Skip to main content

Evidence explained

Who experiences romance scams?

Women and middle-aged adults appear frequently in studied romance-scam reports, but other samples contain many men and younger adults. The evidence does not establish a single victim profile: which scam researchers count, who reports it, and whether they measure contact or financial loss all change the picture (Cross, 2023; Cross et al., 2024; DeLiema & Witt, 2023; Kassem & Carter, 2023).

Explanation updated . This can reflect an editorial correction without a new evidence snapshot.

AI-assisted explanation · Automated editorial and source checks. Preparation and review details

Four different ways to describe victims

Interpretation. A study can describe who appears in reports, who reports losing money, or where reported victims live. These are different questions. Reading the findings together requires keeping each study’s population and outcome attached to its result (Cross, 2023; DeLiema & Witt, 2023; Sinclair et al., 2023).

Four different ways to describe victims
FocusWhat the studies showWhat remains uncertain
Who appears in reports Analysis of complaint records

Among 1,015 complainants, 60% were women and 45% were between ages 45 and 64 (Cross, 2023).

Interpretation. These percentages describe this complaint sample, not demographic differences in population risk (Cross, 2023).

Loss versus an attempt Statistical analysis of complaints

Among romance-scam reporters, adults aged 80 or older had 1.76 times the odds of reporting a loss compared with reporters in their thirties. Here, odds compare the probability of reporting loss with the probability of reporting only an attempt. This does not mean the probability of loss was 76% higher (DeLiema & Witt, 2023).

Interpretation. The comparison concerns people who reported a scam; it does not estimate risk among everyone in those age groups (DeLiema & Witt, 2023).

Sextortion reports Analysis of 258 reports

Among 258 reports involving sextortion—threats to share intimate images—in romance fraud, 76% of complainants were men and 54% were ages 18–34 (Cross et al., 2024).

Interpretation. Only 14% indicated financial loss; the demographic percentages describe all complainants, not specifically those who reported losing money (Cross et al., 2024).

Geographic concentration Three-year analysis of victim records

Some residential postcode areas remained concentrations of reported victims across three years. These are places where victims lived, not locations of offenders or their operations (Sinclair et al., 2023).

The researchers counted reported victims rather than calculating victimization rates relative to each area’s population (Sinclair et al., 2023).

A common characteristic is not necessarily a risk factor

Cross’s 2023 study analyzed 1,015 reports. Of the complainants, 60% said they were in Australia and 31% were overseas, so the demographic percentages do not describe a sample confined to Australian residents (Cross, 2023).

Interpretation. Estimating demographic differences in risk would require comparison data: how many people in each group encountered a scam, how many lost money, and how many never reported. Without those comparisons, using the largest group in a complaint sample to set outreach priorities could overlook people whose experiences are less visible in the records (Cross, 2023; Sinclair et al., 2023).

DeLiema and Witt’s analysis illustrates why a group’s size in the records can be misleading. Adults aged 80 or older made up only 1.4% of romance-scam reporters, yet had higher odds of reporting financial loss than reporters in their thirties. Their small share of reports did not mean that those who reported were less likely to describe losing money (DeLiema & Witt, 2023).

Changing the scam category changes the apparent profile

Cross, Holt, and O’Malley analyzed 258 sextortion-related reports in the context of romance fraud, not an unrestricted sample of sextortion experiences. Only 59% of complainants said they were based in Australia, so the demographic findings should not be read as describing Australian residents alone (Cross et al., 2024).

An Indian study examined 35 cases, with gender recorded for 30 victims: 23 men and seven women. Victims had an average age of 35.88 years. Cases clustered in Delhi, Mumbai and Bengaluru. Media coverage may affect which cases appear. The authors recommend police records, surveys or interviews to check whether this pattern reflects a wider population (Verma et al., 2026).

Education and personality do not provide a dependable checklist

Interpretation. A survey of 609 women who used Facebook across Malaysia, aged 19–50, compared 43 who had experienced a love scam with 566 who had not. Statistical tests did not detect a difference in their knowledge or protective-practice scores. This does not prove the groups had equal scores or that knowledge prevents scams; the comparison concerns only the women surveyed (Abidin et al., 2018).

Drawing on victims’ accounts, a qualitative study discussed recent divorce, infidelity, and prolonged social isolation. Such accounts can help explain how scams unfolded, but do not establish whether people in those circumstances are more likely to be deceived than others. The authors cautioned against assuming that their findings describe victims beyond the people studied (Wang & Topalli, 2022).

