Jana Price, Ph.D.

1
Jana Price, Ph.D.

• Statistical End Play Data Study

2
Estimating Reliability of The On-Wing End Play
Measure

• Reliability the ability of the on-wing end play
  measurement to yield repeatable results
• Measures taken on consecutive occasions were
  correlated
• The analysis data consisted of 852 paired
  measures (Time 1 and Time 2)

3
Measures of Reliability

• Correlation coefficients (r) range from 1.0 to
  1.0
• An r value near 1.0 would suggest high
  measurement reliability
• Lower r values indicate less measurement
  reliability

4
Plot of Raw End Play Data at Time 1 and Time 2
r 0.553
5
End Play Differences Between Time 1 and Time 2
6
Estimating Validity of The On-Wing End Play
Measure

• 157 jackscrew assemblies were returned to
  Integrated Aerospace (IA)
• IA performs a bench-check
• Measures of on-wing end play and bench-checks
  were available for 63 jackscrew assemblies
• NTSB analysis correlated on-wing and bench check
  end play measures

7
Differences Between On-Wing and Bench-Check End
Play
r 0.442
8
Conclusions

• Repeated on-wing measures show low reliability
• Reliability is a necessary condition for an
  accurate measure
• Validity analysis using on-wing and bench-check
  measures showed low validity
• Based on these conclusions, NTSB recommended
  changes to measurement procedures

9
(No Transcript)
10
Recommendation A-01-042

• Revise the end play check procedure to minimize
  the probability of measurement error and conduct
  a study to empirically validate the revised
  procedure against an appropriate physical
  standard of actual acme screw and acme nut wear.
  This study should also establish that the
  procedure produces a measurement that is reliable
  when conducted on-wing