The Arts and Education; The effect of theater on ...



The Arts and Education; The effect of theater on graduation ratesThe Arts and EducationAlexandra Burhorn12/4/2013This paper applies OLS regression techniques to analyze the relationship between high school theater programs and high school graduation rates. The theory is that a high school with a theater program will have a higher graduation rate. 369 high schools in the state of Indiana were originally included in this research. This was later decreased to 361 high schools. Through analyzing OLS results it is established that there is no significant relationship between the presence of a theater program and a high school’s graduation rate.Table of ContentsIntroduction3Literature3Model6Data8Empirical Results10Conclusion19Appendix20References23IntroductionNot many studies look at the connection between high school graduation rates and high school theater programs. If they are analyzed, it is as a secondary objective not the main focus. This project attempts to analyze this connection. I chose this topic of study because of my personal background. I have been involved in the performing arts since I was three years of age. Theater has always been a large part of that. I have heard from multiple sources over the years that students who are involved in theater will perform better in school. I wanted to test this theory.The earliest indications of theater as we know it today are traced back to ancient Egypt from 2800 to 2400 B.C. These were based off of a combination of rituals and myths that already existed. The Greeks performed plays each year during the Festival of Dionysus. The Romans also performed plays. These were based off of religious festivals (Parisot). Theater has been a part of the life of man for many years.Data was gathered electronically from websites that had compiled the information and through a site by site search of high school websites. Through regression analysis I found that the presence of a theater program in a high school had no significant effect on its graduation rate. More specific study could be warranted.LiteratureLiterature that focuses on the arts and their effect on graduation rates in high school is not well documented. However, in Eisner’s (1998) “Does Experience in the Arts Boost Academic Achievement?” he states that the effect of arts on academic achievement seems to be largest when they are specifically targeted towards raising achievement in reading and writing (Eisner, 1998). Studies also state that there is not enough evidence to support a causal link between the arts and academic achievement (Winner et al, 2000, Shulruf, 2011). Winner and Hetland (2000) speculated that this was due to too narrow of a focus on grades and test scores as outcomes measuring academic achievement.Shulruf (2011) looked at the link between extra-curricular activities and educational outcomes. He defined extra-curricular activities as “those which are school-sponsored and external to the core curriculum.” (Shulruf 594). Outcomes were measured in three different categories; school engagement and retention, future aspirations and student self-concept, and academic achievement. Shulruf found that the performing arts had no effect on retention, a result contrary to expectations (Shulruf 602). The overall conclusion of Shulruf’s study was that there was a positive connection between extra-curricular activities and achievement, but that there was not enough evidence to support a causal link between the two (Shulruf 606).Although the results of the three previous studies are pertinent to my topic and are of great interest, it is not clear which methods or what forms of regression were used. It is the exact opposite in McNeal’s 1995 publication “Extracurricular Activities and High School Dropouts”.McNeal’s study (1995) uses dropout rates as the dependent variable, as opposed to graduation rates. In his study McNeal goes into detail about his testing procedures and why he chose them. McNeal uses logistic regression because the linear probability model violates the assumptions of a normally distributed disturbance and homoskedasticity. The logistic regression eliminates this issue because it makes the dependent variable, the probability of dropping out, a continuous latent variable. McNeal runs multiple models to try to explain the connection between his variables of race, gender, age, socioeconomic status, single-headed household, ability, academic