hams.“

$1.:
a .

I; .
5.2.1.14.

 

, : 1 an 7.53;). (25%.. ‘
:5 5M5... J 1 . . .... ‘ J .

.1 .3 e. u ‘
:1: r! .. ‘

 

 

This is to certify that the
thesis entitled

Myers-Briggs Type Indicator Jungian Personality Dimensions
and Strong Interest Inventory Holland Themes: The
Multivariate Analysis of the Relationship Between MBTl
Personality Dimensions and the Six Strong General
Occupational Themes

presented by

Matthew A. Miller

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in Counseling Psychology

 

QMQ %§:th0 PM) .

Major/Professor s Signature

December 13, 2002

 

MSU is an Affirmative Action/Equal Opportunity Institution

 

 

Michigan State

 

LIBRARY

University

 

 

PIACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

MNt15204
09220.9

___1

 

DECbH

 

 

“c2990;

JANA; 6t ZQDE’:

 

H
gi

 

 

 

*WIZ?

fl?

 

 

 

 

 

 

 

6/01 C'JCIRC/DatoDquGS-DJS

MYERS-BRIGGS TYPE INDICATOR IUNGIAN PERSONALITY DIMENSIONS
AND STRONG INTEREST INVENTORY HOLLAND THEMES:
THE MULTIVARLATE ANALYSIS OF THE RELATIONSHIP BETWEEN MBTI
PERSONALITY DIMENSIONS AND THE SIX STRONG GENERAL
OCCUPATIONAL THEMES

By
Matthew A. Miller

A DISSERTATION
Submitted to
Michigan State University

in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Counseling and Educational Psychology

2002

ABSTRACT
MYERS-BRIGGS TYPE INDICATOR JUNGIAN PERSONALITY DIMENSIONS
AND STRONG INTEREST INVENTORY HOLLAND THEMES:
THE MULTIVARIATE ANALYSIS OF THE RELATIONSHIP BETWEEN MBTI

PERSONALITY DIMENSIONS AND THE SIX STRONG GENERAL
OCCUPATIONAL THEMES

By
Matthew A. Miller

The Strong Interest Inventory and the Myers-Briggs Type Indicator are the most
fi'equently used vocational assessment instruments across a variety of settings (Hammer
& Kummerow, 1997). But the literature accumulated on the two instruments thus far has
failed to fully explore the possible relationships between Myers-Briggs Type Indicator
personality dimensions and Strong Interest Inventory themes. The relevance of a
theoretically and methodologically correct approach to the exploration of the relationship
between the two instruments is couched within MBTI theory development and Holland’s
(1985) vocational theory development. The current proposed study seeks to address the
shortcomings reflected in the literature via canonical correlation. Canonical correlation
can be applied to adequately address the following research question: to what extent and
in what ways do MBTI personality dimensions overlap with Strong Interest typology? A
sample of 423 individuals who have taken both the Strong Interest Inventory and the
Myers-Briggs Type Indicator, Form G, was collected. Results indicated that (a)
multivariate analysis of the relationship between MBTI personality dimensions and
Strong interest themes yielded significant amounts of variance explained between the two
variable sets, (b) multivariate analysis of the relationship between MBTI personality

dimensions and Strong interest themes resulted in a more comprehensive explanation of

the relationships between the two variable sets than did bivariate correlation, and (0)
gender affects the relationship between the two sets. Practical and research related

implications of the findings are discussed.

ACKNOWLEDGMENTS

The author would like to thank Dr. Robbie J. Steward, Dissertation Chair and
Academic Advisor, for her ongoing support and mentorship throughout the dissertation
process specifically and the graduate education process generally.

The author thanks the following dissertation committee members for the
contribution of their time and ideas to this manuscript: Dr. Alfiee Breland, Dr. Susan
Frank, and Dr. Lee June. Each of these individuals has also uniquely contributed to the
author’s professional development, for which the author is grateful.

The author thanks Dr. Walter Buboltz, Louisiana Tech University Department of
Counseling Psychology, for serving as a consultant on this manuscript and for his many
investments in the author’s professional development.

The author thanks Ms. Beverly Grunden, Wright State University Department of
Mathematics, and Dr. Ken Frank, Michigan State University Department of Counseling

and Educational Psychology, for serving as statistical consultants on this manuscript.

iv

TABLE OF CONTENTS

LIST OF TABLES

CHAPTER ONE: INTRODUCTION

Strong Interest Inventory Overview

Myers-Briggs Type Indicator Overview

Conjoint Use of the Strong and MBTI—a call for research
Strong Interest Themes and MBTI Personality Dimensions—-
an overview of the research

Strong Interest Themes and MBTI Personality Dimensions—
a call for new research

CHAPTER TWO: REVIEW OF LITERATURE

Historical Context

Person-Environment Model Overview

Strong, MBTI, and the P-E Model

Strong Interest Inventory Overview of Applications and Research
The Strong—studying individual change

The Strong—studying group change

The Strong—studying characteristic interests of people in particular
occupations

The Strong—studying cross-cultural influences

Holland’s Theory—an overview and critique

Strong Interest Inventory and Holland’s Occupational Theory
Using Additional Sources of Information

Myers-Briggs Type Indicator and Vocational Psychology
Myers-Briggs Type Indicator—studying occupational selection
The MBTI—studying career decision making

The MBTI—studying job satisfaction

The Strong and the MBTI: Together in Research

Strong interest themes and MBTI personality dimensions—

a call for new research

CHAPTER THREE: METHODS

Participants
Procedures
Measures
Strong Interest Inventory
Myers-Briggs Type Indicator
Design

vii

catsui—

ON

11

15
15
17
20
21
23
23

24
25
26
32
34
34
35
37
38
42

56
62

62
63

67
77

CHAPTER FOUR: RESULTS 79

Sample Demographics and Distribution 79
Bivariate Relationships Between MBTI dimensions and GOTs:

Full Data Set 80
Canonical Correlation Analysis: Full Data Set 84
To What Extent Do the MBTI Personality Dimensions

and the Strong GOTs Overlap? 85
In What Ways Do the MBTI Personality Dimensions

and the Strong GOTs Overlap: Full Data Set 87
Gender Considerations: Male Demographics and Distributions 91
Bivariate Correlations Between MBTI Dimensions and GOTs:

Male Data Set 91
Gender Considerations: Female Demographics and Distributions 94
Bivariate Correlations Between MBTI Dimensions and GOTs:

Female Data Set 94
To What Extent Do the IVIBTI Personality Dimensions

and the Strong GOTs Overlap: Male Data Set 96
In What Ways Do the MBTI Personality Dimensions

and the Strong GOTs Overlap: Male Data Set 97
To What Extent Do the MBTI Personality Dimensions

and the Strong GOTs Overlap: Female Data Set 100
In What Ways Do the MBTI Personality Dimensions

and the Strong GOTs Overlap: Female Data Set 101
To What Extent Do the MBTI Personality Dimensions

and Strong Interest Themes Overlap: A Summary 104

In What Ways Do the MBTI Personality Dimensions

and Strong Interest Themes Overlap: First Canonical Function Summary 105
In What Ways Do the MBTI Personality Dimensions and

