Simon Honours

1 The effectiveness of Sesame show on 3- to 6-Year_old Children Background/ Literature Review Questions: Since television viewing may have important beneficial as well as harmful effects on the behavior of preschool children. The Sesame Street show has focused on collecting different expression before and after watching the show among preschool children. The paper is interested in studying whether the Sesame show has an impact on children, aged 3 to 6 years old, in terms of their performance on various tests. The assumption would be that the Show helps improve preschool children’s cognitive skills. To do this the paper attempts to answer three questions: 1. Whether the show has effectiveness on improving children’s score for each test? 2. For each subgroups of different viewing frequency, what characterizes the children who have most and least gains in each test? 3. Does the Sesame Street show have significant different effect among 3through 6-year old children? Data: The data is a segment of a large data set with 239 valid cases and 20 variables involved in a national assessment of the education television show, Sesame Street (abstracted from Glasnapp and Poggio, 1985). The variables include ability measure, several pre- and post-test achievement measures, measures of viewing behavior and home environment variables. I got the Sesame dataset from my professor Johnson. As the paper focuses on the impact on the changes of Children’s test scores, the ratio of test improvement was calculated by subtracting each posttest score by its counterpart pretest score which was divided by the pretest score and then multiplied by 100. Also, the “age” variable was converted to a unit of year by divided the actual months by 12. Thus, the paper ended up with 239 valid cases and 13 variables for the study (see Table 1). To answer question 1, we using normality test to see whether the outcome variables have normal distribution, and using multicollinearity to see whether there exist multicollinearity in the regression. And then we take multivariable linear regression to find the impact of all variables on each test. To answer question 2, we also using multivariable linear regression, but we split the variables viewcat in four group to see how many times do children watch the show have best performance, and split the variables viewenc in 2 group to see whether children should be encouraged to watch the show. To answer question 3, we split the data with age and watching frequency, age and watching treatment condition. Then we take there- variable figures to see the performance of children with various age. And we take analysis of variance to find whether there exists significant difference between ages Findings: Form the study, we find the show can improve the children’s performance on the 2 test, but it does not mean that watching the show more frequently can get a higher increase in the test scores. There exist a best watching frequency. And we also find that encouragement is also helpful on each test. Then we study each test score by children’s age with watching frequency and watching treatment condition. And we found that children with little age can get greater performance in the posttest Methods: The paper uses R software for the current study. As to answer each questions, various data analysis methods are adopted. Multivariate linear regression is used to understand the impact of all variables on each test, investigate the effects of the TV show, and identify the characteristics of children with most gains for each test. Analysis of variance is used to identify whether there exist significant difference between ages. Results Effectiveness of the Show Since we want to know whether the show has an effective on improving children’s score for each test, we take a descriptive statistics to all variables. In this paper, we have 13 variables in 239 cases, 6 outcome variables and 7 predictors. Referring to Table 2 of Descriptive Statistics, all six test ratios have negative minimum statistics and their standard deviations are large. Thus, the paper could not simply conclude that the show was effective on bringing up their test scores. However, the means of six test ratios are positive. Thus, there is necessity to look into the effects of other variables on the tests as to conclude the overall effectiveness of the show. In the test of normality (see Table 3), we can see that, for all the independent variables, the p-values for Shapiro-Wilk and Kolmogorov-Smirnov are all less than 0.1. Therefore, we have enough confidence to reject the null hypothesis of normality. However, this is not what we really want. With 239 observations, by central limit