Interpretation. A 2025 review included 22 quantitative studies of cyber-scam victim characteristics; three focused on romance scams. Findings for impulsivity and loneliness varied across the reviewed studies. This broader collection therefore does not give a consistent romance-scam profile or establish a dependable way to identify future victims (Whitty, 2025).

A map shows concentrations of reports, not individual vulnerability

Sinclair, Bland, and Savage analyzed 15,575 victim records across three years in England and Wales after removing duplicate entries, invalid addresses, and records outside the study area. They linked postcodes to Acorn classifications, which describe areas’ demographic characteristics using multiple data sources and estimates, and analyzed broader postcode areas rather than individual addresses (Sinclair et al., 2023).

Some concentrations persisted: of 439 first-year hot spots, 162 remained hot spots in both subsequent years. Those recurring areas contained 17% of all reported victims. Hot spots were more frequently in less affluent communities, but the area classifications did not necessarily describe the individual victim (Sinclair et al., 2023).

Interpretation. Without population-adjusted rates, the map cannot distinguish an area with more residents from one with greater individual risk. Recognition also matters: people who have not realized they are victims fall outside the study’s scope. Geographic targeting based on these records could therefore miss places where victimization remains unrecognized (Sinclair et al., 2023).

Using profiles without excluding people

Interpretation. Outreach and support teams can use the findings to broaden who appears in their materials, separate attempts from financial losses in service records, and clarify the purpose of geographic targeting. These are planning steps, not validated tests of who needs help (Button & Carter, 2024; Kassem & Carter, 2023; Whitty, 2025).

  1. Include people outside the familiar stereotype

    Include men, younger adults and people in same-sex relationships in outreach materials. Men formed most of the sextortion complaint sample and most Indian cases with recorded gender. These patterns may partly reflect who reports a scam or whose case attracts media coverage. They do not establish national differences in risk (Button & Carter, 2024; Cross et al., 2024; Verma et al., 2026).

  2. Separate the measures in service data

    Record scam category, attempted fraud, and reported financial loss separately, using a consistent time period and wording. This would make comparisons easier to interpret and avoid treating a change in the mix of complaints as a change in who is most susceptible (Cross et al., 2024; DeLiema & Witt, 2023; Whitty, 2025).

  3. Check what a geographic target actually represents

    Before using hot spots to allocate outreach, ask whether the objective is to reach many reported victims or identify higher population risk. For the latter, request population-adjusted rates and check individual demographic information rather than assuming neighborhood classifications describe every resident (Sinclair et al., 2023).

Who would this targeting rule leave out?

Before adopting an age, gender, or location rule, ask whether it reflects population risk or merely the composition of a particular reporting sample. Then identify who would miss outreach or support because they fall outside it, especially groups that existing research has studied less often (Button & Carter, 2024; Kassem & Carter, 2023; Sinclair et al., 2023).

About the evidence

This explanation draws on selected sources from RSRC’s full synthesis. The full research record, its source list and version history remain available below.

Explore this topic’s methods and groups studied · Follow topic updates

Scope, preparation and limitations

Codex edited and reviewed this explanation from synthesis version 82. Changed passages were checked against preserved source excerpts; passing source checks were reused for unchanged passages. The editorial judgments are recorded as Codex review, not a new API grading run. They assess communication, not reader-test results or scientific certainty. No human or expert content review of this explanation is recorded.

The full synthesis preserves 25 publication records and 90 evidence statements. These counts do not establish the number of independent studies or the strength of a finding. Practical implications are editorial interpretations, with limitations explained alongside the evidence.

RSRC methodology
Sources cited in this explanation (10)

Links open the publication record or original work. Journal access may vary.

