Document Sample

For years, the Forbes magazine has put the Top 100 Celebrity Power List, where they ran Earnings in Millions, Number of Press Clippings, Number of Magazine Covers and a few Somehow they average all these explanatory variables together and get the top 100 list o Now, I'm just as fascinated with these dazzling celebrites as the next person, so I though interesting to regress their ranking and see how much they are explained by the things m Whenever any normal (non-celebrity) person sees this ranking they ask, "Where did this this really explain anything?" I thought it would be interesting to see if these variables re ranking of celebrities. I thought that Earnings, Number of Press Clippings, and Number o the variables most likely to explain these celebrities power ranking out of the ones provid website. I also thought it would be interesting to look at gender and the role it plays in th notice there are not very many women in the top 50. And, in an effort to keep my project only look at the top 50 of these "powerful" celebrites, which is a large enough sample to population of 100 celebrities. I hypothesize that the earnings of a celebrity per million wi on the Celebrities power ranking. The Effect of Data from www.forbes.com: Forbes Celebrity 100 2001 Earnings Press Magazine Gender 60 Rank Name in millions Clips Covers m=0, f=1 50 Tom 1 Cruise 43.2 11,715 11 0 40 Tiger 30 Ranking 2 Woods 53 47,149 5 0 20 3 Beatles 70 26,142 1 0 10 Britney 4 Spears 38.5 19,607 5 1 0 0 -10 0 Bruce -10 5 Willis 70 8,841 2 0 -20 Michael 6 Jordan 37 28,350 1 0 Backstr eet This scatter plot shows 7 Boys 35.5 11,666 3 0 celebrity ranks on the p 8 'N Sync 42 12,506 5 0 some celebrites the mo Oprah closer their rating is to b 9 Winfrey 150 9,495 0 1 line is negative, implyin Mel closer the celebrity will appears to be several o 10 Gibson 31.8 9,591 4 0 trend line. So, outliers Mike 11 Tyson 48 15,770 0 0 George 12 Lucas 250 4,002 0 0 Stephen 13 King 44 6,747 1 0 Steven Spielber 60 14 g 51 10,950 1 0 50 Michael Schuma 40 Ranking 15 cher 59 8,595 0 0 30 Julia 20 16 Roberts 18.9 10,422 7 1 10 Shaquill 0 17 e O'Neal 24 21,380 2 0 0 Metallic 18 a 28 5,077 0 0 Eddie 19 Murphy 39.5 4,689 0 0 J.K. This scatter plot shows 20 Rowling 36 3,109 0 1 celebrity's power ranki because the relationsh 21 Dr. Dre 31.5 7,157 1 0 the celebrity is to being Regis 1. The basic trend see 22 Philbin 35 10,133 0 0 David the closer they are to n Copperfi variability and several 23 eld 60 2,010 0 0 seemed to be logartihm David logs and run a regress 24 Letterm 20 12,576 2 0 explain my data better Kobe 25 Bryant 20 15,554 2 0 Rosie O'Donn 26 ell 25 7,207 2 1 Tina 27 Turner 31 5,183 0 1 Rush 60 Limbaug 28 h 31 3,486 0 0 50 40 29 Brad Pitt 23.8 6,814 2 0 Ranking 30 Tom 30 Clancy 37 2,761 0 0 20 Howard 10 31 Stern 30 3,988 0 0 Nicolas 0 32 Cage 28.4 4,823 0 0 0 -10 Dixie 33 Chicks 25 5,739 1 1 Jennifer 34 Lopez 14.4 9,109 5 1 This scatter plot shows the Dale of a Celebrity. The slope i Earnhar that the higher the salary, 35 dt 24.5 11,592 0 0 shows that as the celebrity Keanu There seems to be curvatu 36 Reeves 25.5 3,942 0 0 logarithmic also. There als Grant line. Later in the project w 37 Hill 26 11,187 0 0 check for outliers. Lennox 38 Lewis 23 9,807 0 0 John 39 Grisham 28 3,959 0 0 Martin Lawrenc 40 e 33.2 2,560 0 0 Jay 41 Leno 17 9,482 0 0 Siegfrie 42 d & Roy 50 697 0 0 Andre 43 Agassi 17.5 16,390 0 0 Ben 44 Affleck 18.3 5,653 1 0 Robin 45 Williams 17.1 6,075 0 0 Brian Grazer/ Ron 46 Howard 45 554 0 0 47 Kiss 24.5 28,386 0 0 Arnold 48 Palmer 18 7,487 0 0 Ridley 49 Scott 26.2 3,014 0 0 Oscar De Le 50 Hoya 23 4,403 0 0 wer List, where they rank celebrities by their azine Covers and a few other categories.. nd get the top 100 list of powerful celebrities. ext person, so I thought that it would be xplained by the things mentioned above. ey ask, "Where did this come from?" and "Does see if these variables really do explain