Establishing a data collection plan - Part two

3/ Define the sample size

Your data will need to be of a realistic sample size to be valuable – for example if capturing the cost of an inventory item – will this be captured over a week, month or year – ensure that the sample size is representative (i.e. not too small). The data should also be recent to ensure that conditions have not changed since the data was taken.

There are various benefits for analyzing a sample of data rather than the whole – most importantly it is generally quicker. There are various statistical techniques available to determine sample size and it’s worth trying some out to ensure that your approach is credible.

4/ Be sure to understand the operational definitions of the data

One of the worst pitfalls in data collection is misunderstanding the data, this can often occur because of the way that the data is collected and/or recorded – your data needs to be common and consistent – common errors occur where standard collection procedures are not followed which can result in different data being recorded. Watch for things such as:
* Ensuring that the recording process is common
* The team understand the business process that generates that data. For example – where measuring supplier delivery schedule adherence – is the delivery time the day the package arrives in the warehouse or the day that the package is “booked in” on the ERP system (i.e. accepted) – make sure that the approach is consistent and that the team gathering the data understands both the results and the originating business process.

5/ Develop a data collection method and assign owner(s)

Typically many improvement projects rely on data obtained from ERP systems, whilst this may rely on an IT resource, it’s generally a straightforward process to obtain the data – there are of course challenges to be considered such data integrity.

Where ERP data is not available or appropriate – methods such as collection sheets can be used – this generally takes the form of a checksheet that either a change agent can use or accompanies a process. A basic checksheet can take the form of a grid where people count the occurances of what is being recorded/analzed. Where checksheets may be used by multiple individuals ensure that the form cannot be misinterpreted (i.e. different users may record different results) and is easy to use.

Finally – ensure that a member(s) of the project is tasked with collecting the data – taking into account appropriate controls and knowledge of the plan - ensure that no personal bias can be implied. There may be training that is required for the data collectors and you may wish to trial the data collection methods that you have established.

6/ Collect the data

Execute the collection plan and gather the data. Depending on your process data could be obtained immediately or over time – be sure that there is somewhere to store the data and that it is easily retrievable.

7/ Present Findings

Once collected data may go through some form of analysis e.g sorting, summarizing – trend analysis etc. Prior to being presented data is usually used to present an argument and there are a variety of ways of presenting data either as part of a list or diagrammatically.

Summary

Any data that is used within an improvement program should ensure that it is representative of the process and be sufficient to enable conclusions and descions to be made.

A data collection plan can help achieve that goal by providing a clear, timely and accurate information.

Whilst the time taken and resource requirement of a plan may vary from business to business the basic premise remains the same in that to miss it out may cause projects to either produce inaccurate findings or to fail completely.