2.20. A student asks whether R' is a point estimator of any parameter in the normal error regression #

Transcription

1 92 Part One Simple Linear Regression c. Management wishes to estimate the expected service time per copier on calls in which six copiers are serviced. Obtain an appropriate 90 percent confidence interval by converting the interval obtained in part (a). lnterpret the converted confidence interval. d. Determine the boundary values of the 90 percent confidence band for the regression line when X,, = 6. Is your confidence band wider at this point than the confidence interval in part (a)? Should it be? *2.15. Refer to Airfreight breakage Problem I.21. a. Because of changes in airline routes, shipments may have to be transferred more frequently than in the past. Estimate the mean breakage for the following numbers of transfers: X = Use separate 99 percent confidence intervals. Interpret your results. b. The next shipment will entail two transfers. Obtain a 99 percent prediction interval for the number of broken ampules for this shipment. lnterpret your prediction interval. c. In the next several days, thee independent sh~pments will be made, each entailing two transfers. Obtain a 99 percent prediction interval for the mean number of ampules broken in the three shipments. Convert this interval into a 99 percent prediction interval for the total number of ampules broken injhe three shipments. d. Determine the boundary values of the 99 percent confidence band for the regression line when Xh = 2 and when Xh = 4. Is your confidence band wider at these two points than the elapsed time of 30 hours. Interpret your confidence interval. b. Obtain a 98 percent prediction interval for the hardness of a newly molded test item with an elapsed time of 30 hours. c. Obtain a 98 percent prediction interval for the mean hardness of 10 newly molded test items, each with an elapsed time of 30 hours. d. Is the prediction interval in part (c) narrower than the one in part (b)? Should it be? e. Determine the boundary values of the 98 percent confidence band for the regression line when Xh = 30. 1s your confidence band wider at this point than the confidence interval in part (a)? Should it be? An analyst fitted normal error regression model (2. I ) and conducted an F test of PI = 0 versus PI # 0. he P-value of the test was.033, and the analyst concluded H,: PI # 0. Was the a level used by the analyst greater than or smaller than.033? If the a level had been.ol, whar would have been the avvrovriate conclusion? A student asks whether R' is a point estimator of any parameter in the normal error regression # Using the normal error regression model (2.1) in an engineering safety experiment, a researcher found for the first 10 cases that R' was zero. IS it possible that for the complete set of 30 cases R2 will not be zero? Could R2 not be zero for the first 10 cases, yet equal zero for all 30 cases? Explain.

