Firstly, it is necessary to state that managers tend to utilize statistical methods in conducting their marketing and management policy along with finding the relevant answers to their corresponding questions. Particularly for this purpose, the descriptive statistics and its approaches have been designed and implemented in various management procedures in order to achieve the relevant indicators of data characteristics. In this regard, the biggest and most important benefit of descriptive statistics is its great economy of time and efforts in case, where the data sets are comprised of enormous massifs of information. However, in situations when the managers need to describe the corresponding characteristics of a particular sample due to their importance in discussing the overall research population, they tend to use the approaches and methods of inferential statistics instead. In particular, when the managers want to draw more general conclusions about the entire population from a corresponding data taken from only one sample, they use inferential statistics approaches (Tanner and Youssef-Morgan, 2013).
The variety of descriptive statistics methods, indexes and approaches is simply enormous; however, the most important and therefore the most popular appear to be the ones describing the “typicality” or the most prevalent trend for a particular data set along with the indexes illustrating the variability of this data. The indexes of descriptive statistics include the famous mean and standard deviation, resulting in different outcomes, still they effectively complement each other.
Another valuable piece of information concerning statistics, is the concept of probability and its importance for drawing corresponding hypotheses along with their further development. Firstly, it is important to understand probabilities in order to understand the normal distribution in statistical terms. The last also share much in common; in particular, they appear to have just one mode and be symmetrical. Correspondingly, due to these common elements, there is a strong association between the probability and characteristics of normal distributions. In other words, understanding all of these, it is possible to draw corresponding conclusions about the likelihood of specific measures to result in the normal distribution.
Another procedure that needs to be discussed is the formal statistical testing. In particular, the introduction to this statistical mechanism is represented by the z-test, being a relatively easy-to-conduct test that is able to include a number of the same issues coming up in more complicated statistical tests, including statistical significance. The z-test is based upon the sample means distribution mentioned previously; in other words, upon the population based on the means of samples of individuals. The reason for using such approach is based on the provisions of central limit theorem guaranteeing the normal distribution of such population (Tanner and Youssef-Morgan, 2013).
Still, t-tests are having their own advantages over the z-tests. For example, in the one-sample test, samples can effectively be compared to populations without any necessity to use any additional indicators and/or indexes, which may also appear unavailable. Further, the independent samples of t-test do not generally require any additional data in order to evaluate two different samples and analyze their considerable peculiarities and differences. As a result, t-test appears to be the most popular in this regard, proving its statistical importance (Tanner and Youssef-Morgan, 2013).
Consequently, the hypothesis testing has two different predictions based upon the mechanisms and corresponding results of t-tests. In case of one-sample test, the situation is simple – sample either represents the population to which it is compared or nor. For the independent t, the trend is slightly different. In this regard, both samples appear either belonging to their respective populations or do not belong to them at all. Despite the fact that the hypotheses significantly simplify the mechanism of statistical results’ reporting, the utilization of one- against two-tailed tests still complements this process. One-tailed tests result in an alternate hypothesis also being directional. It provides a corresponding prediction of how the mean of the population being illustrated by the first group will be different from the same representation of population offered by the second group (Tanner and Youssef-Morgan, 2013).
Still, in any study evaluating a corresponding group of subjects, the individuals within the same sample group can still react differently to the same stimulus. Correspondingly, such difference result in the respective error variance, being compounded in independent group tests estimating specifically different individuals for each corresponding group. The same trend remains actual regardless of the researchers’ accuracy and carefulness in evaluating the corresponding data. Thus, there is a need to choose a specific statistical test in order to decrease this error variance to possible minimum. For example, the approaches of t-tests and F-tests have their corresponding mechanisms of eliminating the source of this error variance by simply matching subjects according to the most significant characteristics or using the data of the same subjects repeatedly (Tanner and Youssef-Morgan, 2013).
However, in order to evaluate and assess all the results achieved according to the tests above, it is necessary to define the purpose and expected outcomes of the statistical study with further comparison of the expected with resulted outcomes. Moreover, the choice of a particular test depends highly upon personal decisions of the managers, their needs and goals and certainly level of expertise. Still, this paper has briefly discussed the most important and most beneficial tests that almost each professional manager can conduct and provide further analysis.
References
Tanner, D. and Youssef-Morgan, C. (2013). Statistics for Managers. San Diego, CA: Bridgepoint Education.