Strong Interest Themes Overlap: Second Canonical Function Summary 106

CHAPTER FIVE: DISCUSSION 108
To What Extent Do the MBTI Personality Dimensions and Strong Interest
Themes Overlap? 110
In What Ways Do the MBTI Personality Dimensions
and Strong Interest Themes Overlap? 112
Implications to the Practitioner: Canonical Correlation Findings,

Full Data Set 1 17
In What Ways Do the MBTI Personality Dimensions
and Strong Interest Themes Overlap: The Effect of Gender 118
In What Ways Do the MBTI Personality Dimensions
and Strong Interest Themes Overlap: A Summary 125
Conclusion 129

REFERENCES 135

vi

LIST OF TABLES

Table 1: Bivariate Correlations Among MBTI Personality Dimensions:

Full Data Set 81
Table 2: Bivariate Correlations Among General Occupational Themes:

Full Data Set 82
Table 3: Bivariate Correlations Between the MBTI Dimensions and GOT:

Full Data Set 82
Table 4: Significant Bivariate Correlations Between the MBTI Dimensions

and GOT Listed By Absolute Correlation Value: Full Data Set 84
Table 5: Canonical Correlation Analysis: Full Data Set 85
Table 6: Canonical Redundancy Index: Full Data Set 87
Table 7: Inclusion Criteria For First Canonical Function: Full Data Set 88
Table 8: Inclusion Criteria For Second Canonical Function: Full Data Set 89

Table 9: Bivariate Correlations Between the MBTI Dimensions and GOT:
Male Data Set 92

Table 10: Comparison of Significant Bivariate Correlations
Between the MBTI Dimensions and GOTs Listed By Absolute Correlation Value:
Full Data Set and Male Data Set 93

Table 11: Comparison of Significant Bivariate Correlations
Between the MBTI Dimensions and GOTs Listed By Absolute Correlation Value:

Full Data Set, Male Data Set, Female Data Set 95
Table 12: Canonical Correlation Analysis: Male Data Set 97
Table 13: Canonical Redundancy Index: Male Data Set 97
Table 14: Inclusion Criteria For First Canonical Function: Male Data Set 98
Table 15: Inclusion Criteria For Second Canonical Function: Male Data Set 99
Table 16: Canonical Correlation Analysis: Female Data Set 100
Table 17: Canonical Redundancy Index: Female Data Set 101

Table 18: Inclusion Criteria For First Canonical Function: Female Data Set 102

vii

Table 19: Inclusion Criteria For Second Canonical Function: Female Data Set 103

Table 20: Canonical Correlation Summary 105

viii

CHAPTER ONE
INTRODUCTION

The Strong Interest Inventory (Harmon, Hansen, Borgen, & Hammer, 1994) and
the Myers-Briggs Type Indicator (Myers & McCaulley, 1985) are the most frequently
used vocational instruments among career counselors in colleges and universities, among
practitioners in community-based mental health and agencies and in private practice, and
among practitioners in organizational settings (Hammer & Kummerow, 1997). Yet as
more professionals are using both the Strong Interest Inventory and the Myers-Briggs
Type Indicator, questions such as “how might this client be helped by examining her or
his results on both the Strong and the MBTI?” have consistently emerged from the field
(Hammer & Kummerow, 1997). In response to questions such as this, Hammer and
Kummerow (1997) stated, “much more research is needed on the use of the Strong
Interest Inventory and the Myers-Briggs Type Indicator. . .so that our understanding about
the strengths and limitations of using both instruments together can be enhanced” (p.
125)

Although a body of research has accumulated in the last two decades exploring
the relationship between the two instruments, the generalizability of much of the current
research is limited by methodological shortcomings, use of outdated versions of the
Strong Interest Inventory, and lack of consistency across results. Before delving into a
discussion of the literature and shortcomings therein, a review of the basic composition of

the Strong Interest Inventory and the Myers-Briggs Type Indicator will be addressed.

Strong Interest Inventory Overview

The development of the Strong Interest Inventory (Strong) was based upon the
research of E.K. Strong (1943) and John Holland (1973, 1985). E.K. Strong hypothesized
that an individual with interests similar to those of persons working in a given occupation
will be more likely to find satisfaction in that particular occupation than would a person
who does not share common interests with those workers (Buboltz, Thomas & Johnson,
in press). From the Strong’s inception in 1927 until 1974, the instrument was considered
atheoretical (Hammer & Kummerow, 1997). However, with the instrument’s revision in
1974, the work of John Holland was integrated into the Strong and “represented an
important milestone in interest measurement” (Harmon et al., 1994, p. 43).

The Strong provides feedback within four scales: General Occupational Themes,
Basic Interest Scales, Occupational Scales, and the Personal Style Scales (Harmon et al.,
1994). Results on the Strong are structured around Holland’s (1985) theory of interest
typology (Harmon et al., 1994). Among the four feedback scales on the Strong, the
General Occupation Themes (GOT) are most closely and directly related to Holland’s
(1985) theoretical system (Harmon et al., 1994).

Holland’s (1985) theory of vocational interest typology is based on four main
assumptions. First, most people can be categorized according to six interest themes—
Realistic, Investigative, Artistic, Social, Enterprising, and Conventional—and one theme
or combination of these themes may characterize each person (Holland, 1985). The
Realistic individual purportedly gravitates toward vocational and avocational activities
centered on building or repairing, hands on activities, and the outdoors. The Realistic

individual is likely interested in vocations ranging from agriculture, athletics, and/or the

military (Harmon et al., 1994). The Investigative type theoretically is drawn toward
researching and analyzing activities within the sciences (Harmon et al., 1994). The
Artistic type, as the name implies, gravitates toward creating and enjoying the arts
ranging from music and drama to culinary (Harmon et al., 1994). The Social type is
purported to have interests in helping and instructing others and/or religious activities and
are often found in fields such as social work, counseling, and teaching (Harmon et al.,
1994). The Enterprising individual enjoys selling, persuading, and managing via
occupational fields such as law, sales, and merchandising (Harmon et al., 1994). The
Conventional type is hypothesized to prefer processing and organizing activities via
computer activities typically found within data management and office services (Harmon
et al., 1994). Second, occupational environments can be divided into the same six themes
and a particular person dominates each environment (Holland, 1985). Therefore, the
personality composition of co-workers establishes the tone within a given occupation.
Third, people search for environments that let them exercise their skills and abilities,
express their attitudes and values, take on problems and roles they find stimulating and
satisfying, and avoid tasks that they find distasteful or difficult (Holland, 1985). Fourth,
job performance, satisfaction, and stability are influenced by an interaction between a
person’s personality and the characteristics of his or her working environment (Holland,
1985)

Holland’s theory organized the six themes by placing them along the points of a
hexagon. The hexagonal arrangement of interest themes reads R-I-A-S-E-C fi'om left to

right (Harmon et al., 1994). Similar types are located adjacently (e. g., Realistic and

Investigative) whereas those types that are most dissimilar are diametrically opposed

from one another (e. g., Artistic and Conventional) (Harmon et al., 1994).

