Phoenix Recording's Research Report
Kenneth Drake, Elizabeth Navarro, Olive Myers-Rose
QNT/561
September 7, 2015
Heidi Carty, Ph. D.

Phoenix Recording's Research Report
Part 1: Formulating the Research Problem
The Research Problem Phoenix Recording is a record label for various genres of music. The country music genre is fairly small and represents both male and female country artists. The Vice President noticed that the overall profits for the country artists are extremely low. He has decided to cut the production of music budget for one of the artist. The Vice President has hired an exterior team to provide viable information that would determine the reason for reducing any particular artist’s production budget. Part of the exterior team’s research will involve analyzing YouTube’s view rates for each artist’s hit song for the current year. The independent variable in the research is the gender of the artist and the dependent variable is the number of YouTube views. The team will take a random sample of male and female artists. The team realizes that the variables of male versus female country artists may give the record label an overall view for that genre of music videos being viewed.
Research Question Is the difference in the number of YouTube views (DV) based on the gender of the artists (IV)?
Hypotheses

H0: The difference in YouTube views (DV) is not based on the gender of the artist (IV).

H1: The difference in YouTube views (DV) is based on the gender of the artist (IV).

Part 2: Literature Review

Literature Review This article formulates the understanding that when comparing two populations of vast amounts data its statistical study can be supported by the central limit theorem (Tsung-Shan, 2003). Robust statistics is a convenient way of determining a central tendency and its dispersion; it also makes the assumption that there is a small amount of outliers. Team C makes the proposal that the difference in the number of YouTube views is based on the on gender of artists and denotes the null hypothesis as the difference in YouTube views in not based on the gender of the artists and represents the status quo. An article raises the question that the genre of music may influence the number of sales for Phoenix Recording. It brings light that musical preference can also affect the number of YouTube views. The researcher concludes that demographics (independent variables) are inconclusive with the musical preferences of genres of music (dependent variables) (White, 2001). Using this conclusion, the VP of Phoenix Recording may want to randomly sample both male and female artists in country music. The number of YouTube views can further influence the VP’s decision making for cutting production costs. The Nielsen Music 2014 study (2014) displays the successes in sales as a whole in the music industry and also categorizes genre of music successes, successes of artists both male and female, digital and physical record sales by artist, the growth of digital streaming (i.e. YouTube), and the overall highlights of how music was consumed (Nielsen, 2014). The study further entails the growth of streaming music versus the purchasing of music as collective digital and physical whole. Phoenix Recording can thus find great evidence as to why their sales have gone down. The study also indicates that the popularity of musical festivals increasing in sales, a possible alternative for Phoenix Recording to incorporate. Country music has historically been a male-dominated (“Nzherald.co.nz”, 2015). Some believe that country female artists are systematically being shut out of commercial radio not because of gender but because of audience preference. Studies show that there has been a decrease in the number of female-voiced singles from 38% in 1998 to 18% in 2014. There is gender imbalance in country music. A study was performed on a group of male and female young adults to explore music preference. Part of the study explored whether or not a listener's gender had any relation to his or her artist preference as well as if listeners were gender biased when artist preference was involved. The study also considered the listeners gender as it relates to genre preference (Millar, 2008). Country music is not the only genre that is mail dominant so is rock, rap, and EDM. Throughout the 90’s and early 2000’s women album artist continued to rise but decreased in the area of individual country artist during 2012, 2013, 2014 (“Billboard”, 2015).

Part 3: Sampling and Data Collection Plan

Sampling Design

Identifying the appropriate target market is vital for Phoenix Recording. Thus, the parameter of interest would be the YouTube views of the Male and Female country artists Phoenix Recording currently has signed. The suggested best approach is as follows:

