Marketing Research I (uc3m) PDF

TOPIC 2: Marketing Research I; Qualitative and Quantitative Methods © 2017 by Mercedes Esteban-Bravo & José M. Vidal-Sanz All Rights Reserved. No part of this publication can be reproduced, stored, or distributed in any way (either manual, electronic,...

TOPIC 2: Marketing Research I; Qualitative and Quantitative Methods © 2017 by Mercedes Esteban-Bravo & José M. Vidal-Sanz All Rights Reserved. No part of this publication can be reproduced, stored, or distributed in any way (either manual, electronic, recording, mechanical, photocopying or including information storage and retrieval systems), without explicit permission from the authors. Updated in 2020. 1 Contents Theme 2: 1. Introducing Marketing Research 2. Qualitative Research Methods 3. Quantitative Research Methods 4. Surveys 5. Experiments and quasi-experiments 2 1. Introduction to Marketing Research “Marketing research is the function that links the consumer, customer, and public to the marketer through information--information used to identify and define marketing opportunities and problems; generate, refine, and evaluate marketing actions; monitor marketing performance; and improve understanding of marketing as a process. Marketing research specifies the information required to address these issues, designs the method for collecting information, manages and implements the data collection process, analyzes the results, and communicates the findings and their implications. ” (Board director of the American Marketing Association, Approved October 2004). 3  Marketing Research (MR) goal is to acquire data and process it into useful information from the micro and the macro environments, detecting opportunities and threats, guiding decision planning, implementation and control processes. The systematic use of MR to guide marketing decisions provides a competitive advantage. But there are some trade-offs 1. MR has an economic cost; Besides: If MR is erroneously done, or the information is misinterpreted it can lead to errors. There is a small risk of information disclosure to rivals. It can delay relevant decisions. 2. MR bring value to the company. It depends on the relevance of the decision at stake, the uncertainty surrounding the decision outcome, and the ability of the information to reduce it. Both features must be balanced before organizing MR processes. MR can be conducted or commissioned by a wide range of organizations (manufacturers and distributors, non-governmental organizations, political parties, etc.) 4 Marketing Research for the marketing mix Product testing: (is it the taste of this food appropriate?) Product Consumer’s satisfaction with the product Package testing (size, color, shape, ease of use) Brand name evaluation (is it appropriate for the product?) Price What price will consumers pay? Will some segments be willing to pay more? What response should we make to a competitor’s price change? Research on traffic flow patterns in stores and where to place Place products Feelings about suppliers, Satisfaction with ordering procedures? Which are the best locations for a shop? How many consumers recall a brand slogan? Promotion Ad testing. Sales promotion effectiveness? 5  1.1. Types of MR Study A) By information source (primary vs secondary sources) 1. PRIMARY MR. This type of MR collects directly the required data from primary sources (with first hand information). Typically, this research requires some time and higher economic cost, but it is more reliable and useful for the problem at hand than secondary MR. 2. SECONDARY MR. This type of MR collect information from secondary sources (secondary sources are third parts that have previously gathered information on the problem at hand, usually for other purposes). Typically, it is cheaper and faster than primary MR, but less reliable, and often less useful for the problem at hand (they can be outdated, or be focused on a different geographic area) It can involve internal sources (other functions inside the firm, different from marketing) or external sources (commercial studies by universities, governmental or non governmental bodies, statistical data,..). Data can be ready to be used, or require further processing. 6  National and local governments (e.g. statistical compilations, census information, different registers, laws, etc.), regulatory bodies Intergovernmental organizations (eg United Nations, European Commission), and SECONDARY non-governmental organizations (consumer associations). Commercial databases, compilations and directories (for example, telephone directories, lists of manufacturers, competitor reports, specialized business indexes) EXTERNAL Reports of companies specializing in market studies (e.g., A.C. Nielsen, Gallup) and specialized departments in universities. SOURCES Syndicated databases: elaborated by MR organizations that provide information from a common database to different companies that subscribe their services. (Surveys, purchase panels, media panels, scanner sales tracking data). Encyclopedias, books and media; Including newspapers (newspapers, magazines), Radio and Television. SECONDARY budgets, schedules of expenditures, accounting records, salesman reports, data from billing records on shipments, records on product returns, service records, customers correspondence, etc. INTERNAL SOURCES 7 Evaluating Secondary Data Who collected the data? Motive Why were the data collected? What data were collected? Essence How, and when, were the data collected? Do multiple sources indicate the same conclusions? Trust Is the study professionally conducted? 