Essay on Multidimensional Scaling

It is important to use case studies or situations to have a clear understanding of the business world and the problems that it entails. One of these case situations is where a harried director walks into the office with many printouts. He says, You are the market research agent, please give guidelines on the number of these printouts that we are going to sell at the end of the year. In another case, the same harried business director comes into the office and displays some advertising campaigns that have been submitted as proposals. He asks, Which of the three should I use? They are all pretty to me. In the last case situation, he asks Why are our main competitors gaining lots of shares? Are they in possession of better widgets?

Is your time best spent reading someone else’s essay? Get a 100% original essay FROM A CERTIFIED WRITER!

Order now

All the situations given occur every day, especially in the corporate world. All the situations can be dealt with using multivariate analysis techniques. It is upon a market researcher to choose the best technique to apply in solving the business problem. There are three commonly used multivariate techniques, these are, multidimensional scaling, factor analysis and cluster analysis (Jackson, 2015). For instance, over the past few years, there has been a rise in the use of desktop computing. This has resulted in companies coming up with various software used in analysis such as SPSS, SAS, and STATA. From the many softwares, challenges have arisen, such as having difficulties in choosing the best software. It needs one to have a full information of each of the softwares strengths and weaknesses. In such cases, multivariate analysis comes in place with its various techniques. In this paper, we will focus on multidimensional scaling (MDS), a type of multivariate analysis used in solving business problems.

Multidimensional scaling is a technique mostly used when one wants to transform consumer judgments which are similar to factors that are different and far from each other. Some of the multidimensional scaling methods are classical MDS, Metric MDS and Non-Metric MDS (Jackson, 2015). It is used to get the extent of dissimilarity, similarity, and distance. It is a technique that employs perceptual mapping to give a representation of the values that are provided. This technique is useful whenever a comparison of products is made, and the common characteristics of the products at hand are unknown. In this case, the dimensions used in making comparisons include price, brand and market perception of the products. To have accurate results using multidimensional scaling, more than two products are used. The purpose of this is to ensure that the dimensions being observed are different or vary (Jackson, 2015). The dimensions used in comparison can be identified subjectively by the respondents or objectively by the researcher.

When compared to the other two multivariate techniques, multidimensional scaling appears to have the most appropriate procedures that lead to providing solutions to business problems (Zhao, 2014).

. For instance, factor analysis is used when there are many variables that need to be compared. It needs over 50 observations to avoid erroneous results. This is different from multidimensional scaling which involves small samples of data. The other difference is that factor analysis an independence technique which does not have dependent variables. This is because most of the data analyzed are standard and continuous (Richarme, 2016). In multidimensional scaling, variables used have dimensions that depend on each other hence bearing both independent and dependent variables.

Multidimensional scaling still has some differences with cluster analysis. In cluster analysis, the significant data used in factor analysis are reduced into smaller data. This is done by dividing the data into subgroups of the variables be analyzed (Mineo, 2000). For the groups to be accurate, the objects should be similar in one way or the other. This technique is faced with the problem of developing irrelevant variables. This is because one has to identify their similarities and differences before making any analysis. A great difference is exhibited from multidimensional scaling which uses perceptual maps instead of groups. The other differences are that multidimensional scaling uses unrecognized dimensions, whereas cluster analysis uses subgroups, which are a representation of the population (Richarme, 2016). In cluster analysis, the clusters should not be similar, but they should have some reachable aspects. On the other hand, the multidimensional analysis uses Kruskals Stress which is a measure of badness of fit. In Kruskals Stress, 0% is an indication of perfect fit while 20% and over is an indication of poor fit (Richarme, 2016).

Multidimensional scaling has been used by various companies and organizations to solve business related problems. By business problems it means things that have to do with decision-making and ways of having a better competitive advantage. It is all about becoming better and getting more shares when compared to other companies. The Coca-Cola Company is one such company that uses multidimensional scaling to solve business problems that it faces. When the company wants to have information on the perception that customers have on their products, it uses multidimensional scaling (Young, 2000). It also uses it to identify the perception of other companies on their products.

This technique has made it get rewards that came along with the U.S. carbonated soft drink market. The last time it did this it obtained an estimate of around $70 billion. It did the analysis in the year 2009 (Young, 2000). Since then, it has managed to remain at the top when compared to other soft drink companies. Currently, Coca-Cola Company has no problems that have to do with product similarity and differences. This is because it has information on what needs to be done to be unique (product wise). It is this unique aspect that has made it enjoy many customers compared to other soft drink companies that produce similar products (Young, 2000).

In my organization, application of multidimensional scaling would be beneficial in various ways. Since factor analysis cannot easily provide similarities and differences between the factors, multidimensional scaling would be the best way to determine characteristics of our target market. In addition to this, I would solve problems oriented towards selling products similar to other companies. In this case, it would provide the definite similarities using underlying dimensions. With this information, it would be easier to come up with a new product that has qualities and dimensions that are not similar to another companys.

Multidimensional scaling needs a procedure that would lead to identifying similarities and differences of products that exist in the. The products would come from both other companys and us. First, points would be assigned to each of the products (arbitrary coordinates) in a space referred to as p-dimensional. Then distance would be calculated using Euclidean pairs of points to form a Dhat matrix. This matrix would then be analyzed with the D matrix using a stress function (n). The value obtained would be either smaller or larger. If at all the value is smaller than the objects or products under comparison then the products would be deemed to be corresponding or similar to each other.

Later, adjustments can be made on the products or simply new ones would be developed. In the end, business problems that have to do with the market and product similarity would be solved from the results provided by Kruskals Stress and the badness of fit. The other way of using multidimensional scaling in my organization would be to find out the number and nature of dimensions based on consumers perception and use of different brands provided in the marketplace. It would also be used to identify the position of current brands based on the dimensions of tastes and preferences.

References

Jackson, Jo. (2015). Multivariate Techniques: Advantages and Disadvantages. Classroom Synonym. Web 22 Mar. 2017

Mineo, A.M. (2000). Applied Stochastic Models in Business and Industry: Multidimensional Scaling and Stock location assignment in a warehouse and application. International Symposium on Applied Stochastic Models and Data Analysis VIII. Vol. 15 Issue 4 pp. 387-392

Richarme, Michael. (2016) Eleven Multivariate Analysis Techniques: Key Tools in your Marketing Research Survival Kit. Decision Analyst. Web 22 Mar. 2017

Young, Forrest. (2000). Multidimensional Scaling. Encyclopedia of Statistical Sciences: John Wiley & Sons. Vol. 5. Print

Zhao, Yanchang. (2014). Multidimensional Scaling (MDS) with R. RData Mining. Web 22 Mar. 2017