ADMO adheres to the privacy requirements in the Family Educational Rights and Privacy Act (FERPA) of 1974 along with other best practices in order to protect students’ right to privacy. FERPA is a federal law that protects the privacy of student education records and pertains to the release of and access to educational records or any information directly related to a student that are maintained by an educational institution or agency or other party acting on their behalf. The law applies to all schools that receive funds under applicable programs of the US Department of Education.
UW standard practice for protecting personally identifiable data is that information for groups of less than 10 students may not be reported in aggregated tables. For ADMO additional practices have been adopted from Federal guidelines. This includes complementary suppression. Complementary suppression rules are applied in cases where using simple subtraction from the total could allow viewers to back into the suppressed low value count of a group.
Suppression Rules Summary
- Counts less than 10 and greater than 0 are not shown. For survey data since viewers do not have the information as to which students responded to the survey question, counts less than 3 and greater than 0 are not shown instead.
- When separate subgroups are masked and added together, they are totaled into a new subgroup called “All Masked Values. The suppressed subgroups that are included in the “All Masked Values” category are listed below the graph.
- When only one subgroup is masked, complementary suppression is used to prevent users from backing into the masked value of the subgroup with simple subtraction from the overall by masking the smallest unmasked group and adding its value to the “All Masked Values” subgroup.
- Whenever there is more than one “generated” subgroup such as “unknown/non-respondent” or “multiple values reported” and either includes a value less than 10, including 0, then the two subgroups are combined and masked together to prevent a possible non-generated subgroup with 10 or greater students from being suppressed.
- If values for subgroups are present and identical for complementary suppression, then the subgroup with the lowest denominator (or n value) will be suppressed. If denominators are the same for subgroups, then an alpha-numeric value (either a label or an ID value) will be used to determine which of the identical subgroups will be masked with consistency. This alpha-numeric value is used to resolve arbitrary ties and produce consistent results across dashboards.
- In situations where a secondary disaggregation (e.g., Female First Generation) is present, the “All Masked Values” subgroup is created within a relevant generated category if available within the primary disaggregation’s generated subgroup (e.g., “All Other Gender All Masked Values”) to prevent a possible non-generated subgroup with 10 or greater students from being suppressed.
- If an overall disaggregation is masked, all subgroups will be masked as well.
How Suppression and Complementary Suppression Work
Unsuppressed: Counts by Race/Ethnicity
| Ethnicity | Count of Students |
|---|---|
| American Indian/Alaska Native | 6 |
| Asian | 88 |
| Black or African American | 52 |
| Filipino | 37 |
| Hispanic | 46 |
| Pacific Islander or Hawaiian Native | 14 |
| White | 95 |
| Two or More Races | 96 |
| Unknown / Non-Respondent | 50 |
| Multiple Values Reported | 16 |
| Total | 500 |
Table 1
1st Level of Suppression: Suppress counts less than 10 and greater than 0
| Ethnicity | Count of Students | This level of suppression is not enough because someone could find the students in the American Indian/Alaska Native category by subtracting the sum of the other ethnicities from the total. Total (500) – All Available Ethnicity Groups (494) = American Indian/Alaska Native (6) |
|---|---|---|
| American Indian/Alaska Native | * | |
| Asian | 88 | |
| Black or African American | 52 | |
| Filipino | 37 | |
| Hispanic | 46 | |
| Pacific Islander or Hawaiian Native | 14 | |
| White | 95 | |
| Two or More Races | 96 | |
| Unknown / Non-Respondent | 50 | |
| Multiple Values Reported | 16 | |
| Total | 500 |
Table 2
Complementary Suppression: When only one subgroup is masked, suppress the subgroup with the next lowest count
| Ethnicity | Count of Students | By implementing complementary suppression someone could not use simple mathematics to determine the data for the American Indian/Alaska Native category. Masked values will be grouped together and displayed together in the “All Masked Values” category. In this example, “All Masked Values” would include American Indian/Alaska Native and Pacific Islander or Hawaiian Native. |
|---|---|---|
| American Indian/Alaska Native | * | |
| Asian | 88 | |
| Black or African American | 52 | |
| Filipino | 37 | |
| Hispanic | 46 | |
| Pacific Islander or Hawaiian Native | * | |
| White | 95 | |
| Two or More Races | 96 | |
| Unknown / Non-Respondent | 50 | |
| Multiple Values Reported | 16 | |
| All Masked Values | 20 | |
| Total | 500 |
Table 3
Complementary Suppression When Two or More Generated Subgroups Are Present
The example below illustrates how suppression is being implemented when two or more generated subgroups are present. This additional step was implemented to ensure that data for important populations are displayed wherever possible.
