Who gets to feel safe after dark?
Recorded. Experienced. Feared.
In Germany, 11.1% of women report feeling rather or very safe alone in parks after dark, compared with 34.8% of men. A national crime count cannot explain that difference, or the choices people say they avoid.
German national estimates / SKiD 2024
Krumme Lanke station, Berlin, July 2022. Lukas Beck / CC BY-SA 4.0. Cropped, not retouched.
Context photograph, not a measured crime location. The survey estimates are national and do not establish the safety of this station.
What does a safer country let people do?
A country can discuss crime as one number. Its residents cannot experience safety that way. Being at home, waiting for a train and crossing a park alone ask different things of a person, particularly after dark. In Germany's national victimization survey, men's and women's answers separate sharply when the setting changes.
The useful investigation begins with that separation. Is it associated with past experience? How much of an experience becomes visible to the police? And what do people say they avoid? The evidence answers parts of each question. Together, the answers show why reducing recorded crime and widening people's freedom to use public space need to be measured separately.
The same setting does not feel equally accessible.
Compare men's and women's answers about being alone after dark. The gap is not a single national constant: it changes with the setting.
Changing the setting changes the size of the gap.
Share feeling rather or very safe alone after dark. 2024 survey-time perceptions. All settings share a 0-100% scale. Gap = men minus women, in percentage points.
Residential neighborhood: men 82.7%, women 66.1%. The gap is the difference between published point estimates, not an estimated causal effect.Men: No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure. Women: No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure.
People aged 16 or older in private households in Germany; source-defined valid responses. Source sex categories are male and female. Overall sample n is not substituted for item-specific n. A named setting is not the same physical location for every respondent.
All settings: exact estimates and uncertainty
| Setting | Men | Women |
|---|---|---|
| Residential neighborhood2024 survey-time perceptions | 82.7%No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure. | 66.1%No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure. |
| Streets, paths and squares2024 survey-time perceptions | 54.1%No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure. | 26.9%No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure. |
| Public transport2024 survey-time perceptions | 57.2%No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure. | 33.2%No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure. |
| Railway stations2024 survey-time perceptions | 38.5%No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure. | 16.0%No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure. |
| Parks2024 survey-time perceptions | 34.8%No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure. | 11.1%No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure. |
Source: Bundeskriminalamt: SKiD 2024: Bundesweite Kernbefunde, V 1.0.
Settings are selected, not interpolated. Missing confidence bounds are not estimated from rounded percentages. No causal explanation for the sex difference is identified.
The source publishes male/female comparisons. Other gender responses enter population totals but do not support a separate published group because of insufficient sample size. Report: printed p. 62 / PDF p. 65.
In the residential neighborhood, 66.1% of women and 82.7% of men report feeling rather or very safe. In parks after dark, the estimates fall to 11.1% and 34.8%. The lowest safety level and the widest difference are not the same thing: streets, paths and squares have the largest displayed gap, at 27.2 percentage points.
These are national estimates for named types of place, not matched men and women standing on the same street. The gaps are differences between published point estimates. Numerical confidence bounds and item sample sizes were not available in the reviewed safety extract, so their statistical precision cannot be inferred from the apparent precision of a drawn line.
The comparison does not choose among possible explanations. Past experience, anticipated consequences, physical vulnerability, local conditions and patterns of exposure may overlap. The next step is to test the part for which a separate survey asks both questions of the same respondent: experience and current feelings of safety.
Shares feeling rather or very safe are not victimization probabilities. A named setting is not a shared physical location.
Reported experience separates the safety profiles.
In Germany, 16.4% of those reporting the specified household experience felt very unsafe, compared with 4.6% among those not reporting it. The association is clear in the point estimates; its precision is uneven.
Compare the rates. Then reverse the question.
Four response categories. Country-specific, paired-complete estimates with pointwise, survey-design-aware 95% confidence intervals.
Feeling very unsafe: 16.4% with the specified experience; 4.6% without it. This is a descriptive association.
Burglary or physical assault, affecting the respondent or a household member, in the preceding five years. “Yes” means reported to ESS, not reported to police. The offense-family control cannot split this combined item.