8 B) By the type of managerial problem 1. EXPLORATORY RESEARCH (for non-structured decision problems, where the boundaries of the problem are not clearly delimited). The MR goal is to understand the problem better, and to build a general framework of thought. Sometimes this is all we need to make a decision, usualy after this stage we need more specific information and we do: 2. CONCLUSSIVE RESEARCH (for structured decision problems, that have been previously studied, so that the general framework is clearly outlined). The MR objective of the research is to find out and convey final specific details to help decision makers. 9 C) By the type of inquiry method used (already focusing on primary research) 1. QUALITATIVE RESEARCH METHODS The goal is to obtain a qualitative understanding of the phenomenon under study to build an initial understanding (exploratory). The main techniques are: focus groups, in-depth interviews, indirect (projective and creative) methods, unstructured observation and ethnography,. They are largely exploratory. Their goal is to collect verbal and visual (sometimes even numerical) information from a sample of the population, but they do not allow any inference to the population. The procedures are not standardized, they can be adapted to meet the characteristics of the sampled individuals, adaptively to the information being collected. 2. QUANTITATIVE RESARCH METHODS They collect numerical information from a sample, and sample information can be inferred to the population of interest using different statistical and econometric tools. They can involve different categories of statistical data (cross sectional, time series, panel data, spatial data, duration data) They are mostly conclusive. The purpose is to recommend a final course of action. The main techniques are: surveys, structured observation, experiments. The procedures are structured (standardized) to obtain patterns of data under relatively homogeneous conditions, or controlled forms of heterogeneity. 10  QUALITATIVE QUANTITATIVE RESEARCH RESEARCH SAMPLE: Small number of SAMPLE: rigorously non-representative cases controlled to allow statistical DATA COLLECTION inferences PROCEDURES: Unstructured, DATA COLLECTION flexible. PROCEDURES: structured ANALISYS: Largely verbal, (standardized), tightly subjective controlled. ANALYSY: Based on statistical and econometric tools 11 There are several levels of analysis for quantitative research DESCRIPTIVE STATISTICS Data summaries and graphs (somewhat exploratory). UNIVARIATE STATISTICAL Estimation of parameters, density estimation, hypothesis testing. INFERENCES MULTIVARIATE STATISTICAL Study of statistical dependences (estimation, tests, forecasts), linear and nonlinear regression and ANOVA, time series analysis, panel data analysis, INFERENCES dimension reduction methods, cluster and discriminant analysis,… STRUCTURAL ECONOMETRIC Simultaneous equations with endogeneity, (estimation, testing, simulation). MODELS Estimation and testing about “causal” effects of treatments (manipulated EXPERIMENTAL ANALISYS variables) over some other variables. 12 Summarizing: FOCUS GROUPS IN-DEPTH INTERVIEWS QUALITATIVE RESEARCH INDIRECT TECHNIQUES (PROJECTIVE, CREATIVE) UNSTRUCTURED PRIMARY OBSERVATION RESEARCH SURVEYS QUANTITATIVE STRUCTURED OBSERVATION MR RESEARCH EXPERIMENTS Sales volumes, margins, INTERNAL accounting records,…. SECONDARY RESEARCH Publications, EXTERNAL statistics, compilations 13  Traditionally, MR was conceptualized as independent operations for specific problems Nowadays, it is more commonly identified with a sub-function of marketing, and there are stable systems to conduct regularly MR. 1.2. Market Market Information Systems is an organizational unit focused on gathering info in Information a continuous way, using internal and external sources by a stable team. Systems (MIS The MIS could be considered as part of the management information systems; but this is a or MkIS) more diffuse concept (that may include the control of any daily operations and transactions), while the MkIS is specifically focused on information to supporting strategic and tactic marketing decisions. 14  Traditional Data are collected for specific studies Marketing to solve particular problems. Research Marketing It collects information, continuously, Information using internal and external sources, Systems (MIS) with a stable team. 15  MkIS 4 subsystems involved Market Marketing Marketing Internal data Intelligence science/ analytics Research Internal Information from the Specific MR information Data analysis: statistics, external environment projects, using systematically econometrics. Machine collected continuously Qualitative and collected (Sales, learning. (data from customers, quantitative costs, inventories, Sometimes it includes a distributors, suppliers methods (as in financial decision support system and secondary sources) traditional MR) accountancy) (DSS) using optimization and simulation 16  Internal data subsystem, Compilation and systematic management of all the key information generated within the organization. Marketing intelligence subsystem: Systematic compilation of external information from the micro and macro environment, using primary and secondary sources.. Marketing Science /analytics, subsystem specialized on the statistical and econometric data analysis, building models for prediction, and in some cases to optimize decisions and for simulation. Some firms use relatively computer-based algorithms known as decision Support Systems to guide their decisions. It is better to have specialists in optimization than an automated software. Market research subsystem, to conduct the classical MR operations (analyzing specific problems and scenarios). 