Variable pages · IRS 990 Master Concordance
SG_01_FUNDRAISER_ADDR_CITY
Fundraiser city
- Description
- Foreign Address - City
- Table
SG-P01-T01-FUNDRAISERS-INFO(many)- Part
- Part I - Fundraising Activities
- Location
SCHED-G-PART-01-LINE-02B-COL-i- Type
- text
- Scope
- PZ – 990 and 990-EZ
- Database
- F990
- Forms
- SZ
6 / 6
xpaths observed / mapped
69k
filings with a value
2010–2024
tax years observed
0 + 4 reviewed
open flags
Checks
Value checks
| Result | Value check | Detail |
|---|---|---|
| ✓ pass | Values are text, not numbers | 0% of values are numeric |
| ✓ pass | No sign of truncation | longest values do not sit at a common field limit |
Flags
| Flag | Severity | Xpath | Detail | Review |
|---|---|---|---|---|
COLLISION | ⚠ error | IRS990ScheduleG/FundraiserActivityInfoGrp/ForeignAddress/CityNm | 539 filings contain 2+ of its xpaths; most common set: /Return/ReturnData/IRS990ScheduleG/FundraiserActivityInfoGrp/ForeignAddress/CityNm ; /Return/ReturnData/IRS990ScheduleG/FundraiserActivityInfoGrp/USAddress/CityNm | ✓ acceptednoteFalse positive: counted per filing. Recounted per repeating-element row in the TY2012 and TY2019 DuckDB builds: 0 rows hold both the US and the foreign address xpath (different rows of the table hold different address types). build_evidence() now counts collisions per row. |
MANY_NOT_REPEATING | ⓘ info | IRS990ScheduleG/FundraiserActivityInfoGrp/USAddress/CityNm | many-to-one table, but 72% of values sit in no repeating element | ✓ acceptednoteGenuine repeating group: the group element is the repeat_root in filings with 2+ rows and '' in single-row filings, so the high unrepeated share just means most filers report one row. Evidence: repeats in 27.1% of 54,332 filings, max 10/filing, TY2014-2024; repeat_root: ''=72% ; IRS990ScheduleG/FundraiserActivityInfoGrp=28%. |
MANY_NOT_REPEATING | ⓘ info | IRS990ScheduleG/FundraiserActivityInformation/AddressUS/City | many-to-one table, but 71% of values sit in no repeating element | ✓ acceptednoteGenuine repeating group: the group element is the repeat_root in filings with 2+ rows and '' in single-row filings, so the high unrepeated share just means most filers report one row. Evidence: repeats in 28.5% of 9,547 filings, max 10/filing, TY2010-2012; repeat_root: ''=71% ; IRS990ScheduleG/FundraiserActivityInformation=29%. |
MANY_NOT_REPEATING | ⓘ info | IRS990ScheduleG/FundraiserActivityInfoGrp/USAddress/City | many-to-one table, but 71% of values sit in no repeating element | ✓ acceptednoteGenuine repeating group: the group element is the repeat_root in filings with 2+ rows and '' in single-row filings, so the high unrepeated share just means most filers report one row. Evidence: repeats in 27.6% of 4,040 filings, max 10/filing, TY2013-2013; repeat_root: ''=71% ; IRS990ScheduleG/FundraiserActivityInfoGrp=29%. |
Xpaths
| Xpath (under /Return/ReturnData or /Return/ReturnHeader) | Form | Years | Filings | % filers | Repeats in | |
|---|---|---|---|---|---|---|
| x1 | IRS990ScheduleG/FundraiserActivityInformation/AddressUS/City | SZ | 2010–2012 | 9,547 | 1.37% | 28.5% |
| x2 | IRS990ScheduleG/FundraiserActivityInformation/AddressForeign/City | SZ | 2010–2012 | 91 | 0.0124% | 14.3% |
| x3 | IRS990ScheduleG/FundraiserActivityInfoGrp/USAddress/City | SZ | 2013–2013 | 4,040 | 1.33% | 27.6% |
| x4 | IRS990ScheduleG/FundraiserActivityInfoGrp/ForeignAddress/City | SZ | 2013–2013 | 38 | 0.0125% | 10.5% |
| x5 | IRS990ScheduleG/FundraiserActivityInfoGrp/USAddress/CityNm | SZ | 2014–2024 | 54,332 | 0.922% | 27.1% |
| x6 | IRS990ScheduleG/FundraiserActivityInfoGrp/ForeignAddress/CityNm | SZ | 2014–2024 | 698 | 0.0125% | 18.3% |
% filers: percent of in-scope filers whose return has the xpath, in its latest year. Repeats in: share of filings where the element occurs more than once.
Coverage by tax year
| Filings | ’09 | ’10 | ’11 | ’12 | ’13 | ’14 | ’15 | ’16 | ’17 | ’18 | ’19 | ’20 | ’21 | ’22 | ’23 | ’24 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| x1 | 2.5k | 3.3k | 3.7k | |||||||||||||
| x2 | 26 | 31 | 34 | |||||||||||||
| x3 | 4.0k | |||||||||||||||
| x4 | 38 | |||||||||||||||
| x5 | 4.3k | 4.5k | 4.6k | 5.0k | 4.9k | 4.9k | 4.8k | 5.3k | 5.8k | 6.0k | 4.2k | |||||
| x6 | 42 | 50 | 51 | 51 | 54 | 54 | 63 | 79 | 109 | 88 | 57 | |||||
| Fill rate: % of all filings of the return type (largest xpath) | ||||||||||||||||
| 990 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 2 | 1 | 2 | 2 | 2 | 1 | |
| 990-EZ | <1 | <1 | <1 | <1 | <1 | <1 | <1 | <1 | <1 | <1 | <1 | <1 | <1 | <1 | <1 | |
Filings with a value, by xpath (rows, numbered as in the xpath table) and tax year. A change of row from one year to the next is usually the IRS renaming the element (most were renamed in 2013). Hover a cell for the exact count.
Values
| Return | Filings | Average length | Longest |
|---|---|---|---|
| 990 | 67,230 | 9 | 22 |
| 990-EZ | 1,516 | 9 | 22 |
Most common values
| Value | Filings | Share |
|---|---|---|
NEW YORK | 4,503 | 25.4% |
Example filings
| Tax year | Return | Xpath | Organization | Value | |
|---|---|---|---|---|---|
| 2024 | 990 | x6 | THE ROTARY FOUNDATION OF ROTARY | MAKATI CITY | XML |
| 2024 | 990 | x5 | PORTLAND PARKS FOUNDATION | PORTLAND | XML |
| 2013 | 990 | x4 | NATURE CONSERVANCY | Toronto | XML |
| 2013 | 990 | x3 | PENNSYLVANIA INTERSCHOLASTIC | LANCASTER | XML |
| 2012 | 990 | x2 | SEATTLE UNIVERSITY | Toronto | XML |
| 2012 | 990 | x1 | UNITED STATES AMATEUR BOXING FOUNDATION | MT LAUREL | XML |