This Document Contains Chapters 10 to 13 PART II SECTION B TEACHING NOTES FOR CHAPTERS CHAPTER TEN INFORMATION FROM RESPONDENTS: SURVEY METHODS Outline of the Chapter • Discuss the different types of survey methods. • Discuss the process involved using each method. • Discuss the advantages and disadvantages of the survey method. • Discuss future trends in survey methods. • Discuss the problems that the researcher faces in conducting international surveys. Teaching Suggestions This chapter on survey methods has been written with a practical orientation. This chapter should be taught in conjunction with the Chapter 9. The different kinds of survey methods and their advantages, disadvantages and limitations are explained in detail. The instructor can provide an overview of these methods, in a lecture format. Internet surveys are becoming increasingly popular and they probably steal share from mail surveys. This edition of the book offers information related to nature of the e-surveys. Key problems of the e-surveys such as users changing e mail addresses frequently, need for continuous monitoring, difficulty in translation, possibility of lower response rate and hence the statistical value are to be highlighted to the students. These results can be given to the students as the basis for a discussion of the factors underlying the shift to telephone surveying. This can be used to encourage students into identifying the present day trends in surveys. Table 10-1 in the text gives the comparative indices of direct costs per completed interview that should give the students an idea of the costs involved and how budgets can determine survey methods. Questions and Problems 1. The Purchase Intercept Technique is ideally suited to this type of research question. Shoppers are more likely to be able to articulate the reasons for shopping at a particular retail store when they are in the shopping environment. Mall intercepts could also be used but this would yield only general shopping information and not data specific to a particular retail store. 2. A respondent’s willingness to participate in a research study may be affected by his perceptions of surveys, the survey sponsor, the interviewer, the data collection method or the survey subject. Increasing response rates require motivating the respondents to cooperate through some reward which may be tangible or intangible. Students should be encouraged to think of cost-effective ways to increase response levels. 3. The general feeling is that the incentive should be a motivating tool rather than full compensation for time spent. The appropriate size of this motivation will depend on the situation: a $5.00 gift certificate will be necessary for a mail diary panel where the respondent routinely records a large number of details for each of 50 or so purchases. On the other hand, that would be quite inappropriate for a modest 2-3 page mail questionnaire. 4. The appropriate actions during the first 30 seconds of an interview which will maximize the response rate are roughly the same for both telephone and personal interviews. (a) establish credentials, to allay fears that the survey is a disguised sales pitch or that a credit/tax check is being conducted. This is easier to do in a personal interview where documents can be shown. (b) build rapport by sounding warm and enthusiastic and avoiding any mention of threatening words that might indicate a sales pitch. Avoid an introduction that sounds mechanical and rehearsed. (c) give the reasons for conducting the survey in a way that is meaningful to the respondent. Don’t just say, “May we ask you a few questions?” (d) if the respondent sounds doubtful or hurried, find out if there is a better time to conduct the interview and make an appointment, and finally, (e) give a realistic time estimate. 5. The telephone pre-notification approach has relatively few disadvantages other than cost (relative to imply sending out a mail questionnaire), the need for a sampling frame with telephone numbers, and the possibility of sample bias because of the frequent need to compromise with a frame that does have numbers. For example, the appropriate frame may be readers of a special interest magazine. However, the subscription list only gives names and addresses. If it is desired to use this list, then telephone numbers must be obtained from directories or information operations, which is not only time consuming but will miss all subscribers with unlisted numbers. 6. This assignment is made difficult by the very sensitive nature of the topic. If there is any question in the respondent’s mind that the results could be used against them (i.e., to discover which houses in a neighborhood were vulnerable because of a lack of protection devices), then they will simply not return the mail questionnaire. The cover letter should emphasize the reputation of the sponsor, clearly indicate the purpose of the survey and emphasize the lack of identification questions that would permit a trace of the answers to a particular house. A good procedure during class is to have several cover letters read and critiqued, as the basis for preparing a more suitable cover letter. 7a. There are many ways of looking at this problem. The fundamental step is to identify the uses of this chemical and also the user of this chemical (intermediate or industrial user) and identify the end-user. The industrial user is the likely purchaser of this chemical. The survey could be conducted from the industrial user’s or the end-user’s perspective to assess the market potential of this chemical for various uses. To identify various uses, the research firm could carry out a survey amongst the R&D scientists and production managers of select companies to identify the possible uses. The companies chosen could be drawn from a gamut of engineering and chemical industries like Automobiles, Aircraft industry, Kitchenware, to name a few. From this, a list of possible uses can be drawn up and the potential for each use arrived at. The chemical firm can target the most profitable segments depending upon their capability. Assuming, that kitchenware manufacturers find this chemical extremely useful in the manufacture of their utensils after a test production run, then the survey could be done to elicit the responses of the industrial user. From their estimated production for the next few years, the chemical consumption can be worked out and hence the total potential can be determined. Also, the end user can be contacted with the survey on how many of them would purchase this type of kitchenware and at what price levels. This can also give a fair idea of the potential. This information collected from the end users can be used by the researcher of the chemical firm to identify the buyer’s perspective and armed with this knowledge, they can contact the kitchenware firms for their business. b. In case of contacting the R&D and production personnel of various factories, in-depth interviewing through the use of a questionnaire is a possibility. However, executive interviewing is a costly affair in terms of time, effort and money. It is difficult to schedule interviews because of the presence of gatekeepers and also because executive time is hard to get. The distinct advantage of a personal interview is that information received is of high use because the respondents are quite knowledgeable on the subject and hence various lines can be pursued. Mail interviews cannot be done. Telephone surveys can be tried but it is difficult to fix up a telephone interview with a busy executive and due to the fact that it is impersonal, conducting a long interview may be difficult. Focus groups are an excellent alternative but assembling a group of qualified professionals is tough. It is therefore best to choose a personal interview method. c. The students can be asked to design a survey and obtain the desired information. The suggested outline for the process is to identify the respondents whether they are factory personnel or end-user of a particular product. Also, identify the technique to be employed for data collection, as instructions vary with various types of method employed. d. The type of method is to be identified and the respondents are to be identified. The costs can be calculated using the comparative indices of direct costs per interview available. This will give the firm an approximate idea of the costs involved. 8. All the educational institutions are connected by the electronic mail facility. It has become common practice for all professors to include their e-mail addresses on their correspondence. The e-mail has become an excellent mode of communication in terms of economy of time and money and also because of its flexibility. If the e-mail addresses of professors are not known, they can be found out. The researcher can send his questionnaire to the respondents’ computers and get their responses back directly without any human interference at all, thereby reducing the errors committed in the surveys. E-mail surveys will be a very effective and popular survey method in the future. 9. In a computer interactive interviewing, the respondent sits before the computer and responds to the questions as they appear on the screen. The responses are directly entered into the computer by the respondent. The advantages of computer interactive interviewing over traditional interviews are many. It eliminates the use of paper and pencil as used in traditional interviews. It reduces the respondent bias by not allowing the respondent to compare his or her responses with the responses to previous questions. The coding and verification details are greatly simplified and the data can be readily used for analysis purposes. 