WHAT DO BUSINESS PROFESSIONALS SAY ABOUT FORECASTING IN THE MARKETING CURRICULUM?
Forecasting practices and processes have been examined in the 1980s (e.g., Mentzer & Cox, 1984), the 1990s (e.g., Mentzer & Kahn, 1995), and more recently, the 2000s (e.g., Cecatto, Belfiore, & Vieira, 2012; Jain, 2006; McCarthy et al., 2006; Spralls & Wilson, 2016). In the last decade, traditional business practices and routines have dramatically shifted in response to disrupting technologies such as big data, mobile Internet, and the Internet of Things (IoT) (Chen et al., 2015). In addition, quantitative forecasting technique usage has been linked to high rates of technological change (McCarthy et al., 2006). An example here helps to illustrate these insights. The emerging cutting-edge analytics of data and text mining, allow forecasters to capture interesting, novel, and potentially useful patterns from huge data sets (think big data).
Data and text mining have great potential if one considers forecasting to be more than traditional methods based on time series data, extending the forecast concept to predictive analytics. For example, consider the title change of a forecasting text from Business Forecasting with ForecastX[TM] 6th edition to Forecasting and Predictive Analytics with ForecastX[TM] 7th edition (Keating & Wilson, 2019).
The objective of data mining is to build an effective and efficient predictive model that is predictive and that can be generalized to new data (Chen et al., 2015). Data mining can have a direct impact on forecasting effectiveness (i.e., forecasting accuracy, cost reduction, and delivery performance). Indeed, Chen et al. (2015, p. 6) notes that "Retailers collect customer information, related transaction information, and product information to significantly improve accuracy of product demand forecasting, assortment optimization, product recommendation, and ranking across retailers and manufacturers." Such changes in the business environment suggest that changes in the use of forecasting techniques are likely to have taken place (Doering & Suresh, 2016). Despite extant forecasting management research literature (e.g., Cecatto et al., 2012; McCarthy et al., 2006), scholarly inquiry, that focuses on marketing education and the forecasting techniques currently used in U.S. businesses, has been lacking (Beal & Wilson, 2015; Spralls & Wilson, 2016).
The purpose of the current research is to report the results of a study of forecasting technique usefulness to practicing forecasters and gauge their overall satisfaction with various forecasting techniques. Research indicates that there is a link between the forecasting techniques used by applied researchers and their marketing education (Porter & McKibbin, 1988). Researchers generally agree that marketing education drives technique usage by practicing forecasters (e.g., Bellenger, Howell, Wilcox, & Greenberg, 1990; McCarthy et al., 2006).
THE IMPORTANCE OF THIS STUDY
The authors of the present study argue that not only should academe inform industry about forecasting techniques but industry should inform academe as well. This should include providing feedback as to what forecasting techniques work, in various contexts (e.g., type of industry), and those that don't. To help understand what business forecasters think is important, an email survey was used to obtain insights from these professionals. This research is particularly important as some researchers have called for increased analytics in marketing curricula (e.g., Beal & Wilson, 2015; Spralls & Wilson, 2016). More specifically, researchers agree that at least some forecasting should be covered in marketing research, strategic marketing, sales management, and supply chain courses (Beal & Wilson, 2015; Chu, 2007; Smith & Clarke, 2011; Spralls & Wilson, 2016). Based on McCarthy's et al., (2006) finding that marketing related functions are dominate in the forecasting activities of American businesses, forecasting may be a good place to begin to increase analytics in the marketing curriculum.
Additionally, this study is important because, often, innovative forecasting techniques are not adequately diffused into business (Myers, Greyser, & Massy, 1979). Although forecasting techniques have become more sophisticated over the last three decades (Bellenger et al., 1990; McCarthy et al., 2006), not all firms are knowledgeable about the extent to which they are used in their industry.
Three practical objectives are to: (1) uncover how business forecasters evaluate currently available forecasting techniques and how does their forecasting affect their bottom line performance. In this regard it has been shown that three percent increase in forecast accuracy has been estimated to increase profits by two percent (Morris, 2014), (2) foster application oriented forecasting research, and (3) help marketing educators develop curricula based on forecasting techniques actually used in industry (Greenberg, Goldstucker, & Bellenger, 1977; Beal & Wilson, 2015; Bellenger et al., 1990; Spralls & Wilson, 2016; Walker et al., 2009). To accomplish these objectives, the following research questions are put forth:
* What forecasting methods should Marketing majors know as they enter the workforce?
* In what academic departments do professional forecasters think forecasting should be taught?
* Are forecasting certifications valued by industry?
* Is it sufficient to know how to apply a forecasting technique or is understanding the technique essential for recent graduates?
To fulfill the study purpose and address the research questions, we begin with a literature review. Then, the research methodology will be discussed followed by a discussion of survey results.
This study contributes to the forecasting management literature. Relying on Moon (2004) as well as the empirical findings of Doering and Suresh (2016), forecasting management consists of at least four major sets of practices, including functional integration (i.e., communication, coordination, and collaboration), a methodological approach, having a well defined forecasting system, and performance measurements. Within the methodical approach dimension of forecasting management, the present study focuses on two major areas at the firm level.
First, it is concerned with forecasting process quality in general and the level of forecasting technique sophistication in particular. Doering and Suresh (2016) found forecasting process quality to have a strong impact on forecasting management competence. At a more granular level, it is important to identify the forecasting techniques firms use, find useful, and with which they are satisfied/dissatisfied. Based on previous research, it is expected that larger firms will display higher levels of forecasting technique sophistication (McCarthy et al., 2006). Yet, Armstrong (2001) and Cecatto et al. (2012) argue that simple methods are about as accurate as more complex methods and that simplicity fosters practitioner understanding and implementation.
