Questionnaire and Conjoint

Questionnaire Design

The questionnaire is a structured technique for collecting primary data in a marketing survey. It is a series of written or verbal questions for which the respondent provides answers. A well-designed questionnaire motivates the respondent to provide complete and accurate information.

The survey questionnaire should not be viewed as a stand-alone tool. Along with the questionnaire there is field work, rewards for the respondents, and communication aids, all of which are important components of the questionnaire process.

Steps to Developing a Questionnaire

The following are steps to developing a questionnaire - the exact order may vary somewhat.

• Determine which information is being sought.
• Choose a question type (structure and amount of disguise) and method of administration (for example, written form, email or web form, telephone interview, verbal interview).
• Determine the general question content needed to obtain the desired information.
• Determine the form of response.
• Choose the exact question wording.
• Arrange the questions into an effective sequence.
• Specify the physical characteristics of the questionnaire (paper type, number of questions per page, etc.)
• Test the questionnaire and revise it as needed.

Required Information

To determine exactly which information is needed, it is useful to construct tables into which the data will be placed once it is collected. The tables will help to define what data is needed and what is not needed.


Question Type and Administration Method

Some question types include fixed alternative, open ended, and projective:

• Fixed-alternative questions provide multiple-choice answers. These types of questions are good when the possible replies are few and clear-cut, such as age, car ownership, etc.
• Open-ended questions allow the respondent to better express his/her answer, but are more difficult to administer and analyze. Often, open-ended questions are administered in a depth interview. This technique is most appropriate for exploratory research.
• Projective methods use a vague question or stimulus and attempt to project a person's attitudes from the response. The questionnaire could use techniques such as word associations and fill-in-the-blank sentences. Projective methods are difficult to analyze and are better suited for exploratory research than for descriptive or causal research.

There are three commonly used rating scales: graphic, itemized, and comparative.

• Graphic - simply a line on which one marks an X anywhere between the extremes with an infinite number of places where the X can be placed.
• Itemized - similar to graphic except there are a limited number of categories that can be marked.
• Comparative - the respondent compares one attribute to others. Examples include the Q-sort technique and the constant sum method, which requires one to divide a fixed number of points among the alternatives.

Questionnaires typically are administered via a personal or telephone interview or via a mail questionnaire. Newer methods include e-mail and the Web.


Question Content

Each question should have a specific purpose or should not be included in the questionnaire. The goal of the questions is to obtain the required information. This is not to say that all questions directly must ask for the desired data. In some cases questions can be used to establish rapport with the respondent, especially when sensitive information is being sought.

Sensitive questions can be posed in ways to increase response likelihood and to facilitate more honest responses. Some techniques are:

• Place the question in a series of less personal questions.
• State that the behavior or attitude is not so unusual.
• Phrase the question in terms of other people, not the respondent.
• Provide response choices that specify ranges, not exact numbers.
• Use a randomized response model giving the respondent pairs of questions with a randomly assigned one to answer. The interviewer does not know which question the person is answering, but the overall percentage of people assigned to the sensitive question is known and statistics can be calculated.

Form of Question Response

Questions can be designed for open-ended, dichotomous, or multichotomous responses.

• Open-ended responses are difficult to evaluate, but are useful early in the research process for determining the possible range of responses.
• Dichotomous questions have two possible opposing responses, for example, "Yes" and "No".
• Multichotomous questions have a range of responses as in a multiple choice test.

The questionnaire designer should consider that respondents may not be able to answer some questions accurately. Two types of error are telescoping error and recall loss.

• Telescoping error is an error resulting from the tendency of people to remember events as occurring more recently than they actually did.
• Recall loss occurs when people forget that an event even occurred. For recent events, telescoping error dominates; for events that happened in the distant past, recall loss dominates.

Question Wording

The questions should be worded so that they are unambiguous and easily understood. The wording should consider the full context of the respondent's situation. In particular, consider the who, what, when, where, why, and how dimensions of the question.

For example, the question,

"Which brand of toothpaste do you use?"

might seem clear at first. However, the respondent may consider "you" to be the family as a whole rather than he or she personally. If the respondent recently changed brands, the "when" dimension of the question may be relevant. If the respondent uses a different, more compact tube of toothpaste when traveling, the "where" aspect of the question will matter.

A better wording of the question might be,

"Which brand of toothpaste have you used personally at home during the past 6 months? If you have used more than one brand, please list each of them."