10. Conducting research in developing countries has its own advantages and disadvantages. Various factors have to be considered while conducting research in a developing country. Wage costs are typically low in the developing countries and hence the costs can be kept down. However, the availability of skilled research staff is scarce and hence training of the staff becomes important. Developing countries are also typically marked by low levels of telephone ownership, poor communication facilities, absence of mailing lists, poor mail services and high levels of illiteracy. These are some of the problems faced by an international researcher in conducting research in the developing countries. 11. The students should be made to identify the research purpose and the respondents to be surveyed. This will eliminate some methods that are not relevant to the objective being studied and a shortlist of survey methods can be arrived at. In order to identify the optimal survey method in each of the countries, the students should be made to draw up a checklist of the important factors to be considered. This will help in the process of identifying the most suitable survey method. For instance, let us assume that a washing machine manufacturer wishes to capitalize on the economic boom in India and plans to launch his or her range of products in the near future. The manufacturer wants to identify the various aspects of the decision making process of the consumers in the purchase of a washing machine. The respondents are middle to high income groups with equal proportion of men and women. In this context, after consideration of the costs involved in using questionnaires through intercept method, mail survey or a telephone survey seem to be the ideal method. Low levels of telephone ownership in India, poor communication facilities, and high costs make telephone surveys unviable in India. With respect to mail surveys, absence of mailing lists, reluctance of respondents to respond by mail may make a mail survey not so attractive in India. This leaves us with the option of the questionnaires administered through intercept method in upscale consumer durable stores which is suitable in India. 12a.A mail survey is a process involving a series of difficult decisions regarding the identification of study participants and the interview package that consists of an outgoing envelope, cover letter, questionnaire, return envelope and incentives, if any to be used. Any problem on anyone of the above mentioned factors could lead to a low response rate or incomplete responses. Ideal Idaho should have a broad identification of the individuals to be sampled before commencing the data collection exercise. Mailing lists have to be precise, failing which the mail surveys do not serve their purpose. The questionnaire should be concise, well laid out and unambiguous in order to get high response quality. The cover letter is an important component of a mail survey. It should clearly state the objective of the survey and seek cooperation from the respondents. It is prudent to attach a honorarium with the cover letter and the questionnaire as an incentive to ensure that the customers respond to the survey. A pre-addressed, stamped envelope should be enclosed so that the respondents send back the survey. In case, the response rate is not to the expected level, it is better to send in a remainder after a specified interval of time. Even after the remainder is sent and the response rate remains low, a fresh set of questionnaires can be mailed to another set of respondents. Thus the company can collect the relevant data effectively by using a mail survey. b. There are many advantages of conducting a telephone survey. More interviews can be conducted in a given time period because there is no time lost in locating respondents. Also, it is easier to get to interview people over the phone rather than traveling to meet them. If the respondent is not available, callbacks can be made at a very low cost, in a very easy manner. The telephone interviews are marked by speed, absence of administrative problems and cost per completed interview. However, there is an element of sample bias that may creep in the telephone surveys as respondents without telephones or with unlisted telephones are not covered and might be the part of the sample. Also, there can be an amount of impersonality in a telephone interview and hence the interviewer has to rely on verbal cues to judge the understanding and reaction of respondents. The focus groups can be used to ascertain if the consumers would be interested in this unique product offering but it is not a useful method in estimating national sales potential. c. In case of a mail survey, the non response bias is a major factor. It should always be tested whether there exists any differences between the respondents and non respondents to eliminate the chances of biased results. In case of a telephone survey, there could be the possibility of sample bias as only the respondents with telephones are considered. d. The foreign markets to be considered are identified first. A preliminary list of the possible survey methods is considered. The countries’ infrastructural, economic, sociopolitical and cultural data should be studied. Based on these available data, the most appropriate survey method can be identified. An example of this kind is given in Q. 11. 13a. The research purpose of this study would be to identify the reasons for the stagnant sales of art objects in the last two years. The students should be made to identify the possible survey methods, the advantages and disadvantages of the methods with particular emphasis on the following two dimensions: cost and time. An outline of the solution to the problem is given. At a basic level, the personal interviewing of all the specialty store owners is unviable. Mail survey is too impersonal and also the response rate will not be high. The possible survey methods could be Fax surveys or Telephone surveys. b. A foreign market opportunity analysis is to be conducted to identify the possible markets in Europe. The markets could be looked at two levels: identifying the countries that would be receptive to these specialty goods or identifying market segments across countries. For instance, Ms. Gomez could look at individual markets of Britain, France, Italy etc., or the upper income segments of France and Italy. The buyers of such a specialty good would typically belong to the upper middle to higher class of people who have higher income levels and are art lovers. Personal interviewing seems to be an inviable option. The options boil down to mail, fax, mall intercept surveys and telephone surveys. The advantages and disadvantages of each method should be identified. Within the scope of the constraints and after ascertaining that the research method fulfills the research objectives, the final survey method is then selected. c. There are many ways to approach this question. Ms. Judy Flook will need a lot of survey data to understand the buying sentiments. Judy may decide to use a traditional paper and pencil method to understand the buying behavior. However, conducting and processing such a study will take enormous time and costs. If she can persuade her directors to make a one-time investment on computer related equipment, she can resort to computer interviewing. the advantages of computer interviewing are that it is faster, more accurate than using pencil and paper. In Computer interactive interviewing, the respondent directly interacts with the computer and responds to the questions as they appear on the screen. The information is directly input by the respondent. Many hardware and software developments continue to make this form of interviewing more feasible. Since the list of patrons is available and most of the patrons are likely to be residents of the city, it is easier and cheaper to reach them by telephone and inform them of the purpose of the survey. This could be followed up by a questionnaire mailed to them asking them about their attitudes concerning the collection and ways of improving the collection. The questionnaire could contain several open ended questions that the respondent may not have answered during the telephone interview. This will be cheaper and fast because the survey will have focus only within the city limits. PART II SECTION B TEACHING NOTES FOR CHAPTERS CHAPTER ELEVEN ATTITUDE MEASUREMENT Learning Objectives • Define attitudes. • Discuss the concept of measurement and scaling in marketing research. • Discuss the different scales in measurement. • Describe the different types of scales used for measuring attitudes. • Give a description of each of the well‐known scales that are used to measure attitudes. • Describe the guidelines for developing a multiple‐item scale. • List the limitations of attitude scales. • Discuss how to select the most appropriate attitude scale. • Discuss the concepts of reliability, validity, and generalizability. • Describe issues using scales in cross-national research. Teaching Suggestions The instructor has a number of options for using this chapter: In Conjunction with the chapter on questionnaire development, As a basis for discussion of related consumer behavior theory; including consumer decision processes, market segmentation theory, consumer values and beliefs and so forth. For example, it may be desirable to discuss AIO lifestyle (psychographic) measures in this context to clarify the distinction of middle-level beliefs and specific attitudes towards products, persons or situations. Questions and Problems 1. There are severe practical problems encountered in using a sales response criterion for budgeting and evaluation purposes. Sales are a consequence of multiple influences (competitive behavior, changes in other elements of the marketing mix etc.) and very carefully controlled experiments are necessary to separate these effects. To complicate matters advertising does not act immediately, and the length of the lag in effect is often unknown. Finally, a sales response criterion is not helpful in diagnosing the reasons for good or poor sales performance, and suggesting possible improvements. 