track, vocational track, hours worked, sports clubs, fine arts, academic clubs, and vocational clubs. In doing this he is looking at involvement in four different extracurricular activities; sports clubs, fine arts, academic clubs, and vocational clubs. In using a single-activity model McNeal is assessing the impact of student involvement with each activity individually. He also works with a multiple-participation model which looks at the impact of a student being involved in multiple activities. When he obtains the results from these he runs a test to determine if there are mediating variables. Through all of these tests the results that are of interest to my topic are that participation in the arts, and participation in sports clubs decreases the probability of dropping out. Involvement in academic organizations also decreases the probability of dropping out. Vocational clubs had no impact. McNeal found, through his multiple participation model, that the arts decreased the probability of dropping out by fifteen percent. He also found, through the same model, that participation in sports clubs decreased the probability of dropping out by forty percent (McNeal 1995). Overall, McNeal’s study states that the arts do have an impact on the probability of a student dropping out of high school, but that this impact was not near as significant as that of sports clubs.To find information on graduation rates on their own, I looked to Heckman and LaFontaine (2010). This essay gives the history and the methodology behind the graduation rate, which will help me to understand the results of my model(s). In this essay I found that graduation rates are affected by race and gender most often. Heckman and LaFontaine also explain how graduation rates are calculated. In the past GED recipients were counted as part of the high school graduation rate, but this was changed in 2001. Now, as Heckman and LaFontaine put it, “only students who receive a secondary credential that is counted as fully aligned with each state’s academic standard are to be counted as high school graduates.” (Heckman 246). This means that those who pass the GED but do not meet the standards of high school graduation are no longer counted as part of the graduation rate.ModelMy model attempts to assess the effects of various independent variables on the high school graduation rate. In this model, my dependent variable is the graduation rate and the independent variables are; the presence of a theater program, the percentage of non white students, the percentage of females in the senior class, the total enrollment, the student teacher ratio, and the attendance rate. My model is mathematically represented by:GradRatei=β0+β1attenratei+β2theateri+β3%femalei-β4%non whitei+β5Ln(total enrollment)i+β6ratioi+?iIn which:GradRatei = the graduation rate of the ith high schoolattenratei = the attendance rate of the ith high schooltheatersi = a dummy variable equal to 1 if there is a theater program present, zero otherwise%femalei = the percentage of females enrolled in the senior class of the ith high school%nonwhitei = the percentage of non white students enrolled at the ith high schooltotal enrollment i= the total enrollment of the ith high schoolratioi = the student teacher ratio of the ith high schoolThe main hypothesis I wish to test is that the coefficient for the presence of a high school theater program is positive. I predict that the coefficient will be positive, that the presence of a high school theater program will increase the graduation rate at said high school. This is due to my research, as mentioned in the literature, and what I have seen and been told throughout my life. The test looks as follows:Ho: β1≤0Ha: β1>0with the null hypothesis stating that β1 is negative or equal to zero, meaning that if a theater program is present at a high school the graduation rate will either decrease or remain unchanged. In my model, I expect the impact of a higher attendance rate to be positive. If students attend high school regularly then they will learn to be better equipped to graduate.I expect that a one percent increase in the percentage of females in the senior class of the ith high school to have a positive affect on the graduation rate because my research shows this relationship. Heckman and LaFontaine (2010) show in their work that females, overall, have a slightly higher graduation rate than males in the recent past. (Heckman 254). I