Answering in Happyforce is voluntary. That is a design decision â forced participation produces compliance answers, not truthful ones â but it introduces self-selection: those who answer may not be representative of those who do not.
We do not hide this. We mitigate it in three ways.
1. Participation is a first-class metric
We monitor it by team and segment, not only at company level. An aggregate participation of 45% can hide one team at 80% and another at 10%, and those two teams do not read the same way.
2. Sample size travels with the data
Every aggregate is reported alongside the number of responses behind it. A Score computed on 4 responses and one computed on 40 do not deserve the same confidence, and the dashboard tells you so.
3. Silence is signal
Changes in participation itself are information. Disengagement usually shows up as silence before it shows up as a drop in the scores: people stop answering before they answer badly. A participation drop in a specific team is frequently the first indicator that something is going on.
What this means for you
When you look at your data, always look at two things at once: the number, and how many people are behind it. And treat a drop in participation for what it is â data, not a reporting problem.
For reference, sustained median participation across the organisations we measure is 41%, with the top quartile above 56%.Keep reading
Anonymity: why it is a condition of validity â the reason answering is safe.
Why should I pay attention to participation? â how to raise it in practice.
Limitations of the model â self-selection, stated plainly.