2. Measurement is a standardized process of assigning numbers or other symbols to certain characteristics of the objects of interest according to some pre-specified rules. Measurement often deals with numbers because it is easier to conduct data analyses on numbers. The essential characteristics of the measurement process are that there should be one to one correspondence between the symbol and the characteristic in the object that is being measured and that the assignment rules should be invariant over time and objects being measured. Each object should be measured and placed on a continuum depending upon the amount of measured characteristic the object possesses. The process of creating such a continuum is called scaling. The measurement scales are classified into nominal scale, ordinal scale, interval scale and the ratio scale. Nominal Scale: The objects are assigned to mutually exclusive, labeled categories but there is no ordering involved. Sex, geographical location and marital status are nominally scaled i.e. if one entity is assigned the same number as another, the entities are identical. Ordinal Scale: This scale is obtained by ranking objects or by arranging them in order with regard to some criteria (e.g. income). It provides information on the relative preference between objects. Assume that brands of frozen vegetables are rated on quality, from the highest quality to the lowest quality, on a scale from 1 to 5, then the scale is said to follow an ordinal scale. Interval Scale: The numbers used in the scale also represent equal increments of the attribute being measured and the differences between the numbers can be compared. The difference between 1 and 2 are the same as those of difference between 2 and 3 and the difference between 2 and 3 is half that of the difference between 2 and 4. The location of zero is not fixed. Ratio Scale: A kind of interval scale that has a fixed zero point. Interval scale permits us to make comparisons of absolute magnitude. For example, an income of $30,000 is two times as large as $15,000, has a known zero point and therefore a ratio scale. 3. a. Nominal b. Interval c. Ratio d. Ordinal 4. Two types of attitude changes would be of interest to the product manager. The first is whether the attitudes of the baby boomers themselves are changing as they age. The second is a comparison of the attitudes of baby boomers versus the current target group if the ‘product caters to a 40-50 consumer age bracket. Attitude changes about environmental issues would be of interest since they might affect packaging preferences (i.e., biodegradable containers). Attitudes about health, additives, fitness, nutrition, convenience and so on could all have an impact on consumer attitudes about branded food products. The product manager might also conduct a segmentation analysis including changes between the segments of light vs. heavy users. 5. A variety of sources of data could be utilized to generate category descriptions for this purpose: Previous studies of financial institutions focus group interviews in-depth exploratory interviews repertory and techniques (these ask: in what ways is institution A different from B and C? is B different from A and C etc.)? These sources should be followed by exploratory work which tests the ability of the adjectives or phrases to discriminate among financial institutions. 6. If the intent of the research question is to try to predict whether or not smokers will try to quit then both attitudes about smoking and the intention to quit should be researched. Students can use any number of different scales for these questions. Some possible questions that could be asked are: a) Do you believe that smoking in general is detrimental to a person’s health? b) Do you believe that your smoking is detrimental to your health? c) Do you agree/disagree with the following statements? i. Smoking is a dirty habit; ii. Smoking is a difficult habit to break. iii. Smoking should be banned from all public places. iv. Smoking is a pleasurable social habit. v. Smoking is related to stress. d) How often and have you tried to quit smoking in the past? e) Would you like to be able to stop smoking? f) Do you intend to quit smoking some day? g) Do you intend to quit smoking within the next year? 7. In fact, it can be shown that A50 B50, B50 C50 and A50 C50 is consistent with any of the results in the three arrays. If it is A25 B25 C50 then there is a group constituting 50% of the sample that always prefers C over A or B. and is indifferent between And B as a second choice. Similarly, if the preferences in the array are A33, B33, C33 then the result A50 C50 is obtained by having half of those preferring B choosing A as their choice, and the other half choosing C. Finally, A50 B50 C50 is consistent with the pair-wise results if the half of the sample preferring A always prefers B rather than C, and those preferring C always prefers B rather than A. In this last example, B is always a second choice, although there is no evidence to that effect from the pair-wise data. 8. Attitude measures will be good predictors of behavior when the measures are reliable, the attitudes are stable and well-informed, and there is little likelihood of significant extraneous variables influencing the subsequent behavior. As an example of a situation where attitudes are poor predictors consider a person who is poorly informed about a product category or thinks all brands in the category are alike and is thus not involved in differences among brands and is very susceptible to short-term price promotions and out-of-stocks. 9. Validity refers to the extent to which a scale truly reflects the underlying construct it is attempting to measure. For instance, differences in attitude scores should reflect differences among the objects or individuals on the characteristic being measured. There are three approaches to assessing validity. (i) Face Validity refers to whether the contents of the scale fully capture the relevant facets of the construct being measured. Construct Validity relates to what a scale is measuring. It tests whether or not the measure confirms hypotheses generated from the theory based on the concepts. This validity can be measured only after establishing the discriminate and convergent validity. If the scaled measure correlates with other measures of that variable, convergent validity is established. If the scaled measure correlates weakly with other measures that are not measuring the same variable, discriminant validity is established. (iii) Predictive validity refers to how well the scaled values are able to predict some other observable variable that the scaled construct is supposed to influence. The ability of a measure to produce the same results on different tests over time is reliability. An unreliable scale produces wide variations that are not reflective of true differences in attitudes. The three methods for establishing reliability can be classified according to whether they measure stability of results over time or internal consistency of items in an attitude scale. (i) Test-retest reliability: This method is used by measuring a subject’s response on the same scale at two different time periods. (ii) Equivalent form approach: This method is appropriate for attitude measures that presumably measure the same underlying uni-dimensional attitude. Internal consistency: This method compares different samples of items being used to measure a phenomenon during the same time period. 10. Students should be asked to follow the following steps in developing a multi item measure. * Delineate the major characteristics of the given construct from other similar constructs. * Determine clearly what is to be measured. * Generate as many items as possible and evaluate it using experts. * Specify the actionable steps that must be followed in order to determine the type of attitudinal scale to be measured. * Include validation items in the scale. * Administer the items to a sample and evaluate and refine the scales. Optimize the scale length. 