expect a one percent increase in the percentage of non white students at the ith high school to have a negative effect on the graduation rate, because of what I learned through reading the literature. Heckman and LaFontaine also show evidence that non-whites have a lower graduation rate than do white students (Heckman 245).I expect the impact of total enrollment at the ith high school to be non-linear. I expect the relationship to be positive, meaning that as enrollment increases the graduation rate will increase but at a decreasing rate. I expect an increased student teacher ratio to have a negative effect on the graduation rate because if a teacher is spread too thin they cannot teach as effectively. Also, a larger number of students could mean that the school’s assets are now spread thinner as well.DataThe variables for high schools that I used throughout my research are as follows; attendance rate, 12th grade enrollment, theater program, graduation rate, percentage of females in the 12th grade, percentage of non white students, total enrollment, student teacher ratio, the number of non white students, the number of female students in the 12th grade, and 12th grade enrollment. All of these variables were collected for the year 2012 in the state of Indiana.The variables attendance rate, total enrollment, the number of female students in the 12th grade, and 12th grade enrollment were provided in yearly reports that are collected on the Indiana Department of Education’s website. The variable graduation rate is found at Indianapublicmedia’s website and was compiled from data from the Indiana Department of Education. The student teacher ratio is found on and is collected from the Indiana Department of Education.However, the variables percentage of females in the 12th grade, percentage of non white students, and the number of non white students required some calculation. I acquired Excel spreadsheets from the Indiana Department of Education’s webpage that showed me school enrollment by grade and gender, and by grade and ethnicity. From there I was able to calculate the percentage of females in the 12th grade by using the statistics given for that grade for each high school.% females in the 12thgrade=# femalestotal 12th grade enrollment*100I used the same method to calculate the percentage of non white students. I was given the number of students who were identified racially as American Indian, Asian, Black, Hispanic, Multiracial, and Native Hawaiian or Pacific Islander for each high school. I first added all of these together to give me the number of non white students at each high school and then applied the same method as with the percentage of females in the 12th grade. % non white students=#non whitetotal enrollment*100The variable “theater” is a dummy variable that equals one if the high school has been deemed to have a theater program, zero otherwise. This was the most difficult variable to collect information for. I searched the internet for each individual high school’s website and from there I searched through their listed clubs, teacher directories, and calendars for what I deemed evidence of a theater program. Evidence of a theater program included; a drama/theater club or troupe, a teacher in the directory that was listed as teaching “performing arts”, “theater”, or “drama”, or if the calendar showed a date for a school play. I also checked the announcements for mention of club meetings or upcoming plays when there was no other source available. If one of these criteria was met then the school received a 1, and if these criteria were not met the school received a 0.I was forced to eliminate some of the high schools that met my original criteria of having a 12th grade enrollment. This was because they were either special needs schools and therefore did not fit with my sample, or the data was not available and I could not find a suitable substitution. With my first collection of data my sample size was 369 high schools. This included those magnet and charter schools that I could collect data for. A magnet school is defined by the Merriam Webster Online Dictionary as, “a school that has courses in special subjects (such as the arts or technology) and is designed to attract students from all parts of a community”. Even though these are specialized schools they are still public schools and therefore I decided to include them in my sample. Merriam Webster Online Dictionary defines a charter