2-anj-10.1177_26338076251335352 for Lawyers’ strategies for cross-examining complainants of child sexual abuse by Natalie Martschuk, Jane Goodman-Delahunty, Martine B Powell, Nina Westera and Rachel Zajac in Journal of Criminology

Supplemental Material

sj-docx-3-anj-10.1177_26338076251335352 - Supplemental material for Lawyers’ strategies for cross-examining complainants of child sexual abuse

Supplemental material, sj-docx-3-anj-10.1177_26338076251335352 for Lawyers’ strategies for cross-examining complainants of child sexual abuse by Natalie Martschuk, Jane Goodman-Delahunty, Martine B Powell, Nina Westera and Rachel Zajac in Journal of Criminology

Supplemental Material

sj-sav-4-anj-10.1177_26338076251335352 - Supplemental material for Lawyers’ strategies for cross-examining complainants of child sexual abuse

Supplemental material, sj-sav-4-anj-10.1177_26338076251335352 for Lawyers’ strategies for cross-examining complainants of child sexual abuse by Natalie Martschuk, Jane Goodman-Delahunty, Martine B Powell, Nina Westera and Rachel Zajac in Journal of Criminology

	M (SD)	Mdn	Max	% (SD)	% Cases
Inconsistency	27.54 (24.41)	19.5	152	27.15 (13.17)	99.2
With own report	12.08 (14.54)	7	90	11.11 (8.71)	92.5
With another witness	9.70 (10.24)	7	59	10.61 (9.25)	86.7
With other evidence	1.78 (3.70)	0	21	1.76 (4.00)	33.3
With accused	3.98 (6.43)	0	31	3.68 (5.46)	49.2
Plausibility	13.13 (12.28)	9	71	14.42 (10.44)	96.7
Delayed report	3.41 (4.07)	2	24	4.40 (6.45)	70.8
Continued relationship	2.33 (4.69)	0	36	2.21 (3.86)	44.2
Lack of resistance	1.83 (3.16)	0	20	1.91 (2.952)	47.2
Lack of distress	0.63 (1.32)	0	10	0.72 (1.38)	32.5
General plausibility	4.95 (6.55)	3	43	5.18 (5.14)	85.8
Reliability	11.53 (15.95)	7	119	11.26 (09.10)	90.8
Memory	10.83 (15.42)	6	117	10.48 (8.78)	90.8
Situational factors	0.71 (2.27)	0	16	0.78 (2.36)	17.5
Credibility	9.29 (12.11)	5	78	8.86 (8.12)	86.7
Lying	3.35 (5.22)	1	33	2.98 (3.66)	62.5
Ulterior motive	3.64 (5.16)	2	42	3.88 (4.45)	72.5
Previous bad character	2.30 (5.09)	0	26	1.99 (3.89)	38.3
Multipurpose	13.38 (11.80)	10	68	14.57 (8.98)	98.3
Substance use	0.83 (2.77)	0	21	0.94 (2.97)	16.7
Mental health	0.30 (1.3)	0	10	0.31 (1.20)	7.5
Custody dispute	0.24 (1.11)	0	11	0.24 (0.95)	10.0
Collusion or contamination	1.48 (3.40)	0	21	1.90 (4.27)	31.7
Complainant is wrong	10.19 (10.04)	8	68	10.87 (9.25)	95.8
History of sexual abuse	0.33 (1.59)	0	16	0.31 (1.20)	12.5
Miscellaneous	20.51 (11.58)	16	121	23.75 (11.66)	100.0