2 Chapter 2 II!~~.I~,II<.<,.v ilr /2(:q1.(,s.\.io11 cllrtl C~II.I-(,/CII~(III AII(I/I.V~.S Rel'er to Grade puint average Problem I. 19. a. Set up the ANOVA tahlc. h. What ix cxtiniatccl by MSI\' ill your ANOVA table? by MSE? Under what condition do MSK ancl MSE estimate the sanic quantity? c. Co~~cluct all 1.' test ol'wlicthcr 01. not PI = 0. Control the u risk at.01. State the alternatives. clccisio~~ IXIIC. ;111cl co~icl~~sion. el. Whal is (lie ahsolute magnitude of tlie reduction in the variation of Y when X is introcluced inlo the rc~rcsxion moclel? What is the relative reduction? What is the name of the latter ~ncasurc'! c. Ohtain 1. ancl attach the nppropriate sign. I. Wiic ~iic~x~rc. R or I. i s tic ~orc c r - I ~ I Iiieretion I Explain. ::'7_.24. Rel'cr to Copier ~ ~~ai~~tel~ni~cc Pl.ohlc a. Set L I tlic ~ h;~sic ANOVA table ill the li)rnlat ol't;~l>lc 7.2. Which clc~nc~its ol'yo~~r table are additive? Also set 111) the ANOVA t;~hlc in tlic form;~t oftahle 2.3. How do the two tables differ? I,. Co~icI~~ct ;II~ 1" ~CSI to ~Icter~iii~ic WIICIIIC~ or not there is a linear absocii~tion between time spent alicl numher ofcopiers scrvicecl: LIS~ u =. 10. Stale the alternatives. decision r~~le. and co~~clusio~~. c. I3y how ~iiucli. relatively. is tlic total variation in nuliiher ol'~iiin~~tcb s11en1 011 c;iii re~l~~cccl wlic11 the ~i~~~~iherol'col>icrb servicc~l is introd~cecl intotllc ;111;1Iysis'? IS this i~ relatively srn~~ll or I;~lpc rccluction'? Wli;~t is rhc name ol'tliis me:~sure'? el. C;~le~~l;~te 1. ;11icI ~~tt;~cli the ;~pp~.opri:~tc sign. c. Which mcilsurc. I. or K'. has the niorc clear-cut opcn~tional intcrpl-ct;\tion'? '!'7.75. IIcI'cr 10 Ail-fl-eigl~t I,~.eiakagc Problc~n 1.?I. a. Set LIIJ tlic ANOVA table. Which elements are ;1dclitive? h. Concluct an F Icst to clcciclc wlictlic~or not tlicrc is a linciu. ;~ssociatio~i between tlie nl~~nber ol'ti~iics a carton is t~.a~isl'errccl i111cl tlie nun~berol'broke~i anipules: control tlie cu risk nt.o.5. S~i~tc tlic ;~lte~.~i:~tives. clccisio~i I.LIIC, i111ci co~icl~~sio~~. c. Obtain tlie I! statistic for the test in past (b) and elenionstrate nunierically its eclnivalence to the F" slatistic obtainccl in part (b). el. Calculate K' anel 1.. What ~~roporliol~ ol'tlie variation in Y is ucco~~nted for by introducing 0 X into the regl.essio11 moclel? Relkr to Plastic hardness Prublcm ;I. Set LIP tlie ANOVA table. b. Test by means of ;ui F test whether or not there is a linear association between the hardness of tlie plastic ancl tlie elapsecl time. Use u =.01. State the alternatives. decision r~~le. and conclusion. c. Plot tlie deviations Y; -?; against Xi on ;I graph. Plot the deviations?; - Y against Xi on another graph. 11si11: the same scales as for the ti rst praph. From your two ~raphs, does SSE or SSR appear to be the larger component of SSTO? What does this imply about the mag~iitude of R'? d. Calculate R' and I.. %27. Refer to Muscle mass Problem a. Conduct a test todecide whether or not there is a negative linear association between amount of muscle mass and age. Control the risk of Type 1 error at.05. State the alternatives. decision r~~le. and conclusion. What is the P-value of the test'?

3 148 Part One Simple Linear Regression d. Plot the residuals ei against Xi to ascertain whether any departures from regression model (2.1) are evident. What is your conclusion? e. Prepare a normal probability plot of the residuals. Also obtain the coefficient of correlation between the ordered residuals and their expected values under normality to ascertain whether the normality assumption is reasonable here. Use Table B.6 and a =.01. What do you conclude? f. Prepare a time plot of the residuals. What information is provided by your plot? g. Assume that (3.10) is applicable and conduct the Breusch-Pagan test to determine whether or not the error variance varies with the level of X. Use (Y =.lo. State the alternatives, decision rule, and conclusion. Does your conclusion support your preliminary findings in a. Obtain the residuals ei and prepare a box plot of the residuals. What information is provided by your plot? b. Plot the residuals ei against the fitted values pi to ascertain whether any departures from regression model (2.1) are evident. State your findings. c. Prepare a normal probability plot of the residuals. Also obtain the coefficient of correlation between the ordered residuals and their expected values under normality. Does the normality assumption appear to be reasonable here? Use Table B.6 and (Y =.05. *3.7. Refer to Muscle mass Problem c. Plot the residuals e, against?; and also against Xi on separate graphs to ascertain wh any departures from regression model (2.1) are evident. Do the two plots provide the information? State your conclusions. d. Prepare a normal probability plot of the residuals. Also obtain the coefficient of correla between the ordered residuals and their expected values under normality to ascertain whe or not the error variance varies with the level of X. Use (Y =.01. State the alte decision rule, and conclusion. Is your conclusion consistent with your preliminary in part (c)? 3.8. Refer to Crime rate Problem a. Prepare a stem-and-leaf plot for the percentage of individuals in the county having at 1 a high school diploma Xi. What information does your plot provide? b. Obtain the residuals ei and prepare a box plot of the residuals. Does the distribution of residuals appear to be symmetrical? I I i