Myers-Briggs Type Indicator Overview

 

The Myers-Briggs Type Indicator (MBTI) is a measure of personality as defined
by CG. Jung (Myers & McCaulley, 1998). Jung theorized that an individual’s
perceptions of and judgements about events and experiences in life comprise and explain
the development of personality (Hall & Lindzey, 1970). Behavior, therefore, may better
be understood and predicted through the exploration of individual perceptions and
judgements. Jung discovered what he believed to be patterns of perceiving and patterns
of judging across human behavior. J ung's theory of personality describes various
consistent behavioral patterns that result from how individuals assimilate information and
make decisions (Hammer & Kummerow, 1997).

The MBTI represents an attempt to quantify Jungian personality theory (Hammer
& Kummerow, 1997). The MBTI explores an individual’s preferred patterns of
perceiving and judging life events through the analysis of two important dimensions:
attitude and functions (Myers & McCaulley, 1985). Attitude is explored through the
ways in which an individual likes to be energized (Extravert versus Introvert) and the
ways in which an individual deals with the outer world (Judging versus Perceiving).
Extraverts (E) are energized through activities and other people. Introverts (I) prefer time
and space alone and are energized by reflection. Judging (J) types deal with the outer
world in an organized way whereas Perceiving (P) types prefer to be more spontaneous

(Myers & McCaulley, 1985).

Functions are examined through how one takes in information (Sensing versus
Intuition) and how one goes about making decisions (Thinking versus Feeling). Sensing
(S) types take-in information by noticing facts and details first. The Sensing type looks
to the present and what is (Myers & McCaulley, 1985). Conversely, Intuitive (N) types
ofien notice the big picture and the many possibilities therein first. The Intuitive type
looks to the future and what can be (Myers & McCaulley, 1985). The Thinking (T) type
makes decisions relying on objective and rational methodology. This individual prefers
to step back and analyze a situation carefully, evaluating pros and cons from a perceived
rational perspective (Myers & McCaulley, 1985). The Feeling (F) type makes decisions
through a more personal, subjective, and values-oriented approach. This individual
prefers to place himself or herself within a decision and to note where harmony and
disharmony may occur (Myers & McCaulley, 1985).

Through the analysis of attitude and functions, eight personality dimensions (E, I,
S, N, T, F, J, P) are created and divided into four dichotomies (E-I, S-N, T-F, J-P).
Jungian theory hypothesized that everyone uses each personality dimension at different
moments; however, Jung stated that individuals express preferred attitudes and functions
and seek to implement their preferences within their workworld (Myers & McCaulley,
1985). Individual preference patterns along the four dichotomy combine to create a four-
1etter personality type. A total of sixteen personality types exist within the MBTI
classification system (Myers & McCaulley, 1985). Each of the sixteen types is presumed
to possess and express unique characteristics (Myers & McCaulley, 1985).

Related to vocational psychology, differences between the MBTI personality

dimensions are hypothesized to serve as helpful indicators of career choice and

satisfaction (Zunker, 1998). Myers and McCaulley (1985) stated that one of the most
important motivations for career choice is a desire for work that will permit the use of
preferred functions and attitudes. Furthermore, tasks that require preferred functions and
attitudes demand less effort for better performance and provide greater levels of
satisfaction (Myers & McCaulley, 1985). MBTI results are purported to be helpful,
therefore, toward the consideration of career opportunities that utilize preferred attitudes
and functions (Myers & McCaulley, 1998).
Conioint Use of the Strong anAMBTI—a call for research

Both the Strong and MBTI manuals recommend using the two instruments
together (Healy, 2000). Proponents of using the two instruments together have suggested
that MBTI results can assist an individual toward a better understanding of their Strong
results (Healy, 2000). Furthermore, Hammer and Kummerow (1997) argued that the
MBTI and the Strong presented unique but complimentary information. Yet questions
remain regarding the integration of the two instruments (Hammer & Kummerow, 1997).
Hammer and Kummerow (1997), for example, stated that more research is needed on the
use of the Strong Interest Inventory and Myers-Briggs Type Indicator together. Toward
this end, Hammer and Kummerow (1997) stated that a number of topics offer a logical
beginning for future research efforts. One such area highlighted by the authors was the
analysis of the relationship between Holland’s interest themes employed on the Strong
Interest Inventory and Jungian personality dimensions employed on the Myers-Briggs

Type Indicator (Hammer & Kummerow, 1997).

Strong Interest Themesj and MBTI Personality Dimensiokan overview of the resear_ch

Although the existent literature will be reviewed in detail within chapter two of
the manuscript, the present discussion will focus on highlighting the theoretical and
methodological shortcomings inherent within the current literature.

Published explorations of the relationship between Strong interest themes and the
MBTI personality dimensions have typically employed one of two research designs.
First, researchers have compared the relationship between MBTI personality dimensions
(e. g., E-I, S-N, T-F, and J -P) with the six Strong interest themes (e. g., R-I-A-S-E—C).
Second, researchers have compared the relationship between MBTI four-letter types (e.g.,
INTI) and the six Strong interest themes. Within the first methodological paradigm,
researchers have typically employed Pearson correlations or MANOVA to statistically
evaluate the relationship between MBTI personality dimensions and Strong interest
themes. Examples of Pearson correlational analyses include those performed by
McCaulley (1978), Dillon and Weissman (1987), Decola (1992) and Hammer and
Kummerow (1997). On the other hand, Buboltz, Thomas, and Johnson (in press)
performed MANOVA analysis.

All of the aforementioned studies discovered moderate correlations between
MBTI personality dimensions and the Strong interest themes. Only one relationship, a
moderate correlation between MBTI Feeling and Strong Social, was significant across all
studies. The validity of the findings is limited however because the full multivariate
relationships occurring among and between the MBTI personality dimensions and the
Strong interest typologies have been left unexplored or accounted for (K Frank, personal

communication, May, 2001). This likely explains the questionable level of consistency

obtained between research findings. Specifically regarding the multivariate nature of the
relationship among and between the two instruments, exploration into the theoretical
foundations of the Strong Interest Inventory and the Myers-Briggs Type Indicator reveals
the dynamic and interactive nature inherent within the classification systems employed.

To illustrate the multivariate nature of the six Strong interest themes, Holland
(1985) noted that a classification system comprised of only six independent themes was
insufficient to fully represent human and occupational workplace diversity (Harmon et
al., 1994). Holland therefore (1985) expanded his vocational interest classification
system to include combinations of the six themes such as Realistic-Artistic and Social-
Enterprising. Holland (1985) also expanded his theory to include placement of the six
types along a hexagon wherein similar types were located adjacent to one another and
highly dissimilar types were located opposite of one another. Scoring on the Strong was
constructed such that an individual’s score on any single interest theme was only relevant
vis—a-vis other interest scale scores. It is these dynamics of Holland theory and interest
theme combinations that are not and cannot be fully explored using Pearson correlation
or MAN OVA (K. Frank, personal commtmication, May, 2001).

