Understanding Data

Why we need data

Data is used to construct our knowledge, actions and arrangements. In order to gather meaningful data we need matching concepts and some awareness of the context. A height of 170cm may be "tall" or "short" depending on the person's age, gender, race, group....

The application of data may, or may not, be problematic depending on the nature of causal relationships (if any) involved. For many physical phenomena, cause and effect are consistent over place and time. Thus data can be used to make reliable predictions and transfer best practice.

In most social phenomena, the relationships between cause and effect are not consistent over place and time. This fundamental reality is often masked by the fact that some observations can make sense in retrospect (after the event). The thinking error involved is, "because something can now be explained it could have been predicted before it happened".

Rather the following are often true if the phenomena are complex or chaotic:

• cause and effect may not be related at all in any meaningful way
• cause of effect may be remote from each other in place and time
• cause and effect may be related but also inconsistent over place and time - repeated experiments give significantly different results, or small differences result in very different results
• despite our best efforts, outcomes are unpredictable, messy

Data and Complex Phenomena

In complex phenomena such as social activity it is common for patterns to emerge in/from the interactions of the agents. That is, the outcomes are better understood as patterns rather than "products".
[Note: It is more appropriate to use the term 'product' in relation to the outputs of a production process, one which can be properly understood in terms of Input-> Process-> Output (product)]

The use of data in relation to complex phenomena is to enable us to identify patterns, trends and opportunities rather than to manage our endeavours as production activities. Understanding the difference between production and emergence is critical in field such as education.

Of course education and similar endeavours uses processes but they are typically iterative rather than linear, as is typical of production processes.

The implications include

• Fail-safe (fool proof) approaches are rarely available
• Best practice is rarely a valid assessment despite 'proven' examples
• Some approaches may generally work better than others but there are always exceptions
• 'Transfer' of successful practice is not a simple matter - practices need to be continually constructed and reconstructed
• It is best to try safe-fail experiments - small scale changes that can be easily reversed if they fail to deliver the intended outcomes
• While using data may be better than just guessing, it is much better to use 'knowledge' based on experience and relationships informed by agreed data
• Each of the parties has unique knowledge that is critical to the current success of any working relationshi