Multiple regression analysis is an extensive form of statistical analysis used by organizations to predict the value of an unknown variable such as sales using known values of two or more variables such as employee remuneration, advertising costs and number of staff (Koen & Holloway 2014:67). Often, marketing personnel are interested in predicting thesales, potential sales or attitudes of market agents towards a given product. To be able to do this, the following general conceptual framework is first used to define variables:
The model provides an effective framework within which organization can identify and define variables that are vital for any given sales forecast. The centered factors represent the dependent variables, while the peripheral factors are independent. Any of the independent variables can be used to predict the behavior of a specific dependent variable. Where the organization intends to predict purchases by a special group of consumers, purchases is in the centered circle and is therefore a dependent variable. Peripheral factors in the four cornered circles are deemed as the independent variables. Managers are obligated to determine which of the independent variables are most suitable for the forecast. If for instance Microsoft intends to introduce a new line of computers into the market, it might require information on consumers’ attitudes, experiences with substitute products, consumer’s lifestyle and other influences. The conceptual model is however only used for the identification and definition of relevant variables. Once variables are defined, data for each of the variable is obtained from previous records and fitted into different case scenarios to help predict sales for different periods.
Correlation is a statistical measure that is used to define the relationship between two sets of data. Often, its value ranges from +1.0 to -1.0. While a positive figure indicates a stronger relationship between variables, negative digits indicate inverse relationship (Sumn.org n.d:71). A zero coefficient indicates that there is no relationship between the given data sets. As opposed to correlation, causation relies on the principle of cause and effect where one event leads to the occurrence of another i.e. A causes B. It equally does not give the intensity with which the different sets of data are related. Coefficient correlation, r, should not be used to define a cause and effect relationship owing to the problems posed by third variables.
Often, when it is stated that decrease in price would lead to market surplus, it is easier to assume that the correlation between the two variables is equally a cause and effect relationship. This is however not true since there are a number of factors that could equally lead to the stated market surplus. For instance, an increase in total production or a fall in wages could also affect the market in a similar way. Where there are several factors producing a given outcome i.e. multiple causation, attaining perfect causation may not be possible. This is because it would require one to hold constant all irrelevant variables which is unrealistic in real business problems. According to Graham (2003), Correlation is able to draw relationships from different spectrums. These assertions are made based to help compare correlation and Causation when it comes to dealing with variables. Owing to these imperfections in causation, Graham (2003) noted that correlation is deemed as the most appropriate method of determining the relationship between business variables.
As stated earlier, multiple regression analysis is an effective method used by businesses to forecast dependent variables. To be able to determine the organization’s future demand, it is essential for the firm to first select a range of demographic and economic variables that are deemed vital in the industry in which the firm operates, e.g. prices of other related goods, consumer taste and preferences as well as lifestyle of the target market (Koen & Holloway 2014:53). This can be done using a basic regression model for the industry. For instance, if Toyota intends to predict next year’s demand of its premio model, variables such as prices of other similar models by competitors, changing needs of consumers as well as the status of the economy are vital independent variables. The firm would then collect historical data relating to the variables identified. With all the necessary sets of data, it is then prompted to model appropriate scenarios which are then used to quantify the future value of demand.
The most relevant topic in my case relates to the use of quantitative and qualitative data to make business decisions. Often, organizations are faced with a number of critical decision making scenarios. Quantitative as well as qualitative data are often very intensive and are used by managers to make decisions regarding distribution of resources among several departments (Cooper & Schindler 2014:26). The manager would first have to determine which of the departments is most profitable to the firm. He is prompted to use the firm’s accounting data to construct a mathematical formula that can be used to effectively apportion the resources. This concept is equally useful for businesses seeking to expand into new markets.
Cooper, D. R., & Schindler, P. S. (2014). Business Research Methods (12th Ed.). New York, NY:McGraw-Hill Irwin. http://www.acemyhw.com
Graham, A. (2003). Statistics. Blacklick, OH: McGraw-Hill.
Koen,R. & Holloway, J. (2014). Application of multiple Regression Analysis to Forecasting South Africa’s Electricity Demand. Journal of Energy in Southern Africa. Vol. 25. Issue. 5. Retrieved,11 June 2015, from http://web.uct.ac.za/depts/erc/jesa/volume25/25-4jesa-Koen-Holloway.pdf
Sumn.org. (n.d). Correlation vs. Causation. Web. Retrieved, 11 June 2015, from http://docs.sumn.org/SUMN_Toolkit/Correlation_vs_Causation_9-09.pdf