A computer software organization was encountering serious problems in predicting the man-day requirements for implementing large-scale programming projects. As part of a study on how to improve predictions. the company asked 24 programmers to predict the number of man-days for a large project about to be undertaken. The programmers were given full details about the specifications for the project. The programmers were classified by type of experience and amount of experience, and after the project was completed the forecasting errors were obtained. The forecasting errors (man-days predicted minus actual man-days) were:

State the model you will use to analyze the effects of amount and type of experience on the forecasting error.

Test whether the factors have any main effects and whether they interact. Use a level of significance of .05 for each test. Write your ANOVA table. Present all the confidence intervals and summarize your results.

Type of Experience Small-scale systems only Large-scale systems Under 2 years 2-5 years More than 5 years 278 46 196 92 58 58 241 188 83 Amount of Experience 68 57 109 118 88 96 47 52 31 49 38 33 42 31