Monday, December 16, 2019

Statistics for Business (BUS 308: Statistics for Managers, July 22,2019)


Statistics for Business
            Every day organizations collect data from multiple sources such as sales reports, social media clicks, invoices, and productivity reports. Companies, managers, and employees have access to volumes of data, but many do not know how to apply that information in a useful way (Forbes, 2016).  Data analysis skills are essential for numerous job positions and departments from management to marketing. Data analysis can help managers make flow charts for productivity. Marketers can use data to create effective campaigns in the most suitable markets. Entrepreneurs may use analytics when choosing which market to invest in. To make wise decisions, these individuals must know how to test data sets for differences, correlations, strengths, and anomalies. However, one must learn terms to describe the data, how to make inferences from the data, formulate applicable hypotheses, and perform the correct tests. With this fundamental knowledge combined, professionals can make statistically meaningful inferences beneficial to their businesses.
Descriptive Statistics
            Descriptive statistics describe the way data is dispersed on a scatter plot. The mean is the average midpoint of all the data. The median is middle ordinal score. Mode is any number that occurs most frequently in the data (Doctoral…, 2013). Excel can calculate a five-number summary that contains the max, min, and median. Range, variance, and standard deviation are measures of dispersion which explain how the data is spread. The range consists of the highest value and lowest value (Clark, Reich, & Rubenstein, 2000). Variance is the average of squared distances from the mean. Variance is calculated subtracting the mean from a data point, squaring the result, then dividing by the number of the total sample. The F-test in Excel calculates data variances. The standard deviation is the variance squared. In normal distribution, sixty-eight percent of data points fall within one standard deviation one both sides of the mean. Ninety-five percent of the data lies within two standard deviations and 99.7 percent falls within three standard deviations (Doctoral…, 2013). Descriptive statistics tells a lot about the structure of data, and at this point one can formulate questions, but it is not enough information to provide answers.
Inferential Statistics
One uses inferential statistics to create hypotheses and test data for significant relationships. With several excel formulas, one can test sample data against the population data to see if the data sample is standard or rare (Research, 2016). Excel features several tests used for inferential statistics like the t-Test, Analysis of Variance (ANOVA), Chi Square test, regression analysis, and F-tests, but before performing any tests, one must formulate a hypothesis.
The hypothesis contains the null hypothesis, alternate hypothesis, p-value, test name, test statistic, direction, conclusion, and interpretation (Choudhry, 2019). The null hypothesis states the equality value, which essentially means the sample and any relationships are likely to occur within the population. An alternate hypothesis states the sample data is significantly different from the population. If the alternate hypothesis proves true, then the tested data is different from the population and the difference is unlikely due to mere chance.
Test Selection
In order to obtain statistically strong evidence, one must select the appropriate test for the sample data. The type of test is dependent on the number of samples, the type of data collected, and what the tester wants to test. Nominal or ordinal data are tested differently from interval and ratio data that contains numbers. Data may include two samples such as male and female data, or it may have only one sample and multiple variables.  A regression analysis test works with a scatter plot, whereas a Chi-Square test is suitable for data tables (Dr…, 2012). The 5 –number summary gives the range, median, and quartile information. The descriptive statistics data is beneficial when analyzing one-sample data. An employee may use these values when comparing number of items produced on a production line compared to the number of employees working. The Two Sample t-Test tests the sample(s) for mean differences. A manager may use this test to assure that the raises she approves are equal among genders and ages.  The F-test compares the variances of two groups for significant differences. A marketer can use this test to determine if one region’s consumer habits are significantly different than another area. The Pearson Correlation test, regression analysis, and multiple regression test data for correlations. If a business wants to know why their sales of fake skeletons increases in September and October, but is substantially less all other months, then the business will collect different samples in hopes to find a correlation.
Analyzing the Data
With the results of the test or formula, one decides if the null is rejected or accepted.  Generally the p-value is set at 0.05 if a ninety-five percent confidence interval is acceptable for the results. The null hypothesis states that the tested data is equal, while the alternate hypothesis claims that the sample data is significantly different from the population. If the p-value is smaller than 0.05, then the data is significantly larger or smaller than the population. Sometimes the test explains that a relationship exists and other times it may only explain that an abnormality with a small percentage of occurring exists in the sample data. Correlations and patterns are not explained by every test so one must be careful to avoid hasty inferences. For example the mean and standard deviation explain how far the data is spread and where the data is centered, but this information does not explain differences between two data sets. An F-test demonstrates if the variances of two data sets are significantly different, but that does not explain anything about the means. Even when a significant correlation is the result of a linear regression test, the only way to rule out an error in the sample as the cause is to perform a sample effect size test.
Conclusion
With increased online capabilities for consumers and businesses, a large amount of data is available. Many are learning and seeking ways to apply this data in a beneficial way. Statistics teaches us to analyze and perhaps find value within the data. People may use statistics to create a monthly or annual budget to pay off loans faster. A couple may poll their friends to decide what type of food to cater at their wedding. Professionally, statistics can help streamline production, trim expenses, meet seasonal demands, or decide if the company should raise the minimum pay to sixteen dollars per hour. Statistics helps people make wiser decisions.
References
Choudhry, A. (2019). BUS 308 week 2 lecture 2: Examining differences – part 1. [Instructor lecture]. Retrieved from http://login.ashford.edu
Clarke, M. (Writer), Reich, J. M. (Director), & Rubenstein, B. (Producer). (2000). Measures of variability and relative standing [Video file]. Retrieved from https://fod.infobase.com/OnDemandEmbed.aspx?token=10255&wID=100753&plt=FOD&loid=0&w=640&h=480&fWidth=660&fHeight=530
Doctoral Journey, The [Screen name]. (2013, August 26). Descriptive statistics, part 1 [Video file]. Retrieved from https://www.youtube.com/watch?v=8Iklj-lf1fY&feature=youtu.be
Dr Nic’s Maths and Stats [Screen name]. (2012, February 1). Choosing which statistical test to use – statistics help [Video file]. Retrieved from https://www.youtube.com/watch?v=rulIUAN0U3w&feature=youtu.be
Forbes Insights. (2015, April 9). The role of data & analytics today [Video file]. Retrieved from https://youtu.be/fxroi4beKhE
Research by Design [Screen name]. (2016, August 23). Inferential statistics – sampling, probability, and inference (7-5) [Video file]. Retrieved from https://www.youtube.com/watch?v=EPWH91UcpZw&feature=youtu.be



No comments:

Post a Comment