	Wald (df = 1)	p_b	B [95% CI]	SE B	OR [95% CI]
Step 1
Inconsistency strategy	3.98	.045*	−0.04 [−0.09, 0.00]	0.02	0.96 [0.92, 0.99]
Plausibility strategy	0.22	.633	−0.12 [−0.08, 0.05]	0.03	0.99 [0.94, 1.04]
Reliability strategy	2.46	.082*	−0.05 [−0.11, 0.01]	0.03	0.96 [0.90, 1.01]
Credibility strategy	5.04	.023**	−0.07 [−0.14, −0.02]	0.03	0.94 [0.89, 0.99]
Multipurpose strategy	1.23	.291	−0.03 [−0.09, 0.03]	0.03	0.97 [0.92, 1.02]
Constant	4.32	.024	3.11 [0.43, 6.77]	1.49	22.32
Step 2
Inconsistency strategy	9.10	.003***	−0.08 [−0.19, −0.04]	2.74	0.92 [0.87, 0.97]
Plausibility strategy	1.56	.253	−0.04 [−0.14, 0.03]	3.14	0.96 [0.90, 1.02]
Reliability strategy	3.94	.047*	−0.08 [−0.20, −0.01]	3.83	0.93 [0.86, 0.99]
Credibility strategy	6.28	.008***	−0.09 [−0.22, −0.03]	3.58	0.91 [0.85, 0.98]
Multipurpose strategy	3.60	.071	−0.07 [−0.19, 0.01]	3.60	0.93 [0.87, 1.00]
Gender	0.00	.990	0.00 [−1.76, 1.92]	0.62	1.00 [0.30, 3.37]
Abuse type	3.58	.037*	1.07 [0.03, 2.60]	0.56	2.90 [0.96, 8.75]
Relationship	0.01	.964	−0.04 [−1.45, 1.49]	0.55	0.96 [0.33, 2.82]
Trial type	10.96	<.001***	2.28 [1.39, 4.64]	0.69	9.76 [2.53, 37.59]
Complainant age at cross-examination	0.35	.500	−0.02 [−0.10, 0.05]	0.03	0.98 [0.93, 1.04]
Offending timespan	2.30	.152	0.21 [−0.05, 0.81]	0.14	1.23 [0.94, 1.60]
Procedural delay	1.14	.084	0.01 [−0.01, 0.03]	0.01	1.01 [0.99, 1.02]
Constant	6.23	.013	4.15 [0.72, 9.43]	1.66	63.42

Footnotes

Acknowledgements

We pay tribute to our co-author, Nina Westera, who passed away on 25th May 2017. We thank Madeline Bearman, Anne Sophie Pichler, Lauren Vogel, and Sarah Zydervelt for their research assistance, and Chelsea Leach for her contribution to a previous version of the manuscript. Ideas and preliminary data in this article were submitted as part of a report to the Royal Commission into Institutional Responses to Child Sexual Abuse: ; Project title: “An Evaluation of How Evidence is Elicited from Complainants of Child Sexual Abuse” by Martine Powell, Nina Westera, Jane Goodman-Delahunty, and Anne-Sophie Pichler.

Data availability

Transcripts from this study are not available due to the sensitive nature of the data. However, Supplemental Materials 1 contains the coding book and contains the de-identified data file.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and publication of this article.

Ethical approval

Ethical approval to conduct the study was granted by the relevant University Human Research Ethics Committees (Deakin University: 2015-010; Charles Sturt University: HREC No. 2015/031).

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by funding from the Royal Commission into Institutional Responses to Child Sexual Abuse. The views and findings expressed in this article are those of the authors and do not necessarily reflect those of the Royal Commission.

ORCID iDs

Natalie Martschuk

Jane Goodman-Delahunty

Martine B Powell

Rachel Zajac

Supplemental material

Supplemental material for this article is available online.

References

Akhtar

Justice

L. V.

Knott

Kibowski

Conway

M. A.

(2018). The "common sense" memory belief system and its implications. The International Journal of Evidence & Proof, 22(3), 289–304. https://doi.org/10.1177/1365712718784045

Alaggia

Collin-Vézina

Lateef

(2019). Facilitators and barriers to child sexual abuse (CSA) disclosures: A research update (2000–2016). Trauma, Violence, & Abuse, 20(2), 260–283. http://doi.org/10.1177/1524838017697312

American Bar Association. (2018). Child trauma. https://www.americanbar.org/groups/public_interest/child_law/resources/attorneys/child-trauma/

Australian Bureau of Statistics. (2021). Recorded crime—Victims. https://www.abs.gov.au/statistics/people/crime-and-justice/recorded-crime-victims/2021

Benjamini

Hochberg

(1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society, Series B (Methodological), 57(1), 289–300. https://doi.org/10.1111/j.2517-6161.1995.tb02031.x

Browne v. Dunn. (1893). 6 R 67. H.L.

Cashmore

Bussey

(1996). Judicial perceptions of child witness competence. Law & Human Behavior, 20(3), 313–334. https://doi.org/10.1007/BF01499026

Cashmore

Taylor

Parkinson

(2017). The characteristics of reports to the police of child sexual abuse and the likelihood of cases proceeding to prosecution after delays in reporting. Child Abuse & Neglect, 74, 49–61. https://doi.org/10.1016/j.chiabu.2017.07.006

Cashmore

Taylor

Parkinson

(2020). Fourteen-year trends in the criminal justice response to child sexual abuse reports in New South Wales. Child Maltreatment, 25(1), 85–95. https://doi.org/10.1177/1077559519853042

10.