the power Clippings, and Number of Magazine Covers were g out of the ones provided to me on the Forbes nd the role it plays in the power ranking. If you ffort to keep my project manageable, I decided to arge enough sample to represent the whole celebrity per million will have the greatest effect The Effect of Magazine Covers on Power Ranking Magazine Covers Linear (Magazine Covers) 5 10 15 5 10 15 Number of Magazine Covers This scatter plot shows the effect of Magazine Covers on how a celebrity ranks on the power scale. A trend can be seen that for some celebrites the more magazine covers that they have the closer their rating is to being number 1. The slope of the trend line is negative, implying that the more magazine covers, the closer the celebrity will be to the number one ranking. But, there appears to be several outliers in this data, making it hard to fit a trend line. So, outliers will need to be tested for this data. The effect of Press Clips on Power Ranking Press Clips Log. (Press Clips) 0 20,000 40,000 60,000 Number of Press Clipings This scatter plot shows the effect of the number of Press Clips on a celebrity's power ranking. The slope is negative, which makes sense, because the relationship should be the more press clips, the closer the celebrity is to being at the top of the power list, that being number 1. The basic trend seems to be the more press clips on the celebrity the closer they are to number one. There seems to be quite a bit of variability and several outliers. The best fit line for this scatter plot seemed to be logartihmic, so I will convert the data for press clips to logs and run a regression later in the project to see if that helps to explain my data better. Press clips will also be checked for outliers. The effect on earnings on Power Ranking Earnings in millions Log. (Earnings in millions) 100 200 300 Earnings (in millions) his scatter plot shows the effect on Earnings in millions on the Power Rank f a Celebrity. The slope is negative because the relationship should be hat the higher the salary, the higher the power ranking. The trend line hows that as the celebrity earns more money the closer they are to one. here seems to be curvature in the graph, and the best fit line seemed to be ogarithmic also. There also seem to be outliers that are upsetting the trend ne. Later in the project we will calculate the log of Earnings and we will heck for outliers. 11,715 47,149 26,142 19,607 8,841 28,350 11,666 12,506 9,495 9,591 15,770 4,002 6,747 10,950 8,595 10,422 21,380 5,077 4,689 3,109 7,157 10,133 2,010 12,576 15,554 7,207 5,183 3,486 6,814 2,761 3,988 4,823 5,739 9,109 11,592 3,942 11,187 9,807 3,959 2,560 9,482 697 16,390 5,653 6,075 554 28,386 7,487 3,014 4,403 43.2 53 70 38.5 70 37 35.5 42 150 31.8 48 250 44 51 59 18.9 24 28 39.5 36 31.5 35 60 20 20 25 31 31 23.8 37 30 28.4 25 14.4 24.5 25.5 26 23 28 33.2 17 50 17.5 18.3 17.1 45 24.5 18 26.2 23 11 1 5 2 1 3 5 4 2 5 1 6 3 7 5 8 0 9 4 10 0 11 0 12 1 13 1 14 0 15 7 16 2 17 0 18 0 19 0 20 1 21 0 22 0 23 2 24 2 25 2 26 0 27 0 28 2 29 0 30 0 31 0 32 1 33 5 34 0 35 0 36 0 37 0 38 0 39 0 40 0 41 0 42 0 43 1 44 0 45 0 46 0 47 0 48 0 49 0 50 RESIDUAL OUTPUT Observation Predicted Rank Residuals Standard Residuals 1 -4.196714645 5.196714645 0.518937435 2 -5.529570889 7.529570889 0.751893546 3 13.46347981 -10.46347981 -1.044870026 4 9.182997449 -5.182997449 -0.517567653 5 18.94110681 -13.94110681 -1.392141516 6 18.0780916 -12.0780916 -1.206103144 7 20.58552173 -13.58552173 -1.356633234 8 13.19926818 -5.199268179 -0.519192427 9 9.524224834 -0.524224834 -0.052348437 10 19.30253125 -9.302531248 -0.928939153 11 25.2237989 -14.2237989 -1.420370795 12 -3.872394215 15.87239421 1.584997464 13 27.36854184 -14.36854184 -1.434824644 14 24.11987197 -10.11987197 -1.010557777 15 26.81430125 -11.81430125 -1.179761369 16 11.16616326 4.833836739 0.4827009 17 20.77641333 -3.776413335 -0.37710792 18 33.86647744 -15.86647744 -1.584406622 19 32.07300661 -13.07300661 -1.305454114 20 32.27482863 -12.27482863 -1.225749058 21 29.32367069 -8.323670691 -0.83119136 22 30.20333537 -8.203335371 -0.819174826 23 29.84160104 -6.841601043 -0.683193736 24 25.74366295 -1.743662953 -0.174120005 25 24.29665452 0.703345482 0.070235202 26 26.32160146 -0.321601461 -0.03211472 27 32.12881231 -5.128812305 -0.512156792 28 34.12249997 -6.122499967 -0.611385201 29 27.88849367 1.111506334 0.110993634 30 33.44069019 -3.440690186 -0.343583025 31 34.05092627 -3.050926274 -0.304661688 32 33.92095676 -1.920956762 -0.191824343 33 29.96382044 3.036179561 0.303189099 34 18.43755162 15.56244838 1.554046661 35 31.30406013 3.695939874 0.36907194 36 34.84884279 1.151157215 0.114953122 37 31.24232767 5.757672327 0.574953968 38 32.42991228 5.570087718 0.556222003 39 34.40971297 4.590287026 0.45838033 40 34.19327764 5.80672236 0.579852043 41 33.62191644 7.378083559 0.73676621 42 32.20306505 9.796934949 0.978309689 43 30.17914943 12.82085057 1.280274127 44 32.32945447 11.67054553 1.165406102 45 35.26014094 9.739859059 0.972610162 46 33.13428771 12.86571229 1.284753962 47 23.14386548 23.85613452 2.382243801 48 34.41893795 13.58106205 1.356187896 49 35.17911397 13.82088603 1.380136419 50 35.05571268 14.94428732 1.492317869 Now, that we have identified some of the residuals we could elimate them all together or we could turn that data set into binary numbers. Both options do not seem to be viable options for this project because the outliers were not the same in every data set and would do little to help explain the Power Ranking of Celebrities. Even though there are outliers that may be causing the trend line to look more loga am going to convert Earnings and Press Clips into logs so as to control for curvatu make the graphs more linear. I will only convert the X values to log, so I will use wh calls semi-log. This may or may not have any impact on the regression. I cannot c Magazine covers and gender to logarithms because there are X variables of 0. It is possible to take the log of zero. Earnings Log earnings Rank 43.2 1.635483747 1 53 1.72427587 2 70 1.84509804 3 38.5 1.58546073 4 Effect of Log Earnings 70 1.84509804 5 37 1.568201724 6 60 35.5 1.550228353 7 50 42 1.62324929 8 Power Ranking 40 150 2.176091259 9 31.8 1.50242712 10 30 48 1.681241237 11 20 250 2.397940009 12 10 44 1.643452676 13 0 51 1.707570176 14 59 1.770852012 15 -10 0 1 18.9 1.276461804 16 Log Earnings 24 1.380211242 17 28 1.447158031 18 39.5 1.596597096 19 By taking the log of the data for E 36 1.556302501 20 and then re-graphing the data, I c has become more linear and easi 31.5 1.498310554 21 slope is still negative, but the rea 35 1.544068044 22 increased effect on Power Rankin 60 1.77815125 23 increased Earnings in Millions is 20 1.301029996 24 although there are still a few outl 20 1.301029996 25 25 1.397940009 26 31 1.491361694 27 31 1.491361694 28 23.8 1.376576957 29 37 1.568201724 30 30 1.477121255 31 28.4 1.45331834 32 25 1.397940009 33 14.4 1.158362492 34 24.5 1.389166084 35 25.5 1.40654018 36 26 1.414973348 37 23 1.361727836 38 28 1.447158031 39 33.2 1.521138084 40 17 1.230448921 41 50 1.698970004 42 17.5 1.243038049 43 18.3 1.26245109 44 17.1 1.23299611 45 45 1.653212514 46 24.5 1.389166084 47 18 1.255272505 48 26.2 1.418301291 49 23 1.361727836 50 Press Clips Log of Clipping Rank 11,715 4.068742293 1 47,149 4.673472486 2 Effect of Log Press Clip 26,142 4.41733881 3 19,607 4.292411149 4 8,841 5 60 3.946501391 28,350 4.452553063 6 50 Power Ranking 11,666 4.066921972 7 40 12,506 4.097118424 8 30 9,495 3.977494969 9 9,591 3.981863891 10 20 15,770 4.197831693 11 10 4,002 3.602277084 12 0 6,747 3.82911071 13 0 1 10,950 4.039414119 14 8,595 3.934245881 15 Log of Press Cl 10,422 4.017951069 16 21,380 4.330007701 17 5,077 3.705607163 18 In this scatter plot, I can see th 4,689 3.671080233 19 to logarithmic functions, did in helped the graph to become mo 3,109 3.492620722 20 clips before taking the log show 7,157 3.854731017 21 graph may seem to show a littl 10,133 4.005738043 22 still present. 