|Back to Three-Mode Data Sets |

Information on the Parker & Fleishman Complex Tracking Behaviour Data


The complex tracking behaviour experiment by Parker and Fleishman (1960) was a forerunner of the complex flight simulators that are now in existence. A tracking device was constructed "to simulate roughly the display characteristics and control requirements of an airborne radar intercept mission". A control system involving a control stick and rudder pedals as in a standard aircraft control system was manipulated by the subject to control a target dot in the centre of a cathode ray oscillograph, and a needle of a voltmeter was used a side-slip indicator. The 203 subjects had to defeat an induced movement of the dot so as to keep the dot in the centre of the oscillograph. Side-slip error occurred when the subject did not co-ordinate the movement of the stick and the rudder pedals. Measures of horizontal error, vertical error, side-slip error and time-on-target were obtained.

Ten segments of the learning curve were selected by Parker and Fleishman to represent successive stages in learning. The error scores were reversed to produce accuracy measures. Thus there were 4 measurements for each of 10 stages of practice and together they formed a 40x 40 so-called multimode covariance matrix.

Description of the data sets

The file TRACOVAR.ZIP contains a zipped 50x50 covariance matrix as a lower triangular matrix calculated from the published correlations and the standard deviations. Most of the latter were approximately 2. Because the original correlations were only available in two decimals, the covariance matrix is singular.

The file TRACSMOO.ZIP contains a zipped smoothed 40x40 covariance matrix (the smallest eigenvalue was set at .1 and subsequently the covariance matrix was reconstituted). Differences with the original covariance matrix are virtually unnoticeable. The variable Integrated Error Score is not included in the covariance matrix to avoid further singularities (see NOTE below).
NOTE: The Integrated Error Score (IES) is a linear combination of the Horizontal Error Score (HES), the Vertical Error Score (VES), and the SideSlip Control (SSC): IES = 1/2 HES + 1/2 VES + SSC (see Parker and Fleishman, 1960, p. 5)

Variable x Trials (Measure -x- Stages)

(Integrated Error Score 1 through 10)
Horizontal Error Score 1 through 10
Vertical Error Score 1 through 10
Sideslip Error Score 1 through 10
Time-On-Target Score 1 through 10

A large number of background variables are available, but they are not included in the data sets made available hree. They can be found in the original publication or can be supplied upon request.


[Data: Original covariance matrix: TRACOVAR.ZIP]    [Data: Smoothed nonsingular covariance:TRACSMOO.ZIP]


Data are contained in:
Parker, J.F., Jr. & Fleishman, E.A. (1960). Ability factors and component performance measures as predictors of complex tracking behavior. Psychological Monographs, General and Applied, 74(16, Whole 503)

A three-mode analysis of the multimeasure-multistage matrix is contained in:
Tucker, L.R. (1967). Three-mode factor analysis of Parker- Fleishman complex tracking behavior data. Multivariate Behavioral Research, 2, 139-152.

A new analysis of these data with both three-mode principal component analysis and three-mode factor analysis is contained in:
Kroonenberg, P.M., & Oort, F.J. (2003). Three-mode analysis of multimode covariance matrices. British Journal of Mathematical and Statistical Psychology, 56,305-336.
[Colour figures from the Kroonenberg-Oort (2003) paper]

| Centre for Child & Family Studies | Leiden Institute of Education and Child Studies | | The Three-Mode Company | Three-Mode Data Sets | TOP |
P.M. Kroonenberg
Leiden Institute of Education and Child Studies, Leiden University
Wassenaarseweg 52, 2333 AK Leiden, The Netherlands
Tel. *-31-71-5273446/5273434 (secr.); fax *-31-71-5273945
E-mail:kroonenb at fsw.leidenuniv.nl

First version (month/day/year): 04/19/1999;