theorem, we can say that the data is approximately normal. The normality will always be a concern in our further analysis. Then, we take a test of multicollinearity of all predictors by using variance inflation factor (see Table 4). VIF of all variables are less than 2, there is no multicollinearity for predictors. As to determine whether a show is effective or not, the paper focuses on identifying whether viewcat and viewenc have significantly positive impact on difference ratios for six tests. Looking at coefficients and p-values of all independent variables on the variables of test ratios (see Table 5), independent variables have various impacts on various difference ratios. For test for knowledge of body parts, the coefficient of viewcat is positive; the coefficient of viewenc is negative. It means that the show is effective as the more frequently children watched the show under the condition that they were being encouraged, the higher scores they achieved for the posttest. However, both the p-values of viewcat and viewenc are not significant. For test for letters, the show has similar effect, and the p-value of viewcat is significant. For test for forms, similar effect was interpreted and both p-values are insignificant. For test for numbers, the coefficients of viewcat and viewenc are both positive. The coefficient of viewenc 3 tells us that when child encouraged to watch the show, the lower scores they achieved for the posttest. So it means the show was effective as children watched the show frequently but encouraging children to watch the show has a negative impact on their test scores. And only the p-value of viewcat is significant. For test of relation terms, impact of watching frequency is positive and the effection of whether to encourage to watch is negative, and the p-value of viewcat is significant. As for test of skills, the show is effective as children watch the show more frequently and as they encouraged to watch it, but both p-values are insignificant. One note should be taken that even though the p-values are insignificant for either variables of interest for each test, the involvement of covariates in fact mitigates the effect size of the show. Therefore, the paper does not take the p-values for granted. Instead, the paper intends look at each variable separately in the following section. Children’s Characteristics of Watching Frequency In terms of the analysis of test difference ratios by viewcat, viewcat has four ranging from 1 (rarely watched the show), 2 (once or twice a week), 3 (three to five times a week), to 4 (watched the show on average of more than 5 times a week). The means of test difference ratios under each subgroup of viewcat differs across six tests (see Table 6). Presumably, the ratios increase as children watch the show more frequently. Among all the tests, only the mean of difference ratio for letters test increases according to the presumption by viewcat that the more frequently children watch the show the better scores they get. It is interesting that the means of difference ratios for tests for body knowledge, forms, and numbers have their maximum mean values if children watch the show three to five times a week (viewcat= 3), but then the means drop drastically if they watch more frequently (viewcat= 4). Even though children show the highest mean of difference ratio for the test of relation terms when they watch the show the most frequently (viewcat= 4), the tendency of the ratio could not be generalized for the other three subgroups as their means do not move in a linear direction. Finally, the mean of difference ratio of the test for classification skills is maximized as children watch the show once or twice a week (viewcat= 2), which is also against the presumption. Due to the variety of mean distribution across viewcat subgroups, the paper then splits the data by viewcat and looks at characteristics of children, which are believed to impact the effects of the show on children’s test performances. In our new regression, we set viewcat as dummy variables. So we add three new variables to our model. When viewcat1=1, viewcat2=0, viewcat3=0, it means that viewcat=1. When viewcat1=0, viewcat2=1, viewcat3=0, it means that viewcat=2. When viewcat1=0, viewcat2=0, viewcat3=1, it means that viewcat=3. When viewcat1=0, viewcat2=0, viewcat3=0, it means that viewcat=4. We can see the regression results from table 7 are the same as the means of test. In summary, based the tests scores of children who rarely watched the show, watching the show has a positive impact on children’s test scores except the test of forms. What’s more, watched the show more than twice a week can improve the scores best