This brings us to the second dimension considered in the research. That is, that the level of training, education, and documentation in the forecasting process are likely to drive forecasting technique sophistication (McCarthy et al., 2006). Previous research on forecasting practices has focused on: (1) defining the relationships among sales forecasting technique satisfaction, familiarity, application, and usage (Mentzer & Cox, 1984; Mentzer & Kahn, 1995), (2) forecasting techniques within specific industries, functions, and disciplines (Choo, 2000; Peterson & Jun, 1998; Wisner & Stanley, 1994), (3) forecasting practices of international firms (Duran & Flores, 1998; Klassen & Flores, 2001; Lam, 1996; Mady, 2000), (4) forecasting techniques across a broad range of industries (Sanders & Manrodt, 1994; Jain, 2006), and (5) the penetration of marketing research techniques (including forecasting) in various industries (Greenberg et al., 1977).
Regarding satisfaction, McCarthy et al. (2006) found that respondents were most satisfied with exponential smoothing methods, followed by trend-line analysis, decomposition, moving averages, and regression. Based on longitudinal research over the previous 20 years, McCarthy et al. (2006) found that firms tended to depend more on qualitative forecasting techniques, which may be easier to learn and understand. However, quantitative methods have been shown to be more accurate. Additionally, McCarthy et al. (2006), found low levels of technique familiarity, which may partially explain poor forecast accuracy. Decreased forecasting accuracy may be the result of inadequate training and education. Indeed, "Given the wide range of available forecasting techniques, selecting proper forecasting techniques for each class of items has been a challenge for practitioners. This requires proper training of forecasting personnel on statistical aspects and correct use of forecasting methods" (Doering & Suresh, 2016, p.87).
Walker et al., (2009) argue that marketing faculty should not only teach forecasting theory (particularly at the graduate level), but also how to apply forecasting techniques and evaluate their usefulness in various contexts. Nevertheless, research indicates that forecasting is rarely taught as a standalone course in marketing departments (Spralls & Wilson, 2016). Much earlier Wilson and Daubek (1989) found that only 6.4% of marketing faculty teach forecasting as a standalone course. Hanke and Weigand (1994) found that only 47 percent of AACSB accredited schools offered a standalone forecasting course. It is unclear if the results in Hanke and Weigand refer to having a forecasting course in the catalogue or whether it means having the course regularly in course offerings. Thus, it appears that forecasting, as a standalone course, is not a valued component in the marketing curriculum (Beal & Wilson, 2015; Spralls & Wilson, 2016).
It was difficult to obtain good contact information for forecasting professionals. The usual sources for survey email addresses, such as Survey Sampling, did not have what was needed. Eventually a vendor was located who indicated that they could provide an appropriate list. However, the list obtained was not as specific as expected and contained some individuals with little or no forecasting responsibility. For the list provided, an initial email request was sent. followed by two follow-up requests for participation (Dillman, 2000). Simultaneously, Forecast Pro was contacted about accessing their clients. They were helpful by providing access to a link for the survey on their forecasting blog. Eventually, 136 respondents were obtained through these combined avenues.
In addition, a business forecasting discussion board on LinkedIn.com that included a link to the survey instrument (Appendix 1) was established. LinkedIn.com was selected as a sampling frame because it provided access to specific target groups. More specifically, 10 LinkedIn groups that appeared to focus on diffusing and improving the use of forecasting techniques were selected:
1. Institute of Business Forecasting & Planning
2. New Product Forecasting
3. Sales & Operations Planning, Forecasting
4. Demand Planning, Sales Forecasting
5. Retail Forecasting & Planning
6. Biopharmaceutical Product Planning, Forecasting
7. Forecasting Net
8. IEEE Working Group on Energy Forecasting
9. Demand Management & Forecasting
10. Autobox User Group--Forecasting & Time Series Analysis
This resulted in an additional 87 respondents for a total of 223 respondents. A few respondents answered only the initial demographic items or did not complete the survey sufficiently and were not included in the analyses. After data cleansing, there were 219 usable returns. Some studies have found that multi-method survey approaches, such as the approach taken here, produce more reliable data (Dolnicar, Laesser, & Matus, 2009; Greenlaw & Brown Welty, 2009). Yet, others find no difference between mixed-mode formats and conventional methods (Porter & Whitcomb, 2007).
It is conventional to report a response rate, however because the survey link was available to an unknown number of business forecasters there is no way to provide a reasonable estimate. In order to evaluate the representativeness of the mixed-mode sample used in the study, the characteristics of respondents have been compared with the characteristics of respondents in McCarthy et al. (2006). Results show considerable diversity among respondents and similarity to the results of McCarthy et al. in their important 2006 Journal of Forecasting study. This provides evidence that the sample used in the current study is representative of the broader forecasting community.
McCarthy et al. (2006) reported that more than half of their respondents had sales revenue between $101 million and $5 billion. The categories used in the present study make exact comparison difficult but 78% reported revenues in excess of $50 million which suggest similarity. Additionally, McCarthy et al. (2006) reported having more than 50% of respondents from firms with 1,000 or more employees compared to 59.3% in the present study who reported more than 1,000 employees.
It is helpful to provide two additional comparisons of the findings with those from the McCarthy et al. (2006) study. In both surveys, respondents were asked to indicate what departments within their firms "contribute" to the forecast and which "own" the forecast. The departments included in both surveys (in alphabetical order) are:
* Executive Leadership Team
* Product Management
The Spearman rank order correlations between the two studies for both the "contributes" and the "owns" variables were calculated. The correlation for departments that contribute to the forecast is 0.867 (two tailed significance = 0.000). The correlation for departments which own the forecast is 0.783 (two tailed significance = 0.003). Thus, the results of the present study can be viewed as similar to those reported by McCarthy et al. (2006). This provides additional support for the notion that the sample used is representative of the broader population of business forecasters.
