Guidelines for conducting an online survey (Overview)

Online surveys have significant cost advantages over paper-based forms and telephone interviewing. Getting the full benefit from your web survey project takes some planning but is not particularly difficult. Below are a typical steps in the online survey process. Some recommendations seem obvious, but all issues listed are based on experience.

Start with the end (reports) in mind

Figuring out what you want to know is the best way to develop a strategy for the project.

• Do you want to compare average rankings across questions and groups of questions?
• How will the data be broken down? By Department, Geographic location, Age, Gender, etc.
• How much analysis do you want to do on "Write in" text comments?
• Who will be viewing the reports, and how will they be shown?

What questions to ask

It's easy to ask too many questions. After all, you're going through the process of making the survey, why not get as much data as possible? Focus on the important questions that give meaningful answers.

• Long surveys that force people to answer redundant questions have a lower completion rate!
• Start with a full question set and "prune" out the questions that don't have an obvious purpose or sound too much like other questions.
• Avoid asking questions that require long, complex instructions. People don't always read instructions, and sometimes it is the exception rather than the rule. More than 1-2 sentences will get ignored.
• Have multiple people proofread your questions before you start building the web survey.
• Use skip patterns (branching) to move people past questions they do not need to answer, or are unqualified to answer.

Your survey sample or respondent population

In the case of an employee survey, the population you're surveying is fixed, and the only question you need to address is which employees will participate. For other types of surveys, determing (and contacting) the appropriate participants requires more effort.

• Determine how many responses you need to collect. Refer to statistical "sample size calculators" if you are using your sample data to infer characteristics of a population.
• Estimate a response rate. Response rates vary greatly from one survey to the next. It is VERY important to determine factors affecting your response rate and come up with a realistic projection.
• For many surveys, a high response rate is what determines the cost/benefit ratio of the project.
• Develop a plan for increasing your response rate during the survey in case initial returns are less than expected.

Contacting your respondents

How you contact potential participants and motivate them to complete the survey is the biggest factor in raising your response rate.

Know how people will be contacted. Email, telephone call, in person requests, newsletters and web site links are all effective. Which methods work for you depends on the information available and the people you're attempting to survey.

• Is the survey participation mandatory?
• When participation is voluntary, remember that you're asking for a favor when you contact participants.
• What will motivate people to participate in your survey? Busy professionals are less likely to spend their time giving away "free" information.
• Multiple contacts may be necessary to get a response. Consider sending reminder emails and telephone followups to people that don't respond initially.
• Provide a tangible incentive to raise interest. This doesn't need to be complex or expensive. A drawing for a $50 gift certificate may be enough.

Building a web survey form

The web survey form should be attractive and easy to read. Respondents shouldn't have to guess about what a question is asking or how to respond. Questionnaire design is one of the services offered through Survey-Hosting.com.

• Avoid using "free form" write in questions whenever you can.
• Set up Likert Scale (weighted score) questions to gather attitudinal data.
• Use single choice questions whenever you ask people to "pick one response".
• Use Check All that Apply questions only when you want to gauge overall exposure and interest.
• Realize that "specify other" write-ins complicate reporting and may require recoding during analysis.
• Whenever possible, provide lists of responses for locations, departments, agencies, and other organizational breakdowns.
• Plan a testing period where a few respondents take your survey and give you feedback on the organization, wording and usability. Make adjustments before the official launch.

This list is meant to be a general overview of preparing for a web survey. Each type of survey and population has specific requirements and considerations. The experienced professionals at Survey-Hosting.com have the knowledge to work through the details of a new project.

Request a proposal or quote for survey administration, planning design or management services.