To illustrate the multivariate nature of the MBTI personality dimensions, Myers
and McCaulley (1998) discussed the fact that in developing personality theory, “J ung,
Myers, and Briggs were constantly attentive to the dynamic character of the personality
types they were describing. The richness, depth, and breadth of their descriptive systems
result from the dynamic interplay of the functions and attitudes inherent in each type. It
is important to bear this in mind in applying MBTI theory and interpreting the MBTI

personality inventory” (p. 23). As is evident from the preceding citation, the importance

the MBTI authors (Myers & McCaulley, 1985; 1998) placed on the interactive nature of
the MBTI personality dimensions is derived from the evolution of Jung’s own thinking
(Myers & McCaulley, 1998). Initially, Jung’s attempt to explain individual differences in
personality stemmed from his observation that there were two types of people: extraverts
and introverts (Myers & McCaulley, 1998). However, ten years after his initial
observations, Jung noted that these two types did not provide a complete picture (Myers
& McCaulley, 1998). Jung then subdivided his initial extravert and introvert types into
six types by identifying Sensing versus Intuitive type and Thinking versus Feeling type
(Myers & McCaulley, 1998). Myers and Briggs extended Jung’s model by adding the
Judging versus Perceiving dichotomy (Myers & McCaulley, 1998).

In summary, MBTI theory asserts that the interactions of MBTI personality
dimensions result in behaviors that cannot be predicted from exploration of each
preference separately (Myers & McCaulley, 1998). Because each preference in a given
type interacts with all of the other preferences in that type, MBTI personality theory
asserts that any combination of two or more preferences will yield behaviors that are
more than the sum of their parts (Myers & McCaulley, 1998). According to MBTI
personality theory therefore, unique relations exist among the personality dimensions that
suggest characteristics or behaviors not apparent at a univariate or bivariate level of
analysis (Myers & McCaulley, 1998). Myers and McCaulley (1985), for example,
interpreted a high score on MBTI Thinking dimension differently when it was paired
with MBTI Sensing dimension than when it was paired with MBTI Intuition dimension.
Hammer (1996) therefore stated, “researchers are implored to make more use of (MBTI)

preference combinations or (MBTI) whole types in their research” (p. 50). It is these

MBTI personality dimension combinations that are not fully explored when performing
Pearson correlation or MANOVA analyses (K. Frank, personal communication, May,
2001)

The second research design typically employed within the literature is the
comparison of the MBTI four-letter types (e. g., ESTJ) with the six Strong interest
themes. Within this design, researchers have used profile mean analysis (Hammer &
Kummerow, 1997) and chi-square analysis (Buboltz et al., in press) to statistically
evaluate the data. Based upon a sample of 468 college students, Hammer and Kummerow
(1997) calculated the three MBTI four-letter personality types with the highest mean
score within each of the six GOTs and ranked them in descending order. Hammer and
Kummerow’s (1997) findings presented some theoretical conundrums. Furthermore,
several points of disagreement-~as well as some points of agreement--were established
between the authors’ data and that of previous bivariate studies by Dillon and Weissman
(1987), Hammer and Kummerow (1997), and Buboltz et al. (in press). Based upon data
derived from 426 college students, Buboltz, Thomas, and Johnson (in press) employed
Chi-square analysis to determine significant relationships among MBTI four-letter
personality types and Strong interest themes. Chi-square analysis revealed that the
sixteen MBTI types and the Strong GOT types were not independent (Buboltz et al., in
press). Several of the MBTI types appeared to cluster within one or two of the Strong
interest themes (Buboltz et al., in press). Interestingly however, no four-letter MBTI type
and Strong interest theme relationships agreed between Buboltz et al. (in press) and

Hammer and Kummerow (1997).

10

In addition to the questionable levels of consistency between the findings of
Hammer and Kummerow (1997) and Buboltz et al. (in press), the validity of the reported
findings of these two typology studies is limited by at least one important consideration: a
typology approach to the investigation of MBTI personality dimensions and Strong
interest types treats the dimensions as discrete variables as opposed to continuous; it
therefore overlooks the subtle complexity of the potential relationships among MBTI
personality dimensions and Strong interest themes. For example, individuals scoring at
the 5‘h and 35th percentiles of the Thinking-Feeling MBTI scale are treated the same
within a typology analysis (Healy, 2000).

Strong Interest Themes grid MBTI Personality Dimensions—a call for new research

 

Researchers have attempted to gain a better understanding of the relationships
between MBTI personality dimensions and Strong interest themes as well as MBTI
personality types and Strong interest themes via bivariate correlation analysis,
MANOVA, chi square analysis, and profile mean score analysis. By virtue of not fully
exploring or accounting for potential relationships within and between each classification
system (e.g., Pearson Correlation) and minimizing variability within each classification
system (e. g., MAN OVA), a fuller understanding of the potential relationships between
the MBTI personality dimensions and the Strong interest types has not been reached. In
short, bivariate correlational analyses and the use of categorical data are not fully
appropriate toward the development of a deeper understanding of the relationship
between MBTI personality dimensions and Strong interest themes. A gap therefore
exists in the field’s understanding of the relationship between the MBTI personality

dimensions and the Strong interest types. The relevance of exploring this gap hinges

11

upon the fact that the Strong and the MBTI are the two most frequently used vocational
assessment instruments across a variety of settings (Hammer & Kummerow, 1997).
Furthermore, both instruments promote conjoint usage (Myers & McCaulley, 1985;
Harmon et al., 1994).

The current proposed study seeks to distinguish itself fiom previous research
through exploring the relationship between and among MBTI personality dimensions and
Strong interest themes from a multivariate perspective. Specifically, the research design
will uniquely position the analysis to reveal which MBTI and Strong variables most
strongly covary and in-tum, the ways in which MBTI personality dimensions and Strong
interest themes might be combined to maximize covariance. Practically therefore,
results will highlight those variables that “go together” most fiequently. To do so,
canonical correlation will be employed to address the following research question: “to
what extent and in what ways do MBTI personality dimensions overlap with Strong
Interest themes?”

Canonical correlation is a multivariate statistical procedure not often seen within
behavioral sciences literature (Jaccard & Becker, 1990). This lack of use is not related to
its value but more likely due to the lack of understanding surrounding its function and
purpose (J accard & Becker, 1990). Canonical correlation addresses the degree of overlap
between sets of variables. The present research question will therefore be addressed in
three steps. First, canonical correlation will treat the four MBTI dimensions (i.e., E—I, S-
N, T-F, J -P) as one set“ ” and will treat the six Holland codes as one set “Y”. As has
been discussed, the treatment of each as a set versus independent constructs is

theoretically sound. Second, one new variable will be created for set X (MBTI

12

dimensions) based upon those MBTI variables that best correlate with Strong Interest
variables. This variable or variables will be called the function of X {f(X)}. One new
variable will also created for set Y (Holland themes) consisting of those variables that
best correlate with the MBTI personality dimensions. This set will be called the function
of Y {f(Y)}. Third, the correlation between f(X) and f(Y) will be analyzed.

Detailed hypothesis statements will be discussed further in chapter two. Generally
speaking however, the present study hypothesizes that the MBTI personality dimensions
and Strong interest themes interact (i.e., covary) and are potentially correlated to
previously unknown extents. More will therefore be revealed, it is hypothesized,
regarding the relationships between and among MBTI personality dimensions and Strong
interest themes if they are analyzed from a multivariate perspective as opposed to
categorical and/or bivariate perspectives. Formally stated, the hypothesis asserts that the
correlation between the two sets (i.e., MBTI personality dimensions and Strong interest
themes) and the percent of variance accounted for within the canonical correlation
functions will be considerably higher than any percentage of variance accounted for by a
bivariate regression or correlation analysis procedures (K. Frank, personal
communication, May, 2001).