relatively. (Add table)This section composes the results from the multivariate and 4 discriminant analyses as to study the characteristics of children that categorize the children who present most gains in test scores by watching frequency. The paper identifies children’s characteristics for this particular topic as setting and sex. Coefficients and p-values of those characteristics variables on each test difference ratio shows children with diverse characteristics present most gains on tests, but the group pattern is not present for all tests. When children rarely watch the Show (viewcat = 1), children who watch the Show at school have the most gains on tests for forms and relation terms, while girls have most gains on test for numbers. When children watch the Show once or twice a week (viewat = 2), girls have the most gains on test for numbers. When children watch the show three to five times a week (viewcat = 3), girls have the most gains on the test for relation terms. When children watch the Show on average of more than five times a week (viewcat =4), children have the most gains on tests for knowledge of body parts and numbers in specific age. In terms of the analysis of test difference ratios by viewenc, viewenc has two levels, ranging from 1 (child encouraged to watch), to 2 (child not encouraged to watch). Presumably, the ratios increase as children encouraged to watch the show. From the mean of all test by viewenc in Table 8, we can see among all the tests, only the mean of difference ratio for numbers test increases according to the presumption by viewenc that when children are encourage to watch the show the better scores they get. It is interesting that the means of difference ratios for tests for body knowledge, letters, forms, relation terms and classification have their maximum mean values if children are encourage to watch the show (viewenc= 1), and the means increasing mostly in test of letters, increasing slowest in test of knowledge of body parts. For the test of numbers, the tendency of the ratio changes could not be generalized for the other five test as their means increasing when children are encourage to watch the show, which is against the presumption. Due to the variety of mean distribution across viewenc subgroups, the paper then splits the data by viewenc and looks at characteristics of children, which are believed to impact the effects of the show on children’s test performances. In our new regression, we set viewenc as dummy variables. So we add new variables to our model. When viewenc1=1, it means that viewenc=1. When viewenc1=0, it means that viewenc=2. We can see the regression results from table 9 are the same as the means of test. In summary, we can see watching the show have a positive impact on children’s test scores except the test of numbers. Children’s Age by Watching Frequency The study classify each test score by children’s age and watching frequency, and then take box plot of each test score by watching frequency and age. In Figure 1, there is the box plot of dfratiobody by viewcat and age. The horizontal axis is watching frequency. The vertical axis is age, from 3 years old to 6 years old. When children rarely watch the show (viewcat= 1), children of three years old have the greatest median scores, then four years old, third year old and six years old, lastly, five years old have the lowest median scores. When children watch the show once or twice 5 times a week (viewcat= 2), children of four years old have the highest median scores, then five years old, and lastly six years old have the lowest median score. When children watch the show three or five times a week (viewcat= 3), children of three years old have the greatest median scores, then is children of four years old, third is children of six years old, children of five years old have the lowest median scores. When children watch the show more than five times a week (viewcat= 4), children of three years old have the greatest median scores, then is children of five years old, third is children of six years old, children of four years old have the lowest median scores. So we can say, the show have a greater impact on children of smaller age. In Figure 2, there is the box plot of dfratiolet by VIEWCAT and AGE. When children rarely watch the show (viewcat= 1), children of three years old have the greatest median scores, then four years old, third is six years old, children of five years old have the lowest median scores. When children watch the show once or twice a week (VIEWCAT = 2), children of six years old have the greatest median scores, then five years old, children of five years old have the lowest median scores. When children watch the show