Coburn

P. I.

Connolly

D. A.

Woiwod

D. M.

George Alder

Bernstein

D. M.

(2022). Cross-examination may be more detrimental to repeated-event children than single-event children. Memory (Hove, England), 30(6), 715–724. https://doi.org/10.1080/09658211.2021.1909622

11.

Connolly

D. A.

Price

H. L.

Gordon

H. M.

(2009). Judging the credibility of historic child sexual abuse complainants. Psychology, Public Policy, & Law, 15(2), 102–123. https://doi.org/10.1037/a0015339

12.

Cooper

Mattison

(2017). Intermediaries, vulnerable people, and the quality of evidence: An international comparison of three versions of the English intermediary model. International Journal of Evidence and Proof, 21(4), 351–370. https://doi.org/10.1177/1365712717725534

13.

Cossins

Goodman-Delahunty

(2017). The application of the uniform evidence law to delay in child sexual assault trials. In Roberts

Gans

(Eds.), Critical perspectives on the uniform evidence law (pp. 104–124). Federation Press.

14.

Criminal Procedure Act 2009 (VIC).

15.

Criminal Procedure Amendment (Child Sexual Offence Evidence) Act 2023 No. 32 (NSW).

16.

Dallaston

Mathews

(2022). Reforming Australian criminal laws against persistent child sexual abuse. Sydney Law Review, 44(1), 77–109.

17.

Davies

Henderson

Seymour

F. W.

(1997). In the interests of justice? The cross-examination of child complainants of sexual abuse criminal proceedings. Psychiatry, Psychology & Law, 4(2), 217–229. https://doi.org/10.1080/13218719709524912

18.

Denne

St. George

Stolzenberg

S. N.

(2023a). Developmental considerations in how defense attorneys employ child sexual abuse and rape myths when questioning alleged victims of child sexual abuse. Journal of Interpersonal Violence, 38(23–24), 11914–11934. https://doi.org/10.1177/08862605231189512

19.

Denne

St. George

Stolzenberg

S. N.

(2023b). Myths and misunderstandings about child sexual abuse in criminal investigations. Journal of Interpersonal Violence, 38(1–2), 1893–1919. https://doi.org/10.1177/08862605221093679

20.

Easton

S. D.

(2013). Disclosure of child sexual abuse among adult male survivors. Clinical Social Work Journal, 41(4), 344–355. http://doi.org/10.1007/s10615-012-0420-3

21.

Ellison

Munro

V. E.

(2017). Taking trauma seriously: Critical reflections on the criminal justice process. International Journal of Evidence and Proof, 21(3), 183–208. https://doi.org/10.1177/1365712716655168

22.

Enriques

Brown

L. F.

Goodman

Metcalf

Quas

J. A.

Redlich

A. D.

Quas

J. A.

(2023). The developmental science of children in criminal and dependency court. In The Oxford handbook of developmental psychology and law (pp. 85–108). Oxford University Press.

23.

Evidence Act 1995 (Cth).

24.

Field

Katz

(2023). The experiences and perceptions of sexually abused children as participants in the legal process: Key conclusions from a scoping literature review. Trauma Violence & Abuse, 24(4), 2758–2771. https://doi.org/10.1177/15248380221111463

25.

Goodman-Delahunty

Cashmore

Dilevski

(2024). Prosecutorial decision-making in cases of child sexual abuse: Lessons from Australia. In Miller

M. K.

Yelderman

L. A.

Huss

M. T.

Cantone

J. A.

(Eds.), The Cambridge handbook of psychology and legal decision-making (pp. 240–258). Cambridge University Press.

26.

Goodman-Delahunty

Cowdery

(2023). Rights of survivors of child sexual abuse in criminal proceedings in Australia. In Scicluna

Wijlens

(Eds.), Rights of alleged victims in penal proceedings: Provisions in Canon Law and the criminal law of different legal systems (pp. 83–118). Nomos Publishing.

27.

Goodman-Delahunty

Martschuk

(2020). Mock jury and juror responses to uncharged acts of sexual misconduct: Advances in the assessment of unfair prejudice. Zeitschrift für Psychologie, 228(3), 199–209. https://doi.org/10.1027/2151-2604/a000410

28.

Goodman-Delahunty

Martschuk

Cossins

(2017a). What Australian jurors know and do not know about evidence in child sexual abuse cases. Criminal Law Journal, 41(2), 86–103.

