Analyse data for marketing research What is data? 3 Editing data 4 Field editing 4 Central editing (in-house editing) 4 Missing responses 5 Coding data 8 Pre-coded questions 8 Post-coded questions 9 Developing a code frame 9 Transcribing data 11 Computer assisted telephone interviewing (CATI) 11 Data cleaning 11 Statistically adjusting data 12 Data analysis: an introduction 14 Computer data analysis packages Descriptive statistics 14 16 Univariate techniques 16 Multivariate techniques 16 Tabulations 16 Cross-tabulations 18 Correlation 20 Hypothesis testing 20 Other descriptive statistics 20 Other statistical techniques 22 Summary 23 Key terms for your glossary Feedback to Activities and Check your progress exercise 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 25 27 1 2 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot What is data? Data becomes information when it is explained in a meaningful way for decision-makers. When data is first collected, it is called raw data. Raw data consists of responses contained in a questionnaire. It may be recorded in ways such as codes, scales or words. In most cases, this raw data is not in a format suitable for decision-making. Once it is coded, it is known as data. Data is most often analysed by computer. However, it can also be analysed manually, as in the case of qualitative data and especially when there are only small numbers of responses involved. Information refers to the information gained from processing the data and turning it into meaningful information or facts that can be used in decisionmaking. Activity 1: Vital terms For this activity, it would be useful for you to consult textbooks on the topic—preferably the ones suggested for this module. Ensure you understand the vital terms below. Also enter these terms in your personal glossary. 1 data coding _____________________________________________________________________ 2 data matrix _____________________________________________________________________ 3 data reduction _____________________________________________________________________ To check your answers, refer to the feedback at the end of this topic. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 3 Editing data The first step in data preparation is to ensure that questionnaire responses have been recorded accurately and completely. This process is known as editing. Effective editing will identify and rectify any inaccurate or incomplete answers. Editing can be carried out either in the field or at a central office or both. Field editing Field editing is carried out in the field either by the interviewer or, more commonly, by a supervisor or by both. The purpose of a field edit is to detect the most glaring omissions and inaccuracies in the data. The questionnaires are checked for: completeness legibility comprehensibility consistency uniformity. Central editing (in-house editing) Central editing, also referred to as in-house editing, is usually conducted from the fieldwork office. It can begin as soon as questionnaires start coming back from the field and continue until fieldwork is completed. While field editing is important and can quickly identify and rectify obvious problems, central editing by a single editor involves a more thorough and exacting scrutiny and correction of completed data collection forms. The main task of the editor is to identify and correct inconsistent or contradictory responses to ensure there will be no problems at the coding or data entry stage. Editors will sometimes have to make decisions about responses where answers appear to be incorrect. Ideally, they do this by going back to the interviewer to check the correct response. However, if this is not possible, a decision will need to be made about what the correct response would be. 4 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot In these cases the editor must exercise caution and objectivity based on predetermined principles or rules for making decisions. These pre-determined principles should not involve the editor changing or adding anything to the respondents’ answers that would change the results. For instance, it would not be acceptable for an editor to add or change a response to indicate that a respondent preferred one brand to another, if the respondent had not clearly answered this question. In editing questionnaires, mistakes must be rectified. In some cases, the person editing the questionnaire can safely assume what the correct response might be. In other cases, this is not so clear. Where possible, the interviewer should be consulted to establish if they can remember the correct response recorded. It may also be possible to return the questionnaire to the field and contact the respondent to clarify the response. However, in some situations, the questionnaire should simply be discarded. Where there is so much missing information that the questionnaire is invalid, or where responses are obviously incorrect, and then discarding the questionnaire will be the preferred option. Missing responses Perhaps the most difficult editing question to answer is what to do with missing responses from questionnaires. If contacting the respondent is not possible and discarding the questionnaire is not desirable, a substitute answer can sometimes be given by the editor or the researcher. A value such as ‘99’ can be used to show that the respondent did not answer the question. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 5 Activity 2: Common mistakes in questionnaires Some of the more obvious mistakes in questionnaires are outlined below. For each mistake, determine a method for editing the error. 