Population: The population that will be used for the research study are male and female music artists.
Target population The target population is country music male and female artists for Phoenix Recording.
Sampling elements Data mining is the sampling element that will be used because of a large amount of data that has to be collected. Data mining is a way to collect large amounts of data as well as it can be useful when Phoenix Recording is trying to increase its revenue, cut costs, or both (Palace, 2015). In the case of Phoenix Recording the company is trying to cut costs therefore data mining will be effective and cost efficient for gathering the information needed.
Sample size The sampling size must be quite substantial since click data must be quite large to be applicable for the statistical needs. The sample size is calculated at a 95% confidence level with a 5% margin of error thus resulting in the sample size of 384.61. (See Appendix A)
Method of random sampling Stratified sampling is “After a population is divided into the appropriate stat, a simple random sample can be taken with each stratum” (Cooper & Schindler, p. 379, N/D). Phoenix Recording will use stratified sampling because it is the most effective sampling method and it allows the team to research specifically country music artists within the entire music industry of artists. Even though simple random sampling would also work for the stratified method of sampling provides higher statistical precision ("Explorable," 2015).
Validity
The validity would be determined based on the reliability and consistency of data collection. The validity would also be based on the correct information collected based from the target population. The reliability of YouTube’s metrics reports will be extremely vital to determining the most fitting conclusion.
Physical Data Collection Physical data collection will be made by monitoring and recording the total number of YouTube views for each artist (both male and female) per day at the same time.
Data Storage
Data can be stored on various data software systems using computers. The data will be made accessible to all researchers via any computer or device; the data can be accessed through Google Drive or another cloud-based system. The data software system will also be protected with passwords. The virtual drives provide security and ease of access for all team members including Phoenix Recording.

Part 4: Descriptive Statistics and Interpretation

Descriptive Statistics

Phoenix Recordings wants to know if there is a correlation between viewers to male and female country artists. Using stratified sampling techniques, Phoenix Recordings takes the total population of YouTube viewers and segregate into distinct music genres. The target subgroup (strata) is country music.
Numeric Variable: YouTube views (DV) for Male Country Artists (IV)
Distribution is normally distributed

|Central Tendency: Mean = 57.70 |
|Dispersion: Standard deviation = 28.16 |
|Number: 20 |
|Min/Max: 8 YouTube views to 100 YouTube views |
|Confidence Interval: 45.36 to 70.04 |

*See the histogram in Appendix B and descriptive statistics in Appendix C
Numeric Variable: YouTube views (DV) for Female Country Artists (IV)
Distribution is not normally distributed

|Central Tendency: Median = 76.50 |
|Dispersion: Interquartile range 30.50/2 = ±15.25 views |
|Number: 20 |
|Min/Max: 14 YouTube views to 100 YouTube views |
|Confidence Interval: Not applicable (data is not normally distributed) |

*See the histogram in Appendix B, and descriptive statistics in Appendix C. A scatter plot is in Appendix D.
Descriptive Statistics Interpretation
Numeric Variable: YouTube views for Male Country Artists Twenty male country artist subjects were randomly selected. Their YouTube views were observed between 8 and 100 YouTube views. The average YouTube view was 58 views, with a variation of plus or minus 28 views. One half or more had above 58 views. There is a 95% confidence that the population of YouTube view average is between 45 and 70 views.

Numeric Variable: YouTube views for Female Country Artists The data was significantly skewed. Twenty female country artist subjects were randomly selected. Their YouTube views were between 14 and 100 YouTube views, with a variation of plus or minus 15 views. One half or more subjects had above 77 YouTube views or more. The middle half of the subjects’ YouTube views fell between 52 and 82 views.
Part 5: Inferential Statistics and Findings

Determining the appropriate Statistical Tool (Z-Test)

Our research question proposes the YouTube views would be the dependent variable can be measured through a ratio method. The independent variable is the gender of the studied country artists and would be considered to be measured categorical or on a nominal basis. We can test the null hypothesis by using the p-value.
|Hypothesis Test: Independent Groups (z-test) | | |
| | | | | |
| |Male Artists=1 |Female Artists=2 | | |
| |59.57 |62.03 |mean | |
| |28.77 |29.43 |std. dev. | |
| |30 |30 |n | |
| | | | | |
| | |-2.467 |difference (Male Artists=1 - Female Artists=2) |
| | |7.514 |standard error of difference |
| | |0 |hypothesized difference |
| | | | | |
| | |-0.33 | z | |
| | |.7427 | p-value (two-tailed) |
| | | | | |
| | |-17.194 |confidence interval 95.% lower |
| | |12.260 |confidence interval 95.% upper |
| | |14.727 | margin of error |

Hypothesis Test Findings and Answering the Research Question

When testing two independent groups, we are able to conclude that gender of the artists does not depend on the number of YouTube views. We can reject the null hypothesis as it states the number of YouTube views does not make a difference on the gender of the artist. The confidence interval indicates the distribution is skewed from lower -17.194 to upper 12.60. The p-value (two-tailed) is larger than a=.05 which indicates that 0.7427 shows evidence to reject the null hypothesis. One also can conclude that concluding its findings with the sample of 30 does not have enough evidence to make a valid conclusion. Lastly, we can accept the alternative hypothesis because when comparing the artists’ means and dispersion we can conclude a difference. Furthermore, answering the research question is defined by having evidence to see a difference in the number of YouTube views based on the gender of country artists.