three or five times a week (VIEWCAT = 3), children of four years old have the greatest median scores, then five years old, six years old, children of three years old have the lowest median scores. When children watch the show more than five times a week (VIEWCAT = 4), children of three years old have the greatest median scores, then four years old, followed by five years old, children of six years old have the lowest median scores. In Figure 3, there is the box plot of dfratioform by VIEWCAT and AGE. When children rarely watch the show (VIEWCAT = 1), children of four years old have the greatest median scores, then six years old, followed by five years old, children of three years old have the lowest median scores. When children watch the show once or twice times a week (VIEWCAT = 2), children of four years old have the greatest median scores, then five years old, and last six years old. When children watch the show three or five times a week (VIEWCAT = 3), children of six years old have the greatest median scores, then five years old, followed by four years old, and last are children of three years old. When children watch the show more than five times a week (VIEWCAT = 4), children of three years old have the greatest median scores, then is children of five years old, third is children of four years old, children of six years old have the lowest median scores. In Figure 4, there is the box plot of dfrationumb by VIEWCAT and AGE. When children rarely watch the show (VIEWCAT = 1), children of three years old have the greatest median scores, then is children of four years old, third is children of five years old, children of six years old have the lowest median scores. When children watch the show once or twice times a week (VIEWCAT = 2), children of three years old have the greatest median scores, then is children of six years old, children of five years old have the lowest median scores. When children watch the show three or five times a week (VIEWCAT = 3), children of four years old have the greatest median scores, then followed by five years old and six years old, three years old children have the lowest median scores. When children watch the show more than five times a week (VIEWCAT = 4), children of four years old have the greatest median scores, followed 6 by five years old, third is children of six years old, children of three years old have the lowest median scores. In Figure 5, there is the box plot of dfratiorel by VIEWCAT and AGE. When children rarely watch the show (VIEWCAT = 1), children of three years old have the greatest median scores, then is children of four years old, third is children of five years old, children of six years old have the lowest median scores. When children watch the show once or twice times a week (VIEWCAT = 2), children of six years old have the greatest median scores, then is children of four years old, children of five years old have the lowest median scores. When children watch the show three or five times a week (VIEWCAT = 3), children of three years old have the greatest median scores, five years old, four years old, lastly, six years old have the lowest median scores. When children watch the show more than five times a week (VIEWCAT = 4), children of three years old have the greatest median scores, then four years old, five years old, lastly, six years old have the lowest median scores. In Figure 6, there is the box plot of dfratioclasf by VIEWCAT and AGE. When children rarely watch the show (VIEWCAT = 1), children of three years old have the greatest median scores, then four years old, six years old, lastly, five years old have the lowest median scores. When children watch the show once or twice times a week (VIEWCAT = 2), children of four years old have the greatest median scores, then five years old, six years old have the lowest median scores. When children watch the show three or five times a week (VIEWCAT = 3), children of three years old have the greatest median scores, then four years old, six years old, lastly five years old have the lowest median scores. When children watch the show more than five times a week (VIEWCAT = 4), children of three years old have the greatest median scores, then five years old, followed by four years old, lastly, six years old have the lowest median scores. Then we split the data into four sections by watching frequency, and take analysis of variance of all sections. In the test of knowledge of body parts (see Table 10), when VIEWCAT = 1, the p-value in various ages are insignificant, it means 3- to 6-year-old children who rarely watch the show have a similar performance in the posttest. When VIEWCAT = 2 and VIEWCAT = 3, the p-value also insignificant. When