11a.The possible hypotheses for the difference in temperatures could arise from the following explanations: Miscalibrated thermostats in homes Delay in responding to changes in the day time and night time temperatures Error in estimating the desired fuel consumption savings in homes b. The problem with a telephone survey is that there is no way to validate the verbal reports of household temperature. However, one can use the evidence from the personal interview surveys that those who actually had lower temperature readings tended to be higher income, suburban people in professional and managerial occupations. In addition, it was found that people who practice certain conservation behaviors tended to practice others as well, e.g., tanning off lights, using less hot water, dosing off unused rooms and insulating attics, windows and walls. This gives a picture of those people who are likely to give realistic answers to questions about conservation behavior. To understand their conservation attitudes, it is necessary to understand: * knowledge about what to do to save energy; * orientation to the comfort and welfare of the family; * belief in the existence of an oil shortage; * concern over dependence on foreign sources; * belief in the right of citizens to use as much energy as they want. 12. It is helpful to begin discussion of this question by putting the results of the two surveys side-by-side on the board: Among the possible reasons for the difference are: (a) that many respondents rated Ford “only fair”, in the Harris survey, but such a rating would not be sufficient to register disapproval. (b) the Gallup question forces a response into one of two categories, and judging from the high non-response rate of this question, many people are not happy with this limited choice. (c) a president’s performance has to be evaluated on many dimensions, and those people who were unhappy with his performance on several of these dimensions might not want to register overall disapproval. (d) variability due to random sampling error, (e) differences in data collection method, elapsed time of data collection and so forth. The clear implication from these problems is that an accurate picture of people’s attitudes can only be obtained by asking several related questions. Reliance on the results of one question is unwarranted. 13. It is true that in decisional research cases, researchers prefer to have an attitude measure predict some future event. The more the uncertainty of the future events, the greater is the use for predictive validity. The very nature of decisions requires predictions of uncertain future events and hence predictive measure is useful. For instance, assume a brand manager of a toothpaste company wishes to conduct a brand preference study. If a brand measure he or she has developed can predict the future sales, then the measure is valid and is also the most important one as it enhances his decision making capabilities. If it is shown through past sales records that the measure is consistent in predicting future sales, then the measure is reliable. 14a.The same questionnaire cannot be used in all the three countries. The same type of survey method cannot be used in all the three countries in question. The different languages or dialects in the three countries make it impossible to use the same questionnaires. It is important that the survey method to be used is identified and a basic questionnaire be designed in English. This questionnaire has to be translated in the languages of other countries and back translated to check whether it reads the same in English. The scales used to measure the attitudes or preferences have to be chosen after careful consideration of the specific country factors. b. Before developing a scale to be used in international marketing research, a suggested outline of the factors to be considered are given below. * Literacy levels of the countries identified. * Educational levels of the respondents in each country so that the questions can be directed in an appropriate manner. * Cultural factors specific to each country. * Calibration of the response formats. c. Based on the steps outlined in (b) above, the most suitable scale for each country can be identified. 15. Ben Gatsby needs to learn the basic orientation or attitudes of the prospective customers. The study of their attitudes includes the information they have, their feelings of like and dislike and their intentions to behave. Ben Gatsby should understand and influence consumers’ behavior to buy his artifacts. This is possible only through the study of attitudes because attitude is a precursor to behavior. Also attitude questions are easier to ask rather than observing behavior which is influenced by a lot of factors. Attitude studies can also be used to evaluate competitive alternatives and to get an insight into the process by which choice decisions are made. Therefore, Ben Gatsby needs to get an insight into consumer attitudes to understand their buying behavior. 16a.Marketing strategies depend upon understanding attitudes of target consumers and when necessary, attempt to modify or change those attitudes. Measurement of attitudes is much more difficult to measure as it is not observable and present in the minds of the consumer. Attitude scaling is based on various operational definitions created to measure the attitude construct. b. As the name suggests, single item scales are those which have only one item to measure a construct. The various issues that have to be considered regarding the single item scale are given below: * Number of scale categories * Types of poles used in the scale * Strength of the anchors * Labeling of the categories * Balance of the scale PART II SECTION B TEACHING NOTES FOR CHAPTERS CHAPTER TWELVE DESIGNING THE QUESTIONNAIRE Learning Objectives • Discuss the concept of questionnaire design. • Describe how to format question. • Discuss how to deal with sensitive questions. • Discuss how the sequence and layout decisions impact ease of use and order bias. • Discuss the issues of questionnaire design in an international context. Teaching Suggestions This chapter may be used in conjunction with Chapters 8, 9, 10 and 11 as a background source, or treated in depth as a separate subject. If a session is devoted to writing questionnaires most of the effort should be directed to writing and critiquing questions. The questions for discussion are a useful vehicle but virtually any survey offers a fertile source of faulty questions. Below are some further thoughts on ways to help make the point that responses are very sensitive to the wording of the question, and that questions can be asked in many different ways. (a) Impact of different ways of asking the question: As an example of the basic problem, here are two questions that were asked at about the same point in 1969 concerning the president’s plans for withdrawal of troops from Vietnam. The Gallup Poll (June 1969): “President Nixon has ordered the withdrawal of 25,000 United States troops from Vietnam in the next three months. How do you feel about this—do you think troops should be withdrawn at a faster rate? (“Same as now” not presented, but accepted if volunteered.) Responses: Faster, 42 percent; same as now, 29; slower, 16; no opinion, 13. The Harris Poll (September—October 1969): “in general, do you feel the pace at which the president is withdrawing troops is too fast, too slow, or about right?” Responses: Too slow, 28 percent; about right, 49; too fast, 6; no opinion, 18. (b) When the question suggests the answer, The following quotation comes from a column by Daniel Seligman in the May 7, 1979 Fortune page 105. It has now been demonstrated conclusively that a pollster who asks the right questions, or, to be more precise, the wrong questions can get the American people to agree to anything. Consider the following survey. Respondents were asked: “Do you feel economic growth and increasing G.N.P. tends to overproduce products which in turn lead to more waste?” Not wanting to be on the side of waste, 59 percent said yes. The next question asked whether economic growth “makes us far too dependent on physical resources which are running out.” Not wanting to be guilty of such bad planning, 68 percent said yes. Another question asked Americans to ponder whether growth “falsely makes people want to acquire possessions rather than to enjoy nonmaterial experiences.” Being opposed to falsity, 60 percent held to the affirmative yet again. Further, the survey asked: “Of course, what we have asked you about here would mean a very different style of life in America. It would mean buying fewer products, not expanding our standard of living, not working as long hours or weeks, and having less income. But the alternative might mean higher inflation, more shortages, and more recessions. If you had to choose between those two, change in life-style or the risk of more inflation and more recessions, which would you choose?” You will not be stunned to learn that 71 percent came down on the side of a nobler life-style, while only 12 percent opted for more inflation and recession. Based on the results, the researcher concluded that “the age of materialism as we have known it is going to be radically altered.” (c) Creative question writing. This parody was done in 1958 by Alan Barton, but still has lots of potential for humorous insight. It is a series of ways to ask the question, “Did you kill your wife?” 1. Causal Approach: “Do you happen to have murdered your wife?” 2. The Numbered Card: Would you please read off the number on this card which corresponds to what became of your wife? (Hand card to respondent.) 1. Natural death 2. I killed her 3. Other (what?) (Get card back from respondent before proceeding) 3. The Everybody Approach: “As you know, many people have been killing their wives these days. Do you happen to have killed yours?” 4. The “Other People” Approach: (a) “Do you know any people who have murdered their wives?” (b) “How about yourself.” 5. The Sealed Ballot Technique: In this version you explain that the survey respects people’s fight to anonymity in respect to their marital relations, and that they themselves are to fill out the answer to the question, seal it in an envelop, and drop it in a box conspicuously labeled “Sealed Ballot Box” carried by the interviewer. 