About Enhanced Anonymity
The Anonymity Threshold determines how many responses must be included for a given data point before it can appear in your dashboard. This data point can be as broad as a widget or as specific as a bar within a chart.
There are also settings for Enhanced Anonymity. While the Anonymity Threshold is a great way to protect the privacy of employees’ responses, Enhanced Anonymity adds additional layers to filters and widget breakouts that can improve anonymity in certain use cases.
For example, let’s say Barnaby’s team has 15 people. When we look at the engagement scores for his team, we don’t really know how each team member responded to questions about their manager’s effectiveness. However, let’s say there are only 2 women on his team. Regular Anonymity Thresholds will ensure we can’t see the women’s responses directly, however, if we add a gender filter, we can make a pretty good guess what each of the women on his team had to say. Enhanced Anonymity senses disparities of this kind. It ensures that data from groups that don’t meet the Anonymity Threshold is combined with the next smallest group to hide their answers when breaking out data or using filters.
Enabling Enhanced Anonymity
You can enable enhanced anonymity when from the Dashboards tab, or from your dashboard’s anonymity settings. Enhanced Anonymity can be turned on and off at the level of the dashboard.
Enabling Enhanced Anonymity from the Dashboards Tab
- Go to the Dashboards tab.
- Click the dropdown next to the desired dashboard.
- Click Edit Dashboard.
- Determine the Anonymity Threshold. This is the cutoff for non-open text data. This determines how much data has to be collected for a field before it is no longer grouped for anonymity, or how much data must be collected overall before the dashboard can display data in its widgets.
- Select Enable Enhanced Anonymity.
- Click Save.
Enabling Enhanced Anonymity from the Anonymity Tab
- While viewing your dashboard, click Settings.
- Go to the Anonymity tab.
- Enable Turn on Enhanced Anonymity.
- Click Save.
There are some fields participants can be identified by, to which you will want to apply Enhanced Anonymity, and some fields by which participants can’t be identified, to which you will not want Enhanced Anonymity applied. For example, tenure, gender, and the team to which someone belongs can all be used to figure out who they are. However, questions asked in an Employee Experience survey are almost always non-identifiable, with the exception of demographic questions like language, location of office, and age.
When fields are marked as identifiable, the data from groups that do not meet the anonymity threshold will be combined with the next smallest group in order to protect the identities of the respondents. When fields are not selected and are marked at unidentifiable, this grouping will not happen unless you filter or break out by an identifiable field.
In order to edit which fields are marked identifiable and which are not, you will want to do the following:
- Go to Settings.
- Select Anonymity.
- Select fields to mark them as identifiable. This means Enhanced Anonymity settings apply. Deselect fields to mark them as non-identifiable.
- Click Save.
- To return to the original configuration and remove all your changes, click Reset.
Example: In our dashboard, we did not mark engagement questions as identifiable, because they are not demographic and cannot be used to identify their respondents in any way.
Let’s say the dashboard’s threshold is 5. If we made a table displaying how employees responded to a non-identifiable field like “I feel proud to tell people where I work,” answers will not be grouped. See below how “Strongly Disagree” appears, although it only has one response.
Note that the total number of responses in the widget must still meet the threshold. This graph has a total of 90 responses. If it had fewer than five, the graph would be blank because the default behavior of the anonymity threshold is to hide data from widgets that do not meet the threshold.
Setting the Primary Data Source
Otherwise, dashboard filters include data from all data sources in the dashboard data, which means historical data may be included in response counts, and can skew anonymity groupings. For example, if current and historical results are counted for a small team rather than just the current year’s results, the small team seems bigger than it really is, and may not fall below the anonymity threshold. Limiting the filters to the primary data source (i.e., the current project or current year’s data) resolves this issue.
Once you’ve enabled Enhanced Anonymity and added a filter to your dashboard, filter options with results below the threshold will be combined with the next smallest option in the filter before any selections are made. If you have just one group that falls below the threshold, this will be combined with the next smallest group, regardless of whether the next group meets the threshold or not. This is to ensure that even if only one group doesn’t meet the threshold, their data is protected.
