Disclaimer: The following data are based on the information provided in the victim entries, much of it official but some anecdotal. While some may have potential errors or suffer from reporting bias (we've left notes of caution), it should prove insightful nevertheless. Those interested in more in-depth analysis are recommended to apply specific filters before opening this page.
❗ The ages reported are those at the time of the last status update, which is different for each victim. However, in the case of detainees, this age is usually not very different from the victim's age when they were first detained (within 1-2 years, in most cases).
✓ For detained victims, no real biases have been identified, and it is likely that the age distributions are representative of all detainees in Xinjiang.
The "fluid" charts integrate both exact ages and age ranges into a single chart, by assuming uniform distributions for uncertain age ranges (for example, if a victim is reported as being "in his 40s", with the corresponding range of 40-49, "one tenth of a person" is added to the counts for 40, 41, 42, ..., 49). The corresponding statistics are therefore based on expected values and may be seen as estimates. While less intuitive, they are more statistically powerful as they make use of all or most of the data, and not only those victims for whom the exact age is reported. (Only use exact ages:.)
Quick filters post-2017 detainees only
❗ Gender rates for underage victims (typically separated from their parents) are close to 50-50, and very different from the rate for adult detainees, where men make up the large majority. Please apply detainee-only filters if you would like to focus on adults.
✓ There are no expected or observed biases in the reported data, with the data believed to be representative of Xinjiang at large.
Quick filters post-2017 detainees only
❗ Currently, the data reported in our database have a strong regional bias, with the majority coming from the mostly-Uyghur southern Xinjiang. As a result, many of the regions that are traditionally home to other ethnic groups are undersampled, with fewer documented victims from those groups and regions. The data presented here should therefore be seen as a summary of what has been documented, and not as a representative picture of ethnic demographics among victims.
Quick filters post-2017 detainees only serving prison terms
❗ The locations given are the last known or estimated locations ("current" at the latest status update), and as such will correspond to different times for different victims. They should not be taken as representative of the locations *right now*, though for many victims this will likely be the case, since many of those serving long prison terms have remained in the same location, while those who were in camp and released are likely in their home regions (where they were also camped).
❗❗ Owing to location bias in our data sources, which favor southern Xinjiang, the data above should not be seen as representative of Xinjiang in its entirety.
The color depth for a given region corresponds to the concentration of victims there, with the number of victims scaled by the total population.
Quick filters post-2017 detainees only with updates in the past months
❗ The status reported for each victim is that at the time of the latest update, with this time different for each victim. If you are interested specifically in statuses that have been reported more recently, please use the quick filter above with the corresponding time period.
❗❗ Because of both time lag and reporting bias (some sources are long lists of victims in only one form of detention), it is not likely that the data above are representative of victim statuses in Xinjiang in general.
For victims who were put in hard detention, the time "problems started" is when they were initially detained. For victims who were never put in hard detention, the time given is when they were initially subjected to soft detention (having documents taken or being put under house/town arrest, for example). In the case of children, it is when both/only of their parents were detained, or when they were taken to a "boarding school" (in some cases, even when one parent was still at home).
While periods spent in custody and camp are generally not fixed and are taken as those empirically observed, the prison terms are reported as set by the court. As such, these are the hypothetical terms the victims are expected to serve, and also include the time spent in custody. Life sentences are excluded from the statistics (mean, mode, and median) since they do not have a fixed term, while any sentences with trailing months have been rounded to the nearest integer (rounding down in the case of 6 months).
show relative values
Quick filters post-2017 detainees only exclude pre-Chen Quanguo victims
❗ Data on victims come from various sources from various time periods, with the years 2017-2019 especially represented and data since 2020 more scarce. Not all detention forms are represented equally well. Consequently, the data here should not be taken as representative of all detention forms and/or all time periods, and only convey what has been documented.
❗❗ Detention trends changed drastically after Chen Quanguo's assuming the role of Party secretary in late 2016, with much of the international attention on Xinjiang focusing on this period. Please use the quick filter above to exclude victims detained before that if you would like to focus solely on the people who were first detained after the arrival of Chen Quanguo.
✓ The prison term lengths have a very large sample size and are not believed to be subject to much bias, though a significant portion of those reported (close to 20000) date to 2015 and earlier, when the authorities were less strict. ❗❗❗ Consequently, it may be best to consider the prison-term length statistics for before Chen Quanguo and after separately.
The "fluid" chart integrates both exact dates and date ranges into a single chart, by assuming uniform distributions for uncertain ranges (for example, if a victim is reported as detained in April 2018, with the corresponding range of April 1 to April 30, "one thirtieth of a person" is added to the counts for April 1, April 2, April 3, ..., April 30). The corresponding statistics are therefore based on expected values and may be seen as estimates. While less intuitive, they are more statistically powerful as they make use of all or most of the data, and not only those victims for whom the exact date is reported.
Quick filters post-2017 detainees only camp detainees
❗ Any number of reasons can be attributed to a given victim's detention (our database limits to one suspected reason and three official ones). Furthermore, a single individual can contribute to the more general reason categories multiple times (for example, if a victim were detained for two reasons that both belong to the general category of "terrorism", they would add 2 and not 1 to the overall "terrorism" count). Finally, a single reason can belong to multiple general categories. Given all this, the numbers that appear here should be treated as instances of a given reason occurring, and not as the number of victims detained for that reason.
❗❗ The reasons used to send documented victims to camp are generally different from those used to sentence them to long prison terms (the latter originate from China's criminal code). We recommend using the "camp detainees" quick filter above if you'd like to focus only on those victims who have been documented as held at camp (though a small portion of them may have been sentenced also, in a previous or later detention).
✓ There are no expected or known biases in the reported detention reasons, and the data are believed to be representative of Xinjiang at large. ❗❗❗ However, because the reporting of detention types - camp vs. non-camp - may not be sufficiently balanced, it is still best to consider the groups separately.
Quick filters victims in hard detention (custody, camp, prison, labor, quarantine, orphanage)
❗ The absence of reported health issues does not mean the absence of health issues, and the data above should not be interpreted as reflecting the proportions of victims who have problems with their health, but rather as providing a (very) conservative lower estimate. These statistics are intended to draw attention to those victims for whom health issues have been documented, with the quick filter above specifically seeking to identify those who were still in detention at the time of latest status update.
❗❗ Because not all victims have age groups reported, the individual numbers of a given column do not necessarily sum up to the total number of people with that health status, as given in the column label.
Quick filters post-2017 detainees only
❗ The majority of victims do not have their profession reported, and while a large variety of jobs have been covered by those victims for whom it has been, the sample is still relatively small and may suffer from various biases that we've yet to identify.