An article in Communications of the ACM (Vol. 30, No. 5, 1987) studied different

ID: 3062937 • Letter: A

Question

An article in Communications of the ACM (Vol. 30, No. 5, 1987) studied different algorithms for estimating software development costs. Six algorithms were applied to several different software development projects and the percent error in estimating the development cost was observed. Some of the data from this experiment is shown in the table below.

Which factor is the nuisance factor?

Does it have a significant impact on the response(i.e., test the significance of the nuisance factor)?

Project 1(SLIM 2(COCOMO-A) 281 129 396 1306 336 910 3(COCOMO-R) 220 84 458 543 300 794 4(COCONO-C) 225 83 425 552 291 826 5(FUNCTION POINTS) 6(ESTINALS) -20 35-53 170104 199 1244 21 82 2221 905 839 19 1-3412 15 103

Explanation / Answer

ANOVA: Single Factor

SUMMARY

Algorithms

Count

Sum

Average

Variance

SLIM

5312

885.3333

661519.5

COCOMA-A

3358

559.6667

203937.9

COCOMA-R

2399

399.8333

64260.97

COCOMA-C

2402

400.3333

69639.87

FUNCTIONPOINTS

235

39.16667

3581.767

ESTIMALS

435

72.5

10442.7

ANOVA

Source of Variation

P-value

F crit

Between Groups

2989130

597826.1

3.539588

0.012384

2.533555

Within Groups

5066913

168897.1

Total

8056044

The hypothesis under testing is

H0: % Error in estimating software development cost is same in all algorithms

HA: % Error in estimating software development cost may be different in different algorithms.

The p-value = 0.012384 < 0.05 implies that H0 may be rejected means there is evidence to conclude that % Error in estimating software development cost may be vary with algorithm.

Further it is highest in SLIM whereas least in FUNCTION POINTS.