12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. CHAPTER 14 QUANTITATIVE DATA ANALYSIS LEARNING OBJECTIVES After completing Chapter 14 you should be able to: 1. Demonstrate the ability to get data ready for quantitative analysis. 2. Describe the various processes by which one can get a feel for the data in a study. 3. Describe the means by which the reliability and validity of measures can be assessed. INTRODUCTION After quantitative data have been collected from a representative sample of the population, the next step is to analyze them to answer our research questions. However, before we can start analyzing the data, some preliminary steps need to be completed. These help to ensure that the data are accurate, complete, and suitable for further analysis. This chapter addresses these preliminary steps in detail. Subsequently, general guidelines are provided for calculating and displaying basic descriptive statistics. The easiest way to illustrate data analysis is through a case. We will therefore introduce the Excelsior Enterprises case first. EXAMPLE Excelsior Enterprises is a medium-sized company, manufacturing and selling instruments and supplies needed by the health care industry, including blood pressure instruments, surgical instruments, dental accessories, and so on. The company, with a total of 360 employees working three shifts, is doing reasonably well but could do far better if it did not experience employee turnover at almost all levels and in all departments. The president of the company called in a research team to study the situation and to make recommendations on the turnover problem. Since access to those who had left the company would be difficult, the research team suggested to the president that they talk to the current employees and, based on their input and a literature survey, try to get at the factors influencing employees' intentions to stay with, or leave, the company. Since past research has shown that intention to leave (ITL) is an https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 1/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. excellent predictor of actual turnover, the president concurred. The team first conducted an unstructured interview with about 50 employees at various levels and from different departments. Their broad statement was: “We are here to find out how you experience your work life. Tell us whatever you consider is important for you in your job, as issues relate to your work, the environment, the organization, supervision, and whatever else you think is relevant. If we get a good handle on the issues involved, we may be able to make appropriate recommendations to management to enhance the quality of your work life. We would just like to talk to you now, and administer a questionnaire later.” Each interview typically lasted about 45 minutes, and notes on the responses were written down by the team members. When the responses were tabulated, it became clear that the issues most frequently brought up by the respondents, in one form or another, related to three main areas: the job (employees said the jobs were dull or too complex; there was lack of freedom to do the job as one wanted to, etc.), perceived inequities (remarks such as “I put much more in my work than I get out of it”); and burnout (comments such as “there is so much work to be done that by the end of the day we are physically and emotionally exhausted”; “we feel the frequent need to take time off because of exhaustion”; etc.). A literature survey confirmed that these variables were good predictors of intention to leave and subsequent turnover. In addition, job satisfaction was also found to be an important predictor of intention to leave. A theoretical framework was developed based on the interviews and the literature survey, and four hypotheses (stated later) were developed. Next, a questionnaire was designed incorporating well-validated and reliable measures for job enrichment, perceived equity, burnout, job satisfaction, and intention to leave. Perceived equity was measured by five survey items: (1) “I invest more in my work than I get out of it”; (2) “I exert myself too much considering what I get back in return”; (3) “For the efforts I put into the organization, I get much in return” (reversed); (4) “If I take into account my dedication, the company ought to give me better training”; and (5) “In general, the benefits I receive from the organization outweigh the effort I put in it” (reversed). Job enrichment was measured on a four-item Likert scale: (1) “The job is quite simple and repetitive” (reversed); (2) “The job requires me to use a number of complex or higher-level skills”; (3) “The job requires a lot of cooperative work with other people”; and (4) “The job itself is not very significant or important in the broader scheme of things” (reversed). Participants responded to these items on a five-point scale, ranging from “I disagree completely” (1) to “I agree completely” (5). Burnout was measured with The Burnout Measure Short Version (BMS). The BMS includes ten items that measure levels of physical, emotional, and mental exhaustion of the individual. Respondents are asked to rate the frequency with which they experience each of the items appearing in the questionnaire (e.g., being tired or helpless) on a scale ranging from 1 (“never”) to 5 (“always”). Job satisfaction was measured by a single-item rating of “satisfaction with your current job,” using a five-point “not at all–very much” scale. Intention to leave was measured using two survey items: “With what level of certainty do you intend to leave this organization within the next year for another type of job?” (item 1) “for a similar type of job?” (item 2). Participants indicated on a four-point rating scale their level of certainty. Demographic variables such as age, education, gender, tenure, department, and work shift were also included in the questionnaire. The questionnaire was administered personally to 174 employees who were chosen on a disproportionate stratified random sampling basis. The responses were entered into the computer. Thereafter, the data were submitted for analysis to test the following hypotheses, which were formulated by the researchers: H1: Job enrichment has a negative effect on intention to leave. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 2/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach H2: Perceived equity has a negative effect on intention to leave. H3: Burnout has a positive effect on intention to leave. H4: Job satisfaction mediates the relationship between job enrichment, perceived equity, and burnout on intention to leave. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 3/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. It may be pertinent to point out here that the four hypotheses derived from the theoretical framework are particularly relevant for finding answers to the turnover issue. The results of testing the hypotheses will certainly offer insights into how much of the variance in intention to leave can be explained by the independent variables, and what corrective action, if any, needs to be taken. GETTING THE DATA READY FOR ANALYSIS After data are obtained through questionnaires, they need to be coded, keyed in, and edited. That is, a categorization scheme has to be set up before the data can be typed in. Then, outliers, inconsistencies, and blank responses, if any, have to be handled in some way. Each of these stages of data preparation is discussed below. Coding and data entry The first step in data preparation is data coding. Data coding involves assigning a number to the participants' responses so they can be entered into a database. In Chapter 9, we discussed the convenience of electronic surveys for collecting questionnaire data; such surveys facilitate the entry of the responses directly into the computer without manual keying in of the data. However, if, for whatever reason, this cannot be done, then it is perhaps a good idea to use a coding sheet first to transcribe the data from the questionnaire and then key in the data. This method, in contrast to flipping through each questionnaire for each item, avoids confusion, especially when there are many questions and a large number of questionnaires as well. Coding the responses In the Excelsior Enterprises questionnaire, we have 22 items measuring perceived equity, job enrichment, burnout, job satisfaction, and intention to leave, and six demographic variables, as shown in Figure 14.1, a sample questionnaire. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 4/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 5/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 6/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. FIGURE 14.1 Sample questionnaire The responses of this particular employee (participant #1 in the data file) to the first 22 questions can be coded by using the actual number circled by the respondent (1, 2, 3, 1, 4, 5, 1, 3, 3, etc.). Coding the demographic variables is somewhat less obvious. For instance, tenure is a special case, because it is a two-category variable. It is possible to use a coding approach that assigns a 1 = parttime and a 2 = full-time. However, using 0 = part-time and 1 = full-time (this is called dummy coding) is by far the most popular and recommended approach because it makes our lives easier in the data analysis stage. Hence, we code tenure (full-time) with 1 for participant #1. Work shift (third shift) can be coded 3, department (production) 2, and age 54. Gender can be coded 0 (male) Finally, education (less than high school) can be coded 1. At this stage you should also think about how you want to code nonresponses. Some researchers leave nonresponses blank, others assign a “9,” a “99” or a “.” All the approaches are fine, as long as you code all the nonresponses in the same way. Human errors can occur while coding. At least 10% of the coded questionnaires should therefore be checked for coding accuracy. Their selection may follow a systematic sampling procedure. That is, every nth form coded could be verified for accuracy. If many errors are found in the sample, all items may have to be checked. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 7/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. Data entry After responses have been coded, they can be entered into a database. Raw data can be entered through any software program. For instance, the SPSS Data Editor, which looks like a spreadsheet and is shown in Figure 14.2, can enter, edit, and view the contents of the data file. FIGURE 14.2 The SPSS Data Editor Each row of the editor represents a case or observation (in this case a participant of our study – 174 in the Excelsior Enterprises study), and each column represents a variable (here variables are defined as the different items of information that you collect for your cases; there are thus 28 variables in the Excelsior Enterprises questionnaire). It is important to always use the first column for identification purposes; assign a number to every questionnaire, write this number on the first page of the questionnaire, and enter this number in the first column of your https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 8/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. data file. This allows you to compare the data in the data file with the answers of the participants, even after you have rearranged your data file. Then, start entering the participants' responses into the data file. Editing data After the data are keyed in, they need to be edited. For instance, the blank responses, if any, have to be handled in some way, and inconsistent data have to be checked and followed up. Data editing deals with detecting and correcting illogical, inconsistent, or illegal data and omissions in the information returned by the participants of the study. An example of an illogical response is an outlier response. An outlier is an observation that is substantially different from the other observations. An outlier is not always an error even though data errors (entry errors) are a likely source of outliers. Because outliers have a large impact on the research results they should be investigated carefully to make sure that they are correct. You can check the dispersion of nominal and/or ordinal variables by obtaining minimum and maximum values and frequency tables. This will quickly reveal the most obvious outliers. For interval and ratio data, visual aids (such as a scatterplot or a boxplot) are good methods to check for outliers. Inconsistent responses are responses that are not in harmony with other information. For instance, a participant in our study might have answered the perceived equity statements as in Figure 14.3. Note that all the answers of this employee indicate that the participant finds that the benefits she receives from the organization balance the efforts she puts into her job, except for the answer to the third statement. From the other four responses we might infer that the participant in all probability feels that, for the efforts she puts into the organization, she does get much in return and has made a mistake in responding to this particular statement. The response to this statement could then be edited by the researcher. FIGURE 14.3 Example of a possible inconsistent answer It is, however, possible that the respondent deliberately indicated that she does not get much in return for the efforts she puts into the organization. If such were to be the case, we would be introducing a bias by editing the data. Hence, great care has to be taken in dealing with inconsistent responses such as these. Whenever possible, it is desirable to follow up with the respondent to get the correct data, even though this is an expensive solution. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 9/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach Illegal codes are values that are not specified in the coding instructions. For example, a code of “6” in question 1 (I invest more in my work than I get out of it) would be an illegal code. The best way to check for illegal codes is to have the computer produce a frequency distribution and check it for illegal codes. Not all respondents answer every item in the questionnaire. Omissions may occur because respondents did not understand the question, did not know the answer, or were not willing to answer the question. If a substantial number of questions – say, 25% of the items in the questionnaire – have been left unanswered, it may be a good idea to throw out the questionnaire and not include it in the data set for analysis. In this event, it is important to mention the number of returned but unused responses due to excessive missing data in the final report submitted to the sponsor of the study. If, however, only two or three items are https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 10/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. left blank in a questionnaire with, say, 30 or more items, we need to decide how these blank responses are to be handled. One way to handle a blank response is to ignore it when the analyses are done. This approach is possible in all statistical programs and is the default option in most of them. A disadvantage of this approach is that, of course, it will reduce the sample size, sometimes even to an inappropriate size, whenever that particular variable is involved in the analyses. Moreover, if the missing data are not missing completely at random, this method may bias the results of your study. For this reason, ignoring the blank responses is best suited to instances in which we have gathered a large amount of data, the number of missing data is relatively small, and relationships are so strong that they are not affected by the missing data (Hair, Anderson, Tatham & Black, 1995). An alternative solution would be to look at the participant's pattern of responses to other questions and, from these answers, deduce a logical answer to the question for the missing response. A second alternative solution would be to assign to the item the mean value of the responses of all those who have responded to that particular item. In fact, there are many ways of handling blank responses (see Hair et al., 1995), each of them having its own particular advantages and disadvantages. Note that if many of the respondents have answered “don't know” to a particular item or items, further investigation may well be worth while. The question might not have been clear or, for some reason, participants could have been reluctant or unable to answer the question. Data transformation Data transformation, a variation of data coding, is the process of changing the original numerical representation of a quantitative value to another value. Data are typically changed to avoid problems in the next stage of the data analysis process. For example, economists often use a logarithmic transformation so that the data are more evenly distributed. If, for instance, income data, which are often unevenly distributed, are reduced to their logarithmic value, the high incomes are brought closer to the lower end of the scale and provide a distribution closer to a normal curve. Another type of data transformation is reverse scoring. Take, for instance, the perceived inequity measure of the Excelsior Enterprises case. Perceived inequity is measured by five survey items: (1) “I invest more in my work than I get out of it”; (2) “I exert myself too much considering what I get back in return”; (3) “For the efforts I put into the organization, I get much in return” (reversed); (4) “If I take into account my dedication, the organization ought to give me a better practical training”; and (5) “In general, the benefits I receive from the organization outweigh the effort I put in” (reversed). For the first, second, and fourth items, a score indicating high agreement would be negative, but for the third and fifth questions, a score indicating high agreement would be positive. To maintain consistency in the meaning of a response, the first, second, and fourth items have to be reverse scored (note that we are measuring equity and not inequity). In this case, a 5 (“I completely agree”) would be transformed to a 1 (“I completely disagree”), a 4 to a 2, and so forth. Data transformation is also necessary when several questions have been used to measure a single concept. In such cases, scores on the original questions have to be combined into a single score (but only after we have established that the interitem consistency is satisfactory (see Testing goodness of data, later on in this chapter). For instance, because five items have been used to measure the concept “perceived equity”, a new “perceived equity” score has to be calculated from the scores on the five individual items (but only after items 1, 2, and 4 have been reverse coded). This involves calculating the summed score (per case/participant) and then dividing it by the number of items (five in this case). For example, our employee #1 has circled, respectively, 1, 2, 3, 1, and 4 on the five participation in decision-making questions; his (employee #1 is a man) scores on the items, once items 1, 2, and 4 have been reverse https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 11/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach coded, are 5, 4, 3, 5, and 4. The combined score on perceived equity would be 5 + 4 + 3 + 5 + 4 = 21 / 5 = 4.2 ). This combined score is included in a new column in SPSS. It is easy to compute the new variables, using the Compute dialog box, which opens when the Transform icon is chosen (Figure 14.4). https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 12/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. FIGURE 14.4 Transforming data with SPSS Note that it is useful to set up a scheme for categorizing the responses such that the several items measuring a concept are all grouped together. If the questions measuring a concept are not contiguous but scattered over various parts of the questionnaire, care has to be taken to include all the items without any omission or wrong inclusion. GETTING A FEEL FOR THE DATA We can acquire a feel for the data by obtaining a visual summary or by checking the central tendency and the dispersion of a variable. We can also get to know our data by examining the relation between two variables. In Chapter 12, we explained that different statistical operations on variables are possible, depending on the level at which a variable is measured. Table 14.1 summarizes the relationship between scale type, data analysis, and methods of obtaining a visual summary for variables. TABLE 14.1 Scale type, data analysis, and methods of obtaining a visual summary for variables Scale Examples Measures of Measures of central tendency dispersion … for a … for a single single variable variable Visual summary … for a single variable Measure of relation … between variables Visual summary of relation … between variables Mode Bar chart, pie chart Contingency table (Crosstab) Stacked bars, Clustered bars Ordinal Satisfaction rating on a 5Median point scale (1 = not satisfied at all; 5 = extremely satisfied) Semi-interquartile range Bar chart, pie chart Contingency table (Crosstab) Stacked bars, Clustered bars Interval Arithmetic mean Minimum, maximum, standard deviation, variance, coefficient of variation Histogram, scatterplot, box-andwhisker plot Correlations Scatterplots Arithmetic or geometric mean Minimum, maximum, standard deviation, variance, coefficient of variation Histogram, scatterplot, box-andwhisker plot Correlations Scatterplots Nominal Social security number, gender Ratio Age, Sales https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 — 13/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach Table 14.1 shows that, depending on the scale of our measures, the mode, median, or mean, and the semi-interquartile range, standard deviation, or variance will give us a good idea of how the participants in our study have reacted to the items in the questionnaire. These statistics can be easily obtained, and will indicate whether the responses range satisfactorily over the scale. If the response to each individual item in a scale does not have a good spread (range) and shows very little variability, then the researcher may suspect that the particular question was probably not properly worded. Biases, if any, may also be detected if the respondents have tended to respond similarly to all items – that is, they have stuck to only certain points on the scale. Remember that if there is no variability in the data, then no variance can be explained! Getting a feel for the data is thus the necessary first step in all data analysis. Based on this initial feel, further detailed analyses may be undertaken to test the goodness of the data. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 14/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. Researchers go to great lengths to obtain the central tendency, the range, the dispersion, and other statistics for every single item measuring the dependent and independent variables, especially when the measures for a concept are newly developed. Descriptive statistics for a single variable are provided by frequencies, measures of central tendency, and dispersion. These are now described. Frequencies Frequencies simply refer to the number of times various subcategories of a certain phenomenon occur, from which the percentage and the cumulative percentage of their occurrence can be easily calculated. Excelsior Enterprises: frequencies The frequencies for the number of individuals in the various departments for the Excelsior Enterprises sample are shown in Output 14.1. It may be seen therefrom that the greatest number of individuals in the sample came from the production department (28.1%), followed by the sales department (25.3%). Only three individuals (1.7%) came from public relations, and five individuals each from the finance, maintenance, and accounting departments (2.9% from each). The low numbers in the sample in some of the departments are a function of the total population (very few members) in those departments. OUTPUT 14.1 FREQUENCIES From the menus, choose: Analyze Descriptive Statistics Frequencies [Select the relevant variables] Choose needed: Statistics … Charts … Format [for the order in which the results are to be displayed] https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 15/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 16/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. OUTPUT: RESPONDENT'S DEPARTMENT Frequency Percent Valid percent Cumulative percent Marketing 13 7.5 7.5 7.5 Production 49 28.1 28.1 35.6 Sales 44 25.3 25.3 60.9 Finance 5 2.9 2.9 63.8 Servicing 34 19.5 19.5 83.3 Maintenance 5 2.9 2.9 86.2 Personnel 16 9.2 9.2 95.4 Public Relations 3 1.7 1.7 97.1 Accounting 5 2.9 2.9 100.0 Total 174 100.0 100.0 100.0 https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 17/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/6/4@0:0 18/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. Measures of central tendency and dispersion There are three measures of central tendency: the mean, the median, and the mode. Measures of dispersion include the range, the standard deviation, the variance (where the measure of central tendency is the mean), and the interquartile range (where the measure of central tendency is the median). Measures of central tendency The mean The mean, or the average, is a measure of central tendency that offers a general picture of the data without unnecessarily inundating one with each of the observations in a data set. For example, the production department might keep detailed records on how many units of a product are being produced each day. However, to estimate the raw materials inventory, all that the manager might want to know is how many units per month, on average, the department has been producing over the past six months. This measure of central tendency – that is, the mean – might offer the manager a good idea of the quantity of materials that need to be stocked. The mean or average of a set of, say, ten observations, is the sum of the ten individual observations divided by ten (the total number of observations). The median The median is the central item in a group of observations when they are arrayed in either an ascending or a descending order. Let us take an example to examine how the median is determined as a measure of central tendency. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/12/4@0:0 1/13 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach EXAMPLE https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/12/4@0:0 2/13 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. Let's say the annual salaries of nine employees in a department are as follows: $65 000, $30 000, $25 000, $64 000, $35 000, $63 000, $32 000, $60 000, and $61 000. The mean salary here works out to be about $48 333, but the median is $60 000. That is, when arrayed in ascending order, the figures will be as follows: $25 000, $30 000, $32 000, $35 000, $60 000, $61 000, $63 000, $64 000, $65 000, and the figure in the middle is $60 000. If there is an even number of employees, then the median will be the average of the middle two salaries. The mode In some cases, a set of observations does not lend itself to a meaningful representation through either the mean or the median, but can be signified by the most frequently occurring phenomenon. For instance, in a department where there are 10 white women, 24 white men, 3 African American women, and 2 Asian women, the most frequently occurring group – the mode – is the white men. Neither a mean nor a median is calculable or applicable in this case. There is also no way of indicating any measure of dispersion. We have illustrated how the mean, median, and the mode can be useful measures of central tendency, based on the type of data we have. We will now examine dispersion. Measures of dispersion Apart from knowing that the measure of central tendency is the mean, median, or mode (depending on the type of available data), one would also like to know about the variability that exists in a set of observations. Like the measure of central tendency, the measure of dispersion is also unique to nominal and interval data. Two sets of data might have the same mean, but the dispersions could be different. For example, if Company A sold 30, 40, and 50 units of a product during the months of April, May, and June, respectively, and Company B sold 10, 40, and 70 units during the same period, the average units sold per month by both companies is the same – 40 units – but the variability or the dispersion in the latter company is larger. The three measurements of dispersion connected with the mean are the range, the variance, and the standard deviation, which are explained below. Range Range refers to the extreme values in a set of observations. The range is between 30 and 50 for Company A (a dispersion of 20 units), while the range is between 10 and 70 units (a dispersion of 60 units) for Company B. Another more useful measure of dispersion is the variance. Variance The variance is calculated by subtracting the mean from each of the observations in the data set, taking the square of this difference, and dividing the total of these by the number of observations. In the above example, the variance for each of the two companies is: https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/12/4@0:0 3/13 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach 2 2 (30−40) +(40−40) +(50−40) Variance for Company A = 2 = 66.7 3 2 2 (10−40) +(40−40) +(70−40) Variance for Company B = 3 2 = 600 As we can see, the variance is much larger in Company B than Company A. This makes it more difficult for the manager of Company B to estimate how many goods to stock than it is for the manager of Company A. Thus, variance gives an indication of how dispersed the data in a data set are. PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. Standard deviation The standard deviation, which is another measure of dispersion for interval and ratio scaled data, offers an index of the spread of a distribution or the variability in the data. It is a very commonly used measure of dispersion, and is simply the square root of the variance. In the case of the above two companies, the standard deviation for Companies A and B would be √66.7 and √600 or 8.167 and 24.495, respectively. The mean and standard deviation are the most common descriptive statistics for interval and ratio scaled data. The standard deviation, in conjunction with the mean, is a very useful tool because of the following statistical rules, in a normal distribution: 1. Practically all observations fall within three standard deviations of the average or the mean. 2. More than 90% of the observations are within two standard deviations of the mean. 3. More than half of the observations are within one standard deviation of the mean. Other measures of dispersion When the median is the measure of central tendency, percentiles, deciles, and quartiles become meaningful. Just as the median divides the total realm of observations into two equal halves, the quartile divides it into four equal parts, the decile into ten, and the percentile into 100 equal parts. The percentile is useful when huge masses of data, such as the GRE or GMAT scores, are handled. When the area of observations is divided into 100 equal parts, there are 99 percentile points. Any given score has a probability of 0.01 that it will fall in any one of those points. If John's score is in the 16th percentile, it indicates that 84% of those who took the exam scored better than he did, while 15% did worse. Oftentimes we are interested in knowing where we stand in comparison to others – are we in the middle, in the upper 10 or 25%, or in the lower 20 or 25%, or where? For instance, if in a company-administered test, Mr Chou scores 78 out of a total of 100 points, he may be unhappy if he is in the bottom 10% among his colleagues (the test-takers), but may be reasonably pleased if he is in the top 10%, https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/12/4@0:0 4/13 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach despite the fact that his score remains the same. His standing in relation to the others can be determined by the central tendency median and the percentile he falls in. The measure of dispersion for the median, the interquartile range, consists of the middle 50% of the observations (i.e., observations excluding the bottom and top 25% quartiles). The interquartile range is very useful when comparisons are to be made among several groups. For instance, telephone companies can compare long-distance charges of customers in several areas by taking samples of customer bills from each of the cities to be compared. By plotting the first and third quartiles and comparing the median and the spread, they can get a good idea of where billings tend to be highest, to what extent customers vary in the frequency of use of long-distance calls, and so on. This is done by creating a box-and-whisker plot for each area. The box-and-whisker plot is a graphic device that portrays central tendency, percentiles, and variability. A box is drawn, extending from the first to the third quartile, and lines are drawn from either side of the box to the extreme scores, as shown in Figure 14.6(a). Figure 14.6(b) has the median represented by a dot within each box. Side-by-side comparisons of the various plots clearly indicate the highest value, the range, and the spread for each area or city. For a fuller discussion on this, refer to Salvia (1990). FIGURE 14.6 (a) Box-and-whisker plot; (b) comparison of telephone bills in three cities In sum, we have illustrated how the mean, median, and the mode can be useful measures of central tendency, depending on the type of available data. Likewise, we have shown how the standard deviation (and variance, which is the square of standard deviation), and the interquartile range are useful measures of dispersion. Obviously, there is no measure of dispersion associated with the mode. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/12/4@0:0 5/13 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. Relationships between variables In a research project that includes several variables, beyond knowing the descriptive statistics of the variables, we would often like to know how one variable is related to another. That is, we would like to see the nature, direction, and significance of the bivariate relationships of the variables used in the study (i.e., the relationship between any two variables among the variables tapped in the study). Nonparametric tests are available to assess the relationship between variables measured on a nominal or an ordinal scale. Spearman's rank correlation and Kendall's rank correlation are used to examine relationships between two ordinal variables. A correlation matrix is used to examine relationships between interval and/or ratio variables. Relationship between two nominal variables: χ2 test We might sometimes want to know if there is a relationship between two nominal variables or whether they are independent of each other. As examples: (1) Is viewing a television advertisement of a product (yes/no) related to buying that product by individuals (buy/don't buy)? (2) Is the type of job done by individuals (white-collar job/blue-collar job) a function of the color of their skin (white/nonwhite)? Such comparisons are possible by organizing data by groups or categories and seeing if there are any statistically significant relationships. For example, we might collect data from a sample of 55 individuals whose color of skin and nature of jobs, culled from a frequency count, might be illustrated as in Table 14.2 in a two-by-two contingency table. Just by looking at Table 14.2, a https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/12/4@0:0 6/13 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. clear pattern seems to emerge that those who are white hold white-collar jobs. Only a few of the nonwhites hold white-collar jobs. Thus, there does seem to be a relationship between the color of the skin and the type of job handled; the two do not seem to be independent. This can be statistically confirmed by the chi-square (χ2 ) test – a nonparametric test – which indicates whether or not the observed pattern is due to chance. As we know, nonparametric tests are used when normality of distributions cannot be assumed as in nominal or ordinal data. The χ2 test compares the expected frequency (based on probability) and the observed frequency, and the χ2 statistic is obtained by the formula: χ 2 (Oi − Ei) 2 = ∑ Ei where χ2 is the chi-square statistic; Oi is the observed frequency of the ith cell; and Ei is the expected frequency. The χ2 with its level of significance can be obtained for any set of nominal data through computer analysis. statistic TABLE 14.2 Contingency table of skin color and job type Skin color White collar Blue collar Total White Nonwhite Total 30 5 35 2 18 20 32 23 55 Thus, in testing for differences in relationships among nominally scaled variables, the χ2 (chi-square) statistic comes in handy. The null hypothesis would be set to state that there is no significant relationship between two variables (color of skin and nature of the job, in the above example), and the alternate hypothesis would state that there is a significant relationship. The chi-square statistic is associated with the degrees of freedom (df), which denote whether or not a significant relationship exists between two nominal variables. The number of degrees of freedom is one less than the number of cells in the columns and rows. If there are four cells (two in a column and two in a row), then the number of degrees of freedom would be 1: [(2 − 1) × (2 − 1)] . The chi-square statistic for various df is provided in Table III in the statistical tables toward the end of the book. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/12/4@0:0 7/13 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach The χ2 statistic can also be used for multiple levels of two nominal variables. For instance, one might be interested to know if four groups of employees – production, sales, marketing, and R&D personnel – react to a policy in four different ways (i.e., with no interest at all, with mild interest, moderate interest, and intense interest). Here, the χ2 value for the test of independence is generated by cross-tabulating the data in 16 cells – that is, classifying the data in terms of the four groups of employees and the four categories of interest. The degrees of freedom here will be 9: [(4 − 1) × (4 − 1)] . The χ2 test of significance thus helps us to see whether or not two nominal variables are related. Besides the χ2 test, other tests, such as the Fisher exact probability test and the Cochran Q test are used to determine the relationship between two nominally scaled variables. Correlations A Pearson correlation matrix will indicate the direction, strength, and significance of the bivariate relationships among all the variables that were measured at an interval or ratio level. The correlation is derived by assessing the variations in one variable as another variable also varies. For the sake of simplicity, let us say we have collected data on two variables – price and sales – for two different products. The volume of sales at every price level can be plotted for each product, as shown in the scatter diagrams in Figure 14.7(a) and 14.7(b). FIGURE 14.7 (a) Scatter diagram with no discernible pattern; (b) scatter diagram indicating a downward or negative slope Figure 14.7 (b) indicates a discernible pattern of how the two factors vary simultaneously (the trend of the scatter is that of a downward straight line), whereas Figure 14.7 (a) does not. Looking at the scatter diagram in Figure 14.7 (b), it would seem there is a direct negative correlation between price and sales for this product. That is, as the price increases, sales of the product drop consistently. Figure 14.7 (a) suggests no interpretable pattern for the other product. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/12/4@0:0 8/13 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. A correlation coefficient that indicates the strength and direction of the relationship can be computed by applying a formula that takes into consideration the two sets of figures – in this case, different sales volumes at different prices. Theoretically, there could be a perfect positive correlation between two variables, which is represented by 1.0 (plus 1), or a perfect negative correlation which would be −1.0 (minus 1). However, neither of these will be found in reality when assessing correlations between any two variables expected to be different from each other. While the correlation could range between −1.0 and +1.0, we need to know if any correlation found between two variables is significant or not (i.e., if it has occurred solely by chance or if there is a high probability of its actual existence). As we know, a significance of p = 0.05 is the generally accepted conventional level in social science research. This indicates that 95 times out of 100, we can be sure that there is a true or significant correlation between the two variables, and there is only a 5% chance that the relationship does not truly exist. If there is a correlation of 0.56 (denoted as r = 0.56) between two variables A and B, with p < 0.01, then we know that there is a positive relationship between the two variables and the probability of this not being true is 1% or less. That is, over 99% of the time we would expect this correlation to exist. The correlation of 0.56 also indicates that the variables explain the variance in one another to the extent of 31.4% (0.562). We do not know which variable causes which, but we do know that the two variables are associated with each other. Thus, a hypothesis that postulates a significant positive (or negative) relationship between two variables can be tested by examining the correlation between the two. The Pearson correlation coefficient is appropriate for interval- and ratio-scaled variables, and the Spearman Rank or the Kendall's tau coefficients are appropriate when variables are measured on an ordinal scale. Any bivariate correlation can be obtained by clicking the relevant menu, identifying the variables, and seeking the appropriate parametric or nonparametric statistics. EXCELSIOR ENTERPRISES: DESCRIPTIVE STATISTICS PART 1 Descriptive statistics such as maximum, minimum, means, standard deviations, and variance were obtained for the interval-scaled items of the Excelsior Enterprises study. The procedure is shown in Output 14.2. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/12/4@0:0 9/13 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. OUTPUT 14.2 DESCRIPTIVE STATISTICS: CENTRAL TENDENCIES AND DISPERSIONS From the menus, choose: Analyze Descriptive Statistics Descriptives [Select the variables] Statistics … [Choose the relevant statistics needed] https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/12/4@0:0 10/13 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach Output pe1 pe2 pe3 pe4 pe5 jobchar1 jobchar2 jobchar3 jobchar4 burnout1 burnout2 N valid 174 174 174 171 171 172 173 173 173 171 171 N missing 0 0 0 3 3 2 1 1 1 3 3 Mean 2.385 2.270 2.523 2.491 2.552 3.526 3.428 3.549 3.462 2.468 2.444 Std deviation 1.023 0.980 1.126 0.916 0.999 1.097 1.035 1.097 1.081 1.238 1.041 Variance 1.047 0.961 1.268 0.840 0.997 1.204 1.072 1.203 1.169 1.533 1.084 Minimum 1 1 1 1 1 1 1 1 1 1 1 Maximum 5 5 5 5 5 6 5 5 5 5 5 burnout3 burnout4 burnout5 burnout6 burnout7 burnout8 burnout9 burnout10 job sat itl1 itl2 N valid 171 171 173 173 173 173 173 174 173 174 174 N missing 3 3 1 1 1 1 1 0 1 0 0 Mean 2.462 2.532 2.734 2.665 2.584 2.688 2.798 2.270 2.931 2.569 2.615 Std deviation 1.144 1.356 1.028 1.064 1.146 1.129 0.988 1.021 1.169 0.828 0.903 Variance 1.309 1.356 1.057 1.131 1.314 1.274 0.976 1.042 1.367 0.686 0.816 Minimum 1 1 1 1 1 1 1 1 1 1 1 Maximum 6 5 5 5 5 5 5 4 5 4 5 https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/12/4@0:0 11/13 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. The results presented in the table in Output 14.2 indicate that: there are missing observations for every item except for the items pe1, pe2, pe3, burnout10, itl1, and itl2; there are illegal codes for items jobchar1 (a 6 has been entered in at least one cell), burnout3 (again, a 6 has been entered in at least one cell), and itl2 (a 5 has been entered in at least one cell); the responses to each individual item have a good spread. Appropriate actions were taken to correct the illegal entries. A further inspection of the missing data revealed that every participant answered either all or the vast majority of the questions. Therefore, no questionnaires were thrown out. Missing data will be ignored during subsequent analyses. From here, we can proceed with further detailed analyses to test the goodness of our data. TESTING THE GOODNESS OF MEASURES The reliability and validity of the measures can now be tested. Reliability Visit the companion website at www.wiley.com/college/sekaran for Author Video: Reliability. As discussed in Chapter 12, the reliability of a measure is established by testing for both consistency and stability. Consistency indicates how well the items measuring a concept hang together as a set. Cronbach's alpha is a reliability coefficient that indicates how well the items in a set are positively correlated to one another. Cronbach's alpha is computed in terms of the average intercorrelations among the items measuring the concept. The closer Cronbach's alpha is to 1, the higher the internal consistency reliability. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/12/4@0:0 12/13 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/12/4@0:0 13/13 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. OUTPUT 14.2 DESCRIPTIVE STATISTICS: CENTRAL TENDENCIES AND DISPERSIONS From the menus, choose: Analyze Descriptive Statistics Descriptives [Select the variables] Statistics … [Choose the relevant statistics needed] https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 1/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach Output pe1 pe2 pe3 pe4 pe5 jobchar1 jobchar2 jobchar3 jobchar4 burnout1 burnout2 N valid 174 174 174 171 171 172 173 173 173 171 171 N missing 0 0 0 3 3 2 1 1 1 3 3 Mean 2.385 2.270 2.523 2.491 2.552 3.526 3.428 3.549 3.462 2.468 2.444 Std deviation 1.023 0.980 1.126 0.916 0.999 1.097 1.035 1.097 1.081 1.238 1.041 Variance 1.047 0.961 1.268 0.840 0.997 1.204 1.072 1.203 1.169 1.533 1.084 Minimum 1 1 1 1 1 1 1 1 1 1 1 Maximum 5 5 5 5 5 6 5 5 5 5 5 burnout3 burnout4 burnout5 burnout6 burnout7 burnout8 burnout9 burnout10 job sat itl1 itl2 N valid 171 171 173 173 173 173 173 174 173 174 174 N missing 3 3 1 1 1 1 1 0 1 0 0 Mean 2.462 2.532 2.734 2.665 2.584 2.688 2.798 2.270 2.931 2.569 2.615 Std deviation 1.144 1.356 1.028 1.064 1.146 1.129 0.988 1.021 1.169 0.828 0.903 Variance 1.309 1.356 1.057 1.131 1.314 1.274 0.976 1.042 1.367 0.686 0.816 Minimum 1 1 1 1 1 1 1 1 1 1 1 Maximum 6 5 5 5 5 5 5 4 5 4 5 https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 2/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. The results presented in the table in Output 14.2 indicate that: there are missing observations for every item except for the items pe1, pe2, pe3, burnout10, itl1, and itl2; there are illegal codes for items jobchar1 (a 6 has been entered in at least one cell), burnout3 (again, a 6 has been entered in at least one cell), and itl2 (a 5 has been entered in at least one cell); the responses to each individual item have a good spread. Appropriate actions were taken to correct the illegal entries. A further inspection of the missing data revealed that every participant answered either all or the vast majority of the questions. Therefore, no questionnaires were thrown out. Missing data will be ignored during subsequent analyses. From here, we can proceed with further detailed analyses to test the goodness of our data. TESTING THE GOODNESS OF MEASURES The reliability and validity of the measures can now be tested. Reliability Visit the companion website at www.wiley.com/college/sekaran for Author Video: Reliability. As discussed in Chapter 12, the reliability of a measure is established by testing for both consistency and stability. Consistency indicates how well the items measuring a concept hang together as a set. Cronbach's alpha is a reliability coefficient that indicates how well the items in a set are positively correlated to one another. Cronbach's alpha is computed in terms of the average intercorrelations among the items measuring the concept. The closer Cronbach's alpha is to 1, the higher the internal consistency reliability. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 3/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. Another measure of consistency reliability used in specific situations is the split-half reliability coefficient. Since this reflects the correlations between two halves of a set of items, the coefficients obtained will vary depending on how the scale is split. Sometimes split-half reliability is obtained to test for consistency when more than one scale, dimension, or factor, is assessed. The items across each of the dimensions or factors are split, based on some predetermined logic (Campbell, 1976). In almost every case, Cronbach's alpha is an adequate test of internal consistency reliability. You will see later in this chapter how Cronbach's alpha is obtained through computer analysis. As discussed in Chapter 12, the stability of a measure can be assessed through parallel form reliability and test–retest reliability. When a high correlation between two similar forms of a measure (see Chapter 12) is obtained, parallel form reliability is established. Test– retest reliability can be established by computing the correlation between the same tests administered at two different time periods. Excelsior Enterprises: checking the reliability of the multi-item measures Because perceived equity, burnout, job enrichment, and intention to leave were measured with multi-item scales, the consistency of the respondents' answers to the scale items has to be tested for each measure. In Chapter 12, we explained that Cronbach's alpha is a popular test of inter-item consistency. Table 14.3 provides an overview of Cronbach's alpha for the four variables. This table shows that the alphas were all well above 0.60. TABLE 14.3 Reliability of the Excelsior Enterprises measures Variable Number of items Cronbach's alpha Perceived equity 5 0.882 Job enrichment 4 0.844 10 0.813 Intention to leave 2 0.749 Burnout In general, reliabilities less than 0.60 are considered to be poor, those in the 0.70 range, acceptable, and those over 0.80 good. Thus, the internal consistency reliability of the measures used in this study can be considered to be acceptable for the intention to leave measure and good for the other measures. It is important to note that all the negatively worded items in the questionnaire should first be reversed before the items are submitted for reliability tests. Unless all the items measuring a variable are in the same direction, the reliabilities obtained will be incorrect. A sample of the result obtained for the Cronbach's alpha test for job enrichment, together with instructions on how it is obtained, is shown in Output 14.3. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 4/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. OUTPUT 14.3 RELIABILITY ANALYSIS From the menus, choose: Analyze Scale Reliability Analysis … [Select the variables constituting the scale] Choose Model Alpha [this is the default option] Click on Statistics. Select Scale if item deleted under Descriptives https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 5/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach Output Reliability statistics Cronbach's alpha Number of items 0.844 4 Item-total statistics Scale mean if item deleted Scale variance if item deleted Corrected item-total variation Cronbach's alpha if item deleted Jobchar1 10,4393 7,143 ,735 ,777 Jobchar2 10,5318 7,483 ,714 ,788 Jobchar3 10,4104 7,639 ,620 ,828 Jobchar4 10,4971 7,554 ,652 ,814 https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 6/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach The reliability of the job enrichment measure is presented in the first table in Output 14.3. The second table provides an overview of the alphas if we take one of the items out of the measure. For instance, it is shown that if the first item (Jobchar1) is taken out, Cronbach's alpha of the new three-item measure will be 0.777. This means that the alpha will go down if we take item 1 out of our measure. Likewise, if we take out item 2, our alpha will go down and become 0.788. Note that even if our alpha would increase if we would take out one of the items, we would not take it out for two reasons. First, our alpha is above 0.7 so we do not have to take any remedial actions. Second, if we would take one of the items out, the validity of our measure would probably decrease. The items (all of them) were included in the original measure for a reason! https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 7/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. If, however, our Cronbach's alpha was too low (under 0.60) then we could use this table to find out which of the items would have to be removed from our measure to increase the inter-item consistency. Note that, usually, taking out an item, although improving the reliability of our measure, affects the validity of our measure in a negative way. Now that we have established that the inter-item consistency is satisfactory for perceived equity, job enrichment, burnout, and intention to leave, the scores on the original questions can be combined into a single score. For instance, a new “perceived equity” score can be calculated from the scores on the five individual “perceived equity” items (but only after items 1, 2, and 4 have been reverse coded). Likewise, a new “job enrichment” score can be calculated from the scores on the four individual “job