2,010 3.303196057 23 12,576 4.099542529 24 15,554 4.191842095 25 7,207 3.857754522 26 5,183 3.714581209 27 3,486 3.542327383 28 6,814 3.833402129 29 2,761 3.441066407 30 3,988 3.60075515 31 4,823 3.683317262 32 5,739 3.758836225 33 9,109 3.959470702 34 11,592 4.064158372 35 3,942 3.59571662 36 11,187 4.048713638 37 9,807 3.991536175 38 3,959 3.597585502 39 2,560 3.408239965 40 9,482 3.976899951 41 697 2.843232778 42 16,390 4.214578954 43 5,653 3.752278985 44 6,075 3.783546282 45 554 2.743509765 46 28,386 4.453104198 47 7,487 3.874307833 48 3,014 3.479143248 49 4,403 3.643748685 50 There do not seem to be any outl for the gender residual plot, nor s there be any since it is a binary Gender Residual Plot variable. 30 20 Residuals 10 0 -10 0 0.5 1 1.5 -20 Jennifer Lopz Gender There are quite a few outliers in that increase its variability. And Covers Residual Plot a few on the residual plot for Ma Covers are Jennifer Lopez with 5 covers and only a rank of 35, an 30 Robers with 7 magazine covers 25 20 rank of only 16. Most celebrites 15 ranking than them have 2 or less Residuals 10 covers. 5 0 -5 0 5 10 15 -10 -15 Julia Roberts -20 Magazine Covers KISS Clips Residual Plot In the Press Clips Residual P variablity. There are several 30 28,386 press clips with a ran Andre 30 28,386 press clips with a ran Andre 25 Agassi with 16,390 press clip Agassi 20 There is also Tiger Woods wi 15 Residuals clips and his rank is number 10 have less than 14,000 press 5 0 -5 0 10,000 20,000 30,000 40,000 50,000 -10 -15 -20 Tiger Woods Clips Earnings Residual Plot In the Earnings Residual Plot, major outliers that are causing 30 skewed or varied. There is G with 250 million dollars last ye 20 ranked only 12, and then there Residuals 10 Winfrey who earned 150 millio is ranked only 9. 0 0 100 200 300 -10 George Lucas -20 Earnings Oprah line to look more logarmithmic, I to control for curvature and to log, so I will use what Excel egression. I cannot convert X variables of 0. It is not Effect of Log Earnings on Power Ranking Log Earnings Linear (Log Earnings) 2 3 Log Earnings he log of the data for Earnings in Millions graphing the data, I can see that the graph e more linear and easier to interpret. The l negative, but the reationship between the ffect on Power Ranking based on arnings in Millions is easier to see, ere are still a few outliers. ffect of Log Press Clippings on Power Ranking Press Clippings Linear (Press Clippings) 2 3 4 5 Log of Press Clippings atter plot, I can see that the converting Press Clips hmic functions, did in fact control for curvture and e graph to become more linear. The graph of press re taking the log showed some variability, but this y seem to show a little bit more, and outliers are ere do not seem to be any outliers the gender residual plot, nor should ere be any since it is a binary ennifer Lopz here are quite a few outliers in the data at increase its variability. And example of few on the residual plot for Magazine overs are Jennifer Lopez with 5 magazine overs and only a rank of 35, and Julia obers with 7 magazine covers and the nk of only 16. Most celebrites with higher nking than them have 2 or less magazine Julia Roberts In the Press Clips Residual Plot it shows much variablity. There are several outliers. Kiss has 28,386 press clips with a rank of 47. And, Andre 28,386 press clips with a rank of 47. And, Andre Agassi with 16,390 press clips and a rank of 43. There is also Tiger Woods with 47,129 press clips and his rank is number 2. Most celebrites have less than 14,000 press clips. These are some of the most powerful people in the world!???? SCARY! In the Earnings Residual Plot, it shows only 2 major outliers that are causing the data to be skewed or varied. There is George Lucas with 250 million dollars last year and he his ranked