There were respondents from 38 States and 14 countries. The majority of non-USA counties were from the EU (including the UK). In an open ended reply about their undergraduate majors, respondents gave a wide variety of answers. The most common were: Business twenty six (26); Math seventeen (17); Economics fifteen (15); Accounting twelve (12); Finance eleven (11); and Marketing eleven (11). The diversity of the other responses included Chemistry, Education, English, Sociology, Statistics, Zoology, and many more.
The sample had considerable variation across respondents' ages with a range from 24 to 73 with a mean age of 44.17 (s.d. = 10.96). On average they had 12.87 years of forecasting experience (s.d. = 9.22) with a range from 0 to 46. In terms of business size 22% had revenues of 50 million USD or less, 29% were between 51 and 999 million USD, and 49% reported revenues of one billion USD or more. With respect to the number of employees, about 41% reported 1,000 or fewer employees, 28% between 1,001 and 10,000, and 31% over 10,000 employees.
FINDINGS AND DISCUSSION
One area of interest is the depth of forecasting knowledge business forecasters expect when they hire college graduates. As indicated in Table 1, respondents rated four statements, on a one to five importance scale, in which the higher the rating the more important the level of knowledge. What was found is not surprising in terms of the order of the constructs evaluated. An understanding of forecast methods was most important followed by knowing how to develop a forecast, when to use various methods, and understanding trade-offs between methods respectively. Using a single sample t-test, the means for all four constructs are significantly greater than the scale midpoint of three.
Recall, the earlier discussion of respondents' relative rankings of the departments which "contribute" to forecasting and those which 'own' the forecast. In Table 2, the departments are ordered by the mean rating for the degree to which departments contribute to the forecasting effort. This was based on a five-point scale in which one was low and five was high in terms of contribution to the forecasting process. With respect to ownership of the forecast, respondents were asked to indicate the one department that had ownership of the forecast. The percentage response is indicated in Table 2 for each department. As you might expect, the Pearson correlation between the mean ratings and the percentages is high (0.888 with a two tailed significance of 0.000).
As depicted in Table 2, marketing and sales are rated high in terms of contributing to forecasting. Combining sales with marketing, since sales is typically taught within a marketing department, close to 20% of forecast ownership in the business community is within the domain of sales and marketing departments. Interestingly, McCarthy et al. (2006) found that 74% of respondents reported that sales and marketing own the forecast, while Jain (2006) found sales and marketing to account for 29%. Thus, one can argue that forecasting should be included in marketing curricula.
Respondents were asked to indicate the academic departments in which they think forecasting is currently taught as well as the departments in which they think forecasting should be taught. A careful review of Tables 3 and 4 reveals that, for every department, respondents had higher ratings on forecasting should be taught than for forecasting is taught. Overall the mean difference is 0.86. The difference in means was highest for management and marketing departments. This suggests that more teaching of forecasting in marketing departments would be desirable.
Going beyond simple analytics, big data has velocity, variety, and volume. For example, it is estimated that Walmart collects in excess of 2.5 petabytes of data every hour from its customer transactions. Consider that a petabyte is one quadrillion bytes, which translates roughly to about 20 million filing cabinets filled with text (McAfee et al, 2012). As a construct, big data may require forecasting professionals to think beyond traditional quantitative and qualitative forecasting. For this reason the survey included business forecasters' perceptions regarding the extent to which they expect "big data" to influence their approach to forecasting. The findings are shown in Table 5. It appears that the perception is that big data and the Internet of Things are viewed as influencing respondents' approaches to forecasting (both means are significantly above the scale midpoint of three; two tailed significance for both less than 0.00).
John Chambers, Cisco CEO, commented that there were only 1,000 devices connected to the Internet when Cisco was created in 1984. There were more than 10 billion connected devices in 2015. By 2025, there will be 500 billion devices connected to the Internet (Camhi, 2018). The Internet of Things (IoT) provides data on product performance, scheduled maintenance, product/service improvements (Cheng, 2014), and much more related to consumer shopping behavior. The current study sought respondents' perceptions regarding the extent to which the IoT will influence professional forecasting. As shown in Table 5, the IoT is also seen as influencing how forecasting may be done. For comparison, "big data" is the more important of the two based on a dependent sample t-test (one tailed significance = 0.000). The fact that both big data and the IoT are expected to influence how forecasters work in the future, suggests that more discussion of these concepts in marketing courses would be appropriate.
The following question was presented to respondents: What forecasting methods should marketing majors know as they enter the workforce? Respondents were provided a list of 15 forecasting techniques (Appendix 1) and asked to rate each on how important the technique would be for job applicants to know using a one to five importance scale. The higher the rating the more important the method. The list, shown in Table 6, covers a wide variety of quantitative and qualitative techniques. In order to encourage response and keep the study manageable, the number of forecasting techniques considered was limited to the 15 shown.
As indicated in Table 6, professional forecasters would like job applicants to know nearly all of the 15 methods considered. Only the Delphi method and Text Mining failed to have mean ratings significantly above the scale midpoint of three. In general, quantitative (objective) methods were rated higher than qualitative (subjective) methods. The highest rated qualitative method was "Sales Force Composite" which might suggest that marketing departments which have sales majors, minors, and/or concentrations might want to be sure this method is covered in the sales curriculum. As shown in Table 6, the Jury of Executive Opinion, the Naive method, ARIMA, the Delphi method and Text Mining are rated as the five least important methods for job applicants to know. These five methods appear to be generally less important than others as discussed subsequently.
The authors are affiliated with a College of Business that encourages students to pursue professional certifications in areas of interest to them prior to graduation. Is this good advice? Respondents were asked the extent to which having a forecast certification would be useful for a job applicant. As summarized in Table 7, over 50% indicated a forecasting certification would be "Helpful" or "Very Helpful." Less than 4% indicated that such a certification would not be at all helpful. Forecasting certifications are available, at various levels, from the Institute of Business Forecasting and Planning (IBF) and the International Institute of Forecasters (IFF) as well as other organizations.