6. The Kinsey Technique: Stare firmly into respondent’s eyes and ask in simple, clear-cut language such as that to which the respondent is accustomed, and with an air of assuming that everyone has done everything, “Did you ever kill your wife?” Problems of this sort can be avoided through proper pretesting. The instructor may wish to emphasize that it is not just individual questions that need to be pretested, but also the questionnaire as a whole. Questions and Problems 1. Some of the aspects that the students are encouraged to think on when critiquing the questionnaire are: The presence of open-ended vs. close-ended questions In case of close-ended questions, the number and order of response categories The word choice in framing the questions The sequence of questions Bias arising out of question wording and order of questions 2. The order of questions is an issue of concern here because answer to question (a) has an obvious implication for the answer to question (b) and people may want to maintain consistency in their answers. We can reverse the order of questions to avoid this concern. 3. The questionnaire is an important tool to elicit information from the respondents that would aid the researcher in getting an insight on the marketing phenomenon being investigated. Exploratory research is preliminary research conducted to clarify the exact nature of the problem to be solved whereas descriptive research is research conducted to provide information on the relevant characteristics of a group of sample units of interest. Exploratory research involves a flexible and less formal approach in contrast to descriptive research that is less flexible. Therefore, exploratory research typically requires a list of topics to be covered. The descriptive research is more specific as the information requirements are more stringent. This makes the design of a questionnaire an easier task in case of exploratory research compared to descriptive research. 4. It is not necessary to have a question included in the questionnaire because of its relevancy to the researcher’s standpoint. Questions should be chosen such that the respondents are willing to answer them and they are capable of answering them. Care should be taken to ensure that the questions are not sensitive to the respondents, otherwise, the respondents may feel offended and hence refuse to respond. 5. The general advantage of open-response questions is the freedom they offer to respondents and their ability to make the respondents more comfortable with the questioning process. Thus they serve as useful introductory questions. When used to assess saliency they have two additional advantages (1) they establish a broad context for the respondents to discuss things which irritate/bother/annoy them, and (2) they give absolutely no hint as to what specific subjects are being investigated so there is no artificially heightened saliency, which is a problem with questions which ask directly “How concerned are you about the frequency of advertisements on TV?” Unaided recall awareness questions are usually superior, for the brands the respondent recalls are usually in the consideration set and will be very familiar. Thus, it is a persistent finding that “top of mind” awareness (first brand mentioned) is a very good predictor of subsequent brand choice in certain frequently purchased product categories. This question provides a very conservative estimate of the number of brands that a respondent might know or remember if they thought at length about the product category. For this reason, it is a good idea to follow with an aided recall question, to learn whether the respondent had heard of the brand name. Unfortunately, this measure tends to be biased upwards, for respondents want to appear cooperative and informed and will say “yes” to any brand name they have dimly heard about in the past, or which seems reasonably associated with the product category (even if it is not). The randomized response question can be structured as follows:
Flip Coin Question Proportion of Population % of Group Answering Yes
Heads Did you report all your income to federal tax authorities? 50% ?
Tails Is your birthday in June? 50% 10%
Total population 100% 16%
We know that 16(100) = 16 percent of the total sample answered “yes” to something. Since 50% of the sample presumably chose the income tax reporting question we can then estimate the proportion of this group answering yes. For convenience, assume the total sample is 100 people. The 50 x+ .10(50) = .16(100). Solving x = .22. 8. The researcher needs to make a preamble in his questionnaire to request the respondents to participate in the survey. This should be done so that the researcher communicates to the respondent that he/she understands the value of time spent by the respondent in responding to the survey. An assurance is to be given that the invaluable information provided by the respondents is used only for research purposes and will not be used elsewhere. Also, the benefits that will accrue to the respondents should be specified. For instance, assume that a famous soft drinks company wants to conduct a survey of its suppliers to obtain information on the efficiency of its bottling operations. The researcher should make sure that the preamble contains a statement that the information is invaluable for the research and it will be used to aid in improving the efficiency of the bottling operations. The preamble should communicate that participation in the survey will provide the company some information on improving its bottling operations and thereby will benefit the suppliers. 9a. The term “private citizen” is ambiguous and should be defined. Does it, for example, include all civil servants in their private life? This should be specified. Another ambiguous phrase is “read books and articles.” Does it mean both books and articles or does it mean, books and/or articles? b. The term “sometimes” is ambiguous and should be defined in terms of the proportion of times the respondents eat at the same place. The term “same place” is ambiguous too because it could mean either the same location or the same chain perhaps at other locations. For example, when traveling, respondents may eat at the same franchise but at different locations. This problem could be solved by inserting the proper term. Finally, the question is not applicable to all respondents because some respondents may never eat out. A third category such as, “never eat out”, should be included. c. The main problem with this question is that there is no category for those respondents who feel that there is too much government regulation of environmental protection. A better phrasing would include three categories like, “Enough, too much and not enough.” Furthermore, due to the knowledge required of the respondent, some respondents who are less informed than others, may not know how to answer and a “don’t know” category is required. d. This question will be difficult to answer for some types of respondents. Some respondents, like university students living at home may not buy all their own clothes. Other respondents do not buy their clothes at any particular store and instead shop at stores depending upon their seasonal price specials. Still other respondents may evenly divide their shopping between two or three stores. These problems could be avoided if “most” was less ambiguously defined. Instead, the respondents could be asked where they buy “60% of their clothes.” A “never buy clothes” option and a “no particular store” option could be provided. e. This question is very ambiguous. The term “everything possible” could theoretically include shutting down the generating station. In this case, the answer to the question will always be “no.” Perhaps a word like “reasonable” is still vague, although it is closer in meaning to what the questionnaire is asking. Furthermore, a “don’t know” category should be provided, given the technical nature of the question. f. The respondents will have a difficult time providing an accurate estimate concerning the “average” issue. A better approach would be to ask respondents about the “least” issue. This will reduce the memory recall problems. g. Although the researcher has attempted to define the terms used in this question, there are still some ambiguities, which could cause problems for respondents. “Almost as often as the magazine is published” is a subjective assessment. h. The word “hobbies” is too vague. It is not clear whether sports or travel are hobbies. Furthermore many hobbies are performed alone versus being part of a club. It’s not clear whether the researcher is looking for a list of all recreational activities or only those of a particular kind. i. The phrase “some of the things” sounds vague. Respondents will need more guidance in terms of context to answer this question. Converting this open-ended question to a closed-ended one by providing some answer choices (including an “others” category) may be of help. Further, defining the term “lately” is required. This will add to the context of the question by bringing in the time perspective. 10. Car problems may be the result of all three factors and this should be kept in mind by the students in developing appropriate questions. Either/or alternatives may not capture the necessary information. Some suggested sample questions might be as follows: a. Suppose you had 100 points to divide among these three groups. How would you divide the 100 points to indicate how much each is responsible for any problem you have had with your car? _____ manufacturer _____ services _____ yourself. b. We are interested in determining your attitude toward your car. Please use this scale from 1-3 and circle for each item, the number which best describes how satisfied...or...dissatisfied...you are with these aspects of your Car.