enrichment” items, and so on. We have already explained that this involves calculating the summed score (per case/participant) and then dividing it by the number of items. Validity Factorial validity can be established by submitting the data for factor analysis. The results of factor analysis (a multivariate technique) will confirm whether or not the theorized dimensions emerge. Recall from Chapter 11 that measures are developed by first delineating the dimensions so as to operationalize the concept. Factor analysis reveals whether the dimensions are indeed tapped by the items in the measure, as theorized. Criterion-related validity can be established by testing for the power of the measure to differentiate individuals who are known to be different (refer to discussions regarding concurrent and predictive validity in Chapter 12). Convergent validity can be established when there is a high degree of correlation between two different sources responding to the same measure (e.g., both supervisors and subordinates respond similarly to a perceived reward system measure administered to them). Discriminant validity can be established when two distinctly different concepts are not correlated to each other (e.g., courage and honesty; leadership and motivation; attitudes and behavior). Convergent and discriminant validity can be established through the multitrait multimethod matrix, a full discussion of which is beyond the scope of this book. The student interested in knowing more about factor analysis and the multitrait multimethod matrix can refer to books on those subjects. When well-validated measures are used, there is no need, of course, to establish their validity again for each study. The reliability of the items can, however, be tested. EXCELSIOR ENTERPRISES: DESCRIPTIVE STATISTICS PART 2 https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 8/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. Once the new scores for perceived equity, job enrichment, burnout, and intention to leave have been calculated, we are ready to further analyze the data. Descriptive statistics such as maximum, minimum, means, standard deviations, and variance can now be obtained for the multi-item, interval-scaled independent and dependent variables. What's more, a correlation matrix can also be obtained to examine how the variables in our model are related to each other. This will help us to answer important questions such as: How big is the problem? In other words, to what extent are the employees of Excelsior Enterprises inclined to leave? What is the average inclination to leave? What is the nature of the problem? Compare, for instance, the histograms provided in Figure 14.8. The average ITL is the same in both cases. However, the graphs show us that in the first hypothetical histogram ITL is "fairly normally" distributed.1 In the second hypothetical histogram, the distribution is clearly not normal. In fact, it looks bimodal (with two peaks indicative of two modes). The first distribution suggests that most of the respondents are neither bent on leaving nor staying. The bimodal distribution, on the other hand, suggests that one group of employees is not inclined to leave at all, whereas another group is determined to leave the organization.2 FIGURE 14.8 Two hypothetical histograms of “intention to leave” https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 9/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. Descriptive statistics will also help us to answer the following questions: Are the employees satisfied with their jobs? What are the employees' perceptions on job enrichment? How many employees have which degrees of burnout? Is there much variance in the extent to which employees perceive the relationship with the company as equitable? What are the relationships between perceived equity, burnout, job enrichment, job satisfaction, and intention to leave? The answers to these questions will help us (together with the results of our hypotheses tests) to make informed decisions on how we can best solve the problem. Descriptive statistics such as maximum, minimum, means, standard deviations, and variance were obtained for the interval-scaled independent and dependent variables in the Excelsior Enterprises study. The results are shown in Table 14.4. It may be mentioned that all variables except ITL were tapped on a five-point scale. ITL was measured on a four-point scale. TABLE 14.4 Descriptive statistics for independent and dependent variables N ITL Minimum Maximum Mean Std deviation Variance 174 1.00 4.00 2.589 0.769 0.592 Job satisfaction 173 1.00 5.00 2.931 1.169 1.367 Perceived equity 174 1.00 4.60 2.444 0.883 0.694 Burnout 174 1.20 4.60 2.566 0.681 0.463 Jobchar 173 1.00 5.00 3.491 0.888 0.789 From the results, it may be seen that the mean of 2.59 on a four-point scale for ITL indicates that Excelsior Enterprises has a problem with regard to turnover. The minimum of 1 reveals that there are some employees who do not intend to leave at all, and the maximum of 4 reveals that some are seriously considering leaving. Job satisfaction is about average (2.91 on a five-point scale). The mean on perceived equity is rather low (2.44 on a five-point scale), as is the mean on experienced burnout (2.57). Finally, the job is perceived as somewhat enriched (3.49). Table 14.4 also points out that the variance for all the variables is rather high, indicating that participants' answers are not always very close to the mean on all the variables. Table 14.5 provides a more detailed account of employees' intentions to leave. This table shows that a large group of employees seriously considers leaving Excelsior Enterprises! Testing our hypotheses will improve our understanding of why employees consider leaving Excelsior Enterprises and will provide us with useful tools to reduce employees' intentions to leave the company. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 10/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach TABLE 14.5 Frequency table intention to leave Frequency Percentage Valid percentage Cumulative percentage 1.00 9 5.2 5.2 5.2 1.50 16 9.2 9.2 14.4 2.00 33 19.0 19.0 33.3 2.50 40 23.0 23.0 56.3 3.00 39 22.4 22.4 78.7 3.50 27 15.5 15.5 94.3 4.00 10 5.7 5.7 100.0 Total 174 100.0 100.0 The Pearson correlation matrix obtained for the five interval-scaled variables is shown in Table 14.6. From the results, we see that the intention to leave is, as would be expected, significantly negatively correlated to job satisfaction, perceived equity, and job enrichment. That is, the intention to leave is low if job satisfaction and equitable treatment are experienced, and the job is enriched. However, when individuals experience burnout (physical and emotional exhaustion), their intention to leave does not increase (this relationship is not significant; we will have more to say about this in the next chapter). Job satisfaction is positively correlated to perceived equity and an enriched job and negatively correlated to burnout and ITL. These correlations are all in the expected direction. It is important to note that the correlations between the independent variables does not exceed 0.272 for this sample. This is an important finding, because if correlations between the independent variables were very high (say, 0.75 and above), we might run into a collinearity problem in our regression analysis. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 11/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. TABLE 14.6 Correlations between independent and dependent variables Intention to leave Job satisfaction Perceived equity Burnout Job enrichment Intention to leave 1 −0.703 −0.384 0.037 −0.426 Sig. (two-tailed) 0.000 0.000 0.629 0.000 N 174 173 174 174 173 Job satisfaction −0.703 1 .280 −0.242 0.268 Sig. (two-tailed) 0.000 0.000 0.001 0.000 N 173 173 173 173 172 Perceived equity −0.384 0.280 1 0.089 0.272 Sig. (two-tailed) 0.000 0.000 0.241 0.000 N 174 173 174 174 173 Burnout 0.037 −0.242 0.089 1 0.028 Sig. (two-tailed) 0.629 0.000 0.241 N 174 173 174 174 173 Job enrichment −0.426 0.268 0.272 0.028 1 Sig. (two-tailed) 0.000 0.000 0.000 0.719 N 173 172 173 173 0.719 173 After we have obtained descriptive statistics for the variables in our study, we can test our hypotheses. Hypothesis testing is discussed in the next chapter. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 12/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach SUMMARY https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 13/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 14/14 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. CHAPTER 16 QUALITATIVE DATA ANALYSIS LEARNING OBJECTIVES After completing Chapter 16 you should be able to: 1. Discuss three important steps in qualitative data analysis: data reduction, data display, and drawing conclusions. 2. Discuss how reliability and validity have a different meaning in qualitative research in comparison to quantitative research and explain how reliability and validity are achieved in qualitative research. 3. Compare and contrast content analysis, narrative analysis, and analytic induction. 4. Describe the characteristics of big data and explain why big data holds many promises for organizations and managers. INTRODUCTION Qualitative data are data in the form of words. Examples of qualitative data are interview notes, transcripts of focus groups, answers to open-ended questions, transcriptions of video recordings, accounts of experiences with a product on the Internet, news articles, and the like. Qualitative data can come from a wide variety of primary sources and/or secondary sources, such as individuals, focus groups, company records, government publications, and the Internet. The analysis of qualitative data is aimed at making valid inferences from the often overwhelming amount of collected data. BOX 16.1 THE INTERNET AS A SOURCE OF TEXTUAL INFORMATION https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 1/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. Earlier in this book we explained that you can search the Internet for books, journals articles, conference proceedings, company publications, and the like. However, the Internet is more than a mere source of documents; it is also a rich source of textual information for qualitative research. For instance, there are many social networks on the Internet structured around products and services such as computer games, mobile telephones, movies, books, and music. Through an analysis of these social networks researchers may learn a lot about the needs of consumers, about the amount of time consumers spend in group communication, or about the social network that underlies the virtual community. In this way, social networks on the Internet may provide researchers and marketing and business strategists with valuable, strategic information. The possibilities for