- What are disaggregations and what information do I need to define them?
- How do I administer disaggregations?
- What are "cross-disaggregation" and "parallel disaggregation"?
- How many disaggregations can I add?
- Parallel disaggregation: What is a primary disaggregation?
- What is impacted by the primary disaggregation?
- What if the primary disaggregation selected isn't the one I want?
- What happens when I disable a disaggregation for an indicator that's already reported data?
Disaggregations are a way to break down indicator definitions into meaningful categories so you can report data for those categories. For example, # lawyers trained could be disaggregated by sex, allowing you to report separate totals of males and females.
To create or edit a disaggregation, you will need to know:
- What you want the disaggregation to be called. This is how it will appear in dropdowns, MultiMatrix, etc. Common titles are things like Age, Sex, Gender, Ethnicity, Training Type, etc.
- The categories for the disaggregation. For an age disaggregation, the categories might be "under 18", "18-65", and "over 65".
DevResults requires all data to have a geographic disaggregation in the form of assigned locations or administrative levels. You do not need to create separate geographic disaggregations on the site. To find out more, please see Geographies.
Disaggregations can be created, edited, and deleted by going to Program Info > Disaggregations (under the Data Definitions section).
For individual indicators, you can add, remove, and reorder disaggregations in the Disaggregations section of the Indicator Definition tab.
When you assign more than one disaggregation to an indicator, you will have to decide if you want these disaggregations to be cross-disaggregated or parallel disaggregated. This tells the system whether it should combine the disaggregation categories in data entry and display (cross), or whether it should treat them as completely independent and separate categories (parallel):
- For parallel-disaggregation, the total result is reported for each disaggregation, split by its disaggregation categories. The example below shows a result reported for males, females, minors, and adults.
- For cross-disaggregated data, the disaggregation categories are combined to make unique definitions. The example shows a result reported for male minors, male adults, female minors, and female adults.
Technically speaking, for direct entry indicators, DevResults will support many disaggregations provided the number of combined disagregation categories — or columns in a data entry page — is less than 300. (Data table and formula indicators are exempted from this limit.)
For parallel disaggregated data, disaggregation categories are additive, so the sum of each disaggregation's number of categories determines the number of columns in a data entry grid (where rows are determined by the number of geographic disaggregations). For cross-disaggregated data, disaggregation categories are multiplicative. To illustrate, if a direct entry indicator were parallel-disaggregated by sex (male and female — 2 categories) and work status (employed, unemployed, retired — 3 categories), there would be only 5 columns (2 + 3 = 5); if that same indicator were instead cross-disaggregated, it would be 6 columns (2 x 3 = 6).
The difference in this example in the number of columns and cells that must be inputted is negligible, but adding many disaggregations with many categories each can quickly multiply that difference beyond 300 (adding just one more four-category disaggregation would quadruple the number to 24). With a data entry table that large, the page itself becomes unwieldy to scroll and entering data accurately is impractical. For this reason, we limit the number of disaggregation categories to 300 as a practical matter, although users will have difficulty well before that limit.
Generally speaking, while it is considered good practice to have rich data that is sufficiently disaggregated to address important questions, it is not good practice to collect excessively detailed data without a clear analysis plan to use such data. The burden placed on data collectors exceeds the value gained, and the intrusion and risk associated with amassing sensitive data on vulnerable populations becomes untenable.
When you have more than one disaggregation assigned to an indicator and you have chosen to parallel rather than cross-disaggregate, the system has to pick one disaggregation as the "default" for providing summary data for the indicator. We don't want to add the separate disaggregations together, because that very well might double or triple count things. And we don't like to pick the highest total by default, since this might vary between activities and/or reporting periods. So we use the first listed disaggregation as the primary disaggregation. To change your primary disaggregation, you can drag your preferred disaggregation to the top of the list.
For parallel disaggregated indicators, you will only see the primary disaggregation totals in these situations:
- MultiMatrix default totals for the indicator (e.g. when no disaggregations are explicitly selected)
- The indicator's Actuals vs. Targets graph
- The indicator's Reports tab graphs
In both situations, you will only see totals displayed when data has been entered for the primary disaggregation. For indicators where you only have data for the secondary disaggregation(s), you won't see any data for this indicator in the Actuals vs. Targets graph, the Reports tab graphs, or the default view in MultiMatrix. (In MultiMatrix, you can see those totals if you check the box for the secondary disagg).
In the Disaggregations table, hover your mouse over the up/down arrow to the left of the disaggregations. Your mouse will turn into arrows and you can click and drag to reorder this list:
Once you've reordered the disaggregations, the top one listed will become the primary disaggregation and will be used in all summary screens.
This depends on whether the indicator is cross-disaggregated or parallel-disaggregated
For parallel-disaggregated indicators, marking a disaggregation as disabled only does two things:
- It removes the disaggregation as an option in Matrix so that you cannot display it any more as a row or column--it basically appears as if the disaggregation never existed.
- In MultiMatrix, by default it will be hidden.
- If this is a direct entry indicator, it will remove this disaggregation and its categories from all reporting period entry grids and templates.
Note that for parallel-disaggregated indicators, no data deletion occurs when you disable a disaggregation. If you re-enable the disaggregation, all data previously reported against that disaggregation will reappear in MultiMatrix, etc.
For cross-disaggregated indicators, marking a disaggregation as disabled will delete all data previously reported for that indicator. It forms a fundamental change to the entry grids, etc., and--like other fundamental definition changes--we will give you a warning that you are about to delete all your data.
Didn't answer your question? Please email us at email@example.com.