Conclusions from the Study
Interpretation and Findings on the Hypothesis Tests using Z-test

When testing two independent groups, we are able to conclude that gender of the artists does not depend on the number of YouTube views. We can reject the null hypothesis as it states the number of YouTube views does not make a difference on the gender of the artist. The confidence interval indicates the distribution is skewed from lower -17.194 to upper 12.60. The p-value (two-tailed) is larger than a=.05 which indicates that 0.7427 shows evidence to reject the null hypothesis. The null hypothesis is rejected because the z-score less than 1.96 or greater than 1.96. The p-value or level of significance is less than required and also indicates that the null hypothesis should be rejected.
Recommendations based on the Results As Phoenix Recording dilemma was to address the production costs, providing viable information that would determine the reason for reducing any one particular artist’s production budget would rely on the external team’s findings. Our team found convincing data from YouTube views that providing enough data to come to conclusion. The recommendation from analyzing both the YouTube views (dependent variable) based on the gender of the country artists’ (independent variable) leads to the conclusion to cut cost from a male country artist. On the other hand, Phoenix Recording’s Vice President may consider cutting production from both male and female country artists as its research may present to further investigate through other avenues.
Observations and Research Challenges
Formulating Research Hypothesis Truly effective research is more likely to result when a marketing research supplier works collaboratively with its client throughout the research process. Researchers struggle with operational definitions that state specific criteria for testing and measuring (empirical standards) the research questions of its clients Researchers often use a research question rather than a descriptive hypothesis (Cooper, 2011). The research question format is less frequently used with a situation calling for relational hypotheses. With explanatory (causal) hypotheses, there is an implication that the existence of or a change in one variable causes or leads to a change in the other variable (Cooper, 2011). The first challenge was at this level in developing a valid and reliable construct to measure a null hypothesis for Phoenix Recording Studio. The Null Hypothesis is the statement that no difference exists between the parameter and a static being compared to it. The team had the null hypothesis in reverse. Therefore, it did not correctly operationalize a significant difference between YouTube views (clicks) as a function of gender. The client wanted determination if you tube clicks would make a determination of one female artist against that of one male artist. However, it is not possible to calculate statistics on a sample of one so it was suggested that a random sample of male and female artist be used in order to calculate meaningful information or data.
Data Collection In the beginning there were problems identifying valid, reliable information on the company itself and its potential to answer the client’s research question. Therefore, the data used to properly answer the research question was taken from 2014 Nielsen Music Study.
Inferential Statistics There was difficulty in translating the hypothesis test in terms of Z score, or F-value or p-value in determining the acceptance or rejection of the null hypothesis. In running inferential statistics such as an ANOVA or z-test to compare the computed statistics (i.e., F-value or z-score) and the p-value to the corresponding value associated with an alpha of .05. If your p-value is greater than .05 then you reject the null hypothesis.
Implications
The implication of the aforementioned results resulted in not being able to sufficiently determine the research question. There are times when there isn’t enough information to determine or define it. Sometimes the basic data are not available or cannot be verified. Individuals in public policy may use statistical analyses based on data obtained in situations where the statistics presented cannot be validated. As researchers or those that report such data, it is our responsibility to verify the statistics and methodology (i.e., the quality of the data) before we utilize them as secondary data in our research endeavors. Exploratory process would allow the client as well as researcher to make decision on valid and reliable research question, hypothesis, data, and inferential statistics. The insufficient data will not allow researcher to control bias or establish a causal relationship between variables. Business decision are important to the bottom line or ROI, therefore correct identification of research question is critical.
Suggested Recommendations to minimize challenges After comparing each member's sampling designs, the confusion lies in identifying the appropriate design. Perhaps as a team, we may need to include more research questions to appropriately identify the area of study. Exploratory research would have been beneficial perhaps in clarifying any confusion among members. Make use of internal sources of information such as DSS and BIS systems and gatekeepers. Contact You Tube directly in developing the right variables and understand its processes. The better we understand the customer’s need the better it would be to develop the research question and the subsequent test results.
Suggested future research Exploration is particularly useful when researchers lack a clear idea of the problems he or she will meet during the study. Despite its obvious value, researchers and managers alike give exploration less attention than it deserves. There are strong pressures for quick answers. Start research early, interview internal and external experts (Google), discover and analyze primary (google analytics) and secondary sources. Use pattern discovery, target marketing to predict trends and behavior. Exploratory data analysis (EDA) provides a perspective and set of tools to search for clues and patterns in the data. These efforts help define or operationalize the research question for the client.
Descriptive statistics and tables—the first steps in exploring data-used statistical measures to depict the center, spread, and shape of distributions and are helpful as preliminary tools for data description. Data analysis should begin with visual inspection. After that, it is not only possible but also desirable to cycle between exploratory and confirmatory approaches. There is a special challenge to presenting statistical data. While some of the data may be incorporated in the text, most statistics should be placed in tables, charts, or graphs.