  1. Abidin, N. Z., Kamaluddin, M. R., Shaari, A. H., Din, N., & Ramasamy, S. (2018). Pengetahuan dan Amalan Perlindungan Pengguna Facebook Wanita Terhadap Penipuan Cinta di Malaysia [Knowledge and Protective Practice Towards Love Scam Among Female Facebook Users in Malaysia]. Jurnal Komunikasi: Malaysian Journal of Communication, 34(4), 113–133. https://doi.org/10.17576/jkmjc-2018-3404-07
  2. Button, M., & Carter, E. (2024). Relationship fraud: Romance, friendship and family frauds. Journal of Economic Criminology, 4, 100069. https://doi.org/10.1016/j.jeconc.2024.100069
  3. Cross, C. (2023). “I knew it was a scam”: Understanding the triggers for recognizing romance fraud. Criminology & Public Policy, 22(4), 613–637. https://doi.org/10.1111/1745-9133.12645
  4. Cross, C., Holt, K., & O'Malley, R. L. (2024). “If U Don't Pay they will Share the Pics”: Exploring Sextortion in the Context of Romance Fraud. Scams, Cons, Frauds, and Deceptions, 10–31. https://doi.org/10.4324/9781003474982-2
  5. DeLiema, M., & Witt, P. (2023). Profiling consumers who reported mass marketing scams: demographic characteristics and emotional sentiments associated with victimization. Security Journal, 37(3), 921–964. https://doi.org/10.1057/s41284-023-00401-5
  6. Kassem, R., & Carter, E. (2023). Mapping romance fraud research – a systematic review. Journal of Financial Crime, 31(4), 974–992. https://doi.org/10.1108/jfc-06-2023-0160
  7. Sinclair, R., Bland, M., & Savage, B. (2023). Dating hot spot to fraud hot spot: Targeting the social characteristics of romance fraud victims in England and Wales. Criminology & Public Policy, 22(4), 591–611. https://doi.org/10.1111/1745-9133.12629
  8. Verma, S., Gujarathi, P., & Lazarus, S. (2026). Romance Fraud in India: A Qualitative Study of Online–Offline Transitions in the Digital Dating Ecosystem. Journal of Economic Criminology. https://doi.org/10.1016/j.jeconc.2026.100248
  9. Wang, F., & Topalli, V. (2022). Understanding Romance Scammers Through the Lens of Their Victims: Qualitative Modeling of Risk and Protective Factors in the Online Context. American Journal of Criminal Justice, 49(1), 145–181. https://doi.org/10.1007/s12103-022-09706-4
  10. Whitty, M. T. (2025). A systematic literature review of profiling victims of cyber scams: setting up a framework for future research. Cogent Social Sciences, 11(1). https://doi.org/10.1080/23311886.2025.2563781
Full synthesis and revision history

Download the full synthesis PDF

The reviewed evidence describes a heterogeneous, non‑representative literature that supports context‑sensitive descriptive claims about people who appear in complaint, help‑seeker, platform, and purposive‑study samples but does not justify a single, general‑population victim profile. Across datasets there are recurring patterns—substantial representation of middle‑aged and older adults in many traditional romance‑scam complaint and support‑site samples, frequent female predominance in help‑seeking and complaint datasets in several jurisdictions, and repeated qualitative reports of psychosocial themes (romantic idealization, loneliness or recent relationship disruption, willingness to self‑disclose)—but these patterns vary systematically by scam subtype, contact channel, geography, and the sample frame used. Evidence quality and sampling heterogeneity (administrative complaint samples, self‑selected surveys, purposive qualitative samples, platform‑detected accounts) limit generalizability and preclude strong causal claims. The literature therefore supports bounded, context‑dependent statements about who appears as a reported or studied victim and identifies clear research needs: incident‑verified, representative or probability‑based sampling, subtype‑sensitive analysis, cross‑referencing of complaint records with population data, and more rigorous study of psychological and demographic correlates (Bilz et al., 2023; Cross, 2023a, 2023b; DeLiema & Witt, 2023; Gujarathi et al., 2026; Kassem & Carter, 2023; Sinclair et al., 2023; F. Wang, 2026; F. Wang & Topalli, 2022; Whitty, 2013b, 2017).

Scope and common sampling frames

Most empirical work on romance‑scam victims in the reviewed corpus draws on non‑probability or administratively recorded samples (complaint portals, help‑site visitors, platform‑detected accounts, and purposive qualitative interview samples). Authors repeatedly note that these frames principally describe reporters or help‑seekers rather than a population risk distribution and therefore constrain generalisability (Bilz et al., 2023; Cross, 2023a; Kassem & Carter, 2023; Sinclair et al., 2023).

Different sampling frames yield complementary but distinct information. Large administrative complaint datasets provide broad counts and incident-oriented descriptive detail (for example, coded contact channel, reported loss, and geographic location) while smaller qualitative and purposive interview studies supply richer narrative detail about grooming, trust, and psychosocial experience. These sources therefore illuminate different aspects of victim appearance and experience but do so in sample‑conditional ways rather than offering population‑representative prevalence or causal inferences (Bilz et al., 2023; Cross, 2023a; Sinclair et al., 2023; F. Wang & Topalli, 2022).

Age patterns and subtype heterogeneity

A recurrent descriptive motif in many complaint and support‑site samples is greater representation of middle‑aged and older adults (commonly mid‑adult to older‑adult bands) among reported romance‑scam victims and among in‑depth interview participants in traditional romance‑fraud studies; several administrative compilations and qualitative samples show clustering in these age ranges (Chuang, 2021; Cross, 2023a, 2023b; Whitty, 2013b, 2017).