school as, “a tax-supported school established by a charter between a granting body (as a school board) and an outside group (as of teachers and parents) which operates the school without most local and state educational regulations so as to achieve set goals”. This is also still considered a public school and therefore I included charter schools in my sample.After collecting this initial sample I realized that I needed more data. I collected data on the student teacher ratio at each high school in my sample. With this second round of data collection I was unable to find data or a suitable substitute for eight high schools and therefore was forced to cut them from my sample. This left me with a fairly large sample of 361 high schools.Empirical ResultsWhen working with my original data I started with my initial theoretical equation. GradRatei= β0+β1theateri-β2nonwhitei+β3femalei-β4enrollmenti+ β5attenratei-β6enrollment 12thi+?iTable 1 shows the initial OLS results:Table 1. Regression Analysis of Equation 01Dependent Variable: GRAD_RATEMethod: Least SquaresDate: 11/14/13 Time: 10:19Sample: 1 369Included observations: 369VariableCoefficientStd. Errort-StatisticProb.??C-127.573325.48786-5.0052580.0000ATTENDANCE_RATE2.2343320.2679308.3392470.0000ENROLLMENT_12TH0.0290130.0098802.9364470.0035ENROLLMENT_TOTAL-0.0035130.002467-1.4240290.1553FEMALE0.0149430.0127151.1752940.2406NON_WHITE-0.0099090.001870-5.2978650.0000THEATER0.4615631.0353240.4458160.6560R-squared0.293608????Mean dependent var87.72439Adjusted R-squared0.281900????S.D. dependent var11.04778S.E. of regression9.361975????Akaike info criterion7.329977Sum squared resid31728.06????Schwarz criterion7.404166Log likelihood-1345.381????Hannan-Quinn criter.7.359449F-statistic25.07720????Durbin-Watson stat1.865728Prob(F-statistic)0.000000I concluded that only the coefficients for attendance and non white were statistically significant and of the correct sign through a standard t-test. The overall fit of the equation was also poor, the adjusted R2 was fairly low. When I performed a Ramsey Reset test I found that the equation was statistically significant. Table 8. Ramsey Reset Test Results Equation 01Ramsey RESET Test:F-statistic7.583170????Prob. F(3,359)0.0001Log likelihood ratio22.67220????Prob. Chi-Square(3)0.0000Test Equation:Dependent Variable: GRAD_RATEMethod: Least SquaresDate: 12/04/13 Time: 15:16Sample: 1 369Included observations: 369VariableCoefficientStd. Errort-StatisticProb.??C21732.729722.4142.2353220.0260ATTENDANCE_RATE-328.2918147.7930-2.2212950.0270ENROLLMENT_12TH-4.2598361.920859-2.2176720.0272ENROLLMENT_TOTAL0.5160470.2328512.2162130.0273FEMALE-2.1867100.989366-2.2102130.0277NON_WHITE1.4552570.6556012.2197300.0271THEATER-68.1973730.55848-2.2317000.0263FITTED^22.6841701.2619332.1270310.0341FITTED^3-0.0211890.010591-2.0007230.0462FITTED^46.14E-053.30E-051.8599660.0637R-squared0.335704????Mean dependent var87.72439Adjusted R-squared0.319050????S.D. dependent var11.04778S.E. of regression9.116592????Akaike info criterion7.284795Sum squared resid29837.29????Schwarz criterion7.390779Log likelihood-1334.045????Hannan-Quinn criter.7.326897F-statistic20.15794????Durbin-Watson stat1.858407Prob(F-statistic)0.000000The calculated F-statistic is approximately 7.58. The critical F-statistic at a 5-percent level of significance is 2.10. The calculated F-statistic is greater than the critical value; therefore I reject the null hypothesis and conclude that Equation 01 is statistically significant at a 5-percent level of significance.This first equation is statistically significant, but too simple. I felt it was not giving me accurate results. The signs of the coefficients were mostly the opposite of what I had expected and were insignificant. I modified this original model six times more, and added additional data, before I ended with Equation 07. Equations 02 through to 06 can be found in the appendix. Equation 07 is as follows:GradRatei=β0+β1attenratei+β2theateri+β3%femalei-β4%non whitei+β5Ln(total enrollment)i+β6ratioi+?iI included the percentages of females and non white students to adjust for size discrepancies between high schools in my sample. I logged total enrollment because I didn’t believe it to have a linear relationship to the graduation rate. I had squared total enrollment in Equation 05 (Table 5. in Appendix) because I believed that to be the relationship, but the signs of the expected coefficients for total enrollment and total enrollment