VIEWCAT = 4, we can see the p-value between 6-year-old and 3-year-old is significant at 5%, and the scores of 6-year-old children is less than 3-year-old, the p-value between 5 and 4 year old is significant at 5%, and the scores of 5-year-old children is less than 4-year-old, the p-value between 6-year-old children and 4-year-old significant at 5%, and the scores of 6-year-old is less than 4-year-old. In the test of letters (see Table 11), when VIEWCAT = 1, the p-value between 5 and 4 year old is significant at 5%, it means the show have a significant impact between 5 and 4-year-old, and the scores of 5-year-old is less than 4-year-old.When VIEWCAT = 2, VIEWCAT = 3 and VIEWCAT = 4, all of the p-value are insignificant. In the test of forms (see Table 12), when VIEWCAT = 1, 2, 3, 4, the p-value are insignificant. In the test of numbers (see Table 13), when VIEWCAT = 1, 2, 3, the p-value are insignificant. When VIEWCAT = 4, the p-value between 6 and 4, 6 and 5 year old are significant at 10%. And the scores of 6-year-old is less than 4 and 5-year-old. In the 7 test of relation terms (see Table 14), when VIEWCAT = 1, the p-value between 5 and 3-year-old children, 5 and 3-year-old are significant, so the scores of 5-year-old children is significantly less than 4 and 3-year-old. When VIEWCAT = 2, 4, all of the p-value are insignificant. When VIEWCAT = 3, the p-value between 5-year-old children and 3-year-old children, 6 and 3-year-old are significant, so the scores of 3-year-old children is significantly greater than 5 and 6-year-old. In the test of classification (see Table 15), when VIEWCAT = 1, 2, 3, 4, all of the p-value are insignificant. From above, we can say there exist some significant difference between various age with watching frequency, and the show can improve the test scores greatly with little children. As the age increasing, the show will have a lesser impact on children’s cognitive skills. Children’s Age by Watching Frequency We classify each test score by children’s age and watching treatment condition, and then we take box plot of each test score by watching treatment condition and age. In Figure 7, there is the box plot of dfratiobody by VIEWENC and AGE. The horizontal axis is watching treatment condition, 1 means that children are encouraged to watch the show, 2 means that children are not encouraged to watch the show. The vertical axis is age, from 3 years old to 6 years old. When children are encourage to watch the show (VIEWENC = 1), children of three years old have the greatest median scores, then of four, followed by five years old, children of six years old have the lowest median scores. When children are encourage to watch the show (VIEWENC = 2), children of three years old have the greatest median scores, then four years old, five years old, children of six years old have the lowest median scores. So we can say, children who encouraged to watch the show with smaller age will get higher scores. In Figure 8, there is the box plot of dfratiolet by VIEWENC and AGE. When children are encourage to watch the show (VIEWENC = 1), children of three years old have the greatest median scores, then is children of four years old, third is children of five years old, children of six years old have the lowest median scores. When children are encourage to watch the show (VIEWENC = 2), children of four years old have the greatest median scores, then five years old, six years old, lastly, three years old have the lowest median scores. In Figure 9, there is the box plot of dfratioform by VIEWENC and AGE. When children are encourage to watch the show (VIEWENC = 1), children of three years old have the greatest median scores, then four years old, six years old, five years old have the lowest median scores. When children are encourage to watch the show (VIEWENC = 2), five years old have the greatest median scores, then four years, six years old, children of three years old have the lowest median scores. In Figure 10, there is the box plot of dfrationumb by VIEWENC and AGE. When children are encourage to watch the show (VIEWENC = 1), four years old have the greatest median scores, then five years old, three years old, finally, of six years old have the lowest median scores. When children are encourage to watch the show (VIEWENC = 2), children of four years old have the greatest median scores, then five years old, six years old, three years old have the lowest median scores. In Figure 11, there is the box plot of dfratiorel by VIEWENC and AGE. When 8 children are encourage to watch the show (VIEWENC = 1), three years old have the greatest median scores, then four years old, five years old, six years old have the lowest median