In order to gain additional insight, respondents were asked how useful professional forecasters thought the selected 15 methods are and their satisfaction with each. These results are shown in Tables 8 and 9. Interpreting these results is somewhat perplexing. In Table 8 only three methods (moving average, linear regression trends, and combining forecasts) had mean ratings for usefulness significantly above the scale midpoint of three. Although McCarthy et al. (2006) did not include usefulness in their study, the authors found that respondents were most familiar with moving average, among quantitative methods. In the present study, ARIMA (mean 1.99) scored very low in usefulness. This finding is supported by Cecatto et al. (2012), who report that for Brazilian food companies, the major drawback to using ARIMA is difficulty in understanding the method. Surprisingly, the findings of the present study regarding the most useful methods are in direct contrast to the findings regarding respondent satisfaction with the methods. As indicated in Table 9, respondents were generally well satisfied with all methods evaluated with the exception of Text Mining. Additionally, the findings of the present study are in stark contrast to McCarthy et al. (2006), who found that only two techniques (out of 14) were rated satisfactory by respondents.
One possible explanation for seemingly contradictory findings of the present study may be related to complexity of the forecasting technique. Greenberg et al, (1977) found that there is an inverse relationship between research technique complexity and industry penetration (percentage of firms using a technique). Put another way, the percent of firms responding that they use various techniques decreases as the technique becomes more complex. Granted, usage is not the same as usefulness. However, it stands to reason that firms are more likely to use techniques they find useful. The four techniques that scored high in usefulness may be perceived as less complex than some of the other techniques. It could be that some firms may simply lack the capability, understanding, and/or need for the more complex techniques (Greenberg et al., 1977).
Another explanation may be related to how the respondents interpreted the term satisfaction on the questionnaire. Drawing on Anderson, Fornell, and Lehman (1994), satisfaction, in the marketing literature, can be defined as an overall evaluation based on the total consumption experience over time. Many respondents may have formed a favorable opinion of a forecasting technique despite having had limited direct experience with it. Indeed, McCarthy et al. (2006) reported decreased forecasting technique familiarity compared to the Mentzer and Kahn (1995) study. Unlike McCarthy et al. (2006), the current survey did not provide a hyperlink to forecasting technique definitions. This might have diminished, or avoided, what might be a "halo effect" over nearly all forecasting techniques used in the study. Further research is needed to explain these seemingly conflicting findings.
Although this study is largely descriptive, the relationship between the three measures of "usefulness", "satisfaction" and "importance for new employees to know" was investigated. Based on the Spearman rank order correlations between the three pairs of these measures there appears to be reasonable consistency. As shown in Table 10, all three correlations are strong and very significant. Thus, it may be that what appears on the surface to be an inconsistency between usefulness and satisfaction, in reality, may not be that important.
Complementing the seminal work of McCarthy et al. (2006), the present study contributes to the literature on sales forecasting management by examining the usefulness of forecasting techniques rather than simply the percent using forecasting techniques. In order to be found useful (a much higher bar to reach), a technique must first be used. Perceived usefulness refers to individuals' beliefs that using a particular technology may improve their performance (e.g., convenience, satisfaction, economic benefit, and task performance) (Yang & Yoo, 2004). Extending on Venkatesh and Davis (2000), perceived usefulness results from the effects of cognitive instrumental processes that include: (1) benefits resulting from use, (2) cognitive matching of task objectives with the consequences of forecasting technique use, and (3) result demonstrability (i.e., the tangibility of the results of using a particular technique). To our knowledge, the current study is the first study in forecasting management to include usefulness.
The present study revealed that the most useful techniques for forecasting practitioners include moving average, linear regression trends, combining forecasts, and simple exponential smoothing. The finding that moving average is one of the four most useful techniques is supported by a Brazilian food industry study by Cecatto et al. (2012). Unlike Jain and Malehorn (2006), Cecatto et al. (2012) did not find exponential smoothing to be heavily used in the Brazilian food industry.
The findings of the current study (i.e., that the most useful techniques include moving average, linear regression trends, combining forecasts, and simple exponential smoothing) should contribute to better forecasting practices and improved firm/marketing department collaboration, which, in turn, could lead to reduced operational costs (Doering & Suresh, 2016). Indeed, Doering and Suresh (2016), found that forecasting process quality (i.e., model selection, forecast preparation, and performance measurement) directly influences both forecasting accuracy and cost reduction. Studies such as the research presented here-in should help firms' benchmark best practices in their industry and identify areas where additional education and training are needed (Jain, 2006).
This study suggests that, at a minimum, marketing departments should offer courses that include the forecasting techniques identified in this study as most useful to professional forecasters (i.e., moving average, linear regression trends, combining forecasts, and simple exponential smoothing). It is important for all marketing students to understand the most useful forecasting techniques as they enter into the workforce (Beal & Wilson, 2015; Easton, Roberts, & Tiao, 1988; Hardy, 1992; Loomis & Cox, 2003). Additionally, the present study revealed that the Jury of Executive Opinion, the Naive method, ARIMA, the Delphi method and Text Mining are the five least important methods for job applicants to know. Thus marketing departments may want to place less of a focus on these methods. However, it is likely that Text Mining will become increasingly important as advances in the related software make Text Mining more accessible.
Several limitations are evident in the research presented here. First, like other types of survey research, the results are based on the information provided by targeted informants and depend on one survey instrument. Second, it would have been interesting to analyze responses by type of industry and/or product type (Cecatto et al., 2012). These, as well as firm size have been shown to affect research technique penetration. Larger firms report higher usage rates of research techniques in general (Greenberg et al., 1977; Doering & Suresh, 2016). Third, professional forecasters found only moving average, linear regression trends, combining forecasts, and simple exponential smoothing to have mean ratings for usefulness significantly above the scale midpoint of three. Yet, respondents were generally satisfied with all methods evaluated except Text Mining. It would be interesting to conduct in depth interviews to see if the differences observed between usefulness and satisfaction could be reconciled.
