Chapter 9 - Survey Research

Systematic Error - Respondent Error

**If the respondents do not cooperate or do not give truthful answers then two types of error may occur. Non-response error Response bias Types of Response bias

CLASSIFYING SURVEY RESEARCH METHODS

1. Method of communication 2. Structured and disguised questions 3. Classifying surveys on a temporal basis a) cross-sectional studies b) longitudinal studies

ERRORS IN SURVEY RESEARCH B. Systematic Error (2)

1. Respondent Error 2. Administrative Error

The nature of surveys

1. Survey objectives: type of information gathered 2. Advantages of surveys

ADMINISTRATIVE ERROR Interviewer error:

Interviewers may record an answer incorrectly or selective perception may influence them to record data supportive of their own attitudes.

RANDOM SAMPLING ERROR

Most surveys try to portray a representative cross section of a particular target population, but even with technically proper random probability samples, statistical errors will occur because of chance variation. Without increasing sample size, these statistical problems are unavoidable. However, random sampling errors can be estimated (see Chapters 13 and 14).

Systematic Error -Administrative Error

The results of improper administration or execution of the research task are examples of administrative error. Such errors are inadvertently caused by confusion, neglect, omission, or some other blunder. There are four types of administrative error: --Data processing error --Sample selection error --Interviewer error --Interviewer cheating

CLASSIFYING SURVEY RESEARCH METHODS (3 WAYS) *Structured and disguised questions:

A structured question limits the number of responses available; whereas unstructured questions tend to be open-ended which allows the respondent considerable freedom in responding. The researcher can also disguise the questions which is particularly advisable if the subject matter is of a threatening nature. Other questions do not require disguising as it is assumed that the respondent is willing to reveal the information. Questions can be categorized according to their degree of structure and disguise. This helps in the selection of the appropriate communication medium for conducting the survey. However, it is not always easy to categorize the surveys as the categories are not clear-cut and most surveys are a hybrid of structured and unstructured questions.

ERRORS IN SURVEY RESEARCH

A. Random Sampling Error B. Systematic Error

CLASSIFYING SURVEY RESEARCH METHODS (3 WAYS) *Classifying surveys on a temporal basis:

Cross-sectional study: Longitudinal study:

-Respondent Error Response Bias

Response bias occurs when respondents tend to answer in a certain direction. This bias may be caused by an intentional or inadvertent falsification or by a misrepresentation of the respondent's answer. --Deliberate falsification: People may misrepresent answers in order to appear intelligent, to avoid embarrassment, to conceal personal information, to "please" the interviewer, etc. It may be that the interviewees preferred to be viewed as average and they will alter their responses accordingly. --Unconscious misrepresentation: Response bias can arise from question format, question ambiguity or content. Time-lapse may lead to best-guess answers.

RULE-OF-THUMB ESTIMATES FOR SYSTEMATIC ERROR

Sampling error may be estimated using certain statistical tools, but ways to estimate systematic error are less precise. Many researchers have found it useful to use some standard of comparison in order to understand how much error can be expected. For example, one cable TV company knocks down the number of people saying that they intend to purchase the service by a "ballpark 10 percent" because previous experience has indicated a 10 percent upward bias on the intention questions.

THE NATURE OF SURVEYS *ADVANTAGES OF SURVEYS: **TYPE OF INFORMATION GATHERED:

Surveys are quite flexible and, when conducted properly, are extremely valuable to managers. They provide a quick, inexpensive, efficient, and accurate means of assessing information about the population. Surveys can also be poorly conducted and certain errors can occur to render such a survey useless.

CLASSIFYING SURVEY RESEARCH METHODS (3 WAYS) * method of communication

Surveys can be classified according to the method of communication, telephone, mail, or personal interviews.

THE NATURE OF SURVEYS

Surveys require asking people, respondents, for information using either written or verbal questioning. Questionnaires or interviews collect data through the mail, on the telephone, or face-to-face. The more formal term, sample survey, emphasizes that the purpose of contacting respondents is to obtain a representative sample of the target population. Thus, a survey is defined as a method of primary data collection based on communication with a representative sample of individuals.

SYSTEMATIC ERROR

Systematic errors result from some imperfect research design, or from a mistake in the execution of the research. These errors are also called non-sampling errors. A SAMPLE BIAS exists when the results of a sample show a persistent tendency to deviate in one direction from the true value of the population parameter. The two general categories of systematic error are respondent error and administrative error.

ADMINISTRATIVE ERROR Data processing error:

The accuracy of the data processed by computer depends on correct data entry and programming. Mistakes can be avoided if verification procedures are employed at each processing stage.

THE NATURE OF SURVEYS *SURVEY OBJECTIVES: **TYPE OF INFORMATION GATHERED

The type of information gathered depends on a survey's objectives. Most survey research is descriptive research which attempts to identify and explain a particular business activity. Business surveys typically have multiple objectives; few gather only a single type of factual information. Although surveys are often conducted to quantify certain factual information, certain aspects of surveys may also be qualitative. For example, in new-product development the qualitative objective of a survey is often to test and refine new-product concepts.

-Respondent Error Types of Response Bias

There are five specific categories of response bias. These categories overlap and are by no means mutually exclusive. --Acquiescence bias: This is a response bias caused by a respondent's tendency to concur with a particular position. For example, "yea sayers" who accept all statements they are asked about. --Extremity bias: Some individuals tend to use extremes when responding to questions which may cause extremity bias. --Interviewer bias: If an interviewer's presence influences respondents to give untrue or modified answers, the survey will contain interviewer bias. Respondents may wish to appear wealthy or intelligent, or they may try to give the "right" answer or the socially acceptable answer. --Auspices bias: The answers to a survey may be deliberately or unintentionally misrepresented because the respondent is influenced by the organization conducting the survey. --Social desirability bias: This may occur consciously or subconsciously. Answers to questions that seek factual information or matters of public knowledge are usually quite accurate, but the interviewer's presence may increase a respondent's tendency toward an inaccurate response to a sensitive question in an attempt by the respondent to gain prestige in the interviewer's mind.

*Classifying surveys on a temporal basis: Cross-sectional study:

This is the most common type of study in which the data is collected at a single point in time. In such a study, various segments of the population are sampled so that relationships among variables may be investigated by cross-tabulation.

ADMINISTRATIVE ERROR Sample selection error:

This type of error is a systematic error that results in an unrepresentative sample because of an error in either the sample design or execution of the sampling procedure.

ADMINISTRATIVE ERROR Interviewer cheating:

To avoid possible cheating, it is wise to inform the interviewers that a small sample of respondents will be back to confirm that the interview actually took place.

*Classifying surveys on a temporal basis: Longitudinal study:

n longitudinal studies, respondents are questioned at different points in time so that changes occurring can be observed over time. Longitudinal studies which involve two or more samples at different times are called cohort studies because similar people are expected to be in each sample over time. Such studies can also be called tracking studies because they are designed to compare aggregate trends and identify changes. Having two or three different sample groups avoids response bias which might normally result from prior interview, but the researcher can never be sure that the changes in the variable being measured are not actually due to having different people in the sample. PANEL STUDY: This is a longitudinal study which includes gathering data from the same sample over time. The panelists record their purchasing habits in a diary for a set period of time. Panels are generally expensive and, thus, are usually managed by contractors which specialize in maintaining consumer panels. Such panels enable the investigator to keep track of repeat purchases, behavior habits affected by changes in price, special promotions, or other aspects of business strategies.