1 An interviewer mistakenly records a respondent’s year of birth as 1850, or the gender of Mary Smith as male. _____________________________________________________________________ _____________________________________________________________________ 2 Contradictory responses may have been given. The respondent records that they don’t consume soft drinks but they record the brand they most regularly consume. _____________________________________________________________________ _____________________________________________________________________ 3 Income is recorded as monthly income rather than annual income as stated on the questionnaire. _____________________________________________________________________ _____________________________________________________________________ 4 The interviewer’s handwriting is illegible. _____________________________________________________________________ _____________________________________________________________________ 5 The interviewer meant to circle either (3) or (4) but the circle has mistakenly overlapped both codes. _____________________________________________________________________ _____________________________________________________________________ 6 Response is incomplete. When asked to nominate three brands, only one brand has been given. _____________________________________________________________________ _____________________________________________________________________ 6 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 7 In response to the questions about age, occupation and place of residence, the respondent claims to be a 101-year-old, one-legged clown in a travelling circus. _____________________________________________________________________ _____________________________________________________________________ 8 The demographics of the respondent do not match what is required for the sample or quota for the questionnaire. _____________________________________________________________________ _____________________________________________________________________ 9 Inconsistency in responses in which a respondent gives their occupation as a doctor records their income as $15 000. _____________________________________________________________________ _____________________________________________________________________ 10 Too much consistency. In answer to every multiple-choice question the respondent has circled (4). _____________________________________________________________________ _____________________________________________________________________ To check your answers, refer to the feedback at the end of this topic. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 7 Coding data Before any data analysis can begin, a process called coding has to take place. This involves: 1 coding the responses from questionnaires 2 sorting data into categories 3 allocating numerical codes to the categories 4 inputting the codes into a computer. Questions are either pre-coded at the time of questionnaire design or they will require coding after completion of the questionnaire. Pre-coded questions Pre-coded responses are suitable for the following types of questions: scale multiple-choice dichotomous. Pre-coding can be used when the researcher can confidently predetermine one of a number of responses to a question. Personal details such as age, gender or income are often asked in a ‘structured’ or ‘closed’ question format as the researcher knows the respondent will fit into one of the predetermined categories. Sometimes, questions can be partially pre-coded. This can be done through the use of the category ‘other’ where respondents are asked to specify their reason or choice. If a large number of respondents give ‘other’ reasons, then these responses must also be coded at a later stage. Pre-coding saves time and allows for quick and convenient editing of questionnaires and speedy computer input. 8 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot Post-coded questions Post-coding occurs with unstructured questions. The purpose of coding is to reduce the large number of different types of answers from respondents to a more manageable number of categories and to sort them into general categories. These categories are then allocated a numerical code for the purposes of data entry. Developing a code frame This initial step in coding involves developing a code frame. This is usually done in batches of 50 questionnaires. (The total number of questionnaires in a survey could be 500 or more.) 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 9 Activity 3: Developing a code frame Consider the following unstructured question. What do you like about this shopping centre? The range of responses to this question could include: 1 I like it because it is close to home. 2 I can catch public transport here. 3 I like the range of shops available. 4 It’s just my local shopping centre. 5 I can always find what I want. 6 I can catch the train there. 7 I like all the shops that are here. 8 I just come here to meet my sister. 9 It’s just a pleasant atmosphere. 10 The people in the shops are friendly. 11 I can always get what I need. 12 It’s convenient—that’s all. 13 There’s a nice café to have lunch in. 14 It only takes me five minutes to drive here. 15 I meet a lot of my friends here. Can you see any similar responses that might be grouped together? _________________________________________________________________________ _________________________________________________________________________ _________________________________________________________________________ To check your answers, refer to the feedback at the end of this topic. 