Applying Nonlinear Causal Resource Analysis Methods to Incorporation of
Personality Traits in Performance Models for Education and Work
Tara Alavi, George V. Kondraske, Ph.D., Advisor
Department of Electrical Engineering, The University of Texas at Arlington, Arlington, Texas 76019
Abstract
Materials and Methods
General Systems Performance Theory (GSPT) is a novel framework for modeling
systems, tasks, and their interface from a performance perspective. Derived from GSPT,
Nonlinear Causal Resource Analysis (NCRA) is a task analysis and performance
modeling methodology. NCRA provides a way to estimate the degree of performance in
a higher-level task (HLT) supported by a set of lower level basic performance resources
(BPRs). GSPT and NCRA have been applied in a number of fields; we were now
interested in applying them to better understand the relationship between characteristics
such as “grit” and educational success. We identified a published study (the relationship
between cognitive and non-cognitive skills (BPRs) and labor earnings (HLT)) with raw
data that, thus far, most closely matched our needs. Data was extracted from scatter
plots and reanalyzed using NCRA concepts. This yielded scatter plots that exhibited
evidence of the threshold relationship predicted by GSPT and a set of Resource
Demand Functions (RDFs). Results suggest a fundamentally different interpretation
than that assumed with correlation analysis, which is explained by GSPT to be incorrect
for relating performance capacities across hierarchical levels. Our method also provides
a basis to find each individual’s limiting performance resource for a given task.
Published papers pertaining to prediction of educational success or utilizing the
concept of the characteristic called “grit” were identified. Within those papers, we
sought out access to raw data via scatter plots including dependent and independent
variables. Few papers included such data. We used data from a paper (Diaz, Arias &
Tudela; 2012) titled: Does Perseverance Pay as Much as Being Smart?: The Returns
to Cognitive and Non-cognitive Skills in Urban Peru, which included a random
sample of the working-age (14-50) urban population in Peru (n=2,660). Participants
completed surveys that measured cognitive and non-cognitive skills, which comprise
the participants’ lower level performance capacities. Cognitive skills: intellectual
ability was measured by the Peabody Picture Vocabulary Test. Non-cognitive skills:
personality traits were measured by Conscientiousness (a Big Five personality trait),
as well as Grit personality traits (perseverance and will to strive for long term goals).
Published scatter plots positioned the HLT (log earnings) on the Y axis and the BPR
on the X axis. Analyzing the data according to the NCRA model required that the
points be re-plotted so that the HLT is reflected on the X axis and the BPR on the Y
axis. Scatter plots were then generated (one for each lower level performance
resource using the log of labor market earnings as the independent variable in each).
A Resource Demand Function (RDF) was manually added to each scatter plot using
NCRA utilizing the specified NCRA strategy.
Results
Figure 4. Original (left) linear regression scatter plot of log earnings performance vs.
non-cognitive skill GRIT 2: “Persistence of Effort” and corresponding NCRA-based
plot (right). The RDF indicates the minimum amount of GRIT 2: “Persistence of Effort
is required to support a given level of earnings.
Introduction
If there exists a certain combination of skills an individual needs to possess in order to
achieve success in educational contexts or make the greatest amount of money in the
labor market, it would be wise to assume many people would regard this as valuable
information. This is the type of problem of interest in the present study.
General Systems Performance Theory (GSPT) is a novel framework for modeling
systems, tasks, and their interface from a performance perspective. Derived from GSPT,
Nonlinear Causal Resource Analysis (NCRA) is a task analysis and performance
modeling methodology. We applied GSPT and NCRA methods to this problem. While
many others have attempted to develop such models, the universal approach used relies
on statistical, correlation-based regression methods. GSPT explains that correlation is
inappropriate for such tasks and gives rise to an alternate systems engineering method,
specifically NCRA.
A key concept in NCRA is the Resource Demand Function (RDF), which defines the
minimum amount of a basic performance resource (BPR) required to achieve a given
level of higher level task (HLT) performance. A BPR can also limit HLT performance.
BPRs are plotted against the HLT to develop performance models. The correlationbased approach does not properly identify how much of a given BPR is required to
achieve a desired level of HLT performance, nor does it identify the limiting resource.