Appendix A
The sample size is a 95% confidence level and a 5% margin. n = sample size, p = proportion of the population having attribute q = proportion of the population not having attribute = (1-p) op = standard error or proportion (.05/1.96)
[pic]
n = [pic] n = [pic] n = [pic] n = 384.61

Appendix B
Male and Female Country Artists Histograms
[pic]
[pic]

Appendix C
Descriptive Statistics for Male and Female Country Artists YouTube Views
|Descriptive statistics | | |
| | | |
| |Male Artists |Female Artists |
|count |20 |20 |
|mean |57.70 |66.30 |
|sample standard deviation |28.16 |27.12 |
|sample variance |792.85 |735.48 |
|minimum |8 |14 |
|maximum |100 |100 |
|range |92 |86 |
| | | |
|population variance |753.21 |698.71 |
|population standard deviation |27.44 |26.43 |
| | | |
|confidence interval 95.% lower |45.36 |54.41 |
|confidence interval 95.% upper |70.04 |78.19 |
| margin of error |12.34 |11.89 |
|z |1.96 |1.96 |
| | | |
|1st quartile |42.75 |51.75 |
|median |55.50 |76.50 |
|3rd quartile |81.75 |82.25 |
|interquartile range |39.00 |30.50 |
|mode |78.00 |#N/A |
| | | |
|low extremes |0 |0 |
|low outliers |0 |0 |
|high outliers |0 |0 |
|high extremes |0 |0 |
| | | |
|normal curve GOF | | |
|p-value |.7788 |.0143 |
|chi-square(df=2) |0.500 |8.500 |
|E |4.000 |4.000 |
|O(-0.84) |4 |4 |
|O(-0.25) |4 |2 |
|O(+0.25) |3 |2 |
|O(+0.84) |4 |9 |
|O(inf.) |5 |3 |

Appendix D Scatterplot Male Country Artists versus Female Country Artists YouTube Views [pic]

References
Cooper, D., & Schindler, P. Business Research Methods, 11th Edition. [VitalSource Bookshelf version]. Retrieved from http://legacy.vitalsource.com/books/9781308263021/page/378
Explorable. (2015). Retrieved from https://explorable.com/stratified-sampling\
Palace, B. (2015). Data Mining: What is data mining. Retrieved from http://www.anderson.ucla.edu/faculty/jason.frand/teacher/technologies/palace/datamining.htm
Nzherald.co.nz. (2015). Retrieved from http://www.nzherald.co.nz/entertainment/news/article.cfm?c_id=1501119&objectid=11495690
White, C.G. (2001) The effects of class, age, gender and race on musical preferences: An examination of the omnivore/univore framework. Virginia Polytechnic and State University. Retrieved from http://scholar.lib.vt.edu/theses/available/etd-09072001-101445/unrestricted/whitethesis.doc.pdf
Billboard. (2015). Retrieved from http://www.billboard.com/articles/news/6583054/women-in-country-radio-sexist-imbalance
Nielsen. (2014). Nielson music U.S. report. Retrieved from http://www.nielsen.com/content/dam/corporate/us/en/public%20factsheets/Soundscan/nielsen-2014-year-end-music-report-us.pdf
Millar, B. (2008, October). Selective hearing: gender bias in the music preferences of young adults. Psychology of Music, 36(4), 429-445. Gale PowerSearch.
Tsung-Shan, T. (2003). Comparing Two Population Means and Variances: A Parametric Robust Way. Communications In Statistics: Theory & Methods, 32(10), 2013. doi:10.1081/STA-120023263