This action might not be possible to undo. Are you sure you want to continue?
ranging from height and weight to political affiliation and water consumption. Various methods were used through out the paper to gather the information used to answer the questions. Statistical techniques used in this paper range from odds ratios, non parametric techniques such as Wilcoxon Rank-Sum and Pearson s and Spearman s correlations. The findings were discussed in detail with explanations broken down for easy understanding. Intel Corps opinion on the quality of the data was also discussed as well as advice on farthing the findings. Introduction As a consultant from Intel Corp I was hired by the Rothberg Company to determine/answer the following:
1. Is there a relationship between an appropriate combination of a person s height, weight, ring and shoe size? 2. 3. Is there difference in the gross income based on gender? Is there a relationship between political party and if the people feel- President Obama should be reelected-in favor of the health care bill- in favor of the death penalty 4. Is there a relationship in handedness and- being in favor of the death penalty-amount of water consumed? 5. 6. Is there a relationship between education and gross income? Is there a relationship between the major political parties and changing or not changing political party?
I expect to see a relationship in question concerning height, weight ring and shoes size, I also expect to see a relation ship between Political party and if the people feel- President Obama should be re-elected-in favor of the health care bill- in favor of the death penalty, and I also expect to see a relationship between education level and gross income. I expect to see a difference in gross income and gender. I expect to see no relationship with handedness and
being in favor of the death penalty and water consumption. I would also not expect to see any relationship between the major political parties and party changes. In order to test my hypothesis and answer Rothberg Companies questions I was allowed to take a simple random sample of 43 from a population of 1092. I ran a series of statistical analyses on the sample in order to be able to apply it to the entire population.
Method Object(s) of Study The 43 people studied were randomly selected from a population of 1092. The
variables of interest are as fallows: weight, height, ring size, shoe size, gross income, gender, education level, handedness, death penalty, water consumption, health care bill, political party, and Obama re-election. Materials A TI-83 calculator was used to take the sample of people, to store the data and to make graphical displays. Excel was used to view the data given by the Rothberg Company. A digital Camera was used to add the Graphical displays from the TI-83 to the document. Research Design Procedure People were first randomly selected 43 from a population of 1092 given to me by the Rothberg Company. To ensure that the same people were not selected twice, the sample size taken from the data was 50 instead of 43 allowing for duplicates to be replaced. Once I had my 42 people I then started imputing the data about the variables into my calculator. When the
data was finished being input I began to create graphical displays. I to find if it had been input correctly reviewed the input data. No errors were found in the data.
Results It was concluded that the variables of interest height, weight, ring and shoe size, water consumption were Ratio/Interval and thus quantitative and continuous. Variables of interest gender, political party, handedness, education level, death penalty, health care bill, and Obama re-election were Nominal/Ordinal qualitative and discrete. Upon that assumption the data was collected and interpreted. The following information is given as supporting evidence of the results recorded and analyzed by me. Question one was the first to be addressed therefore it was the first reported dataset in the results portion of this paper. Initial findings are as follows:
In Table 1.1 Height, weight, ring and shoe size. The five number summary and additional statistics. MALE then Female
Gross income and Gender was the second area of study and therefore the second reported dataset in the results. Initial findings are as follows:
In Table 2.1 Gross Incomes / Gender. The five number summary and additional statistics Male Summary Statistics: Mean: 40523.07 Q3: 64,000 Stan Dev.: 36704.54 Min:0 Max:120,000 Q1: 3900 Med:45,000
Female Summary Statistics:
Political Party and if a person feels that Obama should be re-elected, the health care bill should be passed and if they support the death penalty are the third area of study. Initial findings are as fallows. Totals of all in favor of each of the above:
Handedness and weather the people are in favor of the death penalty and handedness and water consumption are the fourth area of study. Initial findings are as follows:
Water consumption (Ounces) and Handedness Summary Statistics
It is clear that there is more water consumption by right handed people since there are more right handed than left handed people in this sample. Ambidextrous people were not counted since there wasn¶t any. To give a graphical representation would do no justice but I will just to show what it would look like.