Specified experience: 16.4%. 95% CI 8.2% to 30.3%. Paired n = 207; weight-only effective n = 85.1; design df = 120.
No specified experience: 4.6%. 95% CI 3.1% to 6.6%. Paired n = 2,207; weight-only effective n = 943.3; design df = 425.
Exact ESS values, missingness, modes and precision
All-interviewed sample. Documented mode: Face-to-face interview: Computer-assisted (CAPI/CAMI). Recorded modes: 2,420 face-to-face, 0 video, 0 unavailable. Source records: 2,420. Selected-domain valid safety answers: 2,416; valid experience answers: 2,418. Missing either item: 6 (0.2% weighted).
| Safety answer | Experience group | Share / 95% CI | Paired n / precision |
|---|---|---|---|
| Very safe | Specified experience | 21.2%14.0%-30.7% | 207Weight-only effective n: 85.1; df: 120. |
| Very safe | No specified experience | 29.3%25.6%-33.3% | 2207Weight-only effective n: 943.3; df: 425. |
| Safe | Specified experience | 37.6%27.9%-48.4% | 207Weight-only effective n: 85.1; df: 120. |
| Safe | No specified experience | 47.2%43.2%-51.2% | 2207Weight-only effective n: 943.3; df: 425. |
| Unsafe | Specified experience | 24.8%15.5%-37.3% | 207Weight-only effective n: 85.1; df: 120. |
| Unsafe | No specified experience | 18.9%16.1%-22.1% | 2207Weight-only effective n: 943.3; df: 425. |
| Very unsafe | Specified experience | 16.4%8.2%-30.3% | 207Weight-only effective n: 85.1; df: 120. |
| Very unsafe | No specified experience | 4.6%3.1%-6.6% | 2207Weight-only effective n: 943.3; df: 425. |
Displayed percentages are rounded to one decimal; positive values below 0.1% are labeled <0.1%, not zero. Downloads retain the full-precision proportions and interval bounds.
Logit-transformed proportion plus/minus a design-df Student-t critical value times its delta-method standard error, transformed back to probability. The proportion SE is the maintained full-design Taylor estimate with domain-singleton adjustment. Pointwise sampling uncertainty, not simultaneous intervals, nonresponse bias, reporting error, measurement equivalence, or a causal effect test. Do not add category bounds to manufacture another interval.
Suppress below 50 paired observations, at boundary proportions or without positive design variance. Flag below 100. Kish effective n describes weights only; it does not determine these confidence intervals.
Source-coded gender, prespecified age bands and respondent-described domicile are separate domain cuts, not crossed cells or federal-state estimates. Age below 15 or unavailable is excluded only from age-specific domains. These are separate descriptive cuts, not jointly adjusted effects. Income and health adjustment is not estimated in this release.
Source: ESS ERIC / Sikt: European Social Survey Round 11, integrated edition 4.2.
ESS Round 11, edition 4.2. Thirty participating countries, not an EU-27 total. The four proportions use the same paired domain; intervals are pointwise, not simultaneous. No causal effect or annual personal risk is identified.
The severe end of the response scale is about more than a shift in an average score. In the German paired sample, the published point estimates for feeling very unsafe differ by 11.8 percentage points. The underlying 95% intervals are 8.2-30.3% for respondents reporting the specified experience and 3.1-6.6% for those not reporting it. Those are intervals for the two shares, not a confidence interval for their difference.
The comparison uses both answers from each respondent, with the survey weights and country-specific sample design. The experience question covers the respondent or a household member, combines burglary and physical assault, and looks back five years. A negative answer therefore does not identify a person who has never experienced crime.
Now reverse the question. Because only 8.5% of Germany's paired domain reports that specified experience, about 75.0% of the respondents feeling very unsafe belong to the much larger group answering No. This is a descriptive decomposition of the same weighted estimates, not a new causal model. A confidence interval for the reverse comparison has not been estimated. The result does not say that those respondents have never experienced crime.