only 12, and then there is Oprah Winfrey who earned 150 million last year and is ranked only 9. When examining the regression, I first looked is 0.486131057, meaning that the explanatory SUMMARY OUTPUT ranking of celebrities. I used the adjusted R S explanatory variable. The Significance F is 5. Regression Statistics indicated that this model is statistically signific Multiple R 0.726690816 our regression model to be fairly accurate. My less than 0.05, which means at the 0.05 level R Square 0.528079542 different from zero. Adjusted R Square 0.486131057 Standard Error 10.44974404 Observations 50 ANOVA df SS MS F Regression 4 5498.62823 1374.7 12.5887631 Residual 45 4913.87177 109.2 Total 49 10412.5 Coefficients Standard Error t Stat P-value Intercept 41.15912737 2.920398184 14.094 4.0829E-18 Earnings -0.172347809 0.040484706 -4.2571 0.00010384 Clips -0.000485899 0.000189422 -2.5652 0.01371704 Covers -2.928918648 0.743870326 -3.9374 0.00028326 Gender -1.169116368 4.212586323 -0.2775 0.78264472 Equation: Power Ranking = 41.15912737-0.17234781(Earnings) - 0.0004859(Pre - 2.92891865 (Magazine Covers) - 1.16911637(Gender) The p-value for Earnings is less than 0.05 which means that as earnings increase it wi ranking to decrease (come closer to one) by 0.1723. Earnings has the smallest p the variable that most explains the power ranking. The p-value for press clips is less than 0.05. As the number of press clippings increas cause the the ranking of a celebrity to decrease (closer to one) by 0.0004959. cause the the ranking of a celebrity to decrease (closer to one) by 0.0004959. The p-value of magazine covers is also less than 0.05, meaning that as the number of covers increase, the ranking of the celebrity will decrease (again, closer to one) by 2.9 The p-value for gender is greater than 0.05, so at this level, it means that it has no sta significant barrier on the power ranking of celebrities. Since there did appear to be curvature in my take the log of two of our explanatory variabl above without the log values only explained ranking, I decided to take another regression Earnings and Magazine Covers. SUMMARY OUTPUT When examining the regression, I first looked at the Regression Statistics meaning that the explanatory variables explain 72% Multiple R 0.864423018 higher percent explained than in the regression use R Square 0.747227154 than 0.05. This indicated that this model is statistica Adjusted R Square 0.724758456 regression model to be fairly accurate. The Significa Standard Error 7.6477989 believe that I might have more confidence in this mo except for Gender, less than 0.05, which means at Observations 50 from zero. ANOVA df SS MS F Regression 4 7780.502739 1945.1 33.2563628 Residual 45 2631.997261 58.489 Total 49 10412.5 Coefficients Standard Error t Stat P-value Intercept 147.47781 14.78077979 9.9777 5.5773E-13 Log earnings -39.6358063 4.743043213 -8.3566 1.0494E-10 log clips -15.06270601 3.226425648 -4.6685 2.7493E-05 Covers -2.518800882 0.547950385 -4.5968 3.4757E-05 Gender -1.902418836 3.049351464 -0.6239 0.53585863 Equation: 147.47781-39.6358063(log earnings)-15.062706(log press clips) -2.51880088(Magazine Covers-1.90241884(Gender) The p-values here are much more statistically significant than the ones in the regression above The p-values here are much more statistically significant than the ones in the regression abov much more of an impact on my coefficents of Log earnings, log clips, magazine covers, and ge The p-value for log earnings is much less than 0.05, so therefore we can conclude that an incr a 39.636 decrease in power ranking, making it much closer to one. This is a huge jump from t the regression that did not take the log of earnings. The p-value for earnings in this regression causing it to have the most effect on power ranking. The p-value for log press clips is also much less than 0.05, and we can conclude that an incre cause a 15.063 decrease in power