lecture 4 MARK 360

Warmup-up questions examples:

"How often do you go shopping for casual clothes?" "On what days of the week do you usually shop for casual clothes?"

"Heart of the survey" questions examples:

(Broad question): What improvements can be made to enhance your satisfaction with our company? (Specific question): How do you evaluate the timeliness of our service?

Survey question development principles

-Adopt clear, correct, and concise wording -Avoid common problematic words All (and its variants—always, each, every, never, nobody, only, none, etc.) . And—usually indicates multiple questions are being asked . Like—usually indicates an example is being introduced into the question . Not—avoid it when yes or no questions are being asked If—usually employed with skip patterns, may lead to confusion

Ratio scale

-Each value can be treated as a unique category (as in a nominal measure) -Different values have order of magnitude, such as greater than or less than or equal to (as in an ordinal measure) -Basic mathematical procedures can be conducted with the values, such as addition and division (as with an interval measure) -It allows the value of true zero. An example of a ratio measure is someone's annual income. It may be expressed as either a fraction or percentage

Survey question development principles

-Go through several rounds of question revisions; -Minimize the possibility that respondents misunderstand or must guess the intent/meaning of a question; -Avoid leading, loaded, and double-barreled questions

Nominal measure

-Use descriptors (labels) to place objects into categories -Serve for classification, description, and identificationpurposes

Survey Components and Design

1. screens: First questions asked in your questionnaire 2. warm-ups: Immediately after any screens 3. heart of survey questions: focus on the key issues /questions 4. complicated/difficult to answer questions: Middle of the questionnaire; close to the end 5. demographic Qs and classification: Last section

Rules of thumb

1.Create sections or groups of questions based on the similarity in their content and type. 2.Use transitional statement prior to major sections of questions. The goal is to provide brief instructions on major flow or content change in the survey.

Introductory statement

1.Identifies surveyor/sponsor (identify sponsor only if it is not likely to bias respondents). 2.Explains purpose of the survey (general not specific purpose) 3.Explains respondent selection procedures 4.Requests participation/outlines incentives for participating 5.Anonymity vs. Confidentiality

Characteristics of "Good" Survey Questions

1.Questions are specific, simple, and clear 2.Provide a balanced set of answer options 3.A good question's answer options are mutually exclusive and collectively exhaustive Mutually exclusive—response categories should not overlap Collectively exhaustive—all potential response categories should be included

Common Problematic Survey Questions

A leading question prompts or encourages a desired answer. Example: As a Cadillac owner, you are satisfied with your car, aren't you? .

How many scale positions are appropriate when using interval measures ?

A minimum of 5 scale positions should be used. I strongly recommend you use 7-point scales for a computer-aided survey. If you will launch a cell phone survey, you may consider 5-point scales.

Double-barreled question

A question that asks multiple questions in one. Example: "Were you satisfied with the restaurant's food andservice?"

Does it matter if interval measures have an odd or even number of positions?

An odd number of positions is typically preferable because it allows respondents to choose a "neutral" point on the scale if it happens to best represent their views/opinions. Using an even number of scale points might force respondents to state a preference when they really do not have one.

Solution:

Avoid words such as "or" , "and" that usually indicate multiple questions are being asked.

"Heart of the survey" questions

Create groups of questions based on the similarity in their content and scale type. Use the funnel approach to organize these questions—from broad to specific. Such an approach helps avoid question order bias.

Measurement development

Designing questions and response formats to measure the properties.

ratio ex:

How old are you? (please round to the nearest whole year) _____ years old. What is your approximate weight in lbs? ______ lbs How long have you been working with your present employer? ______ yrs ______ months What is your approximate height in feet and inches? ______ feet ______ inches

Interval measure

Identify ordered relationships just like an ordinal scale BUT distances between values are known or fixed •Descriptors (answer labels) are of equal distances •The distance between adjacent descriptors is usually one scale unit

Should multiple questions be used to measure the same construct or variable?

If a single question can be used to adequately measure a construct/variable (e.g., overall satisfaction with the store), then one question is appropriate. However, if a more detailed understanding of the phenomenon is needed, multiple items may be necessary (e.g., satisfaction with employees, satisfaction with products, satisfaction with location, satisfaction with hours, etc.).