A second comparison makes the boundary more concrete. Among German respondents answering No to that item, 7.9% of women feel very unsafe, compared with 1.1% of men. The pointwise 95% intervals are 5.4-11.5% and 0.4-3.0%, respectively. The difference is present within this stratum; it is not an adjusted gender effect, and the item does not cover every relevant experience or threat.
The analysis is descriptive rather than a causal estimate of what an incident did to a person. Timing, exposure and other differences between respondents remain unresolved. Nor are the 30 participating countries an EU-27 population average. Each selected country retains its own fieldwork, denominator, uncertainty and suppression rules.
Never divide the five-year experience measure by five or subtract it from a safety percentage to label beliefs mistaken.
Experience matters.
The record still misses part of life.
A household experience and a police case are not interchangeable. An incident can go unreported; a response about safety can reflect circumstances beyond that incident. Before using a headline count to settle the argument, inspect what makes it into the count.
The police see a selected part of the experience.
Completed residential burglary is much more visible than an attempted entry. The difference appears within the same survey, before any comparison with police totals.
The crime we experience.
Which specified incidents do survey participants recall, and which became known to the police?
BKA Safety and Crime in Germany (SKiD)What this record includes, and what it misses
- Unit
- Survey-defined incidents or affected people/households, depending on the measure.
- Clock
- SKiD 2024 victimization: January to December 2023.
- Inside the account
- Specified reported experiences, including incidents that did not reach the police.
- Outside the account
- A census of all crime, an exact legal crosswalk to PKS, and unlimited subgroup precision.
Change the survey and police exhibits together; their units remain separate.
The thematic link does not establish equal populations, offense scope, recall periods or counting units.
Inspect the three definitions
- ESS
- Burglary is combined with physical assault in one five-year household-linked ESS question.
- SKiD
- SKiD distinguishes completed residential burglary from attempted entry. Source groupings must be retained.
- PKS
- PKS residential burglary is an administrative offense category, including attempts where the table specifies them.
A completed burglary is more visible than an attempt.
Germany, SKiD 2024. Experiences during 2023; residents aged 16+ in private households. Weighted, outlier-adjusted survey event ratios.
Dots: published estimates. Whiskers: published 95% confidence intervals. Common 0–100% scale.
The exact table also gives a derived remainder: “Not recorded as known within this measure.” It is not an estimate of all crime unknown to police; raw blank and uncertain-response handling has not been fully verified.
Definition check: the report's 47.9% groups entry without theft and attempted entry. The dashboard separates them, as shown here.
95% confidence interval 83.3% to 90.8%. Published BKA 95% confidence bounds; not reconstructed from the event count. Published event n: 1115.
Awareness is reported to the respondent's knowledge, including indirect awareness. Definitions and denominator details are retained below.
Exact values, uncertainty and denominators
| Measure | Reported known | Derived remainder | Uncertainty / base |
|---|---|---|---|
| Completed residential burglary with theftJanuary-December 2023 experiences / SKiD 2024 | 87.4% | 12.6% | 95% confidence interval 83.3% to 90.8%. Published BKA 95% confidence bounds; not reconstructed from the event count. Published event n: 1115.Weighted, outlier-adjusted survey events; police awareness is reported to the respondent's knowledge. BKA describes the displayed event n as experiences with an answer on police awareness. Completed residential burglary with theft. Police awareness includes indirect awareness to the respondent's knowledge. People aged 16 or older in private households in Germany; source-defined valid responses. |
| Entry into dwelling without theftJanuary-December 2023 experiences / SKiD 2024 | 46.8% | 53.2% | 95% confidence interval 40.8% to 52.8%. Published BKA 95% confidence bounds; not reconstructed from the event count. Published event n: 2291.Weighted, outlier-adjusted survey events; police awareness is reported to the respondent's knowledge. BKA describes the displayed event n as experiences with an answer on police awareness. Entry into dwelling without theft. Police awareness includes indirect awareness to the respondent's knowledge. People aged 16 or older in private households in Germany; source-defined valid responses. |
| Attempted entry into dwellingJanuary-December 2023 experiences / SKiD 2024 | 51.5% | 48.5% | 95% confidence interval 39.2% to 63.8%. Published BKA 95% confidence bounds; not reconstructed from the event count. Published event n: 715.Weighted, outlier-adjusted survey events; police awareness is reported to the respondent's knowledge. BKA describes the displayed event n as experiences with an answer on police awareness. Attempted entry into dwelling. Police awareness includes indirect awareness to the respondent's knowledge. People aged 16 or older in private households in Germany; source-defined valid responses. |
Source: Bundeskriminalamt: SKiD dashboard: residential burglary, embedded data.