Handedness and Death Penalty In Favor:
Relationship between education and gross income is the fifth area of study. The Initial results are as follows:
Education level and Gross income summary Statistics and Graphs:
Major Political Parties and and change in affliation:
Results of Further Analysis: My first area of study was height weight ring and shoe size. When performing my Pearson¶s test I removed subjects with unknown or incorrect data (Ex: those who put dates instead of a height). I then took the variables height, weight, ring and shoe size and broke them down by gender. The result for male height and weight was that there was not enough evidence to suggest an association between height and weight based on a p-value of .09468 and an r-value of .3727. Form there I constructed a confidence interval for male height and weight it lead me to state that I am 95% confident that the true association falls between (-.1725, .7431). Form the large gap in that interval I was able to determine that there was an extremely weak positive association between the values. Then for male height and
shoes size I ran the non-parametric Spearman¶s Correlation which gave me a pvalue of .053 which allowed me to reason that there is not enough evidence to suggest an association between height and weight for men. My p-value for height and weight was just above my alpha so there could possibly be a slight association though more advanced tests would need to be made to determine that. For male weight and ring size I also had to do a Spearman¶s Correlation, which gave me a pvalue of .461. This led me to reason that there was not enough evidence to suggest that there was an association between male weight and rings size. For the female data pertaining to height, weight ring and shoe size I dropped subjects who had missing or incorrect data (those who put dates instead of height). For female height and weight I ran a Pearson¶s Correlation, which gave me a pvalue of .5612. This led me to reason that there was not enough evidence to suggest an association between female height and weight. I also constructed and confidence interval for female height. The confidence interval is as fallows (-.4635, .5266). I am 95% confident that the true correlation of female height and weight falls between that interval. For female height and shoe size I had to run a Spearman¶s Correlation since the scatter plot appeared to be monotonic. The Spearman¶s Correlation gave me a p-value of .373, which led me to reason that there was not enough evidence to suggest a correlation between female height and shoe size. For female weight and ring size I ran another Spearman¶s Correlation, which gave me a p-value of .002. This led me to reason that there was enough evidence to suggest an association between female ring and shoe size.
The second area of my study involved Gross Income and Gender. Since the data was not normal I preformed a Wilcoxon Rank-Sum. I reasoned that based on a pvalue of .2407 there was not enough evidence to suggest male and female incomes were equal. From that point I broke income down into categories in order to perform an Odds Ratio. I took the incomes and broke them into categories of 0100,000 and 100,000-250,000. (See Appendix) The odds ratio showed that for every .96 males with a gross income of 0-100,000 dollars 1 female has a gross income of the same amount. The Confidence Interval was (.079, 11.662) his means that since 1 is bounded in the interval it is not statistically significant. Meaning there is only a slight difference between the amount of income of males and females. My third area of study was in political party and Obama re-election, in favor of the health care bill and in favor of the death penalty. For political party and Obama reelection I had to omit those people who associated with the Independent or other parties since there was simply to few of them. Undecided votes in the Republican and Democratic parties were treated as lost information and not counted. I ran an odds ratio and determined that for every .13 Republicans who supported Obama reelection one Democrat supported Obama re-election. With a confidence interval of (.019, .916) the information is statistically significant since one is not bounded within the interval. For Political Party and in favor of the health care bill I again did not included undecided votes and was unable to include to Independent and Other since there were too few of them. I ran and odds ratio and determined that for every .94 Republicans who voted in favor of the health care bill one Democrat voted in favor of the health care bill. The confidence interval was (.161, 5.461) which
means the information is not statistically significant since one is bounded with in the interval. This would suggest that Republican and Democrat opinion did not differ by much when it pertained to favoring the health care bill. For political Party and Death Penalty I could not include Independents since there were too few and all undecided votes were counted as missing data and therefore excluded. I ran a Test of Independence and the assumptions were violated. This led me to run odds ratios. The odds ratio for Republicans and Democrats was that for every .47 Republicans who support the death penalty one Democrat also supports the death penalty. The confidence interval was (.06, 3.648) this means it is not statistically significant since one is bounded within the interval. The odds ratio for Republicans and Other was that for every .63 Republicans who support the death penalty one Other supported the death penalty. The confidence interval was (.04, 9.65) this means it is not statistically significant since one is bounded within the interval. The odds ratio for Democrats and Other was that for every 1.33 Democrats who supported the death penalty one Other supported the death penalty. The confidence interval was (.108, 16.48) this means it is not statistically significant since one is bounded within the interval. My fourth area of study was in handedness and water consumption and handedness and death penalty. Due to the lack of left-handed people in my sample the results may be questionable in both my tests. Ambidextrous people were not represented in my sample there for they were not in my tests. In my Wilcoxon Rank-Sum nonparametric test for Handedness and water consumption I found that there was not enough evidence to suggest that left-handed people had consumed as much water as right-handed people. This was based on a p-value of .5382 and t-value of -