17 MIS continuous process Determine information needs Comunicate the Plan the data findings, and collection process their implications Implement the Analysis and results Data collection 18 1.3. Planning MR MR must be planned in advance. One must 1. First consider the problem at hand, information needs, potential value of information, MR budget and time. 2. Determine the type of data most convenient, and the target population, 3. Define the type of MR study (qualitative focus group, quantitative survey, experiment…), and the data collection method (personal interviews, mechanical procedures, self-administered questionnaire,…), 4. Determine how the sample is selected 5. Specify the research plan (people hires and/or external contractors, facilities, implementation procedures, possible use of pre-tests). 19 2. Qualitative Research It is the main form of exploratory research (after some exploratory research using secondary sources). The goal is to gain a deeper insight on weakly structured problems, and therefore understand better the nature of the problem. It is often used to Diagnose a situation Screening a group of alternatives Generate new ideas Withdraws: Subjective and personal. Does not provide conclusive evidence Results are not directly generalizable to the population. Subsequent research expected 20 A classification of qualitative research procedures: Focus Groups Word association DIRECT In-depth QUESTIONING Interviews Cartoon or Sentence Completion Thematic Apperception Tests PROJECTIVE etc…. METHODS QUALITATIVE INDIRECT RESEARCH CREATIVE Delphi METHODS QUESTIONING METHODS Brainstorming Direct observation UNSTRUCTURED Indirect Observation OBSERVATION Biometry 21  Direct Questioning approaches The purpose of the study is disclosed to the responders. It involve directly asking about feelings, behavior and beliefs (e.g. through a non structured interview). Indirect Questioning approaches The purpose of the study is disguised to the responders, who are asked indirectly. The goal is to get information from their subconscious mind; we distinguish: (1) projective methods collect sensitive information (about feelings behaviour and beliefs indirectly), (2) creative methods lower conscious barriers to enhance creativity. Observational methods, They perform a naturalistic inquiry of observable individual behavior in real world or the lab (but we cannot observe feelings of beliefs). Observation can be unstructured (qualitative) or structured (quantitative), it depends on the specific technique used and the sample design 22 2.1. In-Depth Interviews An unstructured, direct, personal interview in which a single respondent is questioned and probed by an experienced interviewer to uncover underlying motivations, beliefs, attitudes and feelings on a topic. The interviewer plays a crucial role, it must generate a relaxed environment, it must motivate respondents to answer, showing empathy with the respondent. But interviewers cannot accept a simple yes or no answer, but find out the reasons, and detect when probing is required Advantages: depth of insight, detailed responses, no social pressure so it is convenient for sensitive or embarrassing topics (e.g. avoidance of the law), meetings are easy to arrange (just two people) Challenges: lack of structure, length of interview, combined with high costs, and the analysis and interpretation of data is difficult 23  The interview can be unstructured, with free format (as a dialogue), but typically it is semi-structured, following an Question guide, script or agenda (but the interviewer can make detours on interesting features that come out). The interview is recorded (text and image), the test is transcribed in written for further analysis. The interview can include activities (games, board moods, projective tests,..). Several individuals are typically interviewed, the selection tries to cover alternative profiles in the target population. The goal is not represent the population, but to cover most profiles possibly associated to alternative answers. It could target children, or professionals. Typically several interviews are conducted, and the transcriptions are analysed together by an experienced analyst or psychology specialist, searching for interesting individual ideas, and for common patterns of answers, specially in connection with the interviewee characteristics. Costs and time requirements are the main drawbacks of this method. 