Event ratios, not average personal reporting propensity. Event n is not respondent n or effective sample size. Published intervals are not reconstructed. Definitions and methodology.
Among completed residential burglaries with theft, 87.4% were reported as known to police. For attempted entry, the estimate was 51.5%. The 35.9-percentage-point difference is calculated from the published point estimates. It is not an estimate of how much crime the police miss overall.
This selection matters for interpretation. A change in recorded cases can reflect a change in incidents, a change in police awareness, or both. The survey does not disentangle those mechanisms over time, but it establishes why the administrative record cannot be treated as an unfiltered census of experience.
The source also separates entry without theft, at 46.8%, from attempted entry. A published report combines those two categories into a 47.9% figure. That broader category must not replace attempted entry in the comparison above. These are weighted event-level awareness ratios, including incidents the police learned about through someone other than the victim; they are not averages of each person's reporting fraction.
The complement means not recorded as known within this published measure. The complete handling of blank or uncertain awareness counts is not verified. It is neither PKS divided by survey totals nor a count of all hidden crime.
A national count is not a map of everyday safety.
Follow burglary or assault through the police record. The changing count describes an offense category, not who feels able to use a particular place.
Residential burglary, in the police statistics.
Germany. Recorded cases, not people; axis ticks use k for thousands. Each dot retains its offense code and completion-year basis.
Solid / residential burglary · Dashed / dangerous and serious bodily injury, including attempts. Missing years are not connected. The lines show cases, not comparable personal risks.
Residential burglary, including attempts / PKS 435*00. Wohnungseinbruchdiebstahl §§ 244 Abs. 1 Nr. 3 und Abs. 4, 244a StGB. Includes 36,049 attempted cases within this total, not in addition. Police-processing completion year; this legal category is not identical to the survey question.
Exact police counts and offense codes
| Completion year | Cases | Offense code |
|---|---|---|
| 2015 | 167,136 | 435*00 |
| 2019 | 87,145 | 435*00 |
| 2020 | 75,023 | 435*00 |
| 2021 | 54,236 | 435*00 |
| 2022 | 65,908 | 435*00 |
| 2023 | 77,819 | 435*00 |
Source: Bundeskriminalamt: PKS 2023, ZR-F-01-T01: cases time series, CSV v=3; Bundeskriminalamt: SKiD 2024: Bundesweite Kernbefunde, V 1.0.
Wohnungseinbruchdiebstahl § 244 Abs. 1 Nr. 3 StGB darunter:. Includes 71,300 attempted cases within this total, not in addition. Police-processing completion year; this legal category is not identical to the survey question. Neither a change in this count nor a year-label match establishes a change in personal risk or exact equivalence to the survey.
A common collection is not a common crime rate.
Inspect a national record rather than rank countries as safer or more dangerous.
Selected: Germany / Residential burglary
| Country / offense | Recorded offenses | Per 100,000 people | Recording stage |
|---|---|---|---|
| GermanyBurglary of private residential premises / ICCS05012Selected record | 77,819 | 93.62 | output |
| GermanySerious assault / ICCS020111 | 154,541 | 185.93 | output |
| AustriaBurglary of private residential premises / ICCS05012 | 7,584 | 83.3 | output |
| AustriaSerious assault / ICCS020111 | 4,463 | 49.02 | output |
| NetherlandsBurglary of private residential premises / ICCS05012 | 22,060 | 123.85 | input |
| NetherlandsSerious assault / ICCS020111 | 4,610 | 25.88 | input |
Germany uses output-stage recording in the cited metadata. These 2023 observations come from the April 2026 vintage. National legal definitions, reporting and recording stages differ; the population-normalized rate is not personal victimization risk.