.0966. I constructed a confidence interval for left-handed and right-handed peoples water consumption. Left-handed peoples true water consumption should fall between 48 and 64 ounces; right-handed peoples true water consumptions should fall between 36 and 100 ounces. For handedness and the death penalty I did not include those who were undecided on the issue instead I focused on the yes and no responses. I preformed an odds ratio on the data and reasoned that for every .96 left-handed people who voted yes for the death penalty 1 right-handed person voted yes in favor of the death penalty. The confidence interval for my odds ratio was (.087, 10.573) which means it is not statistically significant since one is bounded within the interval. This would suggest that there is not a significant difference in opinion concerning the death penalty when it comes to right and left handed people. My fifth area of study was education level and gross income. I only had one person with a masters and one person with a PhD. so I did not include them in my test. Since the data was not normal I preformed the non-parametric Kurskal-Wallis Test. Based on a p-value of .0012 I was able to reason that there was enough evidence to suggest that education level does play a role in gross income. My sixth and final area of study was major political parties and change in party affiliation. Change in party affiliation will be viewed as those who changed party and those who did not. (Not taking into consideration who they now affiliate with now only that there affiliation changed.) All not applicable answers were view as no change in political affiliation. Thus making it a yes or no question making it possible for me to run an odds ratio on the data. The odds ratio is that for every .18
Republicans who changed party one Democrat changed party. The confidence interval was (.029, 1.07) this means that it is not statistically significant since one is bounded within the interval. Discussion My results for male height and weight were not expected; I had expected to see an association between them. I expected an association because it is reasonable to assume that a taller person may weigh more than a shorter person. I think my result for that test came out showing no association due to the bad data collection. Due to my sample size I had to work with number that were flawed in order to even perform a test. I again expected to see an association between male height and shoe size since it was reasonable to assume a tall person would have a larger foot to support there frame and vice versa for a short person. I think the results of this test showing not association was because of the poor quality of the data and the extremely small sample I took. For male weight and ring size I had expected to see an association between them because a heavier person would have thicker fingers and thus a lager ring size and a vice versa for a short person. Once again my test showed other wise and I believe it to be for the same reasons listed in the above test. My results for female height and weight resulted in no association, which again I would expect to see an association since a tall person in theory should weigh more than a short person. Female height and shoe size also showed no association between them which I would expect to see one because a taller person should have lager feet to balance there frame and vice versa for short people. In female weight and a rings size I did have an association which I would
expect since a heavier person would have thicker fingers and thus a lager ring size and a vice versa for a short person. The results for female height and weight and female height and shoe size probably showed no association for the same reason as the males showed no association that is my sample size being so small the data being so poor. The female weight and ring size were what thought. My results for gender and income were not what I would have expected. I would have expected to see males having more income than females. This might have been true in the past but my results showed that the income levels between males and females are relatively the same indicating that times are changing or have changed. Which now in hindsight seems that it should have been more obvious. While male and female income is not equal it is very close. That¶s why I said times have changed. For political party and Obama re-election, in favor of the health care bill, and in favor of the death penalty my results were in some cases what I expected and in others not what I would have expected to see. With political party and Obama reelection my results showed what I expected to see and that was more Democrats in favor of Obama re-election than Republicans. I expected this because Democrats would naturally vote for a candidate from there party. With political party and in favor of the health care bill I expected to see more Democrats in favor of the health care bill than Republicans. What I saw in my test was that both parties were almost equally in favor of the health care bill. The health care bill being more of a concern of Democrats than Republicans I would have expected to see more Democrats than Republicans in favor of the health care bill instead they were almost equal. I would
attribute that to my small sample size and the poor quality of the data. For political party and death penalty my results were also unexpected. I expected to see more Republicans in favor of the death penalty than any other party but my results showed otherwise. Democrats in tests with Republican and Other showed to be more in favor of the death penalty, which I found to be odd since Democrats do not generally support the death penalty more than republicans. In my test with Republicans and Other I found that Other supported the death penalty more than Republicans. Which is odd since Republicans traditionally support the death penalty more than any other party. I again would attribute these odd results to poor data quality and a small sample size. For handedness and in favor of the death penalty and water consumption my results what I expected to see. I expected there to be no relationship between handedness and death penalty and there wasn¶t. My results showed that there was no significant difference in support of the death penalty between right and lefthanded people. For left and right-handed water consumption I expected to see no relationship between them. My results showed that there was not relationship; they were not equal to one another. This may have been because of my small sample size. In general I would not have thought that those things could be related and I was proved correct. For education level and gross income I expected to see a relationship and I did. It makes sense to think a person with a higher education would make money than someone with less of an education. My results showed this to be true that there was a relationship. I think that if I had better quality of data and know more
advanced techniques that I could have found how much of a difference education level played in gross income. For major political parties and change in political affiliation I expected to see no relationship between them and I was incorrect. It seems as my results showed that more Democrats changed parties than Republicans. I would have expected to see no change in party affiliation since most people choose to only associate with the two major parties. This could have also been a result of my small sample size and poor data collection. The only complaints that I have are that the data was poorly collected with missing and uninterruptable data showing up far to often and that my sample size was far to small to be effective with all this errors in the data. I did the best with the techniques I could apply and think that if I knew more advance techniques I could have kept a smaller sample size and produce more accurate results.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.