Real-time, which means daily

Every StakeholderWatch study collects a fresh sample every day of the year, at the same rate, with the same instrument. We call this real-time research and we mean it literally: a change in what people believe or intend appears in the data on the day it happens, and the estimation that runs every evening carries it onto the dashboard the next morning.

Why daily and not weekly or monthly

Daily collection is not chosen for speed alone. It is the condition for causal estimation. A structural break can only be dated to a day if there is data for every day. An event can only be matched to its effect if the series is continuous around it. A coefficient can only be re-estimated on a rolling window if completes arrive at a steady rate, so that the window always holds a comparable sample. Waves, however frequent, cannot do these things: between two waves there is nothing to date, nothing to match and nothing to re-estimate.

What it looks like on the dashboard

The estimation window is a trailing quarter by default; a year, 180 days, a month or the last 300 completes can be selected. Whatever the window, the model behind the dashboard is re-estimated every evening on the completes inside it. A structural break is reported with its estimated date and the average before and after, and is confirmed by further readings outside the old range before it is shown. The playhead on the timeline lets a user set the dashboard to any past date and read the model as it stood that evening.

Since when

StakeholderWatch has collected daily data for its clients since 2013, several of them continuously since 2015.