In conclusion, we argue that forecasting is an essential within-firm competence for carrying out marketing functions. It follows that there is a demand for graduates with the ability to use forecasting techniques, software packages, and systems. This line of thought suggests that marketing departments should facilitate, foster, and encourage a forecasting competence in undergraduate and graduate marketing students.
RECOMMENDATIONS FOR FUTURE RESEARCH
It is likely that marketing faculty teach the forecasting techniques that they think are important and useful to practitioners. Thus, there is a need for additional research that compares the forecasting practices of business professionals to the perceptions and forecasting teaching practices of marketing faculty regarding: (1) the most important forecasting methods for job applicants to know, (2) agreement on the most useful forecasting techniques in industry, (3) agreement on the most satisfactory forecasting techniques, and (4) where (in what departments) forecasting should be taught. Finally, future research should explore the impact of big data and the Internet of Things on forecasting.
Anderson, E.W., Fornell, C., & Lehman, D. R. (1994). Customer satisfaction, market share, and profitability: findings from Sweden. Journal of Marketing, 58, 53-66.
Armstrong, J. S. (Ed.). (2001). Principles of forecasting: a handbook for researchers and practitioners (Vol. 30). Springer Science & Business Media.
Bellenger, D. N., Howell, R. D., Wilcox, J., & Greenberg, B. A. (1990). The diffusion of marketing research techniques in business: Implications for marketing education. Journal of Marketing Education, 12(1), 24-29. From doi:10.1177/027347539001200104.
Beal, M., & Wilson J. H. (2015). Forecasting in the Marketing Curriculum. International Journal of Business Marketing, and Decision Sciences, 8(1), 109-115.
Camhi, J. (2018). "Former Cisco CEO John Chambers predicts 500 billion connected devices by 2025," Retrieved March 26, 2018, from http://www.businessinsider.com/former-cisco-ceo-500-billion-connected-devices-by-2025-2015-11.
Cecatto, C., Belfiore, P., & Vieira, J. G. V. (2012). Forecasting practices in Brazilian food industry. Journal of logistics Management, 1(4), 24-36.
Chen, F., Deng, P., Wan, J., Zhang, D., Vasilakos, A. V., & Rong, X. (2015). Data mining for the Internet of things: Literature review and challenges. International Journal of Distributed Sensor Networks, 2015, 1-14. doi:10.1155/2015/431047.
Cheng, R. (2014). How much is the internet of everything worth? Cisco says $19 trillion. C/Net. Retrieved March 10, 2016, from http://www.cnet.com/news/how-much-is-the-internet-of-everything-worth-cisco-says-19-trillion.
Chu, S. (2007). Some initiatives in a business forecasting course. Journal of Statistics Education, 15(2), 1.
Choo, L. (2000). Forecasting practices in the pharmaceutical industry in Singapore. Journal of Business Forecasting Methods and Systems, 19(2), 18-22.
Dillman, D. A. (2000). Mail and internet surveys: The tailored design method. New York: Wiley.
Doering, T., & Suresh, N. C. (2016). Forecasting and Performance: Conceptualizing Forecasting Management Competence as a Higher-Order Construct. Journal of Supply Chain Management.
Dolnicar, S., Laesser, C., & Matus, K. (2009). Online versus paper format effects in tourism surveys. Journal of Travel Research, 47(3), 295-316.
Duran, J. A., & Flores, B. E. (1998). Forecasting practices in Mexican companies. Interfaces, 28(6), 56-62. From doi:10.1287/inte.28.6.56.
Easton, G., Roberts, H. V., & Tiao, G. C. (1988). Conference Report: Making Statistics More Effective in Schools of Business. Journal of Business & Economic Statistics, 6(2), 247-260.
Greenberg, B. A., Goldstucker, J. L., & Bellenger, D. N. (1977). What techniques are used by marketing researchers in business? The Journal of Marketing, 62-68.
Greenlaw, C., & Brown-Welty, S. (2009). A comparison of web-based and paper-based survey methods testing assumptions of survey mode and response cost. Evaluation Review, 33(5), 464-480.
Hanke, J., & Weigand, P. (1994). What are business schools doing to educate forecasters? The Journal of Business Forecasting, 13(3), 10.
Jain, C. L. (2006). Benchmarking forecasting practices in America. The Journal of Business Forecasting, 25(4), 9.
Jain, C. L., & Malehorn, J. (2006). Benchmarking forecasting practices: a guide to improving forecasting performance. Institute of Business Forecasting.
Keating, B., & Wilson, J. H. (2019) Forecasting and Predictive Analytics with ForecastX, 7th Ed. McGraw-Hill.
Klassen, R. D., & Flores, B. E. (2001). Forecasting practices of Canadian firms: Survey results and comparisons. International Journal of Production Economics, 70(2), 163-174. From doi:10.1016/s0925-5273(00)00063-3.
Lam, S. S. K. (1996). A survey of the forecasting practices of Hong Kong companies. International Journal of Management, 13, 300-305.
Loomis, D. G., & Cox, J. E. (2003). Principles for teaching economic forecasting. International Review of Economic Education, 1 (2), 69-79.
Mady, M. T. (2000). Sales forecasting practices of Egyptian public enterprises: Survey evidence. International Journal of Forecasting, 16 (3), 359-368. From doi:10.1016/s0169-2070(00)00033-9.
Mentzer, J. T., & Cox, J. E. (1984). Familiarity, application, and performance of sales forecasting techniques. Journal of Forecasting, 3(1), 27-36. From doi:10.1002/for.3980030104.
Mentzer, J. T., & Kahn, K. B. (1995). Forecasting technique familiarity, satisfaction, usage, and application. Journal of Forecasting, 14(5), 465-476. From doi:10.1002/for.3980140506.