The relevance of the present study is as follows: it is the first known study to
explore the multidimensional complexity of the relationships between MBTI personality
dimensions and Strong interest themes from the perspective of (a) the ways in which each
of the two interest and personality sets represented on the Strong and MBTI overlap (i.e.,

covary) and (b) the extent of the overlap between the two sets.

13

Practically, counselors can gain a valid understanding as to what common patterns
exist between MBTI personality dimensions and Strong interest themes. The primary
purpose of the present analysis therefore is to highlight patterns of covariance between
interest and personality. This knowledge could assist the counselor and client alike to
assemble interpretive pieces of each instrument in a theoretically and empirically
validated manner and to apply this knowledge toward therapeutic goals such as increased
self-understanding, increased decision-making skills, and increased occupational

satisfaction.

14

CHAPTER TWO
REVIEW OF LITERATURE

The present chapter will begin by discussing the relationship between counseling
psychology and vocational psychology. The discussion will then ttu'n toward a model—
the Person-Environment (P-E) theory—often used within the field for addressing
vocational and avocational issues. As will be highlighted, use of the Myers-Briggs Type
Indicator (Myers & McCaulley, 1985) and the Strong Interest Inventory (Harmon,
Hansen, Borgen, & Hammer, 1994) is based within the P-E model. An overview of
general research on the Strong will be presented as will an overview of general research
on the MBTI. Thorough discussion will occur around empirical reviews of Holland’s
(1985) theory, upon which interpretation of the Strong rests, and upon existent literature
addressing the conjoint use of the Strong and MBTI. Critical review of the literature will
focus upon literature specifically exploring Strong interest themes and MBTI personality
dimensions. The present chapter will conclude by expanding upon the research questions
proposed and the specific hypotheses therein.
Historigl Contex_t

Alone among the helping professions, counseling psychology devotes significant
attention to the role of work in the lives of people (Dowd, 1998). The profession
originated out of a commitment to assisting individuals with work-related concerns or
questions (Dowd, 1998). In fact, the study of vocational psychology has been called the
single most identifiable characteristic of the field (Chartrand, 1991).

Following the conclusion of World War II and the national rise of affluence

therein, counseling psychology broadened its scope of practice to include personal

15

development in areas other than work (Dowd, 1998). Yet the field’s historical interest in
vocational counseling research and practice endured, due largely to the field’s belief in
the strong overlap between work, interpersonal, and intrapersonal domains (Fitzgerald &
Osipow, 1986).

One of the hallmarks of mental health is the capacity to work and to love
(Blustein & Spengler, 1995). Most researchers would agree that compelling evidence
exists in support of an interrelationship between work and mental health (Blustein &
Spengler, 1995). Studies highlight the reciprocal relationships between unemployment
(Osipow & Fitgerald, 1993), job stress (Osipow, 1979), and work related problems (Herr,
1989 in Blustein & Spengler, 1995) to increased emotional distress. Similar links have
been found between career satisfaction and factors such as anxiety, depression, somatic
difficulties, and self-esteem (Blustein & Spengler, 1995).

In response to these observations, intervention models have been proposed for
infirsing the career counseling processes with greater attention toward increased
intrapersonal and interpersonal awareness (Blustein & Spengler, 1995). The consensus
underlying these models is that effective counseling does not maintain overt distinctions
between career and noncareer domains (Blustein & Spengler, 1995). Counseling
psychologists have been at the forefront of this integration between career and noncareer
domains (Dowd, 1998). The discipline is currently recognized as a profession that assists
relatively well-intact individuals to maximize their intrapersonal and interpersonal
developmental potential across work, academic, and home domains (Dowd, 1998). In so
doing, counseling psychology makes use of interventions designed both to help

individuals adjust to their environment and modify their environment (Dowd, 1998). One

16

intervention model in particular, the Person-Environment theory (P-E), represents a
widely used and widely accepted theoretical approach within the field intended to
promote individual adjustment to vocational and avocational contextual factors as well as
the ability to modify one’s vocational and avocational environments (Chartrand, 1991).
Person-Environment Model Overview

The P-E model is a descendant of the Trait and Factor theory. Widely considered
the original model of vocational psychology, the Trait and Factor theory (Parsons, 1909)
focused upon matching an individual’s traits with the requirements of a specific
occupation (Zunker, 1998). The foundation of the Trait-Factor model can be traced to
Parson’s (1909) assertion that vocational choice involves self-knowledge, occupational
knowledge, and an understanding of the relationship between the two. A model evolved
from Parson’s writings wherein the vocational counseling process was viewed as a
facilitation of self-understanding, realistic planning, and action (Zunker, 1998).

Beginning in the early 1900’s and continuing until the 1950’s, the Trait and
Factor approach was widely considered the hallmark of career counseling (Chartrand,
1991). The popularity of the theory faded, however, as humanistic and developmental
approaches began to permeate counseling psychology and allied fields (Chartrand, 1991).
The Trait and Factor model came to be criticized for its promotion of the following
assumptions: (3) occupational choice is a single event; (b) a single type of person works
in each job; (c) there is a single right goal for every career decision-maker; and (d)
occupational choice is available to everyone (Chartrand, 1991). Criticisms such as these
created the caricature of the model as a “test and tell” counseling approach that unfolded

in “three interviews and a cloud of dust” (Chartrand, 1991, p. 519).

17

The Person-Environment (P-E) model of counseling evolved following the Trait-
Factor model’s decline (Chartrand, 1991). P-E theorists believed in the usefulness of the
Trait and Factor perspective yet recognized a need for a more dynamic interpretation of
the interactive nature of the person and environment relationship (Chartrand, 1991). The
P-E model therefore introduced the following four assumptions pertaining to individuals
selecting and shaping their environment: first, people are viewed as capable of making
rational decisions (Jones, Stefflre, & Stewart, 1970; Chartrand, 1991). In the tradition of
cognitive-behavioral approaches, a cognitive orientation guides the search for
determinants and interventions (Jones et al., 1970; Chartrand, 1991). Second, people and
work environments differ in reliable and meaningful ways (Jones et al., 1970; Chartrand,
1991). This is not meant to imply that a single type of person works in each job but that
important patterns exist and can be used to better organize people and environments
(Chartrand, 1991). Third, increased congruence between personal characteristics and job
characteristics yields greater likelihood of occupational satisfaction (Holland, 1985;
Chartrand, 1991). Therefore, knowledge of person and environment patterns can be used
to inform individuals about the probability of satisfaction in different work settings.
Fourth, the P-E fit is a reciprocal process in which individuals shape the enviromnent and
the environment influences the individual (Round & Tracey, 1990; Chartrand, 1991).
Individuals seek and create environments that allow them to express preferred
characteristics. It is this dynamic reciprocity aspect of the P-E model that most clearly
distinguishes it fi'om the Trait-Factor model (Chartrand, 1991).