29.

Goodman-Delahunty

Martschuk

Cossins

(2017b). Validation of the child sexual abuse knowledge questionnaire. Psychology, Crime & Law, 23(4), 391–412. https://doi.org/10.1080/1068316X.2016.1258469

30.

Grob

Ehlers

Schwabe

(2023). Strong but fragmented memory of a stressful episode. Eneuro, 10(9), 0178-23.2023, 1-6. https://doi.org/10.1523/ENEURO.0178-23.2023

31.

Gudjonsson

G. H.

(1992). The psychology of interrogations, confessions and testimony. John Wiley & Sons.

32.

Hampel

Ainslie-Wallace

Brimer

Kune

(2016). Advocacy manual: The complete guide to persuasive advocacy (2nd ed.). Australian Advocacy Institute.

33.

Hanna

Davies

Crothers

Henderson

(2012). Questioning child witnesses in New Zealand's criminal justice system: Is cross-examination fair? Psychiatry, Psychology & Law, 19(4), 530–546. https://doi.org/10.1080/13218719.2011.615813

34.

Hodgson

Cashmore

Cowdery

Goodman-Delahunty

Martschuk

Parkinson

Powell

M. B.

Shackel

(2020). The decision to prosecute: A comparative analysis of Australian prosecutorial guidelines. Criminal Law Journal, 44(3), 155–172.

35.

Horan

Goodman-Delahunty

(2020). Expert evidence to counteract jury misconceptions about consent in sexual assault cases: Failures and lessons learned. University of New South Wales Law Journal, 43(2), 707–737. https://doi.org/10.53637/FDBI9255

36.

Jay

Evans

Frank

Sharpling

(2022). The report of the Independent Inquiry into Child Sexual Abuse. Independent Inquiry into Child Sexual Abuse. www.gov.uk/official-documents

37.

Kezelman

Stavropoulos

(2016). Trauma and the law: Applying trauma-informed practice to legal and judicial contexts. Blue Knot Foundation.

38.

Leipold

A. D.

Abbasi

A. H.

(2006). The impact of joinder and severance on federal criminal cases: An empirical study. Vanderbilt Law Review, 59(2), 349–402.

39.

Libke v. The Queen [2007] HCA 30.

40.

Martschuk

Cashmore

Hoff

Parkinson

Goodman-Delahunty

Shackel

Cowdery

Powell

M. B.

(2024). The importance of consistency in complainants’ evidence about child sexual abuse. Child Abuse & Neglect, 158, 107095. https://doi.org/10.1016/j.chiabu.2024.107095

41.

Martschuk

Powell

M. B.

Blewer

Goodman-Delahunty

(2022). Legal decision making about (child) sexual assault complaints: The importance of the information-gathering process. Current Issues in Criminal Justice, 34(1), 58–76. https://doi.org/10.1080/10345329.2021.1978213

42.

Mathews

(2019). New international frontiers in child sexual abuse: Theory, problems and progress. Springer.

43.

Mathews

Pacella

Scott

J. G.

Finkelhor

Meinck

Higgins

D. J.

Erskine

H. E.

Thomas

H. J.

Lawrence

D. M.

Haslam

D. M.

Malacova

Dunne

M. P.

(2023). The prevalence of child maltreatment in Australia: Findings from a national survey. Medical Journal of Australia, 218(Suppl 6), S13–S18. https://doi.org/10.5694/mja2.51873

44.

McDonald

J. H.

(2014). Handbook of biological statistics (3rd ed.). Sparky House Publishing.

45.

Mills

Sharman

S. J.

(2017). Characteristics of successful and unsuccessful appeals against conviction for child sexual abuse. Psychiatry, Psychology & Law, 24(5), 655–699. https://doi.org/10.1080/13218719.2017.1291271

46.

Pemberton

Mulder

(2023). Bringing injustice back in: Secondary victimization as epistemic injustice. Criminology & Criminal Justice. https://doi.org/10.1177/17488958231181345

47.

Pichler

A.-S.

Powell

M. B.

Sharman

S. J.

Zydervelt

Westera

Goodman-Delahunty

(2021). Inconsistencies in complainant’s accounts of child sexual abuse arising in their cross-examination. Psychology, Crime & Law, 27(4), 341–356. https://doi.org/10.1080/1068316X.2020.1805743

48.

Powell

M. B.

Goodman-Delahunty

Deck

Bearman

Westera

N. J.

(2022). An evaluation of the question types used by criminal justice professionals with complainants in child sexual assault trials. Journal of Criminology, 55(1), 106–124. https://doi.org/10.1177/26338076211068182

49.