Totally Satisfied Somewhat Satisfied Somewhat Dissatisfied
Its appearance and design 1 2 3
The way it is built 1 2 3
The costs of operating it 1 2 3
Performance 1 2 3
The amounts of service it needs 1 2 3
11. These questions could reflect several different research objectives. If the researcher is interested in the type of respondents’ leisure activities then questions (c) and (e) are unnecessary, since the subject of interest is how respondents spend the free time that they have. If on the other hand, the researcher is interested in respondents’ attitudes about their free time, then questions (b) and (d) should be changed to ask respondents whether they feel they have enough time to engage in hobbies or read magazines of their choice. The kind of hobby or magazine is not relevant to this research objective. 12. Double barreled questions: (i) Are you satisfied with the cost and quality of the food items at Restaurant X ? (ii) Do you prefer sitting out in the patio and eating spicy food? (iii) Are you satisfied with the location and interior design of this restaurant? Corrected version: 13. When a question deals with a sensitive issue or is potentially embarrassing, the respondent may refuse to answer it or may respond incorrectly. Such questions may also irritate the respondents. Therefore, a researcher has to decide whether sensitive questions are absolutely essential in the questionnaire. If they are absolutely necessary, then disguised techniques have to be used. A disguised question is one whose true purpose is not known to the respondents. If the researcher feels that direct questions may not elicit truthful answers, then disguised questions have to be resorted to. As an exercise, a sensitive question can be given to the students and various ways of disguised questioning (viz) the casual approach, the numbered card approach, the everybody approach, the sealed ballot approach, the other people approach can be discussed in the class. An example of a sensitive question and the possible disguised question is given below: Sensitive question: Would you buy food from a local, inexpensive restaurant and pass it off to your guests as food from an upscale, expensive restaurant? Disguised question: Is it alright for people you know or associate to buy food from a local, inexpensive restaurant and pass it off to your guests as food from an upscale, expensive restaurant? 14. Pretesting is extremely important because it ensures that the information to be obtained from the study meet the research objectives. It provides a test to the flow of the questionnaire, the length of time it takes to fill out the questionnaire and the wording of the questionnaire. It helps the researcher identify the potential problem areas as far as the respondents are concerned and to ascertain if the categories are exhaustive. It also helps in identifying the limitations of the questionnaire. A personal interview pretest can use either a debriefing or a protocol approach. In the debriefing approach, the subject is asked questions after the interview. The questions pertain to the respondents’ mindset when he or she answered each question and whether any aspects of the subject were not covered by the interview. In the protocol method, the subject thinks aloud as he or she fills out the questionnaire. The interviewer records these thoughts and at the end of the interview asks for clarification, if necessary. The pretest questionnaire, to be designed for the interviewers should encompass the following broad outline: * Do the respondents infer the meaning to questions as intended by the researcher or do they vary drastically? * Are there obscure questions that are simplified by the respondents to make it easier for them to answer? * is there any question that shows far greater variation than intended? * Are the questions of correct length, unambiguous, not double barreled etc.? * Are the questions arranged logically? * Is the questionnaire of ideal length that it keeps the respondents’ attention and interest? A possible questionnaire for interviewers might be constructed as follows: Please make out a separate questionnaire for each pretest interview you conduct. For all “yes” answers, please specify the question numbers or section and explain what the situation or problem seemed to be: 1. Did any of the questions make R uncomfortable? 2. Did you have to repeat any questions? 3. Did R misinterpret any questions? 4. Which questions were the most difficult or awkward for you to read? 5. Do you dislike asking any specific questions? Why? 6. Did any of the sections in the questionnaire seem to drag? 7. Were there any settings in which you felt that the respondent would have liked the opportunity to say more? 8. Were there any questions where the respondent gave an answer not provided for in the questionnaire? And so on. The specific questions of interest will depend upon the survey and its particular problems. 15a.In any research study, the research objectives have to be identified. Before individual questionnaire items can be written, these objectives have to be translated into information requirements. The items in the questionnaire should aim to elicit the possible answers to the questions implied by the research objectives. The possible information requirements for the researcher in this case are: * General attitudes with respect to scholastic testing system and specific attitudes regarding the existing system. * Satisfaction with the existing trend in rise in incidence of cheating at school. * Sources of influences. * Classification variables like year of study, background information etc., b. There is no ready-made answer to the format to be chosen for questioning as choosing an open or closed ended questions is not an either or distinction. The information requirements decide the individual question format. In case of an open ended question, the respondent is free to express exactly the way he or she feels about the issue and hence can provide the researcher with wide array of information. Also, an open ended question will give the researcher some alternatives that cannot be provided by a closed ended question. The disadvantage is that the coding and processing of questions is extremely time consuming. In case of a closed ended question, the processing of data is simple but the limitation is that some valuable information could be lost. In our case, as the study is exploratory in nature, the questionnaire needs to be flexible and hence it would be better to opt for some open ended questions in conjunction with closed ended questions to gain insights into the problem and to assess the students’ thinking in this critical issue. c. The order of questions should be carefully looked into as they tend to make the questionnaire as easy as possible for the interviewer and also help in maintaining the cooperation of the respondent. Care should be taken that the prior questions do not influence answers to subsequent questions. In other words, order bias should be eliminated. A suggested outline to eliminate order bias is given below: * Opening the interview with a non threatening question * Avoiding sudden shifts in topic and questions to which the respondent identifies the preceding questions as the frame of reference * Proceeding from general topics to the specific topics * Placing the sensitive questions and questions pertaining to demographics to the last. d. (i) Pretesting is extremely important because it ensures that the information to be obtained from the study meet the expectations of the researcher. It enables the researcher to identify the flaws in the questionnaire and rectify them. It helps the researcher identify whether the respondents face any problem as far as the questions are concerned and to ascertain if the response categories are exhaustive. It also helps in identifying the limitations of the questionnaire and test for the flow of the questionnaire. (ii) Care should be taken that the respondents are reasonably representative of the sample population. If the questionnaire is straight forward, a small sample of 15 students is required and for longer questionnaires a sample of about 25 students may be used. e. In the international context, it is often noticed that the researcher has no prior experience with the determinants of response in another country. Open ended questions have to be used with care in the international context after due thought to the educational, literacy and cultural factors in the country chosen for research. In France, use of open ended questions may be desirable as they do not impose any cultural bias by the researcher. Also, the familiarity with all the possible responses is not required. f. It is necessary to have a fair idea of the literacy, demographic, cultural, economic and infrastructural factors of the foreign country. The questionnaire design has to be adapted to the requirements imposed by these factors. 16a.The committee is conducting interviews at homes of the members. This means that the identity of the respondent cannot remain anonymous. The members will want to give socially correct answers and hence overestimate their attendance to the Sunday services. The respondents are also likely to overestimate the dollar value of their donations to the church. 