In addition to shifting the person and environment relationship from static to

dynamic and from one-way to reciprocal, P-E theorists were also concerned with

18

addressing the overlap between work and personal domains, giving greater attention
toward a broad range of presenting client issues beyond career choice or satisfaction
(Rounds & Tracey, 1990; Chartrand, 1991). Rounds and Tracey (1990) asserted that
career counseling was actually a subset of psychotherapy and should not be viewed as an
entity to itself. The authors proposed that viewing career counseling as a form of
psychotherapy would facilitate a better understanding of career change processes
(Blustein & Spangler, 1995). Applying their assertions to the counseling process,
Rounds and Tracey (1990) proposed a P-E framework that integrated components of
decision-making, information processing, and problem solving (Blustein & Spengler,
1995). Based in part upon the work of Rounds & Tracey (1990), Chartrand (1991)
elucidated P-E assumptions, objectives, and methods addressing work and non-work
overlap. According to Chartrand (1991), the P-E model assumes that people and
environments change continually in an on-going adjustment process. The primary goal
of the model therefore is not to simply make a one-time match between an individual and
an environment but to facilitate the acquisition of problem-solving skills (Chartrand,
1991). Problem-solving skills are developed through the teaching of a four step
information-processing system consisting of (a) perceiving and appraising the current
situation, (b) establishing concrete and realistic goals, (c) developing plans and pattern
matching, and ((1) taking action (Chartrand, 1991). Since Chartrand’s (1991) seminal
work, recent advances in P-E theory have addressed the career counseling process with a
greater attention toward the broad range of issues that clients present (Blustein &

Spengler, 1995). Lofquist and Dawis (1991) applied a P-E model to client issues ranging

l9

from career questions, marital counseling, and addiction counseling (Blustein &
Spangler, 1995).

The acquisition of self-understanding, information processing skills, and goal-
setting skills are paramount within a P-E approach (Chartrand, 1991) . In order to
facilitate these objectives, the P-E process has traditionally included the administration of
assessment instruments to assist individuals in understanding their personal traits,
behaviors, and characteristics (Buboltz, Thomas, & Johnson, in press). Lowrnan (1991)
emphasized the assessment of one’s interests, abilities, and personality characteristics to
create a comprehensive approach to career assessment and to provide the individual with
as much information as possible for adequate self-understanding. A wide variety of
career and vocational instruments have been developed in an effort to measure
personality and interest attributes (Dillon & Weissman, 1987). Undoubtedly, two of the
most frequently used (Hammer & Kummerow, 1997) instruments are the Strong Interest
Inventory (Harmon, Hansen, Borgen, & Hammer, 1994) and the Myers Briggs Type
Indicator (Myers & McCaulley, 1985).

Strong, MBTI, and the P-E Model

Assumptions underlying the use of the Strong Interest Inventory (Strong) and the
Myers-Briggs Type Indicator (MBTI) are consistent with the P-E model (Hammer &
Kummerow, 1997). Authors of both the Strong and the MBTI recognized a need to go
beyond a simplistic matching of people with jobs. Authors of both instruments also
believed that people act as change agents in shaping their own careers (Hammer &
Kummerow, 1997). Authors of both instruments believed it is important that clients learn

a career exploration process based upon a self-exploration process rather than simply

20

attain a particular career goal (e. g., choosing a vocation) (Hammer & Kummerow, 1997).
Authors of both instruments believed that the knowledge of individual characteristics can
be used to not only evaluate the match with a possible job but also to inform the
counseling approach used with clients. For example, a study by Kivlighan, Hageseth,
Tipton, and McGovern (1981) found that matching counseling approaches to the client’s
Strong interest themes had a positive effect on counseling outcomes (Hammer &
Kummerow, 1997). Similar evidence has been reported for the application of MBTI
personality dimensions to counselor interventions (Hammer & Kummerow, 1997).
Authors of both instruments also attempted to broaden the scope of their applications
beyond purely career related domains to interpersonal and intrapersonal domains
(Hammer & Kummerow, 1997).
Strong Interest Inventog Overview of Applications and Research

Commensurate with the P-E model, the Strong is purported to be useful as a
vehicle for discussion, a stimulus for self-exploration, a tool to help people understand
themselves and their environments, and a basis for action plan formation (Harmon et al.,
1994). A recent survey of counseling psychologists found that 87 percent of respondents
use the Strong Interest Inventor (Katz, Joyner, & Seaman, 1999). F urtherrnore, they use
it more frequently than any other test (Katz et al., 1999). In fact, the Strong is widely
regarded as the “most scientifically sound, thoroughly researched, and widely used
interest inventory” (Harmon et al., 1994).

The Strong is a questionnaire that inquires about a respondent’s level of interest in
a wide range of familiar occupations, occupational activities, hobbies, schools subjects,

and types of pe0ple (Harmon et al., 1994). For each of the 317 items, the respondent is

21

asked to indicate his or her preferences from among three categories on an answer sheet.
The Strong uses a person’s responses to the items to compare her or his response patterns
to those of people of different types and in different occupations (Harmon et al., 1994).
Most people can answer with little difficulty the Strong’s questions about what they like
and dislike but few may understand how their individual patterns of likes and dislikes are
related to the patterns of people working in various occupations (Harmon et al., 1994).
The purported power of the Strong therefore rests on two assumptions: (1) that the day-
to-day activities typical of a specific occupation are reflected in the interests of the people
who are employed in it and (2) that those who have a similar pattern of interest will be
satisfied in that occupation if they have compatible values and the necessary knowledge
and abilities (Harmon et al., 1994).

Results of the Strong are organized into four areas: 6 General Occupational
Themes (GOT) that reflect the individuals overall orientation to work according to
Holland’s six interest types including Realistic, Investigative, Artistic, Social,
Enterprising, and Conventional; 25 Basic Interest Scales (BIS) that report consistency of
interests or aversions in areas such as art, science, and public speaking; 211 Occupational
Scales (OS) representing the degree of similarity between the individual’s interests and
the characteristic interests of individual’s within those occupations; and 4 Personal Style
Scales that measure aspects of the style in which individual’s prefer in relation to
learning, leading, working, and taking risks (Harmon et al., 1994).

The Strong has been the subject of hundreds of research initiatives spanning over

fifty years (Harmon et al., 1994). Research has focused on individual, occupational,

22

group, institutional, cultural, and societal interest pattern trends and shifts (Harmon et al.,
1994)
The Strong—studying individual change

Harmon et a1. (1994) noted that the Strong has been used to explore case studies
of interest pattern change over time wherein a single person is seen as representing a
generalized phenomenon of interest. For example, Harmon et a1. (1994) cited the case of
an individual whose Strong General Occupational Theme (GOT) scores shifted fi'om
Social to Investigative over a ten-year test-retest trial. Confounding variables influencing
the shift in GOT types included the fact that the individual had, over the course of the
ten-year period, engaged in a different style of collective decision-making within his
engineering research group and had attended training sessions to help him fimction more
effectively within his environment (Harmon et al., 1994). That interest types can change
therefore in response to workplace socialization is an important finding from case study
data. Such findings are supportive of P-E assumptions regarding the reciprocal nature of
person and environment dynamics (Harmon et al., 1994).
The Strong—studying goup change

In a study by the American Association of Medical Colleges (Hutchins, 1964),
2,800 medical students were administered the Strong upon entering and upon graduating
(Harmon et al., 1994). The results indicated that the students demonstrated decreased
interest in both scientific and social interests after having scored very high in these areas
initially and a mild increase in adventuresome interests (Campbell, 1971; Harmon et al.,
1994). These findings have served as a basis for further exploration into issues of

professional socialization. Research however has indicated that most people’s interests

