Defense Acquisition Research Journal Issue 95

Use of Factors in Development Estimates: Improving the Cost Analyst Toolkit

https://www.dau.edu

seven WBS elements (not shown) rejected the null hypothesis, necessi tating nonparametric testing throughout the remainder of the analysis. Nonparametric testing identifies similarities of locations in the data ele ments analyzed. Histograms of the data in this analysis reveal a consistent right-skewed profile. Due to the similarities in the shape of the histograms, the nonparametric tests can be considered to be testing medians (Hollander et al., 2014). Therefore, subsequent discussion of nonparametric results will focus on differences in the medians of the data. Commodity Type The first category analyzed is commodity type. The Kruskal-Wallis test reveals statistical differences between WBS element median values (Table 5). Specific differences are identified within the SE/PM, ST&E, and Site Activation WBS elements.

TABLE 5. KRUSKAL-WALLIS RESULTS FOR COMMODITY TYPE

Null Hypothesis Test Result

WBS Element

Alpha

N

Chi-Square

P -value

SE/PM

0.05

406 49.2441

<0.0001

Reject

ST&E

0.05

374 32.3203

<0.0001

Reject

Training

0.05

192

6.9636

0.2234

Do Not Reject

Data

0.05

267

6.1052

0.2961

Do Not Reject

PSE

0.05

149

2.2603

0.8121

Do Not Reject

CSE

0.05

50

1.0203

0.9609

Do Not Reject

Site Activation

0.05

47

14.4899

0.0059

Reject

Spares

0.05

84

3.7434

0.2905

Do Not Reject

TABLE 6. COMMODITY DIFFERENCES SUMMARY

Electronic/ Automated Software

Aircraft

Missile Ordnance Space UAV

SE/PM

2

1

1

0

0

0

ST&E

2

1

1

0

3

1

Site Activation

1

1

0

0

0

0

After determining that statistical differences exist, the Steel-Dwass mul tiple comparison test is employed to identify which commodity types exhibited differences. The identification of differences through the statisti cal tests tells the analyst that utilizing the more readily available aggregated factors is ill-advised. Rather, it indicates that the analysts should take more

52

Defense ARJ, January 2021, Vol. 28 No. 1 : 40-70