squared showed a possible logarithmic relationship. The coefficient for total enrollment was positive, but so was that for total enrollment squared. This showed me that it was possible that as total enrollment increased so did the graduation rate, but at a decreasing rate.The OLS results for Equation 07 are as follows:Table 7. Regression Results for Equation 07Dependent Variable: GRAD_RATEMethod: Least SquaresDate: 12/04/13 Time: 14:49Sample: 1 361Included observations: 361VariableCoefficientStd. Errort-StatisticProb.??C-184.196028.17509-6.5375480.0000ATTENDANCE_RATE2.6559410.2898529.1630870.0000THEATER0.3905581.0175670.3838160.7013PERFEM0.0016310.0190040.0858070.9317PERNWHITE-0.0451650.009704-4.6541860.0000LOG(TOTAL_ENROLLMENT)3.9256040.7347305.3429220.0000RATIO-0.3614760.091923-3.9323610.0001R-squared0.313333????Mean dependent var87.95402Adjusted R-squared0.301694????S.D. dependent var10.78076S.E. of regression9.008907????Akaike info criterion7.253505Sum squared resid28730.78????Schwarz criterion7.328912Log likelihood-1302.258????Hannan-Quinn criter.7.283485F-statistic26.92228????Durbin-Watson stat1.729953Prob(F-statistic)0.000000The coefficient for attendance rate is statistically significant because of its low probability and high t-statistic. If the attendance rate at a high school were to increase by 1 point the graduation rate will increase by approximately 2.67 points, holding all other variables constant. This matches my original theory regarding attendance rate.The coefficient for the variable “theater” is insignificant because of its low t-statistic. The coefficient can be interpreted thusly, if a high school has a theater program, as defined earlier, their graduation rate can be expected to increase by approximately .39 points, holding all other variables constant. This follows my research and my expectations, but yet the coefficient is insignificant.The coefficient for the percentage of females enrolled in the 12th grade is insignificant. Yet, the positive sign is what I expected through my research on the subject. If the percentage of females enrolled in 12th grade were to increase by one percentage point the graduation rate would increase by approximately .002 points, holding all other variables constant.The coefficient for the percentage of non white students enrolled is significant. It has a fairly large t-statistic. If the percentage of non white students enrolled were to increase by one percentage point the graduation rate would decrease by approximately .05 points, holding all other variables constant. The negative sign of the coefficient is as I expected and what my research predicted.The coefficient for the logged total enrollment is significant. The t-statistic is fairly high and passes a one-tailed t-test. If total enrollment were to increase by one percent the graduation rate would increase by approximately 3.93 points, holding all other variables constant. This is as I theorized.The coefficient for the student teacher ratio is significant because of its fairly large t-statistic. If the student teacher ratio at a high school were to increase by one point then the graduation rate would decrease by approximately .36 points, holding all other variables constant. The negative sign of this coefficient complies with my theory.I performed a Ramsey Reset Test to test the overall significance of this equation. The results are as follows:Table 9. Ramey Reset Test Results Equation 07Ramsey RESET Test:F-statistic27.44043????Prob. F(3,351)0.0000Log likelihood ratio76.06027????Prob. Chi-Square(3)0.0000Test Equation:Dependent Variable: GRAD_RATEMethod: Least SquaresDate: 12/04/13 Time: 15:56Sample: 1 361Included observations: 361VariableCoefficientStd. Errort-StatisticProb.??C25694.534913.5455.2293260.0000ATTENDANCE_RATE-337.024565.06558-5.1797660.0000THEATER-49.560829.574718-5.1762170.0000PERFEM-0.2000520.042568-4.6995930.0000PERNWHITE5.5919851.0940935.1110670.0000LOG(TOTAL_ENROLLMENT)-495.679696.02950-5.1617440.0000RATIO45.908318.8751705.1726680.0000FITTED^22.5454700.5394044.7190430.0000FITTED^3-0.0219130.005126-4.2749090.0000FITTED^46.90E-051.78E-053.8775410.0001R-squared0.443784????Mean dependent var87.95402Adjusted R-squared0.429522????S.D. dependent var10.78076S.E. of regression8.142705????Akaike info criterion7.059432Sum squared resid23272.58????Schwarz criterion7.167157Log likelihood-1264.227????Hannan-Quinn criter.7.102261F-statistic31.11667????Durbin-Watson