scores. When children are encourage to watch the show (VIEWENC = 2), three years old have the greatest median scores, then four years old, third five years old, six years old have the lowest median scores. In Figure 12, there is the box plot of dfratioclasf by VIEWENC and AGE. When children are encourage to watch the show (VIEWENC = 1), children of three years old have the greatest median scores, then is children of four years old, third is children of five years old, children of six years old have the lowest median scores. When children are encourage to watch the show (VIEWENC = 2), children of three years old have the greatest median scores, then is children of four years old, third is children of six years old, children of five years old have the lowest median scores. Then we split the data into two sections by watching treatment condition, and take analysis of variance of all sections. In the test of knowledge of body parts (see Table 16), when VIEWENC = 1, the p-value in various ages are insignificant, it means 3- to 6-year-old children who encourage to watch the show have the similar performance in the posttest. When VIEWENC = 2, the p-value also insignificant. In the test of letters (see Table 17), when VIEWENC = 1 and 2, all of the p-value are insignificant. In the test of forms (see Table 18), when VIEWENC = 1 and 2, all of the p-value also insignificant, it means the factor age can not impact the scores obviously. In the test of forms (see Table 19), when VIEWENC = 1, the p-value are insignificant, when VIEWENC = 2, the p-value between 5 and 4-year-old children is significant, so the scores of 5-year-old children is significantly less than 4-year-old children. In the test of forms (see Table 20), when VIEWENC = 1, the p-value are insignificant, when VIEWENC = 2, the p-value between 4 and 3-year-old children, 5 and 3-year-old children, 6 and 3-year-old children are significant, and the scores of 3-year-old children is more than 4, 5 and 6-year-old children. In test of classification (see Table 21), when VIEWENC = 1 and 2, all of the p-value are insignificant. In summary, we can say there exist some significant difference between various age with watching treatment condition, and the children with little age encouraged to watch the show can get the greater scores. As the age increasing, the show will have a lesser impact on children’s cognitive skills. Conclusion As the analysis shows, we believe that the show is effective to children age from 3 to 6. For different tests, children improve differently regarding their age, sex, frequency of viewing, treatment condition, etc. As the paper tries to find the most important factors for children to gain more in the show, evidence is found that watching the show more frequently does not guarantee a higher increase in the test scores. For body knowledge and number tests, children’s score is highest for a moderate watching frequency (three to five times a week) and drops when watching the shows too frequently. This could be explained as that children may feel tired about those topics when they watch too much shows regarding those topics. For the test of letters and relation, children’s score is highest when they watch the show more than five times a week. When children watch more, 9 they will get more letters and learn more about relational terms. As for the classification skill, watching once or twice would have the best outcomes. This may happen since the topic is the most difficult one among those tests, it is hard for children to understand whatever he/she sees and need some time to understand. We also find it interesting that encouragement is also helpful on score on forms, relational terms, letters, classification and knowledge of body parts. Therefore, parents’ attitude toward the show can really make a big difference. This may happen because the show can be selected by parents, so when they watch, they also learn the way how adult think. Then we study each test score by children’s age with watching frequency and watching treatment condition. And we found that children with little age can get greater performance in the posttest. It may tell us that the show is more suitable to little children, and it is beneficial to encourage little children to watch the show. Appendix A: Tables and Figures Table 1: Variable description Variable Description Name 1= Three to five year old disadvantaged children from inner city areas in various parts of the country 2=Four year old advantaged suburban children site 3=Advantaged rural children 4=Disadvantaged rural children 5=Disadvantaged Spanish speaking children Gender of Child sex 1=male. 2=female. age Ages in Years Scale 1-5 1-2 3-6 10 Frequency of