bit more tricky: select on from set or enter own value in "other" box
For each point in time, the aggregation is determined by
derive N as the number of valid responses at that point in time (# different responders with non-expired answers)
derive the set of possible answers-values (open question: should we include answers that don't appear at this timestamp?)
for each answer-value count the number of responders whose state is equal to this value. Note: the answer of most responders has been given earlier in time.
Display:
State diagram (left side)
Display a ring graph showing the number of responders in each active state.
Development diagram (right side)
x-axis: time (as selected by the time-slider
y-axis: lines for each answer-value
Option 1 it might be useful to have a display option that switches from absolute numbers to percentages.
Option 2 it might also be useful to be able to select different graphing options (in the "Display option" dialogue), e.g.
stacking or overlay lines
smoothing/interpolating or hard corners (lineTension?)
Examples:
For each point in time, the aggregation is determined by
Display:
State diagram (left side) Display a ring graph showing the number of responders in each active state.
Development diagram (right side) x-axis: time (as selected by the time-slider y-axis: lines for each answer-value