16b.The questionnaire design and wording will also affect the response. The second question is really a statement phrased in the form a question because not too many members will answer no to that question. It also induces a guilt feeling in all members who think they do not donate enough to the church. As a result they will overstate their donation amount to the church. The third question is not needed at all as the church is only interested in the amount donated and not the frequency with which the donations come in. The subject of the interview is a sensitive one and hence it may be better for the church to guarantee anonymity to the respondents to avoid stepping on sensitive issues. The interviewer could also induce a lot of bias as he/she is a member of the finance committee in the church and hence may expect all respondents to have the same loyalty to the church that he/she has. 17. Here’s possible questionnaire with a view to uncover the reasons for not buying the product. 1. Do you own a lawn mower? ___Yes ___No Continue if the answer to question 1 is “Yes”, otherwise stop here. 2. What make/manufacturer is your lawn mower? 3. What other brands of lawn mower did you consider before your purchase? 4. Please check all sources of information about lawn mower that you used for your purchase? Magazine Radio Television Internet Other (please specify) 5. You purchased your lawn mower because of (please check top 2 reasons): Type Price/Value Packaging information Manufacturer reputation Sale/promotion Warranty Recommendations Other (please specify) 6. How much did you pay for your lawn mower? Less than $100 $100-$199 $200-$299 $300-$399 $400-$500 over $500 7. Where did you buy your lawn mower from? Large supplier store Special equipment store Discount store Online Direct catalogue Other (please specify) 8. What type of mower is yours? Rotary Cylinder Hover 9. What type of power source does your mower use? Battery Electric Gas 10. Which group best describes your family income? Under $15,000 $15,000-$24,999 $25,000-$34,999 $35,000-$49,999 $50,000-$75,000 over $75,000 18 Suggestions: The research objective has been stated: change the image of a technology feature-led company to one focusing on fashion and brand value. Before coming to the actual design of the questionnaire, it is necessary to determine information requirements. Hypotheses should be stated and questions designed to elicit support for the hypotheses. Following are some sample questions: The survey will be done over phone on a cross cultural basis, so open-ended question is probably the most appropriate format. Do you own a camera? If yes, what brand? What do you generally use your camera for? Why did you select your camera brand? What features of the digital camera are important to you? How much did you pay for your camera? Other questions about demographics of the interviewees. The research is conducted on cross cultural basis, therefore some caution should be taken regarding question formats (e.g. open ended question, direct/indirect question, verbal/non verbal), wordings, consumer behavior under the influence of socio-cultural or economic environment, etc. PART II SECTION C TEACHING NOTES FOR CHAPTERS CHAPTER THIRTEEN EXPERIMENTATION Learning Objectives • Define experimentation and discuss the distinction between experiments and other types of research approaches. • Discuss the concept of causality and the conditions required to infer causality. • Describe the distinction between laboratory and field experiments. • Discuss the concept of validity as applied to experimental research and distinguish between internal and external validity. • Discuss the different threats to validity that are controlled by each type of design. • Discuss the basic issues and terminologies involved in experimental research. • Describe the distinction between classical designs and statistical designs. • List guidelines for conducting experimental research and discuss the limitations of experimental research. • Discuss the advantages and limitations of the different types of experimental designs. Teaching Suggestions The material in this chapter normally requires two full class sessions. Appropriate objectives for these sessions are: 1. To further clarify the distinction between exploratory, descriptive and experimental designs. In particular, it is worth emphasizing that important decisions should be based on relationships that are robust, unambiguous, general (externally valid) and persistent. Only experimental evidence can satisfy each of these criteria. 2. To give students experience in using a checklist of threats to validity to evaluate past or prospective designs. 3. To reinforce the need to make conscious trade-offs between cost, time required and validity, in the recognition that anticipation of the threats to validity means that the extent and nature of the bias can often be incorporated into the evaluation of the results. Our experience has been that these objectives are best achieved by having the students evaluate a large number of designs, and suggest improvements. This is the reason for the seven mini-cases in the questions for discussion section, in addition to the four longer cases dealing with the same issues. One of the objectives of this book is to avoid giving the reader the impressions that marketing research in general and data analysis in particular, is simply a matter of turning a crank or reading a cookbook. There should be a continuing effort to generate and apply theory about the relationships that are hypothesized and that emerge from the data analysis. This chapter serves the important function of indicating the richness of causal relationships that can exist when a third variable is added to the analysis. In particular, an association between two variables can simply be spurious, resulting from the fact that both have a common cause. In data analysis, the researcher should always be on the look out for the possibility that associations may simply be spurious. In fact, one of the conditions for causation is that it is not possible to identify alternative explanations for an association. (The other is that a cause-effect time lag is present). The chapter also presents and discusses intervening variables, additive effects, interactive effects and direction of causation, all in the context of causal arrows. The goal is really to get the student to realize that these different relationships exist and can be analyzed by cross-tabulations so the theory building process will be richer and less constrained. Questions and Problems 1. Descriptive and experimental research are not totally different. Experimentation is designed to test causality whereas descriptive research is primarily used for describing the phenomenon under investigation. Experimental research, by systematic manipulation of causal variables and greater control over extraneous variables permits causal inferences to be made with confidence. Both descriptive and experimental research can measure influence but only experimentation can test the causal relationship between variables. The essential difference between the two types of research is in terms of the confidence level associated with the causal inferences stemming from them. 2. Association measures by themselves do not demonstrate causation because they can be the result of extraneous variables. For instance, let us consider association between the number of prisons and churches in a city. It is impossible to imply the causality between the two. There is the presence of a third variable (i.e.) size of city that influences both the number of churches and prisons. It is not so much that the association is spurious but it is the causal interpretation that is spurious. Association measures can be used in prediction even if no causal analysis is involved. A major task of data analysis is to help the researcher identify spurious associations, determine the nature and extent and to correct them. An intervening variable is very different conceptually from a variable causing spurious association. The spurious association can be distinguished from the intervening variable by determining the direction of causation between the variables. If the third variable seems to intervene between two variable, then the association is not spurious. However, if the third variable either caused the other variables or is associated between the other two variables, then the association is spurious. 3. An association between two variables does not necessarily indicate the presence of a causal association. Extraneous variables could have been responsible for the presence of this association. The students should discuss three principal components for imputing cause and effect relationships, namely, concomitant variation, temporal ordering and control over other extraneous factors. 