The evidence indicates important subtype differences rather than a single age profile. Australian administrative data and conceptual reviews note that traditional romance fraud historically concentrates among older adults while ‘romance baiting’ losses in Australia in 2020 were highest among people aged 25–34; scoping reviews characterize investment‑related relationship scams as demographically more heterogeneous and often shaped by financial aspirations and cryptocurrency interest rather than a pronounced older‑adult skew (Cross, 2023a; Gujarathi et al., 2026; F. Wang, 2026).

Sex and gender composition: patterns and important caveats

Many complaint‑based and help‑seeking datasets in the reviewed literature show majority female composition among reported or studied romance‑scam victims, with some single‑dataset records reporting extreme female predominance; qualitative interview work and platform studies often reflect similar female overrepresentation in those samples (Cross, 2023b; Grace Carvalho Fernandes et al., 2023; F. Wang & Zhou, 2022; Whitty, 2017).

At the same time, clear counterpatterns appear by subtype and reporting frame. Analyses of sextortion‑tagged romance‑fraud reports show a strong male predominance among those complainants and concentration in younger adult age bands; qualitative analyses of sextortion also identify gendered narratives in how sextortion unfolds. A 35-case Indian romance-fraud series also showed male predominance: among the 30 cases with reported gender, 23 victims were male and seven were female. Cases were concentrated in Delhi, Mumbai, and Bengaluru, and the reported mean victim age was 35.88 years. These observations do not establish national gender or age risk; the authors leave unresolved whether the male-victim pattern reflects broader victimization or characteristics of media reporting. Systematic reviews and gap analyses additionally emphasize that research on male and same-sex victims is sparse, so observed female predominance in many samples may partly reflect study and reporting biases rather than universal vulnerability patterns (Button & Carter, 2024; Cross et al., 2024; Kassem & Carter, 2023; Verma et al., 2026).

Socioeconomic, occupational, and psychosocial characteristics

Several studies and syntheses report that many reported victims are educated and financially independent; some regression analyses in the reviewed corpus found a positive association between higher education and identified victim status in analyzed samples. Narrative syntheses likewise describe professional or middle‑income adults appearing in complaint and interview datasets (Feng, 2025; C. Wang, 2022; Whitty, 2017).

Conversely, other records document victims across the socioeconomic spectrum (including low‑income, migrant, and marginalized groups) and highlight that platform user demography, complaint propensity, and help‑seeking behavior affect the apparent education or income patterns in available samples, preventing strong, general claims about socioeconomic risk markers (Franceschini et al., 2024; Somiah, 2023).

Qualitative syntheses and multiple interview‑based studies repeatedly identify psychosocial themes in victims’ narratives that plausibly explain how scams unfold in individual cases: investment of trust, romantic idealization, social isolation or recent relationship disruption, and willingness to self‑disclose personal information (F. Wang & Topalli, 2022; Whitty, 2013a, 2013b).

Contact channel, subtype, and geographic moderators

Multiple analyses show that observed victim composition varies by contact channel and scam subtype: sextortion reports and social‑media‑contact cases tend to differ demographically from classic dating‑portal romance scams, and pig‑butchering/cryptocurrency variants show distinctive patterns that often align with younger, crypto‑interested complainants in some datasets (Cross, 2023a; Cross et al., 2024; Gujarathi et al., 2026; F. Wang, 2026).

Large administrative analyses also document geographic concentration: in one England and Wales series, half of reported victims were concentrated in a minority of outward‑postcode areas (hot spots), though those analyses used absolute counts rather than population‑adjusted rates and authors cautioned against interpreting hot spots as direct measures of relative population risk without cross‑referencing demographic denominators (Sinclair et al., 2023).

Methodological patterns, limitations, and evidence gaps

Across systematic reviews, scoping studies, and primary research the literature is characterized by heterogeneous definitions, inconsistent measurement, and sampling reliance on complaints, help‑site visitors, platform‑detected accounts, and purposive interview samples; these recurring methodological features limit comparability, prevent meta‑analysis in many areas, and constrain population generalizability and causal inference (Bilz et al., 2023; Kassem & Carter, 2023; Koning et al., 2023; Whitty, 2025).

Reviewed syntheses and mapping studies identify specific evidence gaps: the overall evidence on romance‑fraud victim profiles is limited and mixed; studies on victim and offender profiles are sparse and sometimes conflicting (notably regarding age); and there is notably little research on male and same‑sex victims. Several authors and reviews therefore call for more diverse, incident‑verified, and comparative research on victim characteristics (Bilz et al., 2023; Button & Carter, 2024; Kassem & Carter, 2023).