23

as measured by the Strong are stable over short periods of time (Harmon et al., 1994).
Over two or three years, some changes may occur particularly in younger individuals yet
the inventory results generally reflect such changes (Harmon et al., 1994). When samples
tested in the 1930’s and 1940’s are compared with analogous samples tested in the 1970’s
and 1980’s, results indicated that the vocational interest patterns are more stable over
longer time spans than many originally believed (Harmon et al., 1994). That interest
themes can change in response to workplace socialization, particularly when individuals
are younger, yet tend to stabilize as individuals grow older is supportive of P-E
assumptions regarding the reciprocal nature of person and environment dynamics.
Furthermore, that level of reciprocity appears to be moderated by person characteristics
such as chronological age is an important contextual variable to consider within the self-
exploration and information P-E processes (Harmon et al., 1994).
The Strong—studyinmacteristic interests of people in particular occupm

Data from the Strong compiles detailed information about a myriad of
occupations in an effort to help people decide if they would be satisfied with employment
in a given occupation. For example, sampling data on the recent Strong (1994)
demonstrated that female geologists interests were concentrated within Investigative
academic areas such as science and mathematics as well as Realistic interests within
outdoorsettings involving mechanical activities (Harmon et al., 1994). Conversely, their
aversions were concentrated within interest areas traditional to the Social theme such as
social service, religious activities, and traditional home or family activities (Harmon et
al., 1994). They also demonstrated disinterest in Enterprising activities such as sales and

marketing (Harmon et al., 1994). The fact that interest types can clearly be defined and

24

strongly irnbedded support P-E assumptions regarding the importance of self-awareness
as well as the potential impact one might have on their environment and vice-versa
(Harmon et al., 1994).

The Strong—studying cross-culWaLiMumcx

The strong is widely used internationally. Holland and Gottfredson reported in
1992 that the Strong had been translated into seventeen languages (Harmon et al., 1994).
Various studies have used the Strong to compare occupational data from other countries
(Harmon et al., 1994). Examples included research performed with German-speaking
psychologists and accountants (e.g., Lonner, 1968, 1969), Pakastani physicians and
engineers (e.g., Shah, 1971), Irish students (e.g., Hanlon, 1971) as well as Spanish
translations of the Strong (F oud, 1984). The results of these studies demonstrated
similarity of interests across several countries among people in the same occupations
(Harmon et al., 1994). The fact that people tend to select occupations commensurate with
cross-culturally valid interest patterns supports the P-E premise that individuals seek out
and create environments that allow them to express preferred characteristics.

Shifting from a broad overview of general research and uses of the Strong
commensurate with the P-E model, discussion will presently focus on a review of the
theoretical foundations upon which the Strong rests-Holland’s (1959; 1985) theory of
vocational psychology (Hammer & Kummerow, 1997). A review of the evolution of
Holland’s theory, a critical evaluation of Holland’s theoretical constructs, and a brief
overview of the historical integration between Holland’s theoretical premises and the

Strong are paramount are foundational to the present study.

25

Holland’s TheonL—an overview ind critique
In an attempt to deve10p an occupational classification system closely tied to
psychometric research, Holland originally proposed in 1959 six categories of
occupational interests comprised of Realistic, Investigative, Artistic, Social, Enterprising,
and Conventional types (Harmon et al., 1994). The Realistic individual was believed to
possess interests primarily centered on building or repairing and hands on activities
typically within outdoor or natural settings (Holland, 1985). The Investigative type
theoretically was drawn toward researching and analyzing activities (Holland, 1985).
The Artistic type was believed to be primarily interested in creating and enjoying the arts
(Holland, 1985). The Social type was purportedly interested in helping and instructing
others (Holland, 1985). The Enterprising individual primarily enjoyed selling,
persuading, and managing (Holland, 1985). The Conventional type was hypothesized to
prefer processing and organizing activities (Holland, 1985). Most people, according to
Holland, could be characterized according to these six types. Furthermore, most work
environments could be categorized according to the same six types. Holland described
the evolution of his interest typology system as follows:
The formulation for the types grew out of my experience as a vocational
counselor and clinician, and out of my construction of a personality
inventory from interest material. After reviewing the vocational
literature—especially factor analytic studies of personality and vocational
interests—I concluded that it might be usefirl to categorize people into six
types. The present types are analogous in some ways to those proposed
earlier by Adler, Fromm, Jung, Sheldon, and others. They differ from
these earlier typologies in their origin—which is largely our vocational
literature—and in their definitions. The six major factors identified in
Guilford’s comprehensive factor analysis of human interest—mechanical,

scientific, social welfare, clerical, business, and aesthetic—approximate
the present types. To the best of my knowledge, Guilford’s factor analysis

26

is the most explicit forerunner of the present typology (Harmon et al.,
l994,p.52)

Career satisfaction was hypothesized to be the result of being clear about one’s
interests via the demonstration of identifiable type preferences, named differentiation by
Holland, possessing consistent interest themes via the possession of similar type
preferences, named consistency by Holland, and finding an occupational environment
congruent with one’s interest themes, named congruency by Holland. Holland’s theory
therefore is built upon the constructs of congruence, consistency, and differentiation.

Congruence appears to be the most widely studied and debated aspect of his
theory in recent years (Hackett & Lent, 1992). Congruency represents a matching of
one’s dominant interest type(s) with an occupation characterized by parallel interest
type(s). For example, a Realistic-Investigative individual establishes congruency through
placing oneself in an occupational setting characterized by Realistic and Investigative
activities and demands. Conversely, congruency is not achieved if the same individual
were to be located within an environment characterized by, for example, Conventional
and Enterprising activities. In a review of the congruence literature, Spokane (1985)
concluded that most studies found mixed or positive results for the congruency theory
(Hackett & Lent, 1992). Spokane (1985) concluded that, “on balance, congruence is
associated with performance, satisfaction, and stability as predicted by Hollan ” (Hackett
& Lent, 1992, p. 424). Based upon a meta-analytic review of 41 congruence studies,
Assouline and Meir (1987) discovered small effect sizes for the relations of congruence
to achievement and stability; mean correlations were .06 and .15 respectively (Hackett &

Lent, 1992). The mean correlation of congruence to satisfaction was .21 for general

27

occupational categories but rose steeply to .42 when congruence involved the fit between
person and occupation specialty (Hackett & Lent, 1992). Meir (1988) and Meir and
Yaari (1988) found stronger congruence-satisfaction relationships when congruence
involved the fit between personality type and occupational specialty as well (Hackett &
Lent, 1992). Several studies in recent years have reported positive relationships between
congruence level and job satisfaction in adults (e.g., Elton & Smart, 1988; Gottfredson &
Holland, 1990; Rounds, 1990) and educational satisfaction in students (e. g., Gade, Fuqua,
& Hurlburt, 1988 in Hackett and Lent, 1992). In their summary of the congruence
literature, Hackett and Lent (1992) stated, “the preponderance of evidence does suggest a
relationship between Holland’s personality types and choice of environment, which
generally holds for men, women, majority, and minority subjects in the studies we
reviewed” (p. 425). Hackett and Lent (1992) however also noted that the congruence
hypothesis has yielded mixed findings.