McAfee, A., Brynjolfsson, E., Davenport, T. H., Patil, D. J., & Barton, D. (2012). Big data. The management revolution. Harvard Business Review, 90(10), 61-67.
McCarthy, T. M., Davis, D. F., Golicic, S. L., & Mentzer, J. T. (2006). The evolution of sales forecasting management. Journal of Forecasting, 25, 303-324.
Moon, M. A. (2004). What is World-Class Forecasting? A Perspective on 20 Years of Research. In 2004 International Conference Proceedings at APICS-The Educational Society for Resource Management.
Morris, R. (2014, November). The crystal ball. Supply House Times, 86-90.
Myers, J. G., Greyser, S. A., & Massy, W. F. (1979). The Effectiveness of Marketing's" R&D" for Marketing Management: An Assessment. The Journal of Marketing, 17-29.
Peterson, R., & Jun, M. (1998). Forecasting practices in accounting, law, and medical professions. The Journal of Business Forecasting, 17(2), 20.
Porter, L. W., & McKibbin, L. E. (1988). Management Education and Development: Drift or Thrust into the 21st Century? McGraw-Hill Book Company, College Division, PO Box 400, Hightstown, NJ 08520.
Porter, S. R., & Whitcomb, M. E. (2007). Mixed-Mode Contacts in Web Surveys Paper is Not Necessarily Better. Public Opinion Quarterly, 71(4), 635-648.
Sanders, N. R., & Manrodt, K. B. (1994). Forecasting practices in US corporations: Survey results. Interfaces, 24 (2), 92-100. From doi:10.1287/inte.24.2.92.
Smith, J., & Clarke S. (2011). Who should own the business forecasting function? Foresight: The International Journal of Applied Forecasting, (20), 4-7.
Spralls, S. A., & Wilson, J. H. (2016). Forecasting in Marketing Education: Empirical Findings. International Journal of Business, Marketing, and Decision Sciences, 9(1), 97-115, 2016.
Venkatesh, V., & Davis, F. D. (2000). A theoretical extension of the technology acceptance model: Four longitudinal field studies. Management Science, 46(2), 186-204.
Walker, I., Tsarenko Y., Wagstaff, P., Powell, I., Steel, M., & Brace-Govan, J. (2009). The development of competent marketing professionals. Journal of Marketing Education, 31 (3), 253-263.
Wilson, J. H., & Daubek, H. J. (1989). Teaching forecasting: Marketing faculty opinions and methods. Journal of Marketing Education, 11, 65-71.
Wisner, J. D., & Stanley, L. L. (1994). Forecasting practices in purchasing. International Journal of Purchasing and Materials Management, 30(4), 21-29. From doi: 10.1111/j.1745-493x.1994.tb00263.x.
Yang, H., & Yoo, Y. (2004). It's all about attitude: Revisiting the technology acceptance model. Decision Support Systems, 38(1), 19-31.
About the Authors:
J. Holton Wilson received a BS in chemistry and a BA in economics from Otterbein College, and MBA from Bowling Green State University. His doctorate is from Kent State University with major areas in both economics and marketing. He is currently professor emeritus at Central Michigan University in the department of Marketing where he continues to teach Sales and Market Forecasting.
Samuel A. Spralls III is Professor of Marketing at Central Michigan University. His major area of teaching and research is Marketing Strategy. Samuel holds a Bachelor's of Science in Industrial Engineering from the University of Wisconsin-Madison, MBA from the University of Chicago, and Doctorate of Philosophy from Texas Tech University.
Q1 What is your current age?
Q2 Gender: (1) Female (2) Male (3) Prefer not to answer
Q3 How many years of forecasting experience do you have?
Q4 What is your job title?
Q5 In what state are you located (Please use two letter designation.)?
Q6 What was your primary undergraduate major?
Q7 What percent of your time is devoted to forecasting?
Q8 Approximate your firm's dollar volume of sales:
Sales in Under $1 $1 Million - $50 $51 Million - $100 Dollars (1) Million (1) Million (2) Million (3) $101 Million - $99 $1 Billion Prefer not Million (4) or More (5) answer (6)
Q9 Approximate your firm's number of employees:
Number of Under 100 (1) 101 - 1,000 (2) 1,001 - 5,000 (3) Employees (1) 5,001 - 10,000 (4) Over 10,000 (5) Prefer not answer (6)
Q10-13 Please indicate how much you agree or disagree with the following four statements (on a 5-point scale):
It is important that the people my firm hires understand forecasting methods.
It is important that the people my firm hires know how to develop an actual forecast.
It is important that the people my firm hires know when to use various forecasting methods.
It is important that the people my firm hires understand the trade-offs between using different methods (e.g., causal vs. extrapolation).
Q14 What functional areas in your business contribute information to forecasting?: Select all that apply (scaling from "never" to "all the time" on a 5-point scale).