24 2.2. Focus Groups A semi-structured group interview conducted by a trained moderator among a small group of respondents, typically between 1-5 hours. The discussion is usually conducted in a facility with a two-way mirror, it is audio/video recorded, and then transcribed in text. The participants stimulate each other to bring in new ideas. The moderators should enhance spontaneity and passion for the issues under discussion, but covering the question agenda. 25  EACH GROUP HAVE 6-12 PEOPLE IN EACH GROUP HAVE USUALLY, SEVERAL FOCUS GROUPS GATHER THEY CAN INCLUDE PEOPLE (SENSITIVE AND SIMILAR PROFILES, THEY ARE INDEPENDENT FOCUS GROUPS CONSIDERABLE INFORMATION PROJECTIVE METHODS, PROFESSIONAL ISSUES ARE SCREENED BY A SORT ARE CONDUCTED, FIRST IN A SHORTER TIME THAN IN- CREATIVE TECHNIQUES, ETC. DISCUSSED IN THE SMALLER QUESTIONNAIRE, THE RUNNING ONE FOR TESTING, DEPTH INTERVIEWS, AND IT IS GROUPS). MEETING CAN BE MORE THEN OTHERS, STOPPING IF CHEAPER. DIFFICULT TO ARRANGE (AS NEW IDEAS ARE NOT THERE ARE MORE PEOPLE COLLECTED. INVOLVED). 26  The moderator plays an important role, it must be a good observer, flexible and permissive, but keeping the control of the discussion encouraging participants when is needed. An advantage is that participants’ interactions may lead to a snowball of really valuable ideas. A drawback is that the moderator cannot devote much time to explore each idea, so the information is wider but less deep. The analysis of transcription is similar to the in-depth case, it is subjective and do not allow extrapolations to the population. Is probably the most commonly used qualitative MR tool. It serves also to prepare a quantitative survey (extracting vocabulary for the questionnaire), to understand possible attitudes, to detect unsatisfied needs of consumers, to evaluate new product concepts or generate ideas. It is generally exploratory, but sometimes it is conducted after a conclusive quantitative research, in an attempt to get intuitive interpretations for surprising results. Recently, some firms conduct online focus groups (this is cheaper and can cover wide geographic areas), but the interactions are weaker, and this may reduce the spontaneity and creativity. 27 2.3. Projective techniques 28  For example, in many cases they ask These are unstructured and indirect participants to interpret the form of questioning that encourages behaviour, beliefs or attitudes or respondents to project their others, so that in the answer they underlying motivations, beliefs, project their own. Alternatively, they attitudes, or feelings regarding the are sometimes asked to act in a play, issues of concern. The goal is to reach to interpret ambiguous scenarios, the subconscious mind. etc. 29  There are several categories: A) Association Techniques: they try to obtain mental associations between concepts, ideas or between attitudes. For example: word association test. The respondent is presented with a list of words, one at a time and asked to respond to each with the first word that comes to mind. The words of interest, called test words, are interspersed throughout the list which also contains some neutral, or filler words to disguise the purpose of the study. Common answer patterns reflects idea associations. Answering delay is due to some emotional involvement, and no answer is due to high emotional implication level. 30 31 B) Construction Techniques: The subject is required to construct a story, or an art form, from the materials provided. For example thematic apperception test (T.A.T.), where respondents must elaborate an history interpreting cards with ambiguous scenes. Also third person interpretation, explaining the behavior or attitude of someone, or building a mood board (or collage). Note: Some authors distinguish between tests of construction tests are based on materials with a defined theme, such as TAT cards, and tests where the construc on is based on a material with little structure (such as Rorschach's test, based on inkblot interpretation). The former are called thematic techniques, and the second structural techniques. But other authors include structural techniques (such as the Rorschach test) in the association techniques category, since the projection in these cases is quite weak (merely an association between concepts). We follow this idea, using indistinctively the name of construction techniques or thematic. 32 What do you think has happened in this scenario? 33  Who do you think probably drinks more milk, Ms. A or Ms. B? 34 35 C) Completion Techniques: participants are shown some type of incomplete object or stimulus, and they are required to complete it freely but in a rational way. For example, sentence completion, People who drink beer are ______________________ A man who drinks light beer is ___________________ Imported beer is most liked by ___________________ A woman will drink beer when______________________ A variation of sentence completion is paragraph completion, in which the respondent completes a paragraph beginning with the stimulus phrase. Others are unfinished scenario, where respondents are provided with part of a story and required to give the conclusion in their own words, also cartoon completion test where the respondents complete a cartoon 36 D) Ordering or choice Techniques: the respondent is shown a number of alternative objects, then it is required to select some or to rank them using a given criterion. E) Expression Techniques: They require the respondent to create some artistic expression For example: Draw a picture, often used with children. Role playing where participants are required to act as a third person in some context. 