Correlation
Traditional
“Threshold”
GSPT Based
Summary and Conclusions
Our results show that data distributions in scatter plots are all
consistent with GSPT predictions and a lower boundary “threshold”
relationship can be identified, thus supporting the applicability of
GSPT and NCRA cause-and-effect constructs to labor market earnings
and questioning previous performance prediction efforts based on
correlation. From the RDFs, a relatively greater amount of a particular
basic performance resource (conscientiousness, grit, intellectual
ability, etc.) is required to achieve higher earnings. The set of RDFs
provides a model for understanding how personality and cognitive
skills are valued in the labor market, and the way in which they
influence an individual’s socioeconomic success as measured by
earnings. The results from this study could serve to influence policies
that emphasize investments in early childhood interventions to
increase non-cognitive skills as they begin to develop.
We intend to collaborate directly with the authors of the original study
in order to create RDFs for the other measures of cognitive and noncognitive skills. We then plan to share findings with other researchers
to show that the traditional correlation approach is not an accurate way
to analyze relationships between lower level performance resources
and higher level task performance. While we did not yet fully apply
NCRA methods to identify limiting resources for individuals who had
low earnings, this would be the next logical step.
Literature Cited
•
Figure 2. Portion of first page of published paper from which data was
extracted and re-analyzed using GSPT and NCRA concepts.
•
•
Results
Figures 3 through 6 are shown below. Each figure represents a pair of scatter
plots. The left plot shows a scatter plot from the original paper (with the log of
earnings, the measure of Higher Level Task performance) plotted against one
of the skills studied. The right plot is the NCRA-based consideration of the
same data represented on the left, including the corresponding Resource
Demand Function (RDF). Note the switched axes. The set of all NCRA-based
plots form a performance model for work success.
Figure 5. Original (left) linear regression scatter plot of log earnings performance vs.
non-cognitive skill conscientiousness and corresponding NCRA-based plot (right).
The RDF indicates the minimum amount of conscientiousness required to support a
given level of earnings.
Kondraske, G.V. (2011). General Systems Performance Theory and Its
Application to Understanding Complex System Performance. Information •
Knowledge • Systems Management, 10 (1-4), 235-259.
Diaz, J. J., Arias, O., & Tudela, D. V. (2012, November 30). Does
Perseverance Pay as Much as Being Smart?: The Returns to Cognitive and
Non-cognitive Skills in urban Peru.
Duckworth, A. L., Peterson, C., Matthews, M. D., & Kelly, D. R. (2007). Grit:
Perseverance and passion for long-term goals. Journal of Personality and
Social Psychology, 92(6), 1087-1101.
Acknowledgments
I would like to thank Dr. George V. Kondraske for his incredible insight,
knowledge, and guidance. I would also like to thank Dr. Jonathan W.
Bredow, Dr. Kambiz Alavi, Dr. Qilian Liang, and Mr. Mohammadreza
Jahangir Moghadam for providing me with this unique opportunity.
Funding for this project was provided by (NSF grant #EEC-1156801,
REU Site: Research Experiences for Undergraduates in Sensors and
Applications).
For further information
Please contact Tara Alavi at [email protected].
Figure 1. Scatter plots of representative types of performance data spanning hierarchical
levels. The traditional correlation thinking is contrasted with GSPT threshold concepts.
The horizontal axis represents Higher Level Task Performance and the vertical axis
represents one of the multiple lower level Basic Performance Resources that contribute to
Higher Level Task Performance. The curve representing the lower boundary in the GSPTbased plot is the Resource Demand Function (RDF).
Figure 3. Original (left) linear regression scatter plot of log earnings performance vs.
non-cognitive skill GRIT 2: “Consistency of Interest” and corresponding NCRAbased plot (right). The RDF indicates the minimum amount of GRIT 2: “Consistency
of Interest required to support a given level of earnings.
Figure 6. Original (left) linear regression scatter plot of log earnings performance vs.
cognitive skill PPVT score and corresponding NCRA-based plot (right). The RDF
indicates the minimum amount of PPVT required to support a given level of
earnings.
G. Kondraske, Ph.D. (advisor)
Human Performance Institute
Univ. of Texas at Arlington
PO Box 19180
Arlington, TX 76019-0180
E-mail: [email protected]