Question:

1. How well can programmers predict the number of days required to complete a large project? At the start of one of these large projects1 an investigator took 24 programmers and asked them to predict how many programmer days would be required. After the project was fin- ished, the prediction errors were calculated (actual minus predicted programmer days). The investigator is interested in the effect of type of programmer experience {small systems only. both small and large systems) or amount of experience {less than 5 years. 5-10 years. more than 10 years}. The data is given below. Factor 3 (years of experience) Factor A ,1 = 1 j = 2 J = 3 (type of experience) Under 5 Sunder 10 10 or more ." = 1 Small 240 110 56 systems only 206 1 13 60 213' 103 63 225 95 58 i = 2 Small and 3'1 4? 3? large systems 53 52 33 68 31 40 5? 49 45 (a) Use FROG GLM to run a one-way ANOVA and answer the following questions: i. Does Factor A (type) have a signicant effect? Report the F-statistic and p-value of your test. ii. Describe the effect of Factor A on the response by interpreting the output of LSMEANS statement. (b) Use PROC GLM to run a two-way ANOVA without interaction and answer the following questions: i. Does Factor A (type) have a signicant effect? Report the F-statistic and p-value of your test. ii. Describe the effect of Factor A on the response by interpreting the output of LSMEANS statement. (c) Use FROG GLM to run a two-way ANOVA with interaction and answer the following questions: i. Do the two factors interact? ii. Does Factor A (type) have a signicant effect? Report the Fstatistic and pValue of your test. iii. Describe the effect of Factor A on the response by interpreting the output of LSMEANS statement