Source: Eurostat: Police-recorded offences by offence category: selected 2023 cells; Eurostat: Eurostat crime metadata: definitions, reference periods and recording stages; Eurostat: Eurostat crime methodology: administrative sources, ICCS and counting.
The German figures originate in the same administrative system as PKS, not independent corroboration. Germany and Austria record at output stage; the Netherlands at input stage. Laws, definitions, counting and reporting differ; neither counts nor rates are personal victimization risks.
The reviewed series tell two different long-run stories. Recorded residential burglary fell from 167,136 cases in 2015 to 77,819 in 2023, a decline of 53.4%. Dangerous and serious bodily injury, including attempts, rose from 127,395 to 154,541 cases, or 21.3%. Calling both movements the crime trend would discard the distinction that makes them informative.
The reference period matters as well. Burglary was far below its 2015 level but 43.5% above the 2021 trough. The chart preserves both facts and leaves the omitted 2016-2018 observations unconnected. None of these changes identifies a particular person's exposure, the reasons for a survey response, or an effect of policing.
There is a second separation in time. Germany's PKS counts cases in the year police processing is completed. SKiD asks about experiences during a specified calendar year. ESS asks whether a particular household-linked experience occurred during the previous five years. A shared year label does not make these records contemporaneous observations of the same events.
The European record is an additional view of administrative data, not independent confirmation of the German count. Common classifications help organize the comparison, but laws, recording practices and reporting can differ. Absolute counts below are exact administrative quantities; they do not rank countries by personal risk.
Recorded cases describe administrative visibility, not the probability that a particular resident will become a victim.
The journeys not made are missing from the crime count.
Reported precautions and avoidance are outcomes in their own right. They show where concern enters ordinary choices, even though they cannot establish what caused each choice.
Some journeys come with a condition.
Share reporting each behavior very often or frequently, not merely ever. Compare populations within the same behavior; overlapping responses are not pieces of a total.
Avoid leaving home after dark
Go out after dark only with company
Women minus men: 27.3 percentage points. A difference between published point estimates; its confidence interval is not available.
Carry an object to attract attention in an emergency
Women minus men: 8.5 percentage points. A difference between published point estimates; its confidence interval is not available.
Each bar uses the same 0–100% scale. Responses must not be added.
Reports doing this very often or frequently, not merely ever. Behaviors and subgroups overlap; do not add their percentages. 2024 survey-time behavior. People aged 16 or older in private households in Germany; source-defined valid responses. All respondents.
No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure.
Exact behavior estimates and response definitions
| Behavior / population | Share | Uncertainty and definition |
|---|---|---|
| Avoid leaving home after darkAll respondents. 2024 survey-time behavior. | 28.6% | No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure.Reports doing this very often or frequently, not merely ever. Behaviors and subgroups overlap; do not add their percentages. People aged 16 or older in private households in Germany; source-defined valid responses. All respondents. |
| Go out after dark only with companyAll respondents. 2024 survey-time behavior. | 25.6% | No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure.Reports doing this very often or frequently, not merely ever. Behaviors and subgroups overlap; do not add their percentages. People aged 16 or older in private households in Germany; source-defined valid responses. All respondents. |
| Go out after dark only with companyMen. 2024 survey-time behavior. | 11.7% | No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure.Reports doing this very often or frequently, not merely ever. Behaviors and subgroups overlap; do not add their percentages. People aged 16 or older in private households in Germany; source-defined valid responses. Men. |
| Go out after dark only with companyWomen. 2024 survey-time behavior. | 39.0% | No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure.Reports doing this very often or frequently, not merely ever. Behaviors and subgroups overlap; do not add their percentages. People aged 16 or older in private households in Germany; source-defined valid responses. Women. |
| Carry an object to attract attention in an emergencyMen. 2024 survey-time behavior. | 10.7% | No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure.Reports doing this very often or frequently, not merely ever. Behaviors and subgroups overlap; do not add their percentages. People aged 16 or older in private households in Germany; source-defined valid responses. Men. |
| Carry an object to attract attention in an emergencyWomen. 2024 survey-time behavior. | 19.2% | No numerical interval endpoints recovered from the cited report text. No interval was digitized or inferred from the figure.Reports doing this very often or frequently, not merely ever. Behaviors and subgroups overlap; do not add their percentages. People aged 16 or older in private households in Germany; source-defined valid responses. Women. |
Source: Bundeskriminalamt: SKiD 2024: Bundesweite Kernbefunde, V 1.0.