qualitative research on the Internet are unlimited, as the following example illustrates. In an effort to find out what motivates consumers to construct protest websites, Ward and Ostrom (2006) examined and analyzed protest websites. A content analysis revealed that consumers construct complaint websites to demonstrate their power, to influence others, and to gain revenge on the organization that betrayed them. This example illustrates how the Internet can be a valuable source of rich, authentic qualitative information. With increasing usage of the Internet, it will undoubtedly become even more important as a source of qualitative and quantitative information. THREE IMPORTANT STEPS IN QUALITATIVE DATA ANALYSIS The analysis of qualitative data is not easy. The problem is that, in comparison with quantitative data analysis, there are relatively few well-established and commonly accepted rules and guidelines for analyzing qualitative data. Over the years, however, some general approaches for the analysis of qualitative data have been developed. The approach discussed in this chapter is largely based on work of Miles and Huberman (1994). According to them, there are generally three steps in qualitative data analysis: data reduction, data display, and the drawing of conclusions. The first step in qualitative data analysis is concerned with data reduction. Data reduction refers to the process of selecting, coding, and categorizing the data. Data display refers to ways of presenting the data. A selection of quotes, a matrix, a graph, or a chart illustrating patterns in the data may help the researcher (and eventually the reader) to understand the data. In this way, data displays may help you to draw conclusions based on patterns in the reduced set of data. Having identified these general stages, it should be noted that qualitative data analysis is not a step-by-step, linear process but rather a continuous and iterative process. For instance, data coding may help you simultaneously to develop ideas on how the data may be displayed, as well as to draw some preliminary conclusions. In turn, preliminary conclusions may feed back into the way the raw data are coded, categorized, and displayed. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 2/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach EXAMPLE Qualitative research may involve the repeated sampling, collection, and analysis of data. As a result, qualitative data analysis may already start after some of the data have been collected. This chapter will discuss the three important steps in qualitative data analysis – data reduction, data display, and drawing and verifying conclusions – in some detail. To illustrate these steps in qualitative data analysis, we will introduce a case. We will use the case, by means of boxes throughout the chapter, to illustrate key parts of the qualitative research process. Data reduction Visit the companion website at www.wiley.com/college/sekaran for Author Video: Data reduction. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 3/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. Qualitative data collection produces large amounts of data. The first step in data analysis is therefore the reduction of data through coding and categorization. Coding is the analytic process through which the qualitative data that you have gathered are reduced, rearranged, and integrated to form theory. The purpose of coding is to help you to draw meaningful conclusions about the data. Codes are labels given to units of text which are later grouped and turned into categories. Coding is often an iterative process; you may have to return to your data repeatedly to increase your understanding of the data (i.e., to be able to recognize patterns in the data, to discover connections between the data, and to organize the data into coherent categories). https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 4/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach CASE INSTIGATIONS OF CUSTOMER ANGER INTRODUCTION Suppose that you are in a fashion shop and that you have just found a clothing item that you like. You go to the counter to pay for the item. At the counter you find a shop assistant who is talking to a friend on her mobile phone. You have to wait. You wait for a couple of minutes, but the shop assistant is in no hurry to finish the call. This event may make you angry. Waiting for service is a common cause of anger: the longer the delay, the angrier customers tend to be (Taylor, 1994). RESEARCH OBJECTIVE Prior research in marketing has applied appraisal theory to understand why anger is experienced in such situations (e.g., Folkes, Koletsky & Graham, 1987; Nyer, 1997; Taylor, 1994). Appraisal refers to the process of judging the significance of an event for personal well-being. The basic premise of appraisal theory is that emotions are related to the interpretations that people have about events: people may differ in the specific appraisals that are elicited by a particular event (for instance, waiting for service), but the same patterns of appraisal give rise to the same emotions. Most appraisal theories see appraisals as being a cause of emotions (Parrott, 2001). Along these lines, appraisal theory has been used to understand why anger is experienced in service settings. Although appraisal theory provides useful insights into the role of cognition in emotional service encounters, recent research suggests that, although they are clearly associated with anger, none of the aforementioned appraisals is a necessary or sufficient condition for anger to arise (Kuppens, VanMechelen, Smits & DeBoeck, 2003; Smith & Ellsworth, 1987). What’s more, for the specific purpose of avoiding customer anger, appraisal theory is too abstract to be diagnostic for services management. That is, service firm management may benefit more from a classification of incidents that are considered to be unfair (for instance, waiting for service and core service failures), than from the finding that unfair events are generally associated with customer anger. In other words, to be able to avoid customer anger, it is crucial that service firm management knows what specific precipitating events typically elicit this emotion in customers. After all, it is easier to manage such events than the appraisals that may or may not be associated with these particular events. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 5/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. Therefore, this study investigates events that typically instigate customer anger in services. This study builds on a rich tradition of research in psychology that has specified typical instigations of anger in everyday life. In addition, it builds on research in marketing that has identified and classified service failures, retail failures, and behaviors of service firms that cause customers to switch services (Bitner, Booms & Tetreault, 1990; Keaveney, 1995; Kelley, Hoffman & Davis, 1993). METHOD Procedure. Following related research in marketing, the critical incident technique (CIT) was used to identify critical behaviors of service providers that instigate customer anger (e.g., Bitner, Booms & Tetreault, 1990; Keaveney, 1995; Kelley, Hoffman & Davis, 1993; Mangold, Miller & Brockway, 1999). Critical incidents were collected by 30 trained research assistants, who were instructed to collect 30 critical incidents each. In order to obtain a sample representative of customers of service organizations, they were instructed to collect data from a wide variety of people. Participants were asked to record their critical incidents on a standardized form in the presence of the interviewer. This has several advantages, such as availability of the interviewer to answer questions and to provide explanations. Questionnaire. Participants were asked to record their answers on a standardized questionnaire, which was modeled after previous applications of CIT in services (e.g., Keaveney, 1995; Kelley, Hoffman & Davis, 1993). The questionnaire began by asking participants to indicate which of 30 different services they had purchased during the previous six-month period. Next, participants were asked to recall the last negative incident with a service provider that made them feel angry. They were asked to describe the incident in detail by means of open-ended questions. The open-ended questions were “What service are you thinking about?”, “Please tell us, in your own words, what happened. Why did you get angry?”, and “Try to tell us exactly what happened: where you were, what happened, what the service provider did, how you felt, what you said, and so forth.” Sample. Critical incidents were defined as events, combinations of events, or series of events between a customer and a service provider that caused customer anger. The interviewers collected 859 incidents. The participants (452 males, 407 females) represented a cross-section of the population. Their ages ranged between 16 and 87 with a mean age of 37.4. Approximately 2% of the participants had less than a completed high school education, whereas 45.1% had at least a bachelor's degree. The reported incidents covered more than 40 different service businesses, including banking and insurance, personal transportation (by airplane, bus, ferry, taxi, or train), hospitals, physicians, and dentists, repair and utility services, (local) government and the police, (virtual) stores, education and child care, entertainment, hospitality, restaurants, telecommunication companies, health clubs, contracting firms, hairdressers, real-estate agents, driving schools, rental companies, and travel agencies. On average, the negative events that participants reported had happened 18 weeks earlier. Coding begins with selecting the coding unit. Indeed, qualitative data can be analyzed at many levels. Examples of coding units include words, sentences, paragraphs, and themes. The smallest unit that is generally used is the word. A larger, and often more useful, unit of content analysis is the theme: “a single assertion about a subject” (Kassarjian, 1977, p. 12). When you are using the theme as a coding unit, you are primarily looking for the expression of an idea (Minichiello, Aroni, Timewell & Alexander, 1990). Thus, you might assign a code to a text unit of any size, as long as that unit of text represents a single theme or issue. Consider, for instance, the following critical incident: https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 6/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach After the meal I asked for the check. The waitress nodded and I expected to get the check. After three cigarettes there was still no check. I looked around and saw that the waitress was having a lively conversation with the bartender. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 7/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. This critical incident contains two themes: 1. The waitress does not provide service at the time she promises to: “The waitress nodded and I expected to get the check. After three cigarettes there was still no check.” 2. The waitress pays little attention to the customer: she is not late because she is very busy; instead of bringing the check, she is engaged in a lively conversation with the bartender. Accordingly, the aforementioned critical incident was coded as: “delivery promises” (that were broken) and “personal attention” (that was not provided). This example illustrates how the codes “delivery promises” and “personal attention” help to reduce the data to a more manageable amount. Note that proper coding not only involves reducing the data but also making sure that no relevant data are eliminated. Hence, it is important that the codes “delivery promises” and “personal attention” capture the meaning of the coded unit of text. BOX 16.2 DATA ANALYSIS Unit of analysis Since the term “critical incident” can refer to either the overall story of a participant or to discrete behaviors contained within this story, the first step in data analysis is to determine the appropriate unit of analysis (Kassarjian, 1977). In this study, critical behavior was chosen as the unit of analysis. For this reason, 600 critical incidents were coded into 886 critical behaviors. For instance, a critical incident in which a service provider does not provide prompt service and treats a customer in a rude manner was coded as containing two critical behaviors (“unresponsiveness” and “insulting behavior”). Categorization is the process of organizing, arranging, and classifying coding units. Codes and categories can be developed both inductively and deductively. In situations where there is no theory available, you must generate codes and categories inductively from the data. In its extreme form, this is what has been called grounded theory (see Chapter 6). In many situations, however, you will have a preliminary theory on which you can base your codes and categories. In these situations you can construct an initial list of codes and categories from the theory, and, if necessary, change or refine these during the research process as new codes and categories emerge inductively (Miles & Huberman, 1994). The benefit of the adoption of existing codes and categories is that you are able to build on and/or expand prevailing knowledge. BOX 16.3 https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 8/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach DATA ANALYSIS Categorization Qualitative data analysis was used to examine the data (Kassarjian, 1977). As a first step, two judges coded critical incidents into critical behaviors. Next, (sub)categories were developed based on these critical behaviors. Two judges (A and B) independently developed mutually exclusive and exhaustive categories and subcategories for responses 1 to 400 (587 critical behaviors). Two other trained judges (C and D) independently sorted the critical behaviors into the categories provided by judges A and B. Finally, a fifth, independent judge (E) carried out a final sort. https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 9/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. As you begin to organize your data into categories and subcategories you will begin to notice patterns and relationships between the data. Note that your list of categories and subcategories may change during the process of analyzing the data. For instance, new categories may have to be identified, definitions of categories may have to be changed, and categories may have to be broken into subcategories. This is all part of the iterative process of qualitative data analysis. BOX 16.4 RESULTS Categories Participants reported a wide range of critical behaviors that made them angry. Some of these behaviors were closely related to the outcome of the service process (e.g., “My suitcase was heavily damaged”). Other behaviors were related to service delivery (e.g., “For three days in a row I tried to make an appointment […] via the telephone. The line was always busy”) or interpersonal relationships (e.g., “She did not stir a finger. She was definitely not intending to help me”). Finally, customers got angry because of inadequate responses to service failures (e.g., “He did not even apologize” or “He refused to give me back my money”). These four specific behavior types represent the four overarching categories of events that instigate customer anger. Two of these categories were further separated into, respectively, three categories representing service delivery or procedural failures (“unreliability,” “inaccessibility,” and “company policies”) and two categories representing interpersonal relationships or interactional failures (“insensitive behavior” and “impolite behavior”). The main reason for this was that the categories “procedural failures” and “interactional failures” would otherwise be too heterogeneous with respect to their composition and, more importantly, with respect to ways of avoiding or dealing with these failures. For instance, avoiding anger in response to unreliability (not performing in accordance with agreements) will most likely call for a different – and maybe even opposite – approach than avoiding anger in response to company policies (performing in accordance with company rules and procedures), even though these failures are both procedural; that is, related to service delivery. Sometimes you may want to capture the number of times a particular theme or event occurs, or how many respondents bring up certain themes or events. Quantification of your qualitative data may provide you with a rough idea about the (relative) importance of the categories and subcategories. BOX 16.5 Table 16.1 indicates that “price agreements that were broken” (category “unreliability”, subcategory “pricing”) was mentioned 12 times as a cause of anger. Hence, broken price agreements represent 1.35% of the total number of critical behaviors (886) and 2% of the https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 10/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach total number of the reported critical incidents (600). The sixth column indicates that nine participants mentioned broken price agreements as the sole cause of their anger, whereas three participants mentioned at least one additional critical behavior (column 7). https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 11/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. TABLE 16.1 Instigations of anger in service consumption settings (Sub)category (Sub)category definition No. of No. of behaviors behaviors in % of behaviors No. of behaviors in % of incidents No. of behaviors in singlefactor incidents No. of Example(s) behaviors in multifactor incidents Procedural failures Unreliability Service firm does 156 not perform the service dependably. 17.61 26.00 73 83 Delivery promises Service provider does not provide services at the time it promises to do so. 11.74 17.33 42 62 Wait for appointment with dentist, physician, or hairdresser, or on a plane, train, or taxi. Service provision Service provider 40 does not provide the service that was agreed upon. 4.52 6.67 22 18 Client receives different car than agreed upon with car rental company or different apartment than agreed upon with travel agent. Bicycle repairers, car mechanics, or building contractors carry out different work than agreed upon or work that was not agreed upon with their clients. 104 https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 12/18 12/25/2017 MBS Direct: Research Methods For Business: A Skill Building Approach (Sub)category (Sub)category definition No. of No. of behaviors behaviors in % of behaviors No. of behaviors in % of incidents No. of behaviors in singlefactor incidents No. of Example(s) behaviors in multifactor incidents Procedural failures Pricing Price agreements are broken. 12 1.35 2.00 9 3 Customers experience difficulties with engaging in the service process. 47 5.30 7.83 17 30 Communicative inaccessibility Inaccessibility via telephone, fax, email, and/or the Internet. 26 2.93 4.33 9 17 “For three days in a row I tried to make an appointment with my physician via the telephone. The line was always busy.” Physical inaccessibility of service elements Customers experience difficulties with accessing a certain element or part of the service. 12 1.35 2.00 4 8 Check-in counter of an airline company, cashpoint of a supermarket, service desk of a holiday resort, or baggage claim at an airport. PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. “After a party we called a cab. We were with a party of five. A van would take us home for a fixed, low price. However, upon arrival, the driver asked the regular clock price.” Inaccessibility https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 13/18 12/25/2017 (Sub)category MBS Direct: Research Methods For Business: A Skill Building Approach (Sub)category definition No. of No. of behaviors behaviors in % of behaviors No. of behaviors in % of incidents No. of behaviors in singlefactor incidents No. of Example(s) behaviors in multifactor incidents Procedural failures Physical inaccessibility of service provider Difficult physical accessibility of service provider because of inconvenient locations or opening hours. 9 1.02 1.50 4 5 Company policies Service provider's rules and procedures or the execution of rules and procedures by service staff is perceived to be unfair. 76 8.57 12.67 45 31 Rules and procedures Inefficient, illtimed, and unclear rules and procedures. 66 7.45 11.00 38 28 https://mbsdirect.vitalsource.com/#/books/9781119266846/cfi/6/42!/4/2/16/22/2/2/2/2@0:81.4 “It was three o'clock on a Saturday afternoon and the dry cleaner was already closed.” “It turned out that the [Cystic Fibrosis] foundation used unfair procedures for assigning families with cystic fibrosis to vacations. For example, some families were invited for years in a row even though this is not allowed.” 14/18 12/25/2017 (Sub)category MBS Direct: Research Methods For Business: A Skill Building Approach (Sub)category definition No. of No. of behaviors behaviors in % of behaviors No. of behaviors in % of incidents No. of behaviors in singlefactor incidents No. of Example(s) behaviors in multifactor incidents Procedural failures PRINTED BY: matthew.phillips5@snhu.edu. Printing is for personal, private use only. No part of this book may be reproduced or transmitted without publisher's prior permission. Violators will be prosecuted. “Only two days before our wedding my wife was ordered to leave the country by the immigration office.” “I went to the local administration to report a change of address. At the same time I wanted to apply for a parking license. In that case you must dr...
Purchase answer to see full attachment