ranking, making it closer to the number one ranking. This a the regression above. The p-value for Magazine Covers is still less than 0.05, and it still has about the same coefficie in magazine covers will have a 2.51 decrease in power ranking (making it closer to one). The p-value for Gender is still greater than 0.05, in this regression much greater than 0.05 and statistical significance. he regression, I first looked at the Adjusted R Square term which meaning that the explanatory variables explain 49% of the power es. I used the adjusted R Square because I have more than one le. The Significance F is 5.91E-07 which is less than 0.05. This model is statistically significant and we can have confidence in del to be fairly accurate. My p-values are all, except for Gender, ich means at the 0.05 level they are statistically significantly Significance F 5.91595E-07 Lower 95% Upper 95% Lower 95.0% Upper 95.0% 35.27714374 47.04111099 35.2771437 47.04111099 -0.253888189 -0.09080743 -0.25388819 -0.09080743 -0.000867416 -0.00010438 -0.00086742 -0.00010438 -4.427150327 -1.43068697 -4.42715033 -1.43068697 -9.65370039 7.315467655 -9.65370039 7.315467655 rnings) - 0.0004859(Press Clips) .16911637(Gender) at as earnings increase it will cause the ings has the smallest p-value, so it is er of press clippings increase it will one) by 0.0004959. one) by 0.0004959. aning that as the number of magazine (again, closer to one) by 2.9289. , it means that it has no statistically ppear to be curvature in my scatter plots and we did wo of our explanatory variables and the regression e log values only explained 49% of the celebrity power d to take another regression using the log values for agazine Covers. gression, I first looked at the Adjusted R-Square term which is 0.72478456, natory variables explain 72% of the power ranking of celebrities. This is a much d than in the regression used above. The Significance F is 6.49E-13 which is less d that this model is statistically significant and we can have confidence in our fairly accurate. The Significance F is much lower in this regression, causing me to e more confidence in this model than the one above. My p-values, again, are all, than 0.05, which means at the 0.05 level they are statistically significantly different Significance F 6.48876E-13 Lower 95% Upper 95% Lower 95.0% Upper 95.0% 117.7077926 177.2478273 117.707793 177.2478273 -49.1887853 -30.0828273 -49.1887853 -30.0828273 -21.56106056 -8.56435146 -21.5610606 -8.56435146 -3.622429561 -1.4151722 -3.62242956 -1.4151722 -8.044127688 4.239290016 -8.04412769 4.239290016 6(log press clips) 1884(Gender) ones in the regression above not using logs, and have ones in the regression above not using logs, and have ps, magazine covers, and gender. we can conclude that an increase in earnings will cause This is a huge jump from the miniscule coefficient in or earnings in this regression is again the lower p-value can conclude that an increase in press clips will number one ranking. This again, is a huge jump from has about the same coefficient as before. An increase aking it closer to one). much greater than 0.05 and therefore has no 2 1 0 1 0 0 2 0 0 0 0 0 0 0 1 1 5 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 Now when looking at Hollywood Power Lists, at least the ranking with some confidence. The explanatory v ranking. The regression without using logs explained taken when using the log of Earnings and Press Clips Earnings, Press Clips, and Magazine Covers of celebr ranking of celebrities in these Hollywood Power LIsts a little miffed that only 5 or 6 women were on the list a significance, did not seem to have any affect on expla still think "the list" is a little gender biased). There were a few problems with this project, there see sets, especially the data set for press clips, that was v model of some of the data in per capita terms might h and explanatory variables. Because of some of the ex