viewing 1=rarely watched the show 2=once or twice a week viewcat 3=three to five times a week 4=watched the show on average of more than 5 times a week Setting in which Sesame Street was viewed, setting 1=home 2=school Treatment condition viewenc 1=child encouraged to watch 2=child not encouraged to watch Mental age score obtained from administration of the peabody Peabody Picture Vocabulary test as a pretest measure of vocabulary maturity Difference ratio of pretest and posttest score on dfratiobody knowledge of body parts(%) Difference ratio of pretest and posttest score on dfratiolet letters(%) Difference ratio of pretest and posttest score on dfratioform forms(%) Difference ratio of pretest and posttest score on dfrationumb numbers(%) Difference ratio of pretest and posttest score on dfratiorel relational terms(%) Difference ratio of pretest and posttest score on dfratioclasf classification skills(%) Table 2: Descriptive Statistics Maximu variables N m site 239 5.00 sex 240 2.00 age 240 6.00 viewcat 240 4.00 setting 240 2.00 viewenc 240 2.00 peabody- dfratiobody- dfratiolet- dfratioform- dfrationum b- Minimu m- -45.00 -100.00 -100.00 Mean Std - -100.00 66.378 - 1-4 1-2 1-2 27-99 (-45,288) (-100,457) (-100,750) (-100,1200) (-100,400) (-55,1100) Skewnes s 0.253 -0.075 -0.637 - Kurtosi s -0.982 - -1.325 -1.849 - 5.973 58.838 11 dfratiorel dfratioclasf 240 240 - -100.00 -54.55 - Table 3: Test of Normality Kolmogorov-Smirnov(lilliefors) outcomes df D statistic Sig dfratiobody-E-09 dfratiolet-E-09 dfratioform-E-16 dfrationumb-E-16 dfratiorel-E-16 dfratioclasf-E-16 - df- - - Shapiro-Wilk W statistic Sig-E-E-E-E-E-E-16 Table 4: Test of Mulitcollinearity VIF SITE 1.164 SEX 1.017 AGE 1.256 VIEWCAT 1.253 SETTING VIEWENC- Table 5: Regression Results Outcome Predictor coeffecience Std.Error (Intercept- SITE- SEX - AGE - dfratiobody VIEWCAT- SETTING - VIEWENC - PEABODY - (Intercept- SITE - SEX- AGE - dfratiolet VIEWCAT- SETTING- VIEWENC - PEABODY- (Intercept- SITE- SEX- AGE - dfratioform VIEWCAT- SETTING - VIEWENC - PEABODY - t value- -1.111 - -1.221 -0.136 - - - - - -1.707 -0.402 -2.945 PEABODY 1.168 Pr(>|t|- *** . *** ** ** *** * . ** 12 dfrationumb dfratiorel dfratioclasf (Intercept) SITE SEX AGE VIEWCAT SETTING VIEWENC PEABODY (Intercept) SITE SEX AGE VIEWCAT SETTING VIEWENC PEABODY (Intercept) SITE SEX AGE VIEWCAT SETTING VIEWENC PEABODY - -9.423 - - - -2.5067 -2.6322 - - - -8.4558 -7.9396 -0.5304 - - -0.718 - - - -0.291 -0.309 - - - -0.65 -0.618 -1.395 - ** * . ** ** * * *** ** Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Table 6: The mean of each test score by watching frequency VIEWCAT VIEWCAT1 VIEWCAT2 VIEWCAT3 VIEWCAT4 total dfratio body- dfratio let- dfratio form- dfratio numb- dfratio rel- dfratio clasf- Table 7: regression by VIEWCAT Outcome dfratiobody Predictor (Intercept) SITE SEX AGE coeffecience- -5.1745 -6.7726 Std.Error- t value- -1.0940 -1.4630 Pr(>|t|) 3.46E- *** . 13 dfratiolet dfratioform dfrationumb dfratiorel VIEWCAT1 VIEWCAT2 VIEWCAT3 VIEWCAT4 SETTING VIEWENC PEABODY (Intercept) SITE SEX AGE VIEWCAT1 VIEWCAT2 VIEWCAT3 VIEWCAT4 SETTING VIEWENC PEABODY (Intercept) SITE SEX AGE VIEWCAT1 VIEWCAT2 VIEWCAT3 VIEWCAT4 SETTING VIEWENC PEABODY (Intercept) SITE SEX AGE VIEWCAT1 VIEWCAT2 VIEWCAT3 VIEWCAT4 SETTING VIEWENC PEABODY (Intercept) SITE SEX -6.8872 - . -6.4772 -0.6935 - - -30.1463 -72.2494 -43.1194 -1.2194 . 5.0177 - - -10.8363 -0.1376 . -21.1993 -8.1758 - -8.5173 -30.1964 -41.0285 - - - - - - - - -0.8880 - . -1.1920 -0.1230 - - -2.728 -3.905 -2.722 -0.079 . 0.387 - - -0.72 -0.009 . -1.721 -0.637 - -0.647 -2.344 -1.902 - - - - -E-05 1.93E- - - -E- *** *** ** *** ** * . ** ** * . ** *** 14 dfratioclasf AGE VIEWCAT1 VIEWCAT2 VIEWCAT3 VIEWCAT4 SETTING VIEWENC PEABODY (Intercept) SITE SEX AGE VIEWCAT1 VIEWCAT2 VIEWCAT3 VIEWCAT4 SETTING VIEWENC PEABODY -15.9244 -23.1788 -19.9472 -20.2932 . -3.2228 -4.5768 - - -15.5513 - -6.4366 . -8.3349 -4.2074 -0.5622 - - - -2.16 -1.878 -1.887 -1.96 . -0.373 -0.508 -3.69 2.81 - -1.397 - -0.412 . -0.638 -0.309 -1.472 - - - * . . . *** ** Table 8: the mean of each test score by treatment condition VIEWENC VIEWENC 1 VIEWENC 2 total dfratiobod y dfratiole t dfratioform dfrationum b dfratiore l dfratioclasf 27.5861 97.8841 61.5468 61.8187 33.3224 47.6634 22.1307 66.1375 48.7679 74.3438 24.0447 37.4489 25.6003 86.3278 56.8950 66.3781 29.9452 43.9452 Table 9: regression by VIEWENC Outcome dfratiobody dfratiolet Predictor (Intercept) SITE SEX AGE VIEWCAT SETTING VIEWENC1 VIEWENC2 PEABODY (Intercept) SITE SEX AGE Estimate- -5.2296 - - NA - - -29.3332 Std- NA- Error- -1.111 - - NA - - -2.653 t value- NA 2.34E- Pr(>|t|) *** . *** * ** 15 dfratioform dfrationumb dfratiorel dfratioclasf VIEWCAT