Unsuppressed: Counts by Gender
| Gender | Count of Students |
|---|---|
| Female | 25 |
| Male | 13 |
| Unknown / Non-Respondent | 5 |
| Multiple Values Reported | 17 |
| Total | 60 |
Table 4
Suppressed: Suppress counts less than 10 and suppress the value of generated subgroups for complementary suppression instead of following the rule above
| Gender | Count of Students | In this instance, the multiple values reported subgroup is masked even though it had a value greater than 10 and was not the next smallest group. This ensures that the Male subgroup is displayed. |
|---|---|---|
| Female | 25 | |
| Male | 13 | |
| Unknown/Non-Respondent | * | |
| Multiple Values Reported | * | |
| All Masked Values | 22 | |
| Total | 60 |
Table 5
Suppression for Secondary Gender Disaggregation in the Cohort View of Student Success Metrics
A secondary disaggregation on gender was included to the Cohort View. Data is displayed in order to determine possible equity gaps for additional subgroups (e.g. female veterans vs male veterans).
Unsuppressed Primary Disaggregation: Counts by First Generation Status – Prior to Secondary Gender Disaggregation
| First Generation | Count of Students |
|---|---|
| First Generation | 80 |
| Not First Generation | 75 |
| Unknown / Unreported | 5 |
| Total | 160 |
Table 6
Suppressed: Suppresses counts less than 10, and when only one subgroup is masked, suppress the subgroup with the next lowest count
| First Generation | Count of Students | Unknown/Unreported is masked since <10. The next highest subgroup, Not First Generation, is also masked for complimentary suppression rules. |
|---|---|---|
| First Generation | 80 | |
| Not First Generation | * | |
| Unknown/Unreported | * | |
| All Masked Values | 80 | |
| Total | 160 |
Table 7
Unsuppressed Secondary Disaggregation: Counts by First Generation Status further disaggregated by Gender
| Gender | First Generation | Count of Students |
|---|---|---|
| Female | First Generation | 45 |
| Female | Not First Generation | 30 |
| Female | Unknown/Unreported | 1 |
| Male | First Generation | 30 |
| Male | Not First Generation | 35 |
| Male | Unknown/Unreported | 3 |
| All Other Values | First Generation | 5 |
| All Other Values | Not First Generation | 10 |
| All Other Values | Unknown/Unreported | 1 |
| Total | 160 |
Table 8
Second Level Suppression: Within the secondary gender disaggregation subgroups, implement the standard suppression and complementary suppression logic for the primary disaggregation. Then sum together all masked values for display within the “All Other Values” category as “All Masked Values.”
| Gender | First Generation | Count of Students | Female Not First Generation is masked due to complementary suppression with Female Unknown/Unreported. Same case with Male First Generation and Unknown/Unreported. All Other Values First Generation and Unknown/Unreported are masked because values are <10. All masked subgroups are summed together and reported as All Masked Values in the All Other Values subgroup. |
|---|---|---|---|
| Female | First Generation | 45 | |
| Female | Not First Generation | * | |
| Female | Unknown/Unreported | * | |
| Male | First Generation | * | |
| Male | Not First Generation | 35 | |
| Male | Unknown/Unreported | * | |
| All Other Values | First Generation | * | |
| All Other Values | Not First Generation | 10 | |
| All Other Values | Unknown/Unreported | * | |
| All Other Values | All Masked Values | 80 | |
| Total | 160 |
Table 9