stat1.811013Prob(F-statistic)0.000000The calculated F-statistic is approximately 27.44. The critical F-statistic is 2.10; because the calculated F-statistic is greater than the critical I reject the null hypothesis and conclude that this model is statistically significant overall. Also, the adjusted R2 is fairly high for a cross-sectional analysis and the signs of the coefficients are as predicted.I then looked at the possibility of serial autocorrelation in this final model.Table 10. Correlation Statistics Equation 07ATTENDANCE_RATETHEATERPERFEMPERNWHITELOG(TOTAL_ENROLLMENT)RATIOATTENDANCE_RATE?1.000000?0.026823?0.042798-0.202037-0.027206?0.050459THEATER?0.026823?1.000000?0.075961?0.100471?0.301177?0.010372PERFEM?0.042798?0.075961?1.000000?0.006225-0.007205-0.033857PERNWHITE-0.202037?0.100471?0.006225?1.000000?0.015496?0.002024LOG(TOTAL_ENROLLMENT)-0.027206?0.301177-0.007205?0.015496?1.000000?0.263368RATIO?0.050459?0.010372-0.033857?0.002024?0.263368?1.000000These results show no signs of serial auto correlation. The values are all under .5, meaning that less than fifty percent of the observed values for the variables are correlated. Most of the variables are showing less than thirty percent correlation. I also performed an LM test for serial auto correlation. The results are as follows:Table 11. LM Test Results Equation 07Breusch-Godfrey Serial Correlation LM Test:F-statistic2.186400????Prob. F(3,351)0.0894Obs*R-squared6.622318????Prob. Chi-Square(3)0.0850Test Equation:Dependent Variable: RESIDMethod: Least SquaresDate: 12/04/13 Time: 16:17Sample: 1 361Included observations: 361Presample missing value lagged residuals set to zero.VariableCoefficientStd. Errort-StatisticProb.??C-0.09111028.20344-0.0032300.9974ATTENDANCE_RATE0.0117680.2899590.0405860.9676THEATER0.1901661.0162030.1871340.8517PERFEM-0.0006980.018925-0.0368980.9706PERNWHITE0.0032330.0097650.3311220.7408LOG(TOTAL_ENROLLMENT)-0.2483380.740684-0.3352830.7376RATIO0.0245640.0928810.2644670.7916RESID(-1)0.1178040.0550352.1405260.0330RESID(-2)0.0202860.0555050.3654780.7150RESID(-3)0.0676810.0565491.1968430.2322R-squared0.018344????Mean dependent var-5.07E-14Adjusted R-squared-0.006826????S.D. dependent var8.933517S.E. of regression8.963956????Akaike info criterion7.251610Sum squared resid28203.73????Schwarz criterion7.359336Log likelihood-1298.916????Hannan-Quinn criter.7.294440F-statistic0.728800????Durbin-Watson stat1.967385Prob(F-statistic)0.682493The calculated value is approximately 6.62. The critical value for a five percent level of significance is 5.99. The calculated value is greater than the critical. Therefore, I reject the null hypothesis and conclude that no serial auto correlation exists.It is quite possible that heteroskedasticity exists in my model because of it being cross-sectional. I performed a White test to test this theory, the results are as follows:Table 12. White Test Results Equation 07Heteroskedasticity Test: WhiteF-statistic12.59059????Prob. F(26,334)0.0000Obs*R-squared178.6865????Prob. Chi-Square(26)0.0000Scaled explained SS1190.277????Prob. Chi-Square(26)0.0000Test Equation:Dependent Variable: RESID^2Method: Least SquaresDate: 12/04/13 Time: 16:37Sample: 1 361Included observations: 361Collinear test regressors dropped from specificationVariableCoefficientStd. Errort-StatisticProb.??C69760.4419874.373.5100700.0005ATTENDANCE_RATE-1660.150430.3828-3.8573790.0001ATTENDANCE_RATE^210.920542.4090374.5331560.0000ATTENDANCE_RATE*THEATER68.3106417.858483.8251110.0002ATTENDANCE_RATE*PERFEM0.2203450.9973350.2209340.8253ATTENDANCE_RATE*PERNWHITE0.9267190.3871212.3938780.0172ATTENDANCE_RATE*(LOG(TOTAL_ENROLLMENT))-65.7507514.59338-4.5055190.0000ATTENDANCE_RATE*RATIO-3.1537722.424620-1.3007280.1942THEATER-5346.5201749.891-3.0553450.0024THEATER*PERFEM-1.1614784.214509-0.2755900.7830THEATER*PERNWHITE1.3097781.3838260.9464910.3446THEATER*(LOG(TOTAL_ENROLLMENT))-125.788655.85898-2.2518960.0250THEATER*RATIO-17.0967410.63148-1.6081240.1088PERFEM-102.642592.18119-1.1134860.2663PERFEM^2-0.0056390.007538-0.7481040.4549PERFEM*PERNWHITE-0.2470870.076490-3.2303320.0014PERFEM*(LOG(TOTAL_ENROLLMENT))5.6661074.0615501.3950600.1639PERFEM*RATIO2.8913400.7192654.0198550.0001PERNWHITE-51.0733037.57416-1.3592660.1750PERNWHITE^2-0.0026250.003844-0.6828530.4952PERNWHITE*(LOG(TOTAL_ENROLLMENT))-1.0989061.104296-0.9951190.3204PERNWHITE*RATIO-0.9647180.247253-3.9017510.0001LOG(TOTAL_ENROLLMENT)4215.2121335.5173.1562390.0017(LOG(TOTAL_ENROLLMENT))^2108.601929.968433.6238770.0003(LOG(TOTAL_ENROLLMENT))*RATIO19.600676.6962122.9271290.0037RATIO28.37107217.57910.1303940.8963RATIO^20.8315250.2072344.0124990.0001R-squared0.494976????Mean