10 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot Transcribing data Once the responses from the questionnaires have been edited and coded, the data is ready for input into the computer. Transferring the raw data from the questionnaires into a computer program is known as transcribing. Transcribing of data provides information for researchers to statistically describe various results from the survey. Computer assisted telephone interviewing (CATI) CATI is a procedure in which interviewing respondents and entering data is conducted simultaneously. The computer screen replaces the paper questionnaire and there is no need to physically record the responses on the questionnaire. Instead, the interviewer sits at a computer and records the responses directly into the questionnaire on-screen while conducting the interview with the respondent. One question at a time is shown on-screen and questions are skipped as appropriate. The computer also analyses the data as the interview is being conducted. CATI is a popular form of interviewing as it cuts down on coding and data entry time as well as ensuring that data quality is enhanced. Data cleaning While the data has already been checked for consistency and correct answers during the editing process, the computer program further enhances the editing process during the data entry stage. For example, the computer can easily identify any responses that are out of range: if there are five pre-coded categories but the number 6 appears as a response, then the computer will identify this incorrect response as 6 is out of the range for responses. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 11 Statistically adjusting data Weighted data Data can be adjusted statistically to enhance the quality of the analysis by a procedure known as weighting. Weighting is often used to make the sample more representative of the population of interest. For example, if males 18–24 were of particular interest to the researcher but the sample provided insufficient respondents within this age group, a weighting factor may be assigned to this quota to reflect the significance of this group of respondents. Variable re-specification When the initial run of computer tables has been completed, the researcher can make some decisions about the information. For example, some of the code frames may be changed or amalgamated to form more meaningful information. If there is a table showing computer ownership cross-tabulated by age groups, the researcher may decide to amalgamate some of the age groups. In addition, the researcher may consider that there is not enough difference between the age groups 51–65 and 65+ to warrant separate categories. You can therefore amalgamate or collapse the two codes into one new code, identified as 51+. There is little point in having additional categories if the information gained offers no meaningful insights into the subject. 12 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot Activity 4: Transcribing data What would you do if you were presented with the three problem situations below: 1 What can be done if the income bracket $100 000+ is of particular interest to the researcher, but there are insufficient of these respondents in the sample? _____________________________________________________________________ 2 If initial computer tabulations indicate little difference between respondents attending the cinema once a week and those respondents attending twice a week, what can be done to make the information more manageable and meaningful to the decisionmakers? _____________________________________________________________________ 3 If the editor of a particular questionnaire fails to identify a respondent who claims not to own a motor vehicle but nominates a model that they own, where will this most likely be identified? _____________________________________________________________________ To check your answers, refer to the feedback at the end of this topic. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 13 Data analysis: an introduction Data analysis is about turning the raw data from the questionnaires into information that is presented in a useable and easily understood format. Analysis is also sometimes defined as the process of resolving complex issues into their simplest and most manageable elements. Ultimately, analysis should provide the researcher and the client with the direction to make effective decisions. As a decision-maker, which of the following would you prefer to know? Respondent 1 answered ‘yes’ to Question 1, respondent 2 answered ‘no’, respondent 3 answered ‘no’ and so on. 47% answered ‘yes’ and 53% answered ‘no’. Analysis turns your collected data into understandable information. The purpose of data analysis is to produce information that will assist decisionmakers. In deciding how the data will be analysed, the researcher must consider: what information is required what information will be most meaningful and useful in the decision process. In selecting a data analysis strategy, factors such as research design, measurement scales and other characteristics of the data have to be taken into consideration. Different statistical techniques are appropriate for different types of data analysis. Computer data analysis packages Several computer packages are available for data analysis. Four popular packages used by the market research industry are SPSS (Statistical Package for the Social Sciences), SAS (Statistical Applications System), Minitab and Surveycraft. If you are purchasing a package for your firm, check whether you can buy the package outright. For many practitioners, a simple Microsoft Excel spreadsheet may suffice. 