Engineering (1) Executive leadership team (2) Finance (3) Forecasting (4) Logistics (5) Marketing (6) Planning (7) Project Management (8) Production (9) Purchasing (10) R&D (11) Sales (12) Other: (13)
Q15 What one functional area "owns" the forecasting function in your company?: Engineering (1) Executive leadership team (2) Engineering (1) Executive leadership team (2) Finance (3) Forecasting (4) Logistics (5) Marketing (6) Planning (7) Product Management (8) Production (9) Purchasing (10) R&D (11) Sales (12) Other: (13)
Q16 How frequently do you think forecasting is currently taught within the following university departments (scaling from "never" to "always" on a 5-point scale)?: Accounting (1) Economics (2) Engineering (3) Finance (4) Forecasting (5) Information Systems (6) Logistics (7) Management (8) Marketing (9) Statistics (10) Other: (13)
Q17 In what departments do you think forecasting should be taught within universities (scaling from "never" to "always" on a 5-point scale)?: Accounting (1) Economics (2) Engineering (3) Finance (4) Forecasting (5) Information Systems (6) Logistics (7) Management (8) Marketing (9) Statistics (10) Other: (13)
Q18 To what extent is a forecasting certification helpful to a job applicant?: Not at All Helpful (1) A Little Helpful (2) Neutral (3) Helpful (4) Very Helpful (5)
Q19 We realize that some of the techniques below may not be familiar to you. Nevertheless, how important is it for job applicants to know each of the following forecasting techniques (scaling from "very unimportant" to "very important" on a 5-point scale)?: ARIMA (Box Jenkins) (1) Causal Multiple Regression (2) Data Mining (3) Delphi method (4) Holt's exponential smoothing (5) How to combine forecasts (6) Jury of executive opinions (7) Linear regression trends (8) Moving averages (9) Naive method (10) Sales force composite (11) Simple exponential smoothing (12) Text Mining (13) Time series decomposition (14) Winters' exponential smoothing (15) Other: (16)
Q20 Considering only the last five years, has the importance of forecasting to your firm (Select only one.): Increased (1) Decreased (2) Stayed the same (3) Not sure/not applicable (4)
Q21 Given the menu of forecasting techniques below, how knowledgeable are you about each technique (scaling from "not at all knowledgeable (1)" to "very knowledgeable (5)" on a 5-point scale)?: ARIMA (Box Jenkins) (1) Causal Multiple Regression (2) Data Mining (3) Delphi method (4) Holt's exponential smoothing (5) How to combine forecasts (6) Jury of executive opinions (7) Linear regression trends (8) Moving averages (9) Naive method (10) Sales force composite (11) Simple exponential smoothing (12) Text Mining (13) Time series decomposition (14) Winters' exponential smoothing (15) Other: (16)
Q22 Please indicate how frequently you use each of the forecasting methods below (scaling from "never (1)" to "always" (5) on a 5-point scale): ARIMA (Box Jenkins) (1) Causal Multiple Regression (2) Data Mining (3) Delphi method (4) Holt's exponential smoothing (5) How to combine forecasts (6) Jury of executive opinions (7) Linear regression trends (8) Moving averages (9) Naive method (10) Sales force composite (11) Simple exponential smoothing (12) Text Mining (13) Time series decomposition (14) Winters' exponential smoothing (15) Other: (16)
Q23 How satisfied are you with the overall performance (considering trade-offs between accuracy, ease of use, cost, and time) of the forecasting techniques below (scaling from "very dissatisfied" to "very satisfied" on a 5-point scale?: ARIMA (Box Jenkins) (1) Causal Multiple Regression (2) Data Mining (3) Delphi method (4) Holt's exponential smoothing (5) How to combine forecasts (6) Jury of executive opinions (7) Linear regression trends (8) Moving averages (9) Naive method (10) Sales force composite (11) Simple exponential smoothing (12) Text Mining (13) Time series decomposition (14) Winters' exponential smoothing (15) Other: (16)
Q24-25 Please indicate how much you agree or disagree with the following two statements (scaling from "strongly disagree" to "strongly agree" on a 5-point scale): (1) The emergence of big data will impact the way I approach forecasting. (2) The "Internet of Things" will impact the way I approach forecasting.
Table 1 Importance of Knowledge for New Hires N Mean (*) It is important that the people you hire understand 192 3.99 forecasting methods It is important that the people you hire know how 192 3.90 to develop an actual forecast It is important that the people you hire know when 192 3.69 to use various forecasting methods It is important that the people you hire understand tradeoffs between using different methods (e.g. 191 3.65 causal vs. extrapolation) Std. Dev. t-ratio It is important that the people you hire understand 1.016 13.50 forecasting methods It is important that the people you hire know how 1.053 11.84 to develop an actual forecast It is important that the people you hire know when 1.129 8.47 to use various forecasting methods It is important that the people you hire understand tradeoffs between using different methods (e.g. 1.146 7.84 causal vs. extrapolation) (*) Scale 1 to 5 with higher values representing greater importance. All means are significantly greater than the scale midpoint (3) at 0.05 level; 2 tailed. Table 2 Functional Areas That Contribute to Forecasting and Those Which "Own" the Forecast Contributes Contributes Percent: Owns Department Mean (*) Std. Dev. Forecast Forecasting 4.69 0.744 28.6 Planning 3.83 1.173 22.0 Sales 3.77 1.231 10.7 Marketing 3.58 1.158 8.9 Executive Leadership Team 3.31 1.143 4.8 Finance 3.31 1.298 11.9 Product Management 2.93 1.342 1.8 Production 2.73 1.328 1.2 Logistics 2.72 1.391 4.8 Purchasing 2.53 1.323 3.0 R&D 2.22 1.169 1.8 Engineering 1.96 1.177 0.6 (*) Scale 1 to 5 with higher values representing a greater contribution. Table 3 Think Forecasting is Taught in Departments Mean (*) Std. Dev. Think