dependent var79.58665Adjusted R-squared0.455663????S.D. dependent var296.6469S.E. of regression218.8636????Akaike info criterion13.68662Sum squared resid15999033????Schwarz criterion13.97748Log likelihood-2443.435????Hannan-Quinn criter.13.80226F-statistic12.59059????Durbin-Watson stat1.938370Prob(F-statistic)0.000000The calculated test statistic is approximately 178.69. The critical value at a five percent level of significance is 31.4. This leads me to reject the null hypothesis and conclude that it is likely that I have heteroskedasticity in my model.To correct for this heteroskedasticity I used the White corrected standard errors and obtained the following results:Table 13. Regression Results with White Corrected SE Equation 07Dependent Variable: GRAD_RATEMethod: Least SquaresDate: 12/04/13 Time: 16:44Sample: 1 361Included observations: 361White Heteroskedasticity-Consistent Standard Errors & CovarianceVariableCoefficientStd. Errort-StatisticProb.??C-184.196053.93369-3.4152310.0007ATTENDANCE_RATE2.6559410.5657364.6946670.0000THEATER0.3905581.1670690.3346490.7381PERFEM0.0016310.0119200.1367990.8913PERNWHITE-0.0451650.021912-2.0612390.0400LOG(TOTAL_ENROLLMENT)3.9256041.2856053.0535070.0024RATIO-0.3614760.214564-1.6846950.0929R-squared0.313333????Mean dependent var87.95402Adjusted R-squared0.301694????S.D. dependent var10.78076S.E. of regression9.008907????Akaike info criterion7.253505Sum squared resid28730.78????Schwarz criterion7.328912Log likelihood-1302.258????Hannan-Quinn criter.7.283485F-statistic26.92228????Durbin-Watson stat1.729953Prob(F-statistic)0.000000The coefficients were not affected by this change. The t-statistics for attendance rate, % non white, logged total enrollment, and ratio decreased. This decrease did not cause the coefficients to be insignificant. The t-statistics for % female and theater increased. This increase was not significant enough to cause the coefficients to be statistically significant. My previous conclusions and interpretations still stand.ConclusionThe presence of a theater program at a high school has no significant effect on that high school’s graduation rate. Although this is not what I theorized I believe the result to be true. The results of this study can be applied when discussing what programs can and cannot be cut from a high school’s curriculum. Cutting the theater program will not affect graduation rates significantly, although students may object. These results can also be used in discussions of high school resource allocation.If this study were to be repeated, I would recommend the addition of a variable that encompasses sports participation. I could not find sufficient data to include in my study, but I still feel it is relevant. My literature sources also mention the validity of this variable. I also would have liked to include private high schools in my study. The data was not available given my constraints. This study could be modified or extended to include more observations. This study was limited by the time and resources allotted.AppendixTable 2. Regression Results for Equation 02Dependent Variable: GRAD_RATEMethod: Least SquaresDate: 11/14/13 Time: 10:21Sample: 1 369Included observations: 369VariableCoefficientStd. Errort-StatisticProb.??C-119.120725.11226-4.7435280.0000ATTENDANCE_RATE2.2374770.2638278.4808350.0000ENROLLMENT_12TH-0.0077090.017299-0.4456130.6561ENROLLMENT_TOTAL-0.0059320.003021-1.9634240.0504FEMALE0.1250440.0398843.1352000.0019NON_WHITE-0.0062280.002088-2.9829080.0031THEATER0.5505841.0213700.5390640.5902ENROLLMENT_TOTAL_2-1.01E-066.30E-07-1.5967440.1112PERFEM-0.1784090.061885-2.8829340.0042PERNWHITE-0.0374620.011173-3.3530210.0009R-squared0.332017????Mean dependent var87.72439Adjusted R-squared0.315271????S.D. dependent var11.04778S.E. of regression9.141853????Akaike info criterion7.290330Sum squared resid30002.88????Schwarz criterion7.396313Log likelihood-1335.066????Hannan-Quinn criter.7.332432F-statistic19.82655????Durbin-Watson stat1.881339Prob(F-statistic)0.000000Table 3. Regression Results for Equation 03Dependent Variable: GRAD_RATEMethod: Least SquaresDate: 11/14/13 Time: 10:23Sample: 1 369Included observations: 