4 150 Part One Simple Linear Regression *3.13. Refer to Copier maintenance Problem.-I.20. a. What are the alternative conclusions when testing for lack of fit of a linear regression function? b. Perform the test indicated in part (a). Control the risk of Type I error at.05. State the decision rule and conclusion. c. Does the test in part (b) detect other departures from regression model (2.1), such as lack of constant variance or lack of normality in the error terms? Could the results of the test of lack of fit be affected by such departures? Discuss Refer to PIastic hardness Problem a. Perform the F test to determine whether or not there is lack of fit of a linear regression function; use cr =.01. State the alternatives, decision rule, and conclusion. b. Is there any advantage of having an equal number of replications at each of the X levels? Is there any disadvantage?. c. Does the test in part (a) indicate what regression function is appropriate when it leads to the conclusion that the regression function is not linear? How would you proceed? Solution concentration. A chemist studied the concentration of a solution (Y) over time (X). Fifteen identical solutions were prepared. The 15 solutions were randomly divided into five sets of three, and the five sets were measured, respectively, after 1, 3, 5, 7, and 9 hours. The results follow Yi: a. Fit a linear regression function. b. Perform the F test to determine whether or not there is lack of fit of a linear regression function; use cr =.025. State the alternatives, decision rule, and conclusion. c. Does the test in part (b) indicate what regression function is appropriate when it leads to the conclusion that lack of fit of a linear regression function exists? Explain Refer to Solution concentration Problem a. Prepare a scatter plot of the data. What transformation of Y might you try, using the prototy patterns in Figure 3.15 to achieve constant variance and linearity? b. Use the Box-Cox procedure and standardization (3.36) to find an appropriate powe transformation. Evaluate SSE for,i = -.2, -. 1,0,.I,.2. What transformation of Y i suggested? c. Use the transformation Y' = log,, Y and obtain the estimated linear regression function the transformed data. d. Plot the estimated regression line and the transformed data. Does the regression line ap to be a good fit to the transformed data? e. Obtain the residuals and plot them against the fitted values. Also prepare a normal probabi plot. What do your plots show? f. Express the estimated regression function in the original units. 10 years ago. The data are as follows, where X is the year (coded) and Y is sales in thou

5 X,: Y,: :I. Prepare u scatter plot of the dnta. Does a linear relation appear adeq~~ate here? b. Use the Box-Cox procedul- *id standardiz:~tion (3.36) to tind an appropriate power trnnsforination of Y. Evaluate 4S.A or 1 =.3,.4..5, What tfiuisformation of Y is suggested? c. Use the t~.anstor~nation Y' = fi and obtain the estimatetl linenrregrcssion function for the t~msformecl dam. d. Plot the estimated ~.cgression line and the transformed data. Does the regression line appear to be a good ti1 to the transformed data? e. Obtain the residu:lls:uid plot them against the titted values. Also prepare anormnl probability plot. What do yo~~r plots show'? t'. Express the estimated regression function in the original units Production time. In ;I ~n;uiuh~cturing study, the procluction times l'or I I I recent production runs were obtained. The table below lists forcach run the production time in IIOLII.S (Y);111d [he production lot size (X). E Y,: b. Use the trnnsfor~natio~i X' = and obtain the estimated linear regression function for the ( i. >.. trnnsfor~ne data. z:~. j.;: I i c. Plot the estimated regression line and the transformed data. Does (lie regression line appear. i s i-~ +.>. to be a good tit to thc transformed data? % d. Obtain the residuals ;uid plot them against the fitted values. Also prepare :I normill probability plot. What do your plots show? e. Express the estimated regression function in the original ~~nits..; : k:: I I I-I.~ A student ti tted a linear regression filnction for a class assignment. The student plotted the residuals e, against Yj and found a positive relation. When the residuals were plotted against the fitted values p;. the student found no relation. How could this difference arise? Which is the more meaningful plot? i.20. If the error terlns in a regression model are independent N(0, a'), what can be said about the error terms after transformation X' = I / X is used? Is the situation the same after transformation Y' = I/ Y is used? ;.?I. Derive the result in (3.29).