As filters are added or removed, enhanced anonymity will take these into account and change the groupings accordingly.
Example: In the screenshot below, we are trying to filter by department. This is a small company, so Finance and Human Resources have very small teams, below the Anonymity Threshold we have set.
You’ll see that Finance and Human Resources are under the header Grouped for Anonymity. If I try to select just one, they will both automatically be selected. If I try to deselect one, both are deselected. This prevents users from figuring out the values of below-threshold groups.
Once you turn on Enhanced Functionality, this feature will interact with your Dashboard’s Anonymity Threshold to determine the data displayed in widgets where data has been broken out into certain groups. This includes Simple Charts where an X-Axis Dimension has been defined, Demographic Breakout Widgets, Heat Map Widgets, and any other widget configuration that isolates groups that may be smaller than the Anonymity Threshold.
The exceptions to this rule include widgets broken out by org hierarchy. Some widgets (Heat Map, Demographic Breakout) support a One Level Below breakout that shows data for each child unit of the currently selected unit in the Org Hierarchy filter. Other widgets (Response Rates, Comparison, Bubble Chart) support drilling down into the hierarchy, showing data for each unit and allowing the user to select them. For these cases, no grouping is applied to breakouts.
Example: Let’s pretend the Anonymity Threshold for the following example is 5.
In the screenshot below, we have set the X-Axis Dimension of a Simple Chart to be the country where the employee’s office is located. Japan, Mexico, and Poland have very small offices, to the point where at least two of them have fewer than 5 people working there. As a result, their responses have been combined.
If you were to change the metric to an average engagement score or an NPS, enhanced anonymity would prevent you from figuring out the smallest office’s data by not allowing dashboard users to isolate that office’s data. This is useful in cases where, for example, we do not want the ratings of each member of the smallest office to be easily calculated.
The more you break out data, the smaller each category can get, and the more categories that get grouped together under anonymity. And because there are now two dimensions to the breakout, there may be different combinations of categories that need to be grouped together for anonymity. Thus, categories grouped for anonymity will be labeled Grouped for Anonymity, and you can hover over them to determine what specific categories were grouped.
Example: This dashboard has a threshold of 5. In the graph below, we broke out the countries our employees work in by attrition risk. We can see a light green block for Mexico in the Low Risk bar, but not in the High Risk bar.
When we highlight the Grouped for Anonymity blocks for each bar, we discover that there’s a difference: Mexico was grouped with Japan and Poland in the High Risk bar, but only Japan and Poland were grouped in the Low Risk bar.
Look at the screenshot below. In High Risk, Mexico has no individual data ( – ), but Japan + Mexico + Poland shows data, since these are grouped together for anonymity (5). Under Low Risk, Mexico meets the threshold, so it does not need to be grouped, and shows individual data (17), while Japan + Poland are grouped together (5).
You’ll also notice that in the legend, anonymity groups are labeled as Grouped for Anonymity, not a compound name. This is to take into account how groupings might change based on how multiple breakouts interact, and to prevent labels that are too long.
Response Rates Behavior
Response rates display how many responses you received, and what percentage of your participant list has completed the survey. This kind of data is reported by Participation Summary and Response Rate widgets.
By default, the dashboard considers response rates to be identifiable information. This means response rate data can be used to identify participants, and so to protect your participants, response rates are subject to being grouped by anonymity.
Example: When you create a Response Rate widget, you can add a Field to break out the widget. Below, we have broken out our response rates by country, which has resulted in Australia, Germany, and Mexico being grouped for anonymity.
Thus, response rate widgets display the same breakout behavior other widgets do.
If the response count is below the threshold, you will not see data in the Participation Summary widget.
Org Hierarchy Filters
If the invited count (i.e., the number of direct reports and the expected count in or below the unit) is less than the anonymity threshold, we gray out org hierarchy units that fall below the threshold.