369VariableCoefficientStd. Errort-StatisticProb.??C-122.939825.38842-4.8423580.0000ATTENDANCE_RATE2.1795410.2663878.1818490.0000ENROLLMENT_12TH0.0390370.0088794.3967230.0000ENROLLMENT_TOTAL-0.0027180.002877-0.9449990.3453NON_WHITE-0.0063900.002113-3.0244630.0027THEATER0.6779771.0330010.6563170.5120ENROLLMENT_TOTAL_2-8.41E-076.35E-07-1.3251640.1860PERFEM0.0059630.0195070.3056990.7600PERNWHITE-0.0363150.011303-3.2129680.0014R-squared0.313728????Mean dependent var87.72439Adjusted R-squared0.298477????S.D. dependent var11.04778S.E. of regression9.253282????Akaike info criterion7.311922Sum squared resid30824.36????Schwarz criterion7.407307Log likelihood-1340.050????Hannan-Quinn criter.7.349813F-statistic20.57163????Durbin-Watson stat1.857172Prob(F-statistic)0.000000Table 4. Regression Results for Equation 04Dependent Variable: GRAD_RATEMethod: Least SquaresDate: 11/14/13 Time: 10:24Sample: 1 369Included observations: 369VariableCoefficientStd. Errort-StatisticProb.??C-119.499225.21089-4.7399820.0000ATTENDANCE_RATE2.2054150.2643648.3423510.0000ENROLLMENT_12TH-0.0087350.017359-0.5032110.6151NON_WHITE-0.0074270.002005-3.7051260.0002THEATER0.3531451.0204300.3460740.7295ENROLLMENT_TOTAL_2-1.57E-065.62E-07-2.7944010.0055PERFEM-0.1397070.058894-2.3721950.0182PERNWHITE-0.0337950.011059-3.0558970.0024FEMALE0.0984780.0376672.6144140.0093R-squared0.324844????Mean dependent var87.72439Adjusted R-squared0.309841????S.D. dependent var11.04778S.E. of regression9.178032????Akaike info criterion7.295591Sum squared resid30325.06????Schwarz criterion7.390976Log likelihood-1337.036????Hannan-Quinn criter.7.333482F-statistic21.65128????Durbin-Watson stat1.867804Prob(F-statistic)0.000000Table 5. Regression Results for Equation 05Dependent Variable: GRAD_RATEMethod: Least SquaresDate: 11/18/13 Time: 20:27Sample: 1 369Included observations: 369VariableCoefficientStd. Errort-StatisticProb.??C-134.458325.38282-5.2972160.0000ATTENDANCE_RATE2.3089840.2658778.6844010.0000ENROLLMENT_12TH0.0431580.0088724.8646350.0000THEATER0.7219991.0444890.6912460.4899ENROLLMENT_TOTAL_2-1.06E-066.38E-07-1.6636070.0971PERFEM0.0047190.0197210.2393020.8110PERNWHITE-0.0524900.010069-5.2128920.0000ENROLLMENT_TOTAL-0.0053470.002773-1.9280940.0546R-squared0.296290????Mean dependent var87.72439Adjusted R-squared0.282645????S.D. dependent var11.04778S.E. of regression9.357118????Akaike info criterion7.331593Sum squared resid31607.59????Schwarz criterion7.416380Log likelihood-1344.679????Hannan-Quinn criter.7.365275F-statistic21.71362????Durbin-Watson stat1.807691Prob(F-statistic)0.000000Table 6. Regression Results for Equation 06Dependent Variable: GRAD_RATEMethod: Least SquaresDate: 11/18/13 Time: 20:49Sample: 1 361Included observations: 361VariableCoefficientStd. Errort-StatisticProb.??C-183.676528.79743-6.3782250.0000ATTENDANCE_RATE2.6563690.2902989.1504920.0000ENROLLMENT12TH0.0004650.0051600.0901260.9282THEATER0.3869411.0197850.3794340.7046PERFEM0.0017080.0190500.0896510.9286PERNWHITE-0.0453230.009874-4.5899860.0000LOG(TOTAL_ENROLLMENT)3.8249481.3374142.8599570.0045RATIO-0.3610210.092191-3.9160150.0001R-squared0.313349????Mean dependent var87.95402Adjusted R-squared0.299732????S.D. dependent var10.78076S.E. of regression9.021554????Akaike info criterion7.259022Sum squared resid28730.12????Schwarz criterion7.345202Log likelihood-1302.253????Hannan-Quinn criter.7.293285F-statistic23.01274????Durbin-Watson stat1.730806Prob(F-statistic)0.000000ReferencesEisner, E. W. (Jan., 1998). Does Experience in the Arts Boost Academic Achievement?. Art Education, Vol. 51(No. 1). Retrieved from , J. J., and P. A. LaFontaine. (2010). The American High School Graduation Rate: Trends and Levels. The Review of Economics and Statistics, Vol. 92(No. 2) Retrieved from , R. B. Jr. (Jan., 1995). Extracurricular Activities and High School Dropouts. Sociology of Education, Vol. 68(No. 1). Retrieved from , E., & Stokes, K. (2013). Indiana 2012 Graduation Rates: Search For Your High School’s Numbers. StateImpact. Retrieved from , K., & Robinson, S. (2005). Origins of Theatre. Theatre History. Retrieved from , B. (February 2011). Do Extra-Curricular Activities in Schools Improve Educational Outcomes? A Critical Review and Meta-analysis of the Literature. International Review of Education, Vol. 56. Retrieved from , E., & Hetland, L. (Autumn-Winter, 2000). The Arts in Education: Evaluating the Evidence for a Causal Link. Journal of Aesthetic Education, Vol. 34(No. ?). Retrieved from (2013). Indiana High School Rankings. SchoolDigger. Retrieved from ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download