forecasting is taught in forecasting departments 4.62 .746 Think forecasting is taught in statistics departments 3.85 .966 Think forecasting is taught in finance departments 3.19 .989 Think forecasting is taught in economics departments 3.17 1.030 Think forecasting is taught in marketing departments 3.04 .994 Think forecasting is taught in logistics departments 3.02 1.090 Think forecasting is taught in management departments 3.00 .966 Think forecasting is taught in accounting departments 2.42 .943 Think forecasting is taught in information systems departments 2.32 .897 Think forecasting is taught in engineering departments 2.18 .952 (*) Scale 1 to 5 with higher values representing a greater likelihood. Table 4 Think Forecasting Should be Taught in Department Mean (*) Forecasting should be taught in forecasting departments 4.89 Forecasting should be taught in statistics departments 4.42 Forecasting should be taught in finance departments 4.23 Forecasting should be taught in management departments 4.21 Forecasting should be taught in marketing departments 4.13 Forecasting should be taught in economics departments 4.06 Forecasting should be taught in logistics departments 3.84 Forecasting should be taught in accounting departments 3.34 Forecasting should be taught in information systems departments 3.17 Forecasting should be taught in engineering departments 3.13 Std. Dev. Forecasting should be taught in forecasting departments .380 Forecasting should be taught in statistics departments .752 Forecasting should be taught in finance departments .836 Forecasting should be taught in management departments .824 Forecasting should be taught in marketing departments .861 Forecasting should be taught in economics departments .946 Forecasting should be taught in logistics departments 1.053 Forecasting should be taught in accounting departments 1.215 Forecasting should be taught in information systems departments 1.028 Forecasting should be taught in engineering departments 1.171 (*) Scale 1 to 5 with higher values representing a greater likelihood. Table 5 Will Big Data and the Internet of Things Influence Forecasters' Approach to Forecasting Mean (*) Std. t-ratio Dev. Big data will influence how I approach forecasting 3.95 0.924 14.25 The "Internet of Things" will influence how I approach forecasting 3.49 1.007 6.74 (*) Scale 1 to 5 with higher values representing greater influence. Table 6 Importance for Job Applicants to Know Various Forecast Methods Mean (*) Std. Deviation t-ratio Moving Average (**) 4.20 .866 18.39 Linear Regression Trends (**) 4.07 .971 14.64 How to Combine Forecasts (**) 4.06 .973 14.49 Simple Exponential Smoothing (**) 3.88 .954 12.20 Data Mining (**) 3.87 .938 12.29 Time Series Decomposition (**) 3.66 1.098 7.89 Causal Multiple Regression (**) 3.64 1.027 8.22 Holt's Exponential Smoothing (**) 3.49 1.060 6.03 Sales Force Composite (**) 3.44 .972 5.84 Winters' Exponential Smoothing (**) 3.43 1.111 5.09 Jury of Executive Opinion (**) 3.34 .991 4.40 Naive Method (**) 3.33 1.062 4.03 ARIMA (**) 3.16 1.036 2.07 Delphi 3.07 .899 0.95 Text Mining 2.91 .987 -1.18 (*) Scale 1 to 5 with higher values representing greater importance. ( (**) = mean is significantly greater than the scale midpoint (3) at 0.05 level; 2 tailed). Table 7 Extent to Which a Forecasting Certification is Helpful to a Job Applicant Scale Item (*) Frequency Percent Not at all helpful 7 3.7 A little helpful 29 15.4 Neutral 57 30.3 Helpful 74 39.4 Very helpful 21 11.2 (*) Scale 1 to 5 with higher values representing greater helpfulness. Table 8 Usefulness of Various Forecast Methods Mean (*) Std. Deviation t-ratio Moving Average (**) 3.76 1.213 8.29 Linear Regression Trends (**) 3.37 1.346 3.57 Combining Forecasts (**) 3.37 1.362 3.55 Simple Exponential Smoothing 3.13 1.430 1.22 Data Mining 2.86 1.490 -1.21 Causal Multiple Regression 2.65 1.354 -3.37 Time Series Decomposition 2.60 1.438 -3.70 Holt's Exponential Smoothing 2.45 1.392 -5.23 Jury of Executive Opinion 2.41 1.319 -5.84 Winters' Exponential Smoothing 2.33 1.446 -6.11 Sales Force Composite 2.30 1.314 -6.98 Naive 2.26 1.379 -6.99 ARIMA 1.99 1.170 -11.33 Delphi 1.83 1.098 -13.96 Text Mining 1.64 1.000 -17.68 (*) Scale 1 to 5 with higher values representing greater usefulness. ( (**) = mean is significantly greater than the scale midpoint (3) at 0.05 level; 2 tailed). Table 9 Satisfaction with Various Forecast Methods Mean (*) Std. Deviation t-ratio Linear Regression Trends (**) 3.91 .821 13.40 Moving Average (**) 3.82 .953 10.66 Combining Methods (**) 3.78 .832 11.08 Simple Exponential Smoothing (**) 3.72 .790 10.77 Time Series Decomposition (**) 3.68 .846 8.70 Causal Multiple Regression (**) 3.66 .879 8.44 Data Mining (**) 3.60 .783 8.57 Holt's Exponential Smoothing (**) 3.59 .778 8.02 Winters' Exponential Smoothing (**) 3.58 .831 7.26 ARIMA (**) 3.44 .752 6.01 Jury of Executive Opinion (**) 3.36 .894 4.47 Naive (**) 3.31 .871 3.89 Sales Force Composite (**) 3.22 .783 3.04 Delphi (**) 3.22 .642 3.41 Text Mining 3.07 .729 0.98 Scale 1 to 5 with higher values representing greater satisfaction. ( (**) = mean is significantly greater than the scale midpoint (3) at 0.05 level; 2 tailed). Table 10 Spearman Rank Order Correlations Useful Satisfaction Spearman's Rho Usefulness Correlation 1.000 Coefficient Sig. (2-tailed) Satisfaction Correlation .950 (**) 1.000 Coefficient Sig. (2-tailed) .000 New Correlation .982 (**) .939 (**) Employees Coefficient Sig. (2-tailed) .000 .000 New Employees Spearman's Rho 1.000 (**) Correlation is significant at the 0.05 level (2-tailed).
|Printer friendly Cite/link Email Feedback|
|Author:||Wilson, J. Holton; Spralls, Samuel A., Iii|
|Publication:||International Journal of Business, Marketing, and Decision Sciences (IJBMDS)|
|Date:||Dec 22, 2018|
|Previous Article:||DOES THE MARKETING MIX AFFECT STREET FOOD CONSUMPTION IN THAILAND?|
|Next Article:||SMALL BUSINESS: TRADE SHOW FACTORS OF SUCCESS.|