Identifying Time … When?

Why is when important?

Facts change over time ... and without information about time, then the data serve little valuable purpose.

It is just plain ridiculous the amount of data that exist without adequate time labeling.

Think of the number of photographs that have been taken of the situation in Haiti … millions of images, almost all without much reference to the exact time when they were taken. Perfectly nice people are taking pictures to record their concern about the disaster and the scale of the crisis … but without a time dimension these data are dangerous. Please also see the observation under Place … The Spatial Attribute

The natural sequences

There are many different time periods that may be used. The choice depends on the natural characteristic of what is being measured.

• By hour ... to show what happens at different times during a 24 hour period
• By day ... to show what happens from day to day
• By month ... to show changes month by month including seasonality
• Year on year ... to show how things progress over the longer period

The case of malaria control

Malaria is a parasitic disease of humans transmitted by mosquitoes. Humans live many decades. Mosquitoes live only days. The parasite life cycle is also days. When malaria control interventions are done on a human timetable they do not work very effectively … but with due respect for the biology of the mosquito and the parasite impressive performance is possible. This requires timely data … very detailed every day. Totally possible but rarely done because metrics rarely are used to optimize performance in the ORDA world!

For example issues like seasonality, or time of day, all have a bearing. Cause and effect may be identified by paying careful attention to the timeline. Time series trends are great indicators of progress ... or not. Time series are simple, clear and powerful. While it is possible to do advanced statistical manipulation ... simple and clear time series tables and charts work very powerfully as well.

A plot of a single parameter shows how this parameter has changed over time ... but in isolation does not show what might have been the cause of any changes. Plotting multiple variable may show something about cause and effect. While this may be done by simple visualization for a couple of variables, a more rigorous mathematical approach is needed for large scale multivariate analysis.

Seasonality

Many data series have an annual pattern arising because of the natural seasons. A time series about weather shows a seasonal pattern … and everything in the economy that depends on weather is influenced by this and the data reflect a similar seasonality.

When there are little data, there is a practice of using seasonality to misinform about progress and performance. Some of this is done because the data collectors and analysts are inexperienced, but some is done knowingly. It is a reprehensible practice.