The steps in Designing a Survey

1. Establish the Goals - What is the intent?
2. Decide on the Segment - Who the survey will target
3. Choose an Interview Methodology - Ways to conduct the survey
4. Craft the Questions - What and how to ask
5. Conduct the Survey - Ask the questions
6. Data Analysis - Comprehend the results

1. Establishing the goals

The first step in any survey is to focus on isolating the goals of the study. The goals set will help in determining what questions to ask, and how to ask them. Ambiguous goals may result in unhelpful and confusing results - or a major waste of time. Spend time on establishing the specific goals for the study.

2. Survey sample

There are two components in detemining the population segment to conduct your survey.

Sample size - how many people to interview

Practicality and logistical concerns limits the number of people that can be interviewed. Statistically, it is possible to represent a population based on a number of samples. The larger the number of people interviewed, the more likely the sample is likely to accurately reflect the population.

Target population - who to interview

Although larger sample sizes may appear more credible, do not be tempted to make up numbers. Instead, focus on the most representative samples of the intended population to obtain the answers that are relevant to your study. Irrelevant samples not taint the results, but are also a hassle to weave out.

3. Choose the interview methodology

There are several methods to conduct your survey (email or Internet surveys, face-to-face interviews and phone questionaires) each having their advantages and disadvantages. Consider the importance of the following aspects when choosing a method to conduct your survey:

• Ability for samples to see, feel or use the product
• Ability to include incentives to complete the survey
• Time/cost of conducting the survey (manpower, or logistical considerations)
• Ability to find, or control the sample completing the survey
• Speed of disseminating the survey
• Delay in receiving responses
• Comfort level to reveal personal or sensitive information over chosen medium
• Ability to gauge intangible responses
• Required literacy level of respondents

More to consider…

Speed	Email and website surveys, are usually quick to collect responses, while mail surveys have the longest delays. Email and website surveys are usually quicker to deploy.
Cost	Email and website surveys are the most affordable methods when conducting surveys for large samples, but excessive for small samples. Conversely, face-to-face interviews are the most expensive for large samples.
Reach	Though email, website and mail surveys can reach large populations, usually only the literate will respond. Face-to-face interviews may be more effective in reaching the general public.
Sensitive Questions	People are less likely to give answers to sensitive questions during face-to-face interviews. Email and website surveys provide some level of anonymity encourages people to answer such questions.
User Perception	Email and website surveys may include catchy media to engage the respondents. Face-to-face interviews however give users a hands-on opportunity with the product.

4. Craft the questions

Here are some points to note when crafting the questions for the survey.

a. Start with an introduction or welcoming message

Setting the mood right will encourage people to take your survey. State who you are, and why you are conducting the survey. People have to know who and why they are taking time to help you out.

b. Avoid biased questions

Do not field questions that may encourage respondents to respond in a certain way. Also, avoid emotionally charged words.

Biased question: Do you agree that the current traffic situation is a problem?

Unbiased question: Is the current traffic situation a problem?

c. Ask one question only

Questions must only serve to answer one question. To avoid implicit answers, do not ask questions that has more than one question in it. Instead, if field both questions separately.

Multiple questions: Do you drive a car or a motorcycle?

d. Write short and concise questions

Questions that are long and wordy may appear confusing to respondents, resulting in unreliable answers. Also, put questions across using simple English. As a guideline, compose questions that are less than 15 words long.

e. Do not force an answer

Allowing for “Don’t know” or “Not applicable” choices presents respondents with a chance to deny answering the question, instead of forcing them to settle with a “best guess answer.” Allow all possible, and mutually exclusive answers.

f. Ask the simple ones first

Leave the more lengthy, difficult, or sensitive ones to the back. The more questions people answer, they are less likely to give up. Users may give up answering the survey if they are forced to provide sensitive information right up in the beginning.

4.5 Pre-Test your Survey

Pre-testing your survey will allow you to better identify a problem. Send out your survey as a test to a small number of people before publishing it to an audience. After they have completed the survey, brainstorm with them to understand their problems with answering any of your questions. Ask if they were able to understand the question correctly, and how they felt about the question. Questions which usually garner the same answers may indicate that the question was crafted wrongly.