Avoidance may change exposure, but these cross-sectional responses do not identify a causal effect on victimization.
Avoidance complicates the simple story in which people encounter a fixed level of risk and then decide how worried to be. Someone who stays away from a place changes their exposure to it. That is a possible mechanism, not a causal result identified by this survey, but it explains why an incident count alone cannot measure access to public space.
In SKiD, 39.0% of women report going out after dark only with company very often or frequently, compared with 11.7% of men: a 27.3-percentage-point difference. Separately, 28.6% of all respondents report frequently or very often avoiding leaving home after dark. These are reports of behavior, not estimates of journeys lost or proof that a precaution prevented an incident.
Compare the same behavior across groups before comparing unlike behaviors. One person can report more than one precaution, and the all-respondent reference overlaps the sex-specific groups. These percentages must not be added into an overall avoidance rate.
The practical distinction is between preventing harm and expanding people's ability to move through everyday life. Both deserve evidence. A low recorded count does not tell us whether people changed their routines to achieve it, while a high reported avoidance share does not prove that a precaution prevented an incident.
Reported behavior is not proof that a precaution prevented an incident. Exposure and causality remain unresolved.
A safer place should also be a place more people can use.
The evidence supports a more demanding account of public safety. Men and women describe different conditions for using the same types of setting. Specified past household experiences are associated with a different distribution of present feelings. Police visibility varies within an offense family, and people report changing what they do.
For policymakers, a falling case count answers an important question, but not the whole one. Evaluation should keep recorded offenses, survey experiences, setting-specific safety and reported avoidance visible as separate outcomes. Improvement in one should not be presumed to settle the others.
This analysis cannot identify which intervention would close a particular gap, or how accurately a respondent's concern reflects their personal risk. It does establish what a credible follow-up needs to measure: fewer harmful experiences, and fewer people feeling that ordinary public space is closed to them.
Take this record with you.
Your selections travel in the URL. The downloadable record preserves the selected measures without merging their denominators.
Every account has a boundary.
Reviewed evidence (JSON) / Exact values (CSV) / Derived comparisons R2 (JSON) / Source fingerprints and reproduction
ESS: paired domains, survey design and suppression
The standalone Round 11 edition 4.2 analysis uses the two observed answers together, anweight and the reviewed full sampling design. Country-specific proportions carry pointwise logit-delta-t 95% confidence intervals, independently checked against R survey. Suppress domains below 50 paired observations, boundary proportions and nonpositive design variance; flag domains below 100. Kish effective n is weight-only, not the variance estimator. Two independent runs are byte-identical. Eleven German subgroup cuts are descriptive, not jointly adjusted models; income and health adjustment is not estimated.
Four clocks, not one survey year
SKiD 2024 asks about victimization in calendar year 2023. SKiD 2020 used November 2019 to October 2020. ESS victimization is a rolling five-year lookback; perceived safety is answered at the interview. Germany's ESS fieldwork ran from 9 May to 21 December 2023. PKS uses an administrative completion-year basis. Survey names, interview dates and experience periods stay separate.
Counts, proportions and uncertainty
Administrative counts and survey proportions never share a quantitative axis. Published survey confidence bounds and source event counts are retained where available; missing bounds remain unavailable. Reporting uses weighted, outlier-adjusted event totals, not the average of personal reporting fractions. The complement is labeled not recorded as known within this measure: the full handling of raw blank and uncertain awareness counts is not public.
Selection, arithmetic and estimation
Controls select reviewed records without refitting the source survey estimators or changing their weights. Stated differences and the reverse conditional are deterministic arithmetic from those records. The reverse comparison uses the same weighted paired domain and must reconcile to its published all-paired share; it is withheld if any required cell is suppressed. No interval for these new derivatives or causal model is estimated. Portable outputs retain the source, unit, method and these limits.
