Data collection is often seen as the most crucial activity in the statistical cycle because anything that affects data collection can significantly impact the quality of data. For example, large scale non-response (survey participants’ failure to complete the form or specific questions in a form) can lead to inaccurate estimates and thereby incorrect conclusions.
Data collection consists of two interrelated activities: managing respondents and managing the collection process.
Respondent management refers to the management of relationships with respondents in order to maximise their participation in surveys. Collection process management refers to providing the right climate, tools and support strategies to achieve acceptable quality and timely data.
4.1 MANAGING RESPONDENTS
The collection of accurate, timely and reliable data depends on the goodwill and cooperation of respondents. This can be maximised by assuring respondents that the data will be used only for the purpose for which it is collected and by assuring the total confidentiality of their response.
4.1.1 Initial Contact
It is strongly recommended that respondents are contacted using an introductory letter or other means to advise their selection and seek their participation and cooperation. The introductory letter should also briefly state the purpose of the data collection, the intended use of their data and assure them of the confidentiality of their responses. Contact details of the collection agency should also be provided for respondents to ask for assistance or further details. The letter should also contain the actual or likely dates of the survey or data collection.
An introductory letter is an effective method of preparing respondents, regardless of how you plan to collect the data. Introductory letters are often referred to as "warm contact" as they help to manage respondents expectations.
Privacy issues can also be covered in the introductory letter. See Chapter 11 – Confidentiality and Privacy for further information on privacy and confidentiality issues in statistical collections.
4.1.2 Collecting the Information
The timing of data collection can impact on the quality of data and the response rate. There are times of the year when it will be difficult to contact respondents or for respondents to complete your survey. For example, families may be away from home during school holidays or businesses may be closed during Christmas, New Year holiday season. Attempting to collect data during these periods period is likely to yield lower response rates. Business data collection should make allowances for the end of the financial year period.
In telephone or personal interviews, you may not be able to contact the respondent at the first attempt. A contact strategy should be in place that allows for a sufficient number of calls or visits made to contact each respondent. Higher contacts will usually increase the level of response. You may need to allow for up to 5 phone calls or visits to maximise your response rate.
After sending out the questionnaire or survey forms to respondents, further contacts may be necessary to ensure that data is received within the specified timeframe or to seek clarifications on data provided by respondents.
In a mail based collection, one or more reminder letters may be sent out. These letters should state the purpose of the letter and details such as the due date for return of forms etc. Such letters can be sent two to three weeks apart depending on the timeframe and urgency.
When completed forms are received, you may notice missing values, unexpected responses or other errors. These types of errors should be detected early in the data processing period by manual or automatic means. Time and resources should be allocated to rectify such errors depending on their likely impact on the data. Some errors may not need correction or can be corrected easily. Some errors, however, may warrant contacting respondents to clarify or confirm the responses. While such follow-up contacts are necessary in some cases, these contacts should be kept to the minimum.
4.1.4 Dealing with Complaints
No matter how well you have planned your collection, it is likely that you will find some respondents unhappy with certain aspects of your collection or collection process. You should have a clear strategy and process to deal with unhappy respondents and to resolve with their complaints.
All staff with direct respondent contact should be trained in complaints handling procedures. Respondents should be also advised on their options if they are not satisfied with the reply they receive or if the complaint could not be resolved between the respondent and the collection agency.
4.2 MANAGING THE COLLECTION PROCESS
The quality of data depends on the knowledge, understanding and skills of staff involved in the collection process. Staff should be trained in advance or on an ongoing basis on procedures for which they are responsible. For example, staff responsible for follow-up should be trained in relationship management skills including complaints handling. Likewise, data processing staff should be trained in identifying errors and inconsistencies and taking corrective actions. For staff handling client queries or complaints, a list of frequently asked questions and suggested replies can often help.
Training should also include general issues such as the purpose of the survey, the scope and coverage, an overview of the sampling methodology and familiarity with questionnaire or survey forms.
Role specific training for staff might include:
· field procedures
· handling queries and complaints
· data quality control
4.2.2 Using Codes for Respondents (Response Codes)
While you may develop the survey frame, methodology, sample etc., with great care, real world factors and situations may interfere or influence the collection and outcome. For example, some selected businesses may cease operation, households may not be contactable or people may be ill. There may be response problems such as very low levels of response, or partially completed forms.
Developing a simple coding system for responses will allow you to manage and analyse these problems more effectively for follow-up, reporting or other purposes. To develop a set of codes simply identify the basic problems likely to occur and assign a code for each problem or category of problems. For example, you may decide to code a normal response having no problems as 1, businesses that ceased operation as 2, businesses which refuse to respond as 3 etc.Such a “response coding” system can be also useful in data processing. For example, a business which has ceased operating should not have any employees; these businesses could be removed from inclusion in editing if a specific code was assigned to non-operating businesses.
4.2.3 Systems to Support the Collection Process
Systems to statistical collections can vary significantly depending on the collection method and available resources. They can be quite complex integrated purpose built systems, or as simple as a spreadsheet. A separate system for the despatch and receipt of survey forms is an example of a large and integrated collection support systems. Regardless of the system used, the receipt of survey forms and responses should be regularly monitored so that steps can be taken if necessary to achieve a higher response rate and greater data quality. Regular monitoring can also assist in prompt follow-up action with difficult respondents.
Where available, the collection support systems can be supplemented by other administrative databases in an organisation. For example, some organisations may have an administrative database containing a history of contacts with each of their clients. This would be useful to the collection support, especially for administrative collections.