37  These methods are classical in strategic marketing. Are particularly useful to study ideas and expectations for a group of people. Types: Brainstorming. Individual opinions and ideas are freely exposed without any critique. A variety of procedures can be used to stimulate participation and construction, lowering mental 2.4. Creative barriers (e.g., circulating anonymous pieces of papers with alternative ideas, contributing to them, finally sharing the Techniques results in group and voting for the best ones) Delphi. Some consensus between experts is attained from different individual approaches. They provide a numerical outcome forecast on a complex issue. The facilitator collect answers, and the median is computed. In a second round the experts receive this information, then they are asked to review their initial forecast, the process is iterated until it nearly converges, avoiding “ego clashes” in arguments. 38  2.5. Unstructured Observation Methods Esta foto de Autor desconocido está bajo licencia CC BY-SA 39 A) Direct Observation These methods record behavioral patters of people, objects and events in a systematic manner to obtain information about the phenomenon of interest. They are accurate, though sometimes the interpretation (why) of the observed behavior can be subjective. For example: Mistery shopper (an observer disguised as a shopper: watching grocery store shoppers approach a product category, shop cleanliness, or product replenishment levels, and sales assistant behavior). It can be based on human observers, or mechanical systems (e.g., behavior in a social network, behavior in a shop recorded by videocamera). In some cases there is interaction between researchers and studied population, in others not. It is generally combined with other methods. It is not very expensive 40 41  For example, tracing pedestrian traffic in different streets can help to find attractive commercial locations Streets with higher volume of weekly pedestrians in Madrid with potential taxonomy Pedestrians, potentials From 10.001 to 20.000 > 20.000 Sales point A Sales point B 42 B) Indirect Observation Sometimes the actual behavior is difficult to observe, but we can observe clues or leftovers to infer the behavior. For example garbage analysis, studying the garbage of families we can infer the actual consumption. Measuring the occupancy of a parking slot we can infer the number of shoppers in a large mall. The supermarket hallways with more worn floor provide evidence of a high circulation of customers. The wared off floor in aisles in a supermarket provide evidence of customers high circulation. Also cookies in a computer provide evidence of previous WWW pages visited. 43  These techniques observe biological response of participants when they are exposed to specific stimulus. C) Biometry They can measure the level of attention, interest, and in some cases some basic form of mental process. 44  Galvanometer It is an instrument for detecting and measuring electric currents. It can be used to track the level of general arousal when the participant is exposed to a marketing stimulus The level or arousal induces small variations in out sweet, which affects skin electrical conductivity. Measuring it we can track the arousal level. There are new galvanometers similar to electronic watches, that can be carried by customers during their shopping experiences, etc. 45  Eye trackers (Eye-movements recorders): a device to measure either the point of gaze (where one is looking) or the motion of an eye relative to the head. It is used to study the experience of viewing pictures of advertisements, signs, packages or shelf displays. It is commonly used. 46 The heat maps below show how scanning patterns changes with a new design. Instead of scanning the entire package (old design to the left), respondents’ gaze were more focused to key elements (new design to the right). These heat map and gaze plot visualizations show that the new 3D label on a shampoo bottle (the bottles to the left on the shelf) attracts more attention than the old label (the bottles to the right on the shelf) where gaze patterns are less focused and more outspread. 47  Voice Pitch Analysis: Measures indicate emotional reactions through physiological changes in a person’s voice (tension in Larynx) Esta foto de Autor desconocido está bajo licencia CC BY-SA 48 Neuromarketing (or neurophysiology applied to MR). Based on neuro- psychology, using brain images. (different techniques TAC, RMN, PET, fMRI, etc.). Esta foto de Autor desconocido está bajo licencia CC BY-SA 49  3. Quantitative Research Conclusive research methods aimed to collect numerical information from a sample, analyze the data and infer the results to the population. It can collect data using questioning or observation methods, but the procedures are highly structured (standardized systems). The method used to select the sample of participants is crucial, if the sample is not controlled the collected data cannot be used to extrapolate inferences (and the results must be regarded as exploratory) The metrics/measures must be carefully planned, to ensure that collected numerical data do not include systematic errors, and the variance of measurement errors is relatively small. 