No district-level inference
National survey results do not support a street map or a local fear estimate. Named settings are survey categories, not measured routes or a shared physical location. The opening evidence stage repeats the reviewed setting comparison; it does not simulate a neighborhood.
A standalone publication
This investigation has its own article, evidence inputs and route. The separate Europe, Measured story program is not modified by this publication.
| Account | Reference clock | What does not follow |
|---|---|---|
| PKS | Year police processing was completed. | Not necessarily the year an incident occurred. |
| SKiD 2024 experiences | January to December 2023. | The survey label is not the experience year. |
| ESS experience | Five years before each respondent's interview. | Not an annual personal probability. |
| Perceived safety | Answered during the survey's fieldwork. | Not a contemporaneous count of incidents. |
SKiD 2024: Bundesweite Kernbefunde, V 1.0
Bundeskriminalamt / V 1.0, April 2026. Retrieved 2026-09-19.
Limited factual extracts from the April 2026 report. Safety: printed p. 62 / PDF p. 65. Behavior: printed pp. 69-70 / PDF pp. 72-73. Conflicting overall avoidance figures are excluded.
Reuse: BKA report-specific reproduction terms; not a standard open-data license. Reproduction, including extracts, requires BKA attribution. The report is not for sale and must not be used for election campaigning.
SKiD dashboard: residential burglary, embedded data
Bundeskriminalamt / Web resource frozen in edition 2026-09-19-r1. Retrieved 2026-09-19.
Fetched in memory only; selected factual cells/definitions retained, not the complete resource.
Reuse: BKA website copyright terms; no open-data license verified. Website content is copyright-protected; copying is limited to private use except where permission or statutory rights apply. Do not label it CC BY or open data.
SKiD dashboard: assault, embedded data
Bundeskriminalamt / Web resource frozen in edition 2026-09-19-r1. Retrieved 2026-09-19.
Fetched in memory only; selected factual cells/definitions retained, not the complete resource.
Reuse: BKA website copyright terms; no open-data license verified. Website content is copyright-protected; copying is limited to private use except where permission or statutory rights apply. Do not label it CC BY or open data.
SKiD dashboard: violence and sexual offences, embedded data
Bundeskriminalamt / Web resource frozen in edition 2026-09-19-r1. Retrieved 2026-09-19.
Fetched in memory only; selected factual cells/definitions retained, not the complete resource.
Reuse: BKA website copyright terms; no open-data license verified. Website content is copyright-protected; copying is limited to private use except where permission or statutory rights apply. Do not label it CC BY or open data.
SKiD dashboard help: periods, denominators and intervals
Bundeskriminalamt / Web resource frozen in edition 2026-09-19-r1. Retrieved 2026-09-19.
Fetched in memory only; selected factual cells/definitions retained, not the complete resource.
Reuse: BKA website copyright terms; no open-data license verified. Website content is copyright-protected; copying is limited to private use except where permission or statutory rights apply. Do not label it CC BY or open data.
SKiD glossary: Anzeigequote versus Anzeigeneigung
Bundeskriminalamt / Web resource frozen in edition 2026-09-19-r1. Retrieved 2026-09-19.
Fetched in memory only; selected factual cells/definitions retained, not the complete resource.
Reuse: BKA website copyright terms; no open-data license verified. Website content is copyright-protected; copying is limited to private use except where permission or statutory rights apply. Do not label it CC BY or open data.
SKiD key findings: publication date and burglary aggregation note
Bundeskriminalamt / Web resource frozen in edition 2026-09-19-r1. Retrieved 2026-09-19.
Fetched in memory only; selected factual cells/definitions retained, not the complete resource.
Reuse: BKA website copyright terms; no open-data license verified. Website content is copyright-protected; copying is limited to private use except where permission or statutory rights apply. Do not label it CC BY or open data.
SKiD copyright and citation terms
Bundeskriminalamt / Web resource frozen in edition 2026-09-19-r1. Retrieved 2026-09-19.