When asking about the frequency of use, the questions should avoid ambiguous words such as "sometimes", "occasionally", or "regularly". Rather, more specific terms such as "once per day" and "2-3 times per week" should be used.


Sequence the Questions

Some neutral questions should be placed at the beginning of the questionnaire in order to establish rapport and put the respondent at ease. Effective opening questions are simple and non-threatening.

When sequencing the questions, keep in mind that their order can affect the response. One way to correct for this effect is to distribute half of the questionnaires with one order, and the other half with another order.


Physical Characteristics of the Questionnaire

Physical aspects such as the page layout, font type and size, question spacing, and type of paper should be considered. In order to eliminate the need to flip back and forth between pages, the layout should be designed so that a question at the bottom of the page does not need to be continued onto the next page. The font should be readable by respondents who have less-than-perfect visual acuity. The paper stock should be good quality to project the image that the questionnaire is important enough to warrant the respondents' time. Each questionnaire should have a unique number in order to better account for it and to know if any have been lost.


Test and Revise the Questionnaire

The questionnaire should be pre-tested in two stages before distributing. In the first stage, it should be administered using personal interviews in order to get better feedback on problems such as ambiguous questions. Then, it should be tested in the same way it will be administered. The data from the test should be analyzed the same way the administered data is to be analyzed in order to uncover any unanticipated shortcomings.

Different respondents will answer the same questionnaire differently. One hopes that the differences are due to real differences in the measured characteristics, but that often is not the case. Some sources of the differences between scores of different respondents are:

• True differences in the characteristic being measured.
• Differences in other characteristics such as response styles.
• Differences in transient personal factors such as fatigue, etc.
• Differences in situation, such as whether spouse is present.
• Differences in the administration, such as interviewer tone of voice.
• Differences resulting from sampling of items relevant toward the characteristic being measured.
• Differences resulting from lack of clarity of the question - may mean different things to different people.
• Differences caused by mechanical factors such as space to answer, inadvertent check marks, etc.

Conjoint Analysis

When asked to do so outright, many consumers are unable to accurately determine the relative importance that they place on product attributes. For example, when asked which attributes are the more important ones, the response may be that they all are important. Furthermore, individual attributes in isolation are perceived differently than in the combinations found in a product. It is difficult for a survey respondent to take a list of attributes and mentally construct the preferred combinations of them. The task is easier if the respondent is presented with combinations of attributes that can be visualized as different product offerings. However, such a survey becomes impractical when there are several attributes that result in a very large number of possible combinations.

Fortunately, conjoint analysis can facilitate the process. Conjoint analysis is a tool that allows a subset of the possible combinations of product features to be used to determine the relative importance of each feature in the purchasing decision. Conjoint analysis is based on the fact that the relative values of attributes considered jointly can better be measured than when considered in isolation.

In a conjoint analysis, the respondent may be asked to arrange a list of combinations of product attributes in decreasing order of preference. Once this ranking is obtained, a computer is used to find the utilities of different values of each attribute that would result in the respondent's order of preference. This method is efficient in the sense that the survey does not need to be conducted using every possible combination of attributes. The utilities can be determined using a subset of possible attribute combinations. From these results one can predict the desirability of the combinations that were not tested.

Steps in Developing a Conjoint Analysis

Developing a conjoint analysis involves the following steps:

1. Choose product attributes, for example, appearance, size, or price.
2. Choose the values or options for each attribute. For example, for the attribute of size, one may choose the levels of 5", 10", or 20". The higher the number of options used for each attribute, the more burden that is placed on the respondents.
3. Define products as a combination of attribute options. The set of combinations of attributes that will be used will be a subset of the possible universe of products.
4. Choose the form in which the combinations of attributes are to be presented to the respondents. Options include verbal presentation, paragraph description, and pictorial presentation.
5. Decide how responses will be aggregated. There are three choices - use individual responses, pool all responses into a single utility function, or define segments of respondents who have similar preferences.
6. Select the technique to be used to analyze the collected data. The part-worth model is one of the simpler models used to express the utilities of the various attributes. There also are vector (linear) models and ideal-point (quadratic) models.

The data is processed by statistical software written specifically for conjoint analysis.

Conjoint analysis was first used in the early 1970's and has become an important marketing research tool. It is well-suited for defining a new product or improving an existing one.