50 Esta foto de Autor desconocido está bajo licencia CC BY-SA Types of quantitative research methods: Survey methods, collect numerical information asking a controlled sample of individuals about relevant topics (e.g. purchase intentions, preferences, attitudes). Structured observation, for collecting numerical information about observable behaviour under structured conditions (e.g. scanner sales data from stores, ratings for share audience), and Experiments, collecting numerical data through controlled simulations conducted either in laboratory environment or in the field (e.g. in a shop). The goal is to study cause-effect relations under specific treatments. (e.g., advertising copy test, product tests). 51  3.1. Surveys Surveys are methods for collecting structured information by interrogating individuals in a sample, where the sample is selected in a controlled manner from a target population. The objective is to make statistical inferences from the sample to the population. Surveys are often classified by the sample extension and size. (For example, local, regional, national, or international basis), Surveys also vary in the way of questioning (face-to-face, postal, telephone, internet, streets or malls interception) Omnibus survey: This is a special case where several firms cooperate to develop a survey (often following a proposal by a marketing research company), sharing the collected data and the costs. Static surveys are called cross-sectional studies. Other surveys are dynamic because they are done regularly over time. We distinguish: Panels or longitudinal studies: these are dynamic surveys where a sample of selected respondents are regularly questioned in waves (with small changes due to attrition). Dynamic cross sectional surveys the sample varies independently for each wave. 52 Esta foto de Autor desconocido está bajo licencia CC BY-NC-ND There is no best form of survey; each has advantages and disadvantages. each type allows the collection of different amounts of information, and they vary in: The amount of data they can collect, The allowed complexity in the questions, The quality in the data (errors) The level o sampling control, and have different response rates (% of respondents that accept the invitation to participate) The different costs The time required for study completion Before planning a survey, we have to consider our budget, the available time, and the amount of information required (it depends on the use, the type of decision involved) 53 Survey design process Research Data report Questionnaire collection, analysis Sampling design Survey type specification Information needs, Budget and Time constraints 54 Basic types of surveys Personal or At home Face-to-face Streets. Malls Telephone STATIC or AD-HOC (Cross-section) Postal (Mail) SURVEYS Internet Panel (longitudinal) DYNAMIC Consecutive Cross-sections 55 Personal or Face-to-face interview There exist direct contact between the interviewer and the interviewee. It can be carried out in different places (home, street, mall) Advantages Disadvantages  High response rate (due to the physical  Higher economic and time costs than contact). In Door-to-door (at home) it can be other methods, specially for done by appointment, and respondents are more geographically disperse populations relax and have more time. Possibility of errors biasing the results  Support materials can be used (visual images, due to : product samples,..)  Interviewer influence  There are no third-persons involved  Question formulation (induction)  Answer transcription  The interviewer can gather additional information, and solve doubts adapting the questions to the intellectual level of the interviewee (explanations and support) 56  Personal in a Mall, shop Respondents are intercepted while they are shopping in malls Advantages Disadvantages  Especiallyappropriate when the  sample could be not respondents need to see, handle, or representative from the consume the product. total population Low costs A variation is the purchasing intercept method, clients are interviewed exactly if and when they choose some product category. 57  Telephone interviews Interviewers have only verbal contact with the interviewees through telephone line. For years, it was the dominant survey technique. Low cost, using a call center. Computer-assisted telephone interviewing (CATI). Advantages Disadvantages Quick completion of the process Questionnaires should be typically Allows adaptive questionnaires with branching shorter questions Possible responses are limited Low costs, even with a large geographical area (sophisticated response scales are not Traditionally high response rate (slightly lower than usable) that of personal interviews), is now being reduced Questions with visual helps cannot be The interviewer can solve doubts. used (e.g. packages or pictures) The interviewer's bias is reduced a little, since he is Still there is a possibility of not physically present. interviewer bias Easily managed (collection, recording and validation). 