Powell

M. B.

Roberts

Guadagno

(2007). Particularisation of child abuse offences: Common problems with questioning child witnesses. Current Issues in Criminal Justice, 19(1), 64–74. https://doi.org/10.1080/10345329.2007.12036415

50.

Powell

M. B.

Thomson

D. M.

(1996). Children's memory of an occurrence of a repeated event: Effects of age, repetition, and retention interval across three question types. Child Development, 67(5), 1988–2004. https://doi.org/10.2307/1131605

51.

Quas

J. A.

Goodman

G. S.

(2012). Consequences of criminal court involvement for child victims. Psychology, Public Policy, & Law, 18(3), 392–414. https://doi.org/10.1037/a0026146

52.

Quilter

McNamara

(2023). Experience of complainants of adult sexual offences in the District Court of NSW: A trial transcript analysis (Crime and Justice Bulletin No. 259). NSW Bureau of Crime Statistics and Research.

53.

Royal Commission into Institutional Responses to Child Sexual Abuse. (2017). Criminal justice report: Parts VII to X and appendices. https://www.childabuseroyalcommission.gov.au/sites/default/files/file-list/final_report_-_criminal_justice_report_-_parts_vii_to_x_and_appendices.pdf

54.

Saks

M. J.

Spellman

B. A.

(2016). The psychological foundations of evidence law. New York University Press.

55.

Shackel

(2009). Understanding children's medium for disclosing sexual abuse: A tool for overcoming potential misconceptions in the courtroom. Psychiatry, Psychology & Law, 16(3), 379–393. https://doi.org/10.1080/13218710903040280

56.

St. George

Denne

Stolzenberg

S. N.

(2022a). Blaming children: How rape myths manifest in defense attorneys’ questions to children testifying about child sexual abuse. Journal of Interpersonal Violence, 37(17–18), NP16623–NP16646. https://doi.org/10.1177/08862605211023485

57.

St. George

Denne

Stolzenberg

S. N.

(2022b). “This incident happened when there were 10 people in the house?” Exploring a framework to categorize defense attorneys’ plausibility questioning in CSA trials. Psychology, Crime & Law, 30(6), 556–582. https://doi.org/10.1080/1068316X.2022.2104277

58.

Stolzenberg

S. N.

Lyon

T. D.

(2014). How attorneys question children about the dynamics of sexual abuse and disclosure in criminal trials. Psychology, Public Policy, & Law, 20(1), 19–30. https://doi.org/10.1037/a0035000

59.

Tabachnick

Fidell

(2018). Using multivariate statistics (7th ed.). Pearson.

60.

Tanford

Penrod

(1984). Social inference processes in juror judgments of multiple-offence trials. Journal of Personality and Social Psychology, 47(4), 749–765. https://doi.org/10.1037/0022-3514.47.4.749

61.

Tanford

Penrod

(1985). Decision making in joint criminal trials: The influence of charge similarity, evidence similarity, and limiting instructions. Law & Human Behavior, 9(4), 319–337. https://doi.org/10.1007/BF01044474

62.

Temkin

(2000). Prosecuting and defending rape: Perspectives from the bar. Journal of Law & Society, 27(2), 219–248. https://doi.org/10.1111/1467-6478.00152

63.

Wallis

C. R. D.

Woodworth

M. D.

(2020). Child sexual abuse: An examination of individual and abuse characteristics that may impact delays of disclosure. Child Abuse & Neglect, 107, 104604. https://doi.org/10.1016/j.chiabu.2020.104604

64.

Yehia

(2010). Cross-examination of children. The Public Defender.

65.

Zajac

O’Neill

Hayne

(2012). Disorder in the courtroom? Child witnesses under cross-examination. Developmental Review, 32(3), 181–204. https://doi.org/10.1016/j.dr.2012.06.006

66.

Zajac

Westera

Kaladelfos

(2017). A historical comparison of Australian lawyers’ strategies for cross-examining child sexual abuse complainants. Child Abuse & Neglect, 72, 236–246. https://doi.org/10.1016/j.chiabu.2017.07.010

67.

Zydervelt

Zajac

Kaladelfos

Westera

(2017). Lawyers’ strategies for cross-examining rape complainants: Have we moved beyond the 1950s? British Journal of Criminology, 57(3), 551–569. https://doi.org/10.1093/bjc/azw023

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB

0.07 MB