14 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot These packages have some differences but they all provide a means of turning raw data into information in a useable format. Essentially SPSS allows for more complex statistical analysis; however, Surveycraft is very popular as it provides all the main types of analysis that are of interest to marketers. Activity 5: Statistical packages Visit the websites below for descriptions of the basic operations of some of the major statistical packages. www.spss.com www.sas.com/products/stat/ Have a look around and check out some of the differences between them. Almost all the major packages, including Microsoft Excel, have the same range of basic statistics. Where they vary is in their specialist statistical tools. Investigate what sort of additional support you can get for each program. Also have a look at www.intelliquest.com for further resources. Using a statistical package If you have a statistical package such as SPSS, install the program and load one of the trial data sets you would have received with it. As we move through this topic, try out some of the data analysis techniques we have discussed. The information you obtain from these statistical tools can sometimes be confusing. If you want to learn what the numbers mean then read through your textbook(s). It has very comprehensive content on most of the data analysis techniques we will discuss. The help files contained within SPSS are also an excellent resource. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 15 Descriptive statistics Descriptive statistics are the most basic form of data analysis. They aim to describe the data in its simplest form. Descriptive statistical techniques can be classified as one of the following: univariate multivariate. Univariate techniques Univariate techniques (that is, techniques referring to one variable only) are used in analysing data when each respondent is analysed using one measurement only, for example, age. Univariate techniques are also used when a respondent is analysed using more than one variable; for example, age and income, but again, these variables are analysed separately. Multivariate techniques Multivariate techniques refer to the analysis of two or more variables at the same time, for example, age and income. These techniques would allow us to identify the ownership of personal computers among people aged 35–50 within the income bracket $50 000 to $100 000. They also allow us to compare different groups. For example, we can compare the characteristics of people who own computers with those people who do not own computers, across multiple variables. Tabulations Usually the first analysis of the data comprises a computer count of the number of specific responses to each question and a calculation of the percentages. This computer output is usually known as tabulation. Percentages are widely used in computer analysis as they are simple to calculate and easy to understand. 16 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot Example An example of computer tabulation from a survey of 500 may look something like the one below. 1 Which age group do you fall into? Less than 18 (1) 120 responses 24% 18–34 (2) 80 responses 16% 35–50 (3) 70 responses 14% 51–65 (4) 150 responses 30% Over 65 (5) 80 responses 16% 500 responses 100% Total You can see from this tabulation that 150 respondents are aged between 51 and 65 and that this group represents 30% of the total number of respondents. This is an example of a simple tabulation or a frequency count, as it counts the number of responses to each code. Responses to a scaled question would generally be tabulated in the same way; however sometimes it may be more useful to combine some of the categories when reporting findings. Example 1 Do you think we should abolish daylight savings? Strongly disagree 40% Disagree 30% Don’t know 10% Agree 10% Strongly agree 10% In this instance you may want to ‘collapse’ the categories of ‘disagree’ and ‘strongly disagree’, and likewise for those who agree or strongly agree with the statement. You could then report that ‘70% of those surveyed disagreed with the idea of abolishing daylight savings, while 20% of those surveyed thought it should be abolished’. This allows for clearer communication of the findings, rather than just reporting raw figures. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 17 Cross-tabulations Cross-tabulation of data covers two or more variables at the same time. This can give us a table that shows the joint distribution of the two or more variables. The category responses from one question are cross-classified with the category responses from another question. Many market research surveys do little more than cross-tabulations simply because they provide a large amount of useful information with no need for complex calculations or statistics. Example The Swallow Drinks Company may want to know how many people in the each age group have purchased a single serve drink in the last three months. To do this they would cross-tabulate the results of a question about purchase frequency with a question about age of respondents, as follows. Yes No Total <18 18–34 35–50 51–65 >65 Total 70 110 120 30 20 350 78% 96% 96% 33% 25% 70% 20 5 5 60 60 150 22% 4% 4% 67% 75% 30% 90 115 125 90 80 500 From the above cross-tabulation, you can gain meaningful information such as: 18 96% of respondents between the ages of 18 and 34 have purchased a single serve drink in the last three months 75% of respondents over the age of 65 have not purchased a single serve drink in the last three months. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot Activity 6: Jonah’s Jam patrons Look at the data set below. Jonah’s Jam Company are exploring the age of the people who patronise their retail shop and what methods they use to pay. Payment method Age Payment method Age Cheque 18 Cheque 66 Cheque 27 Credit card 58 Credit card 95 Credit card 57 Credit card 32 Credit card 74 Cash 16 Cash 40 Cheque 80 Cheque 44 Cash 45 Credit card 55 Credit card 32 Credit card 71 Cheque 51 Cash 13 Cheque 70 Cash 24 Credit card 61 Cash 23 Cheque 47 Cheque 26 Carry out a cross-tabulation on the data to see if there are any differences between the age of the shopper and how they pay for their purchase. Use the table below for your crosstabulation. To check your answers, refer to the feedback at the end of this topic. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 19 Correlation Marketers often want to know the relationship between two factors that may have an effect on their market. Examples of two factors marketers might want to examine are: the relationship between perception of quality and sales the relationship between advertising and sales. In these cases, a statistical technique known as product moment correlation is used. This technique is used to illustrate and measure the relationship between two factors. For example, use of this technique would measure how strongly levels of advertising are linked to changes in sales levels, and the extent to which one would change of the other was changed. The relationship between these two factors is known as covariance. Hypothesis testing A hypothesis is simply a belief that is held at the start of the study that the researcher wants to prove or disprove. Here is an example of a hypothesis: 50% of my customers are aged over 45 The research then sets out to prove if this statement is right or wrong. Other descriptive statistics These basic statistics are used to represent some common forms of summarising information, or to make a lot of data actually mean something. Common descriptive statistics include: 20 mode: the value that occurs the most often median: the midpoint—the value below which the values in a distribution fall mean: the arithmetic average range: the distance between the smallest and the largest values in a frequency distribution standard deviation: based on the deviations from the mean, quantitative index of the distribution’s spread or variability. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot Activity 7: Analysing Jonah’s jam shoppers Using the data you previously grouped, calculate the mean, median, standard deviation and range for the age of the shoppers at Jonah's Jam Shop. 1 mean _____________________________________________________________________ 2 median _____________________________________________________________________ 3 range _____________________________________________________________________ 4 standard deviation _____________________________________________________________________ To check your answers, refer to the feedback at the end of this topic. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 21 Other statistical techniques Most of the other statistical techniques used in data analysis are designed to prove (in one way or another) that the results produced in your sample would be the same, or similar to, the results that would occur if you administered the survey or interview to the entire population. The process of testing how closely the results from the sample match the results from the population is called significance. For example, you may have found that 74% of your sample purchase their music on the Internet. A test of significance would measure how confident you were that 74% of the population would purchase their music on the Internet. Significance is usually measured in terms of a percentage. If you were testing at 95% significance, it means that the result you obtained from your sample is within plus or minus 2.5% of the population. If you were testing at 99% significance, that means the results you obtained were accurate for the population, plus or minus 0.5%. Many of the remaining statistical techniques are connected to the process of testing significance. Some are designed to predict a future occurrence based on past data while others explore the relationships that exist between variables. Your suggested readings have a number of chapters on these techniques, and you should have an understanding of what these techniques are and what they are used for. If you have a version of SPSS and have installed it, you’ll find all of these tools on the pull-down menu. Load the sample data sets and look at the output produced using the different analysis techniques. 22 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot Summary In this topic we have been introduced to some basic concepts of statistical analysis. You should now be familiar with concepts such as data coding, data matrix and data reduction, and be able to perform them. You should also be able to use the techniques associated with tabulations, crosstabulations and correlations. Check your progress Here are 10 multiple-choice questions to test your knowledge of what you should have learnt in this topic. If you answered fewer than eight out of ten questions correctly, then you should go back over the section. 1 The purpose of _______ is to ensure completeness, consistency, and readability of the data to be transferred to data storage. (a) editing (b) coding (c) keypunching (d) checking. 2 Pre-coding is desirable because it: (a) allows data from questionnaires to be entered directly (b) facilitates statistical analysis (c) reduces the length of questionnaires (d) eliminates the need for interviews. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 23 3 The transformation of raw data (by rearranging, ordering, or manipulating) into a form that will make them easy to understand is: (a) inferential statistics (b) univariate statistics (c) data processing (d) descriptive analysis. 4 A survey samples both men and women. The researcher wishing to analyse the data by separating the groups by sex will perform: (a) an analysis of central tendency (b) a simple tabulation analysis (c) a gender analysis (d) a cross-tabulation analysis. 5 The choice of the method of statistical analysis depends upon all of the following except: (a) the question to be answered (b) the number of variables (c) the scale of measurement (d) the questionnaire design. 6 One of the simplest techniques for describing sets of relationships is: (a) cross-tabulation (b) chi-square (c) regression (d) correlation. 24 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 7 A correlation coefficient: (a) is a statistical measure of co-variation between variables (b) indicates the magnitude of the relationship between variables (c) indicates the direction of the relationship between variables (d) all of the above. 8 Constructing a frequency table: (a) is one of the most common means of summarising data. (b) begins by recording the number of times a particular value occurs. (c) is the basis for construction of a percentage distribution. (d) all of the above. 9 Which of the following is not a measure of central tendency? (a) range (b) mode (c) median (d) mean. 10 Counting the number of responses to a question and putting them into a frequency distribution is a: (a) summary statistic (b) univariate analysis (c) simple tabulation (d) percentage. To check your answers, refer to the feedback at the end of this topic. Key terms for your glossary Research terminology is often very specific and frequently uses words in a different way to their common meaning. Write definitions for these key terms and place them in your personal glossary. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 25 coding editing data cleaning transcribing data analysis descriptive statistics mean median mode range standard deviation cross-tabulation hypothesis testing chi-square correlation co-efficient simple regression multiple regression 26 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot Feedback to Activities and Check your progress exercise Activity 1 Your definitions should mention the following 1 data coding: create a code that can be applied to the possible responses to a questionnaire 2 data matrix: coded raw data from a survey 3 data reduction: condenses the data matrix using measures that characterise the data set If you haven’t already done so, enter these terms in your personal glossary. Activity 2 1 You can safely assume that the date should be 1950 and that the gender should be female. Change the responses accordingly. 2 This situation is more difficult. If possible, contact the interviewer for clarity and check for consistency with other questions. You will need to make a decision on how to change this information once you have checked these sources. 3 Change the monthly income to annual income (multiply by 12). 4 Check with interviewer for translation. This should be picked up during the field edit and the interviewer asked to write more clearly. 5 Check with interviewer. If this is not possible, make a decision based on consistency with other answers. 6 Record that only one brand was given. Do not be tempted to add other brands to complete the answer. 7 If this is the calibre of all of the answers in the questionnaire then discard it. There is always at least one joker in every project. 8 Discard the questionnaire. This respondent is not part of the research study. 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 27 9 You could simply accept the inconsistency or check with the interviewer. In this case, you could assume that the income should be $150 000. 10 Again you could simply accept the answers. If you suspect that the respondent was uninterested and simply selected 4 to complete the questionnaire quickly, then discard it. Activity 3 From these responses a number of different categories could be created to develop a code frame. 1 Convenience or proximity: Numbers 1, 4 and 12 all relate to the proximity and convenience of the shopping centre. 2 Ease of travel: Numbers 2, 6, and 14 all relate to how easily customers can travel to the shopping centre. 3 Good range of shops: Numbers 3, 5, 7, 11 and 13 all relate to the appeal of the shopping centre. 4 Other: Numbers 8, 9 and 15 don’t relate to any of the above categories and they all say something slightly different. Activity 4 1 The sample can be weighted. 2 Code frames can be collapsed. 3 During data cleaning. Activity 6 Your table should look like this. Age group Cheque Under 18 18-30 Cash Credit card 2 3 31-40 2 1 2 41-50 2 1 51-60 1 3 Over 60 3 3 Total 9 6 8 Overall most respondents preferred to pay by cheque or credit card. The majority of respondents aged over 51 preferred to pay by credit card, while respondents aged under 30 preferred to pay by cash or cheque. 28 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot Activity 7 1 mean: 46.9 years 2 median: 46 years 3 range: 82 years 4 standard deviation: 22.4 years Note that the years are calculated to one decimal place: this is because any more information is not useful to the decision-maker. Check your progress 1 (a) 2 (a) 3 (d) 4 (d) 5 (d) 6 (a) 7 (d) 8 (d) 9 (a) 10 (c) 9775J: 11 Marketing Research OTEN, 2002/122/5/2003 P0026682 LO:2002_313_011_ot 29