4a. Experimental treatments: The alternative manipulations of the independent variables being investigated are experimental treatments. For instance, in an advertising experiment, low exposure level and high exposure level are all experimental treatments. b. Manipulation: The creation of different levels of the independent variable is known as manipulating the variable. In an experiment, the independent variable is being manipulated and the effect of manipulation on the dependent variable is observed. c. Extraneous variable: Variables other the manipulating variable that affect the response of the test units and hence the results of the experimental research. They are known as confounding variables as they may interfere with the changes in the dependent variable and thus confound the experiment results. d. Selection bias: If an experimental group is systematically different in some relevant way from the population being studied it invalidates the result of the study and is known as selection bias. e. Randomization: A procedure in which the assignment of subjects and treatments to groups is based on chance. Randomization ensures control over extraneous variables and increases the reliability of the experiment. f. Blocking: Even after the adoption of random assignments of subjects and treatments to groups, it is possible at times for the experimental groups to differ in a systematic manner on some relevant variable. Blocking is a procedure by which a non-manipulated variable is introduced into the experiment to ensure that the groups are equalized on that variable. g. Matching: The procedure by which the assignments of subjects to groups that ensures each group of respondents on the basis of pertinent characteristics. Matching helps reduce the experimental error that arises out of the selection bias. 5. The discussion in this case should center around the difference between control and generalization. It should be emphasized that there is a tradeoff between control and generalization of the results that helps the researcher in choosing the type of research. Laboratory experiments have greater control than field experiments. Field experiments have more generalizability than lab experiments. Laboratory experiments are conducted in artificial settings with all the extraneous factors virtually controlled whereas field experiments are conducted in natural setting. There is an inherent limitation to a lab experiment. Due to the artificiality of the setting, there is a question whether the established causal link will hold in natural conditions. Similarly, though the results of a field experiment are generalizable, it cannot be stated confidently that the manipulated causal variable is responsible for the effect. In this case, the need for control in the experiment is greater than the need to generalize the result. Hence all the extraneous influences are shut out and the effect of the advertisement on the brand recognition and purchase behavior (due to the favorable attitude) is studied. In case of a field experiment, it is not possible to attribute the purchase behavior only to the effect of the advertisement. Hence a laboratory experiment would have been preferred by the researcher. 6. The students should be exposed to the difference between demonstrating that the effect was due to manipulation versus generalizing that effect to a larger population. Internal validity refers to whether the criterion variable was the result of the manipulation whereas external validity is concerned with the generalization of the results to the larger population. Field experiments tend to have lower internal validity and higher external validity in contrast to the laboratory experiments. The various threats to internal validity are outlined: History, Maturation, Testing, instrumentation, Statistical regression, Selection bias, Mortality and Selection-maturation interaction. 7. The experiment may accurately determine that 60% or whatever preferred the new formulation, but several serious issues of external validity remain. The major problem is that the perception of the image of the product as personified by the name and package can swamp taste differences especially for those that do not have well defined senses of taste (which are most people—a substantial number of people cannot distinguish between 7-Up and Cola and certainly can’t distinguish between colas). Thus, it can be argued that a blind taste test has very limited external validity and usefulness. 8. The textbook contains a summary of most of the terms, and detailed definitions and illustrations can be found throughout the chapter. 9. Experimentation differs from the other research approaches in that in an experiment, the researcher manipulates the independent variable(s) before measuring the effect on the dependent variable. In this case, the experimentation is not possible due to the fact the independent variable has already been conducted and hence measurement is not possible. Descriptive research is primarily used for describing the phenomenon under investigation. Hence descriptive research is most suitable wherein the researcher can use the company’s internal data to ascertain past sales. The sales performance of the individuals who attended the sales training program can be found out and compared with their past performance to see if the sales training had an effect on the employees. Also, a survey can be conducted amongst the participants of the sales training to elicit their views on their pretraining sales efficiency and post training sales efficiency. A multi-item scale can be developed to measure the employee attitudes. However, it should be remembered that descriptive research does not offer much confidence in making causal inferences (i.e.) sales training improved sales performance. 10. The key principle in this experimental study is manipulation of the exposure to rap music (X) followed by observance of the behavior (Y). If change in the manipulation causes the behavior to change in a hypothesized way, then we can safely attribute that X caused Y. In our case, two groups of teens can be identified: one group with a confirmed record of violent behavior ( could be taken from juvenile homes) and another without record of violence. Various questions can be asked to the two groups on their exposure to rap music and the association between the action and the outcome can be established. The study can be done in various cities across the United States in order to generalize the results of the findings. The limitations of experiments are that it is costlier than either survey research or observation based research. Also problems could be encountered in implementation of the experiments. It may be extremely difficult to obtain cooperation from the respondents or the group of respondents may be mistakenly included in a test. 11. The students should be made to go through the distinction between laboratory and field experiments. The advantages and disadvantages of both the categories are identified. Then, the different ways of conducting this experiment are considered and the pros and cons are assessed and the type of experiment to be conducted is determined. 12. True experimental designs offer the most effective control of internal validity. 13a.This is an example of the one-group, before-after design. The attitudes and perceptions of the smokers are measured and then the group is exposed to the literature. At the end of stipulated time period, the attitudes and perceptions are measured again. This establishes the effectiveness of the literature campaign to stop smoking. b. The before measure may alert the respondents to the fact that they are being studied. This may prompt them to give socially desirable responses and behavior. Before measure can also lead to heightened attention and may lead to discussion with friends and hence modification of behavior is possible. There is also the chance that the respondents might drop out of the study due to various reasons. Also, the change in the interviewer or change in wording might lead to a bias in the responses. 14. The internal validity threats possible are discussed with the students. The adoption of a random assignment procedure will help in that every member of the universe has an equal probability of being chosen for that assignment. In this case, Cheryl Martin is entrusted with the responsibility of assessing women’s attitudes to pantyhose. She can take two categories of women, the working women and the non working women. In each category, she can study the attitudes of the various age categories. The study can be done across cities thereby ensuring randomization and hence threats to internal validity are eliminated. 15a.In trend studies, the measures over time come from a succession of separate random samples from the same population and yield much of the basic information on which marketing decisions are made. The data collected from trend studies can only be analyzed in the aggregate form in which they are collected. The availability of data across a number of time periods means that maturation (changes in the respondents that are a consequence of time) is not likely to be an effect. However, history ( events external to the environment that affect the responses of the people involved in the experiment) and instrumentation (the change in measuring instrument) remain as possible threats to validity. b. Of the possible threats to validity in trend studies, effect of history is difficult to rule out. If this design is to be used, there must be continuing sensitivity to plausible competing explanations. This should be done before the experimental treatment, so that the data is available to confirm and disconfirm the specified hypotheses. 16a. Independent variable: Exposure to anti-gun ownership commercials Dependent variable: Attitude towards gun ownership 16b.Design used: Two group-after only design 16c. Not possible to ascertain that the test group and the control group are exactly identical. Instructor Manual for Marketing Research V. Kumar; Robert P. Leone; David A. Aaker; George S. Day 9781119497493
Download
Close
Close