58  Postal or Mail interviews There is no verbal interaction between the researcher and the respondent. Self –administered questionnaire by postal letter. Advantages Disadvantages Slow (~2 months to complete) Low cost per completed Low answer percentages (10-15%) interview No one is available to help interpret Low potential for instructions or questions. interviewer bias. Don’t know/blank responses are more High flexibility of data frequent. collection. Complex questionnaire are not generally recommended. 59  Web interviews Advantages Disadvantages Relatively high speed of data collection, Cost effective High possibility for respondent Geographic flexibility, Worldwide misunderstanding Visual and interactive. Special Features: It Possible self-selection bias, unless allows graphics and streaming media respondents are invited (recipients who High versatility of questioning sign up at a special Web site thus "opt- Questionnaire length: it can be individualized in.” base on previous respondents’ answers. When opt-in not used, cooperation The software can assure none item non- is low (email is seen as “spam”) and response refuse None interviewer influence of answers. This method excludes people without Respondent can be anonymous Internet access, others lack powerful Ease of Callback or Follow-up (if the e-mail PCs with high-speed Internet address is known) connections. Some respondents are relatively unsophisticated computer users 60 A comparation of Ad-hoc interviews methods Postal Mail Telephone Web Personal Personal Mall- Door to door intercept Complexity and Not much Substantial Very flexible Very flexible The most flexible flexibility Information Substantial low Substantial larger limited amount Information Good for Reduce Good for Possibility of The environment quality embarrassing ambiguities embarrassing interviewer bias reduce its value questions questions Sampling Good Aceptable It can be High, Good Low control or Low (self- selection) % answers 10% 60%-80% 20-50% 80% 80% Speed Slow Fast Medium Relatively fast Very fast costs cheap medium cheap expensive Medium-expensive 61  Based on structured systems to gather numerical data about behavior. Sometimes the whole target population is observed, but usually only a sample is observed. Then the choice of the sample is crucial to allow inferences, it must be controlled. Data are typically recorded in panels. The sample is periodically studied in waves along some period. The sample is kept stably (though it could be some partial replacement of units due to attrition). 3.2. Structured Observation Consumer panels (daily purchases reported by a sample of consumers) Retail panels (scanner sales data from check-out counters from a sample of stores) Audience panels (e.g., television ratings for share audience, or newspapers circulation measured at newsvendor sites) Internet flow traffic panels 62 Commercial Examples  Consumer panels: one of the most widespread ones is http://www.acnielsen.es  Retail panels (a representative sample of the different types of retails stores), a relevant example is also http://www.acnielsen.es  Audience panels: see the examples by http://www.kantarmedia.com/es 63  3.3. Experiments Experimental Analysis is a methodology to study association (causal) relations between some variable (treatment) controlled by the researcher and at least one dependent variable. Usually, the data are obtained using data generated from an experiment under controlled conditions. The researchers administer a specific treatments (e.g. the exposure to different products or advertisements) to some of the experimental units. The aim of the study is to test for changes in the mean of the dependent variables under different treatments. A change in the mean is interpreted as evidence of a causal relationship. 64 Esta foto de Autor desconocido está bajo licencia CC BY-NC-ND  Often we consider two groups, a control group that receives no treatment and another exposed to the treatment, and the analyst studies the difference between the mean response in both groups. Sometimes there are a large number of treatments, then we would have many groups besides the control one. In some cases a measurement of the response variable is performed before and after treatment application (e.g., conducting two surveys) computing the difference for each individual. The mean difference is compared between groups (for example, ask about the intention of purchase before and after a particular ad, the mean increment is compared across treated and control groups looking for differences in the mean difference) There are several possibilities when designing an experiment. The most basic, rather problematic, are known as quasi-experiments. Denote treatment by X and observation by O. There are several quasi-experimental approaches (not recommended): A) “One group Pretest-Posttest Group Design” Experimental group: O1 X O2 Treatment mean effect is measured (O2- O1 ) Problem: differences can be caused by subjects tired with the experiment in the second measurement. B) “Static groups comparison” Experimental group: X O1 Control group: O2 Treatment mean effect is measured (O1- O2 ) Problem: There could be initial differences between people in the experimental group and the control. 