Ideally, organizations administer surveys to solicit employee opinion on various initiatives and activities so that positive improvements can be made. All too often, however, the gathered information is sitting in files or databases without any further use One reason might be the lack of analytical capabilitywithin the organization, but a similarly plausible reason may be the inadequate quality of the data.

Accenture research of the use of analytics in organizations has found that there is often a lack of good data gathered.

There are several survey design steps that can be followed to substantially improve the quality of the resulting data that will enable the organization to make informed decisions:

•  Establish the purpose of the survey. Does the management want to assess the outcomes of some initiative or learn about employees’ perception of/attitude toward some organizational practice, or solicit ideas for improving some organizational aspect? Most importantly, will the survey results be used to inform any decision, policy, or practice? Purpose will determine the format and contents of the survey.
• Clarify the ultimate big-picture question to be answered at the end. In other words, always start with the desired end state and then work backwards. What exactly do you want to learn? What concepts are involved? How are they defined? For example, if the question that needs to be answered at the end is “How to gain employee trust in company’s leaders?” start with clarifying what exactly trust means and how it looks like. Relevant literature search as well as consulting organizational members can help here.
• Identify observable and measurable aspects related to the end-question. For example, in the case of employee trust, identify the antecedents or conditions that may lead to trust. Knowing how employee trust manifests itself helps in selecting the most appropriate antecedents.
• Make sure all concepts of the survey and the links among them are theoretically justified. In our case of employee trust, the selection of antecedents should not be arbitrary. They must be empirically proven in previous research or at least based in theory. In the latter case, one objective of the survey will be to test the relationship between trust and the selected antecedents. This step involves an extensive literature search.
• Identify the measures of the survey concepts. When the end-state and its antecedents are clarified, search the literature for existing measures, that is, questions or statements that have been proven to explain the given concepts. It is possible that such measures do not exist or the existing measures have been developed for industries and situations which are very different from those of your organization. Then new measures need to be developed through broad consultation with organizational members. Developing new measures is a mix of science and art. Nevertheless, the process involves defined steps so that the resulting measures are reliable and valid.
• Decide what kinds of statistical analysis will be performed with the obtained data. That’s right, the data might not be available for several months at this stage yet, but depending on the future analysis, the measurement scales need to be selected right now. Inappropriately formatted data might not allow the researchers to perform the desired analysis, be it descriptive or inferential.
• Decide upon the measurement scales. When the analytical framework is developed, the most appropriate scales must be selected. Survey items can be rated on a Yes/No scale or five point scale, for example. Also multiple-choice lists or ranking lists are possible to use.
• Review the overall formatting and organizational aspect of the survey. People do not like to fill out excessively long surveys. Thus, they should not be longer than absolutely necessary. How will anonymity and confidentiality of individual responses be ensured? More about survey design can be found in Dean Wiltse’s article The Employee Survey: What’s In It For Me?
• Pre-test the survey and do the necessary corrections. Ask some five to ten employees who have not been involved in the survey development to fill out the survey and provide feedback on the clarity of items, ease of filling out the survey, etc. There probably has not been a survey development genius whose work did not need a single correction as a result of a pre-test.
• Choose the most appropriate survey administration mode. Paper based surveys are still used, but the conveniently available online tools make the process much more effective.
• Last but not least, obtain management commitment. If this is going to be a corporate-wide survey, senior executive commitment will open the doors to the necessary resources such as IT support, ability to use the intranet and employees’ time during working hours to fill out the survey. A message from senior executive team will also give the survey credibility and encourage more people to respond. Employees will believe that their opinion will be heard. It will be easier to put the survey results to use as well.

This entry was posted on Tuesday, July 20th, 2010 at 1:52 am and is filed under Survey research. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.