curvature of the graph. I chose to use the semi might also have been used, although, I did not have m how people were ranked is still unclear...How did Tom the number one rating in magazine covers? How in th earned 250 million dollars a year!? (That is A LOT of m that were available to me, such as web hits, and tv an believe them to have a great explanatory effect on the All in all, most of my hypotheses at the beginning of t analysis. Earnings, magazine covers, and press clipp power list a celebrity is ranked, and all of the outcome thought gender might play a role, but my suspicions w My hypothesis that Earnings would have the greatest regression analysis using log values and in the regre like George Lucas with 250 million dollars and Oprah the top 15, it would seem from my regression analysis with a measly 43 million would have to take a back se cannot account for the factor of attractiveness, and m ranking! ywood Power Lists, at least the Forbes Hollywood Power list, we can look at nfidence. The explanatory variables seemed to have some impact on the without using logs explained almost 50% of the ranking and the regression of Earnings and Press Clips explained almost 75% of the ranking. So, d Magazine Covers of celebrities, seem to have a pretty vital impact on the hese Hollywood Power LIsts. Gender, the variable that I put in because I was or 6 women were on the list and I hoped it would have some statistical m to have any affect on explaining the power ranking of celebrities (although I tle gender biased). ms with this project, there seemed to be heteroscedacity in some of the data set for press clips, that was very hard to control for, so perhaps recasting the a in per capita terms might have helped to increase the validity of the data set, s. Because of some of the extreme variablity, it was difficult to control the chose to use the semi-log to control for the curvature, but the quadratic form ed, although, I did not have much success with that method. And, some of is still unclear...How did Tom Cruise get to be number one, when he only had magazine covers? How in the world is George Lucas number 12 when he s a year!? (That is A LOT of money!) There were other explantory variables such as web hits, and tv and radio hits per celebrity, but I truly did not eat explanatory effect on the power ranking. otheses at the beginning of the project were confirmed by my regression azine covers, and press clippings all do have an effect on how high on the anked, and all of the outcomes of the regressions, logically make sense. I y a role, but my suspicions were not confirmed by the regression analysis. ngs would have the greatest effect on the power ranking was confirmed in the g log values and in the regression without using log values. Although, people 50 million dollars and Oprah with 150 million dollars are year, were certainly in from my regression analysis that they would be higher up and Tom Cruise would have to take a back seat to these high powered stars. But, I guess one ctor of attractiveness, and maybe that's how Tom Cuise got his number one

DOCUMENT INFO

Shared By:

Categories:

Tags:

Stats:

views: | 1 |

posted: | 8/2/2012 |

language: | |

pages: | 115 |

How are you planning on using Docstoc?
BUSINESS
PERSONAL

By registering with docstoc.com you agree to our
privacy policy and
terms of service, and to receive content and offer notifications.

Docstoc is the premier online destination to start and grow small businesses. It hosts the best quality and widest selection of professional documents (over 20 million) and resources including expert videos, articles and productivity tools to make every small business better.

Search or Browse for any specific document or resource you need for your business. Or explore our curated resources for Starting a Business, Growing a Business or for Professional Development.

Feel free to Contact Us with any questions you might have.