SETTING VIEWENC1 VIEWENC2 PEABODY (Intercept) SITE SEX AGE VIEWCAT SETTING VIEWENC1 VIEWENC2 PEABODY (Intercept) SITE SEX AGE VIEWCAT SETTING VIEWENC1 VIEWENC2 PEABODY (Intercept) SITE SEX AGE VIEWCAT SETTING VIEWENC1 VIEWENC2 PEABODY (Intercept) SITE SEX AGE VIEWCAT SETTING VIEWENC1 VIEWENC2 PEABODY - NA- - - NA - -9.423 - -17.2185 NA - - - NA - - - - NA -0.5304 - NA- NA- NA- NA- NA 0.3802 - NA- - - NA - -0.718 - -1.158 NA - - - NA - - - - NA -1.395 Figure 1: Box plot of dfratiobody by VIEWCAT and AGE 1.79E- NA- NA- NA- NA- NA 0.16437 *** * . ** *** * . ** ** * * *** ** 16 Figure2: Box plot of dfratiolet by VIEWCAT and AGE Figure3: Box plot of dfratioform by VIEWCAT and AGE Figure4: Box plot of dfrationumb by VIEWCAT and AGE 17 Figure5: Box plot of dfratiorel by VIEWCAT and AGE Figure6: Box plot of dfratioclasf by VIEWCAT and AGE 18 Table10: Analysis of variance in dfratiobody by VIEWCAT and AGE VIEWCAT=1 VIEWCAT=2 VIEWCAT=3 VIEWCAT=4 AGE- diff -30.3055 -22.2134 - -18.1997 -37.5715 - - -23.3076 - -24.0522 -42.5294 -60.7065 -18.4772 -36.6543 -18.1772 lwr - - - -30.4301 -80.5375 -84.6924 -40.7802 - - - - - -58.8592 -81.2553 -54.3399 -70.0194 -86.7486 - -36.9036 -71.0169 -50.1638 upr- -7.8237 -0.0508 - p adj- Table11: Analysis of variance in dfratiolet by VIEWCAT and AGE VIEWCAT=1 VIEWCAT=2 VIEWCAT=3 VIEWCAT=4 AGE- diff -26.9414 -86.5848 -87.9076 -59.6434 -60.9662 - -21.3787 - -79.3122 -49.4459 -76.0186 lwr - - - - - - -47.6898 - - - - - -92.7514 - - - - upr- - p adj- - - -26.5726 - -91.6941 - - - - - - Table12: Analysis of variance in dfratioform by VIEWCAT and AGE VIEWCAT=1 VIEWCAT=2 VIEWCAT=3 VIEWCAT=4 AGE- diff- -27.6839 - -4.6202 -55.4762 - -32.4364 - -15.3880 -44.5271 -80.0943 -29.1391 -64.7063 -35.5671 lwr - - - - - - -41.7052 - - - - - - - - - - - -75.6675 - - upr- p adj- Table13: Analysis of variance in dfrationumb by VIEWCAT and AGE VIEWCAT=1 VIEWCAT=2 VIEWCAT=3 AGE- diff 60.1816 -63.0734 -80.8333 - - -17.7599 -46.5515 - lwr - - - - - - - - - - - - upr- p adj- 20 VIEWCAT=4 - -14.2858 -56.3341 - -28.8604 -11.5072 -81.4056 -69.8984 -75.4533 - - -59.6842 -66.9237 - -56.4954 - - - - Table14: Analysis of variance in dfratiorel by VIEWCAT and AGE VIEWCAT=1 VIEWCAT=2 VIEWCAT=3 VIEWCAT=4 AGE- diff -70.7623 - - -60.3221 -73.4044 -13.0823 - -94.4769 -97.8871 - -3.4102 -23.6291 -20.2190 -51.9011 -44.5028 - -32.3024 -39.7006 lwr - - - - - - -57.7910 - -99.7125 - - - -34.0245 -90.5463 -84.0294 - - - -50.2227 - - upr 53.0449 - - p adj- Table15: Analysis of variance in dfratioclasf by VIEWCAT and AGE VIEWCAT=1 VIEWCAT=2 AGE- diff -23.8001 -66.1010 -50.8333 -42.3009 - -58.8742 lwr - - - -95.5460 - - - upr- p adj- 21 VIEWCAT=3 VIEWCAT=4 - - -44.5169 -81.5966 -79.3021 - -14.3774 -16.6719 -3.5651 -6.4738 -39.7464 -2.9087 -36.1813 -33.2726 - - - - - -37.2882 - -99.1753 - - - -42.4994 - - - Figure7: Box plot of dfratiobody by VIEWENC and AGE Figure8: Box plot of dfratioletby VIEWENC and AGE - 22 Figure9: Box plot of dfratioform by VIEWENC and AGE Figure10: Box plot of dfrationumb by VIEWENC and AGE Figure11: Box plot of dfratiorel by VIEWENC and AGE 23 Figure12: Box plot of dfratioclasf by VIEWENC and AGE Table16: Analysis of variance in dfratiobody by VIEWENC and AGE VIEWENC VIEWENC=1 VIEWENC=2 AGE- diff -22.3979 -39.6680 -40.3574 -17.2701 -17.9594 -0.6894 -15.1736 -21.8684 -31.2353 -6.6948 -16.0617 -9.3669 lwr -99.1468 - - -36.5036 -61.1555 -42.2059 -67.5427 -71.8893 - -28.4894 -68.4308 -59.3877 upr- p adj- Table17: Analysis of variance in dfratiolet by VIEWENC and AGE VIEWENC VIEWENC=1 VIEWENC=2 AGE- diff -77.9398 -98.0086 - -20.0688 -54.6144 - -0.5174 -73.6217 -22.9202 -96.0245 -73.1043 lwr - - - -61.8002 - - - - - -88.9321 - - upr- p adj- 24 Table18: Analysis of variance in dfratioform by VIEWENC and AGE VIEWENC VIEWENC=1 VIEWENC=2 AGE- diff -8.9701 -40.2679 -53.9966 -31.2978 -45.0265 - -2.0114 - lwr - - - -75.5492 - - -63.6484 -61.2492 -95.1318 -42.9468 -98.4910 -92.0690 upr- p adj- Table19: Analysis of variance in dfrationumb by VIEWENC and AGE VIEWENC VIEWENC=1 VIEWENC=2 AGE- diff- -20.2272 -48.5917 - -6.6451 -71.9281 - - -65.2830 lwr -73.2040 -91.9293 - -51.2012 - -95.2235 - - - - - - upr- - Table20: Analysis of variance in dfratiorel by VIEWENC and AGE VIEWENC VIEWENC=1 VIEWENC=2 AGE- diff- -20.2272 -48.5917 -28.3645 -98.1097 - - -18.4057 lwr -73.2040 -91.9293 - -51.2012 - -95.2235 - - - -47.6997 upr- -27.7206 -49.2825 - p adj- p adj- 25 6-4 6-5 -48.0505 -29.6448 - -96.8776 - - Table21: Analysis of variance in dfratioclasf by VIEWENC and AGE VIEWENC VIEWENC=1 VIEWENC=2 AGE- diff 16.2443 -10.5491 -36.5496 -26.7934 -52.7939 -26.0005 -45.5056 -68.4081 -71.3131 -22.9025 -25.8075 -2.9050 lwr - - - -74.0320 - - - - - -61.5738 - -91.6596 upr- p adj-