66 Extraneous Variation can damage the experiment. There are often unidentified sources of variation in the dependent variable (unknown regressors). We can handle it as follows: 1.If they are hold constant, the change in mean is just due to the treatment. Therefore, the traditional response was: "Hold Experimental Conditions Constant," 2.In practice this is impossible to achieve. As a consequence, nowadays experimental design is based on Randomization. Randomization is the process by which experimental units (the basic objects upon which the study or experiment is carried out) are allocated to treatments randomly. Units are located in such a way that each treatment is equally likely to be applied to each unit. Not only treatment but any other known feature is assigned randomly to each individual (the interviewers, the date and time, facilities, and sometimes the order of questions ,... ). A control group is included to compare the results of individuals subject to a treatment, with the results for individuals not exposed to any treatment 67 A true experiment often used is the method C) “Test-Posttest Group Design with random assignment” or "classic" experimental design. It compares the posttests between the treatment group and the control group. If we study several treatments, we use a different experimental group for each one. Denote randomization by R, treatment by X and observation by O Experimental group: R X O1 Control group: R O2 Treatment mean effect is measured by: (O1- O2 ) Drawback: are differences caused just by the treatment X or there are other unobserved variables? 68 The most common true experimental design is D) “Pretest-Posttest Group Design with random assignment” or "classic" experimental design. It compares the posttests between the treatment group and the control group. Experimental group: R O1 X O2 Control group: R O3 O4 Treatment mean effect is measured with “Difference in differences”: (O2- O1 )-(O4- O3 ) 69  We must use statistical procedures to test if the means for the different treatment-groups are equal or not. These tests are based on ANOVA procedures (equivalently regression with dummy variables) and their refinements. More generally, if we pool all observations, experiments can be studied using regression models. The dependent variable is the response variable (or a difference in measures) for the individual. Treatments are represented using discrete or continuous regressors. After estimating a regression model, we can study treatment causality testing the global significance of the regressors. If the treatments are significant, then we should chose to apply the treatment with best response (for our marketing objectives). 70  Another true experimental design is the Solomon Four- Group Design which is more sophisticated in that four different comparison groups are used. Experimental group 1: R O1 X O2 Control group 1: R O3 O4 Experimental group 2: R X O5 Control group 2: R O6 It is used to study if the pre-test measure makes individuals sensible to the treatment generating the differences. We study (O2- O1 )-(O4- O3 ) but also we can measure pretest influence (O2- O5 ) and (O4- O6 ). Other comparisons can be considered 71  72 Actually, we could have several types of independent variables (regressors): Treatments: Treatments can be quantified with continuous variables, but usually they are described by categorical variables (e.g. product type, and package type) called factors. Each factor is described by a group of dummy variables, called levels of the factor. Setting a particular combination of factor we define a treatment. Factors and Blocks: Other causal variables controlled by the researcher when the experimental units are chosen (e.g. the sex of the Blocks participants, the educative level, etc.) or experimental (e.g., the shop where the product test is conducted). These extra factors are called "blocks." Note that randomization avoids biases, but it can increase the variability of the dependent variable. Introducing these block variables in the model we can reduce the variability of the experimental units.  When the experiment is performed in the laboratory, one must be cautious with the results. The findings are not completely reliable if extrapolated beyond the laboratory. laboratories do not always reproduce the Limitations: complexity of the real world. In the real world there may be more feasible behavior alternatives (e.g., more brands), and therefore the consumer response to advertising treatments can be different. In experiments outside the laboratory is not always easy to ensure randomization which eliminates the effect of other variables. 73  A falsified or poorly explained experiment can produce results that cannot be reproduced One example is the 1957 experiment about subliminal advertising at a New Jersey movie theater, by James Vicary. Vicary displayed the messages “Drink Coca-Cola” and “Hungry? Eat Popcorn” for only 1/3000th of a second at a time, imperceptible for viewers. The effect was a 18.1% increase in Coca-Cola sales, and 57.8% increase in popcorn purchases. Rogers, Stuart (1993), “How a Publicity Blitz Created the Myth of Subliminal Advertising,” Public Relations Quarterly, Winter, pp. 12-17. 74 Esta foto de Autor desconocido está bajo licencia CC BY-NC-SA UC3M UC3M