Fetched in memory only; selected factual cells/definitions retained, not the complete resource.
Reuse: BKA website copyright terms; no open-data license verified. Website content is copyright-protected; copying is limited to private use except where permission or statutory rights apply. Do not label it CC BY or open data.
PKS 2023, ZR-F-01-T01: cases time series, CSV v=3
Bundeskriminalamt / PKS 2023 time-series release; download parameter v=3. Retrieved 2026-09-19.
Fetched in memory only; selected factual cells/definitions retained, not the complete resource.
Reuse: BKA website copyright terms; no open-data license verified. Website content is copyright-protected; copying is limited to private use except where permission or statutory rights apply. Do not label it CC BY or open data.
PKS 2023 official time-series downloads
Bundeskriminalamt / Web resource frozen in edition 2026-09-19-r1. Retrieved 2026-09-19.
Fetched in memory only; selected factual cells/definitions retained, not the complete resource.
Reuse: BKA website copyright terms; no open-data license verified. Website content is copyright-protected; copying is limited to private use except where permission or statutory rights apply. Do not label it CC BY or open data.
BKA website copyright terms
Bundeskriminalamt / Web resource frozen in edition 2026-09-19-r1. Retrieved 2026-09-19.
Fetched in memory only; selected factual cells/definitions retained, not the complete resource.
Reuse: BKA website copyright terms; no open-data license verified. Website content is copyright-protected; copying is limited to private use except where permission or statutory rights apply. Do not label it CC BY or open data.
Police-recorded offences by offence category: selected 2023 cells
Eurostat / CRIM_OFF_CAT, structure 34.0; updated 2026-04-29T11:00:00+0200. Retrieved 2026-09-19.
Unmodified, filtered official API response.
Reuse: Eurostat statistical-data reuse policy / Commission Decision 2011/833/EU. Statistical data and metadata may be reused with source acknowledgement, subject to stated exceptions. Editorial EU-owned website content is CC BY 4.0. The selected EU-country data are not among the listed third-country exceptions.
Eurostat crime methodology: administrative sources, ICCS and counting
Eurostat / Web resource frozen in edition 2026-09-19-r1. Retrieved 2026-09-19.
Fetched in memory only; selected factual cells/definitions retained, not the complete resource.
Reuse: Eurostat statistical-data reuse policy / Commission Decision 2011/833/EU. Statistical data and metadata may be reused with source acknowledgement, subject to stated exceptions. Editorial EU-owned website content is CC BY 4.0. The selected EU-country data are not among the listed third-country exceptions.
Eurostat crime metadata: definitions, reference periods and recording stages
Eurostat / Metadata certified/updated 2026-04-29. Retrieved 2026-09-19.
Fetched in memory only; selected factual cells/definitions retained, not the complete resource.
Reuse: Eurostat statistical-data reuse policy / Commission Decision 2011/833/EU. Statistical data and metadata may be reused with source acknowledgement, subject to stated exceptions. Editorial EU-owned website content is CC BY 4.0. The selected EU-country data are not among the listed third-country exceptions.
Eurostat copyright and reuse notice
Eurostat / Web resource frozen in edition 2026-09-19-r1. Retrieved 2026-09-19.
Fetched in memory only; selected factual cells/definitions retained, not the complete resource.
Reuse: Eurostat statistical-data reuse policy / Commission Decision 2011/833/EU. Statistical data and metadata may be reused with source acknowledgement, subject to stated exceptions. Editorial EU-owned website content is CC BY 4.0. The selected EU-country data are not among the listed third-country exceptions.
European Social Survey Round 11, integrated edition 4.2
ESS ERIC / Sikt / Round 11 / edition 4.2. Retrieved 2026-09-19.
Country-specific paired-complete estimates use analysis weights and the full survey design. Pointwise 95% logit-delta-t intervals; two independent executions and 1,804 independent R reference checks. No pooled Europe mean or causal inference.
Reuse: CC BY-NC-SA 4.0. ESS-derived aggregates retain this attribution, noncommercial and share-alike license. Respondent records are not redistributed.