Regression Analysis Models for Marketing Decision Making - Research Paper Example

Regression Analysis Models For Marketing Decision Making Lecturer Table of Contents Introduction 3 Regression Definition 3 Problem Statement 4 Objectives and Aims 4 Challenges 5 Data 5 Results 9 Discussion of results 15 Conclusion 15 Bibliography 16 Appendix 17 Regression Analysis Models For Marketing Decision Making Introduction Decision-making is the core of business, and correct decisions result into businesses thriving. Marketing decisions in a company are made based on information that is as a result of data interpretation or analysis. Analysis of quantitative data which is mainly measurable is done through several statistical models. The research seeks to identify various regression analysis models being applied by businesses in trying to get information from quantitative data and the information being used in making a business decision especially in the marketing department. Regression Definition Regression analysis is a statistical technique that determines linear relationships between two or more variables. Businesses mainly use regression as a causal inference and for predictions. The major regression models available are linear regression model, non-linear regression model, logistic regression and multinomial logistic regression. Simple regression models use only two variables to achieve a particular statistical result. Multiple linear regression is a regression that applies more than two variables. Logistic regression procedures in quantitative statistics will produce all predictions, residuals and influence statistics. Logistic regression also produces goodness-of-fit tests using sales and marketing data in the case where it has to make predictions for the marketing department. The goodness-of-fit tests are created at the individual case level, and this is regardless of methods of data insertion and whether or not the number of covariate patterns is lesser than the total number of instances in question. On the other hand, multinomial logistic regression procedure aggregates all cases internally to form subpopulations with identical covariate patterns for the predictors, residuals, and goodness-of-tests. Non-linear regression is a quantitative statistical method of finding a nonlinear model of the relationship between the dependent variable and a set of several independent variables. Current non-linear models can be used to estimate models with arbitrary relationships between dependent and independent variables. Iterative estimation is mostly used to achieve non-linear regression. Problem Statement Predicting future marketing trends is business is an essential requirement for the management if they have to beat the competition. This is because a lot of the data is available for use by business nowadays due to advancements in technology like the web that collects a lot of statistical data for analysis. The primary problem facing businesses is identifying the optimal data analysis model to use in the analysis the quantitative data and getting valid information for predicting the future marketing trends. Objectives and Aims The research seeks to define and identify the optimal regression analysis models available today. The paper will describe several procedures available in using the regression models to perform data analysis. The paper will then simulate some marketing data to identify the best regression that a business can employ within various marketing situations. The paper will consider and determine all the assumptions that businesses are allowed to apply in using the regression models. The paper will suggest the optimal or the best regression model that a mid-sized business entity can apply depending on the several amount of data and variable they can acquire. Finally, the paper will suggest and analyze how several software application like SPSS can be used in using the optimal regression model identified in the research. Challenges The research will mostly use on-line articles in studying the various regression models based on previously published research without the use of actual sample data in testing the regression models. Acquiring statistical software application will also become a challenge due to most of them being commercially available. Data It is axiomatic to squabble that marketing plays a fundamental role in the growth of business. Any enterprise that intends to present good results must be able to market its products. The public can only understand what is being sold when an extensive marketing that results to building of brand image is done. However marketing is of no sense if the marketer does not understand its market. The choice of marketing platform to be used in marketing depends on the targeted audience. It is important therefore that the marketer understands its audience. Consumers will be willing to consume a product depending on their marginal propensity to consume which is also influenced by the consumer’s disposable income. Therefore before a business launches its product, it is prudent at the disposable income of the targeted market. In the United States of America for instance, an understanding of the same, will only be possible when a researcher understands what the professionals are paid. The analysis of their incomes will inform the marketing decisions on whether to launch a specific product and prices to charge on the goods sold to customers. (Mcdonald, 1992) The data is mainly statistical analysis of salary earned in US top 30 professional. The list profession includes Physician, Psychiatrist, Pharmacist, Judge, Dentist, Astronomer, and Petroleum engineer, Aerospace Engineer, Mathematician, Historian, Geologist, Statistician, Civil Engineer, Physical Therapist, Economist, Architect, Sociologist, Computer Systems Analyst, Actuary, Chemist, Veterinarian, Technical Writer, Librarian, Dental Hygienist, Purchasing Agent, Licensed Nurse, Optometrist, Paralegal Assistant, Computer Programmer and Fashion Designer. Mainly this is secondary data analysis project. All the data used in this report collected from United States Department of Labor-Bureau of Labor Statistics. S.I Profession Salary,2001 Salary,2011 Year of experience Education level 1 Physician $140,376.00 $192,065 20 6 2 Psychiatrist $113,570.00 $160,242 18 6 3 Dentist $110,790.00 $142,144 15 5 4 Judge $79,540.00 $113,439 13 5 5 Petroleum Engineer $81,800.00 $109,147 11 4 6 Pharmacist $72,830.00 $109,070 11 4 7 Astronomer $76,390.00 $105,233 11 4 8 Optometrist $87,980.00 $96,163 10 3 9 Aerospace Engineer $71,380.00 $95,130 9 3 10 Mathematician $73,230.00 $94,178 9 3 11 Economist $72,350.00 $87,240 9 3 12 Actuary $74,720.00 $87,204 9 3 13 Geologist $64,120.00 $81,274 8 2 14 Veterinarian $69,150.00 $81,198 7 2 15 Computer Systems Analyst $63,710.00 $77,153 7 2 16 Civil Engineer $61,000.00 $76,139 6 2 17 Physical Therapist $59,130.00 $74,104 7 1 18 Statistician $57,080.00 $73,208 6 1 19 Architect $59,590.00 $73,193 6 1 20 Computer Programmer $64,536.00 $71,176 5 1 21 Sociologist $56,560.00 $70,122 5 1 22 Chemist $55,880.00 $68,195 5 0 23 Dental Hygienist $56,770.00 $67,107 5 0 24 Historian $44,850.00 $63,208 5 0 25 Technical Writer $51,650.00 $63,170 4 0 26 Fashion Designer $56,340.00 $61,291 2 0 27 Purchasing Agent $45,130.00 $55,165 3 0 28 Librarian $43,750.00 $54,197 3 0 29 Paralegal Assistant $39,220.00 $47,153 2 0 30 Licensed Nurse $31,490.00 $40,090 2 0 After that data entered in SPSS & Excel program for further analysis. All the graph of the report will be generated by Microsoft Excel and all the statistical work will be completed by SPSS program. (United States Department of Labor-Bureau of Labor Statistics, 2015) An understanding on how the salaries/annual income of the professionals gives an idea of the possible purchasing power of the populace and the amount they might be willing to consume on various products. Therefore when doing marketing, an idea will be skewed towards specific demands of the customers. Understanding the income will also help understand the development in the economy and the possible problems faced by the people. Annual Income of Petroleum Engineer was $81,000 in 2001.Later on in 2011 annual income of Petroleum Engineer increased by 33%.In 2001 number cars & transport vehicle was much lower than current situation. Number of cars & transport vehicle increased heavenly. This creates demand for petroleum product. These also open a good number opportunity for petroleum engineer. It means that the purchasing of the people has gone up making to buy cars and afford transport vehicles. Hypothesis H0: There is a relationship between professional’s income and years of experience H1: There is no relationship between professional’s income and years of experience H0: There is a relationship between professional’s income and education level H1: There is no relationship between professional’s income and education level The study will use one sample T-Test which is relevant because it will indicate to us whether or not the existing difference between the average salaries reflects the real difference on the entire population from where the group is sampled. Correlation analysis will also be used. It is relevant as it will show as how strong on weak the salary is statistically correlated. It will mainly consider Pearson correlation, sum of square and cross products and also the covariance. (McClave & Benson, 1988) The analysis of the study will use Statistical Package for the Social Sciences as a data analysis and data management product. The tool is produced by the IBM SPSS Inc. Some of the features that are associated with the SPSS include ANOVA, descriptive statistics, factor analysis, frequencies, cluster analysis, charts, and cluster analysis among others. The tool will be used given that it is easy to understand and provides various options in understanding data within the social sciences. (Norusis, 2011) Results Descriptive statistics Table 1 below presents a summary statistics for the variables. From the table, it can be seen that the average employee salary in 2011 was $8,279.93 with an average experience of 7.77 years in the job. Majority of the employees were at graduate level.   Salary,2011 Year of experience Education level count 30 30 30 mean 86,279.93 7.77 3.07 sample standard deviation 32,918.37 4.49 1.51 sample variance 1,083,618,779.10 20.19 2.27 minimum 40090 2 1 maximum 192065 20 5 range 151975 18 4 confidence interval 95.% lower 73,988.01 6.09 2.50 confidence interval 95.% upper 98,571.85 9.44 3.63 half-width 12,291.92 1.68 0.56 skewness 1.58 1.05 -0.12 kurtosis 3.04 1.04 -1.40 coefficient of variation (CV) 38.15% 57.85% 49.14% The boxplots below shows that there exist outliers in the two variables in the data set. The outliers can be observed in the salaries and in the years of experience. Correlation The study sought to find out whether there exists any correlation between the dependent variable and the explanatory variables. The table below presents the correlation matrix. The correlation coefficient between salary and years of experience is 0.982; this shows that there exists a very strong linear relationship between salary and years of experience. Similarly, with a correlation coefficient of 0.814 between salary and education level, it is evident that there is a strong positive relationship between an individuals’ salary and the education level. Lastly, it can also be observed that there exists a strong positive relationship between the two explanatory variables (education level and work experience). Salary,2011 Year of experience Education level Salary,2011 1.000     Year of experience .982 1.000   Education level .814 .873 1.000 30 sample size ± .361 critical value .05 (two-tail) ± .463 critical value .01 (two-tail) Scatterplots The two scatterplots presented below further confirm the presence of positive linear relation between the dependent variable (salary 2011) and the explanatory variables. Regression results In this section, the study sought to explain how the explanatory variables (years of experience and education level) predict the dependent variable (salary 2011). The following equation is estimated for predicting the dependent variable. Where,  is the coefficient for the constant term,  is the coefficient for the years of experience and lastly  is the coefficient for the education level. To check for the goodness of fit for the model, the study looked at the p-value for the F-statistics. The p-value is Read More