Ratio

It is highest level of measurement and has all the features of above measures. Additionally, it can take on the value of a true zero and may be expressed in a fraction or percentage

Solutions:

Keep questions clear and simple; Don't include problematic words (e.g., don't you, aren't you, etc.) to indicate a specific answer; Provide all options to a question or offer Other and make your survey easy to answer

factors to consider when selecting the type of measurement scale that will be used to assess a variable or construct:

Nature of the property being measured E.g., what type of scale(s) will you use to measure a person's Marital status? E.g., what type of scales will you use to measure attitudes towards an ad? Why? Scale used in previous research studies Market researchers don't have to be always creative when conducting a research project. If your company has been using the same scales to measure consumers' loyalty to your brand in annually, you should continue to use these scales in the future. This will help you compare loyalty across years. Data collection mode: e.g., paper and pencil, phone interview, or online interview. Ability of participants: will your participants understand a 7-point scale?

Ordinal measure ex:

Please rank each car color in terms of your preference, where 1 is most preferred, and 5 is least preferred. •Red •Black •Silver & grey •Natural White

Solution:

Pretesting your survey to make sure every respondent has a way to answer honestly.

Systematic error example:

The way a question is worded might lead all respondents to provide satisfaction scores that are, on average, 1 point higher than their actual level of satisfaction with the company. 3-very satisfied 2-moderately satisfied 1-satisfied

Observed response =

Truth + Systematic Error + Random Error

Interval example

Would you like to watch the movie Mulan this weekend? 1- Very unlikely 2- Unlikely 3-Somewhat unlikely 4-Neutral 5-Somewhat likely 6-Likely 7-Very likely

Ordinal

a measure that can produce rankings or an order among response options.

Nominal

a measure that is used to label or categorize response options to a survey question. Each value can be treated as a unique category.

Interval

a measure where the distance between response options, has an actual meaning, and is of an equal interval. This type of measure has fixed measurement units.

Properties

are specific features or characteristics of an object that can be used to distinguish it from another object (Example: Age, education level, gender, and habits of a typical Mac user).

reliability and validity

are used to describe the extent to which a measure is free from random and systematic errors and thus offer evidence of a measure's quality.

A loaded question

buried in its wording a sneaky presupposition, or it might make reference to supposedly universal beliefs or rules of behavior. ex:Where do you enjoy drinking beer?

Random error example:

despite no change in an individual's actual level of satisfaction with a company, an individual may rate his/her satisfaction level with a company as 3 on one day and then as 6 a week later.

Warmup-up questions

follow the introductory statement and give respondents a "flavor" of what the survey is about. These questions are -Simple -Interesting -Non-threatening

"Heart of the survey" questions

follow warm-up questions and focus on the key issues/questions that are the focus of the study.

Nominal ex:

gender, ethnic group, colors, names, religion (Buddhist, Christian, etc. ), political orientation (republican, democratic, libertarian, green)

Objects

include consumers, brands, stores, advertisements, etc.

Measurement

is the process of assigning values to the amount of properties of objectsin survey research.

Systematic error

refers to a constant bias in a measure; it is called systematic because it affects measures in a constant way. It can be caused by design flaws and researchers.

Random error

refers to a temporary bias in a measure that may be due to individual (e.g., mood) and/or situational characteristics

Validity

refers to the extent to which a measure is free from random and systematic error. Hence, if a measure is not reliable it cannot be valid.

Reliability

refers to the extent to which a measure is free from random error. More precisely, a measure is reliable if it is not influenced by the transitory and/or temporary factors that cause random error.

Truth

refers to what we are truly interested in measuring (e.g., how satisfied are customers with our company?).

Quantitative Research

research involving the use of structured questions in which response options have been predetermined and a large number of respondents involved. Example: survey research, behavioral experiment. (descriptive and causal)

Introductory statement

verbal/textual statement that communicates key facts about the survey to potential respondents. The statement should be put before the formal questionnaire.

Common interval measures

•A semantic differential scale is a type of interval measure that contains a series of bipolar adjectives for properties of the object. •The Likert scale is a type of interval measure that describes the intensity of agreement or disagreement on a statement by using 5 or 7-point scale.

Research objectives in a survey

•You may need to identify a number of different constructs and variables in order to measure a research objective •A survey can include multiple different research objectives