Some administrative and review‑level items explicitly recommend methodological improvements that would clarify profile‑related questions: for example, cross‑referencing victim records with demographic segmentation data to refine targeting and using population‑adjusted rates rather than absolute counts when assessing geographic concentration (Sinclair et al., 2023).

Interpretation, limits on inference, and research priorities

Interpretation of observed victim patterns should be framed as sample-conditional: complaint and help-seeking datasets tell us about who reports or seeks help, not about unconditional population susceptibility. Where multiple records converge (for example, female predominance in several complaint samples or older-adult clustering in many traditional romance-fraud reports), the synthesis treats those as descriptive, contextual findings that require subtype- and frame-sensitive qualification rather than general risk determinants. Knowledge and protective practices also require cautious interpretation. In a survey of 609 female Facebook users aged 19–50 from across Malaysia, independent-samples t-tests found no significant differences in knowledge or protective-practice scores between the 43 respondents with personal love-scam victimization histories and the 566 without such histories. This women-only sample cannot establish female predominance among victims, and the nonsignificant comparisons do not establish equivalent scores, predictive performance, or a protective effect of knowledge. The study used Pearson correlation to examine knowledge–practice relationships, not a causal or prospective test of prevention (Abidin et al., 2018; Bilz et al., 2023; Cross, 2023b; DeLiema & Witt, 2023; Kassem & Carter, 2023; F. Wang & Zhou, 2022; Whitty, 2017).

The reviewed corpus supports several priority directions repeatedly recommended by authors: clearer differentiation of scam subtypes in analysis (for example, distinguishing classic dating-portal romance fraud from sextortion and pig-butchering/crypto investment scams), improved linkage of complaint and platform records to population denominators, and expanded study of understudied victim groups (including men and sexual-minority victims) to reduce current coverage bias. For the Indian male-victim pattern specifically, the authors recommend police records, victimization surveys, or interviews to distinguish broader victimization trends from media-reporting characteristics; these are proposed checks, not completed validation. Outreach should therefore avoid treating that case series as a national risk profile or excluding groups that are less visible in a particular reporting frame (Cross et al., 2024; Gujarathi et al., 2026; Kassem & Carter, 2023; Sinclair et al., 2023; Verma et al., 2026).

References

This AI-assisted synthesis is based on reviewed evidence records. It may contain errors or omissions. Follow the publication and DOI links, consult the original works, and make your own judgment about the evidence.

Revision history

  1. Version 82

    Current public version

    Incremental evidence update. A small update is warranted. The Indian case series adds a male-predominant pattern outside the existing sextortion example, but its metropolitan concentration, missing gender information, and unresolved media-selection effects must remain explicit. The Malaysian survey adds a bounded finding of no significant knowledge or protective-practice differences by personal victimization history, not evidence of equivalence or protection. The central conclusion that the lite...

    • Publications: 25 (+2)
    • Evidence statements: 90 (+5)
    • Cited sources: 25 (+2)

    Sources added: Knowledge and Protective Practice Towards Love Scam Among Female Facebook Users in Malaysia; Romance Fraud in India: A Qualitative Study of Online–Offline Transitions in the Digital Dating Ecosystem.

    The approved evidence set changed while the cited-source list remained the same.

  2. Version 81

    Automated evidence-grounded revision attempt 2. Initial version.

  3. Version 80

    Automated evidence-grounded revision attempt 1. This synthesis is an audited replacement of the prior draft. Preserved text passages identified by the audit were reproduced verbatim with exactly the listed evidence IDs. Other assertions were narrowed or removed where the audit found insufficient direct support. The synthesis emphasizes sample‑conditional interpretation, subtype heterogeneity, and the methodological limitations and gaps explicitly cited in the reviewed evidence records.

    • Publications: 23 (-5)
    • Evidence statements: 85 (-21)
    • Cited sources: 23 (-5)

    Sources removed: A Sinister Fattening: Dissecting the Tales of Pig Butchering and Other Cryptocurrency Scams; Grooming in online dating romance frauds and scams: A scoping review; Is There a Scam for Everyone? Psychologically Profiling Cyberscam Victims; Love, Lies, and Larceny: One Hundred Convicted Case Files of Cybercriminals with Eighty Involving Online Romance Fraud; and 1 more.

View full revision history

Looking for reporting or support information? Visit the RSRC resource list. RSRC does not provide direct support.

Request a correction