Consistency is said to occur when similar Holland interest types characterize an
individual’s personality Strong results. Inconsistency occurs when dissimilar types
characterize an individual’s personality type. Holland’s six personality types are
arranged along the points of a hexagon according to the acronym RIASEC. Holland
personality types nearest each other on the hexagon (e. g., Realistic and Investigative) are
theoretically more similar than are non-adjacent types. Furthermore, those types that are
diametrically opposite (e. g., Realistic and Enterprising) on the hexagon are theoretically
highly dissimilar. Most correlational research seems to corroborate Holland’s idea of
typology consistency. In a recent reference sample of 9, 467 women and 9,484 men,

Realistic and Investigative were highly correlated (.53) as were Enterprising and Social

28

(.42), Enterprising and Conventional (.41) and Social and Artistic (.33). Conversely, the
lowest rates of correlation existed among those types located opposite one another on the
hexagon. For example, Realistic and Social (.06), Enterprising and Investigative (-.01),
and Conventional and Artistic (-.04) (Harmon et al., 1994).

Holland’s idea of typology consistency has not been accepted without debate.
Consistency has generated the greatest amount of cross-cultural validity questions
(Rounds & Tracey, 1996). Most reviews on Holland’s hexagonal structure conducted in
the 1980’s and early 1990’s found support for the ordering of RIASEC (Rounds &
Tracey, 1996). For example, Hansen (1987) said that, “generally, the structure of
interests of international and cross-ethnic populations seem to correspond to Holland’s
model almost as well as does the structure of interests of Whites” (Rounds & Tracey,
1996, p. 311). Summarizing the literature, Holland stated that the “ordering of RIASEC
types is similar even when the data, sexes, and cultures vary” (Rounds & Tracey, 1996. P.
311). In the early to middle 1990’s however, the belief emerged that perhaps confidence
in the structure of Holland’s model and its applicability to the culturally diverse was
premature. As contradictory research emerged fiom F ouad & Dancer (1992) and
Swanson (1992), theorists and researchers adopted a less accepting outlook on the
structure of Holland’s theory and began proposing alternative structures (Rounds &
Tracey, 1996). Gati (1991) claimed that Holland’s order did not adequately account for
the relationships among the six types (Rounds & Tracey, 1996). Rather, according to
Gati (1991), a three-group partition [(RI) (AS) (EC)] demonstrated a better fit with cross-
cultural populations (Rounds & Tracey, 1996). According to Gati’s (1991) model,

correlations between types belonging to the same category (e. g., R, I) were greater than

29

correlation’s between pairs of types outside the category (e.g., R, A) (Rounds & Tracey,
1996).

Both Holland’s and Gati’s models have since received criticism for the
relationship ascribed between the A and S types (Rounds & Tracey, 1996). Several
researchers have noted the cross—cultural discontinuity between Artisitc (A) and other
Holland types. Rounds and Tracey (1996) attempted to address this discrepancy through
adjusting Gati’s (1991) model so that the A type existed within its own category and the
S type was added to the EC category [i.e., (RI) (A) (SEC)]. To test their hypothesis,
Rounds and Tracey (1996) conducted a structural meta-analysis to evaluate the fit of
Holland’s (1985) model, Gati’s (1991) model, and their own to cross-cultural
populations. From the meta-analysis, samples were derived representing individuals
from various cultures within the US. (N=l6; Afiican-American, American Indian, Asian-
American, and Hispanic), eighteen foreign countries (N=75), and individuals of European
descent within the US (N=20). Gati’s (1991) model and Rounds and Tracey’s (1996)
model achieved the best fit to the data. Both models represented the international
samples and the US. “benchmark” samples equally well and showed a better fit to the
international samples than did Holland’s RIASEC model (Rounds & Tracey, 1996).
Across the 18 countries in the international sample, both Gati’s (1991) and the Rounds’
and Tracey (1996) models fit equally well for 12 countries. According to Rounds and
Tracey (1996), none of the differences between these two models presented “a
compelling reason to recommend in general one model over the other” (p. 325). Rounds
and Tracey (1996) suggested that the results might be used to (a) counterpoint previous

conclusions supporting Holland structure across cultures and to (b) warn publishers,

30

authors, and assessors that the knowledge pertaining to consistency within Holland’s
model may not be applicable to cross-cultural groups. In 1998, however, Day and
Rounds seemed to offer contadictory evidence. Day and Rounds (1998) asserted that
people of different ethnicity and sexes hold the same cognitive map of the workworld
when the structures of preferences are explored. Furthermore, Day and Rounds (1998)
stated that the “RIASEC arrangement of general occupational themes prevails
legitimately. Previous research suggesting otherwise probably suffered multiple perils of
. . . small sample size, unrepresentative sampling, selective attrition. . .and attenuated
categorization” (p. 735).

A third important Holland construct-one closely linked with consistency--is
differentiation. Differentiation occurs when an individual clearly expresses a preference
for a Holland personality type(s) in relation to other types. For example, differentiation
occurs when an individual scores high on a Realistic interest pattern and scores low on a
Social interest pattern. Theoretically, consistency and differentiation will be evidenced
only when an individual’s self-identity is well developed. As Holland (1985, p. 73)
acknowledged, differentiation and consistency have had a “checkered research career—
about as many negative as positive findings” (Hackett & Lent, p. 425). A review by
Hackett and Lent (1992) corroborated this trend; neither of those constructs has been
found to uniformly predict career choice stability. Holland (1985, 1987b) argued
however that well-designed studies that closely followed his theory and employed
defensible outcome measures have produced the most findings in support of his theory
whereas studies lacking these qualities have often produced disconfirrning results

(Hackett & Lent, 1992).

31

In summary, the Holland hypothesis that seems to have received the most
consistent support involves the prediction that persons tend to select or prefer
environments that match their interest (Hackett & Lent, 1992). Conversely, “research has
not been entirely kind to the theory” (Hackett & Lent, 1992, p. 425). As Holland (1985)
acknowledged, the constructs of differentiation and consistency have had a “checkered
research career—about as many negative as positive findings” (p. 73). Yet Holland’s
theory has had “an undeniable impact” on psychology (Hackett & Lent, 1992, p. 425).
According to a review by Hackett and Lent (1992), the theory’s relative simplicity and
parsimony have made it accessible to professionals and layperson alike. Furthermore,
commensurate with a P-E approach, its classification scheme provides a useful
framework for organizing information about persons, environments, and their interactions
(Hackett & Lent, 1992).

Strong Interest Inventory LndI-iollanfi’s Occupational Theogy

At its inception in 1927, the Strong was an empirical yet atheoretical instrument
(Harmon et al., 1994). A criticism of the Strong in its early development was the fact that
it lacked a parsimonious organizational structure (Harmon et al., 1994). In 1974, the
General Occupational Themes (GOT) were first introduced into the Strong and
represented the merger of Holland’s work with that of E.K. Strong (Harmon et al., 1994).
The GOTs assess an individual’s interest patterns and yield a Holland personality theme.
The GOTs therefore represent a direct link to Holland’s theory on the Strong. The
merger between Holland’s theory and the Strong was widely considered a powerful

enhancement to the instrument (Harmon et al., 1994).

32

The GOT results are purported to be particularly usefirl when counseling or
advising beginning college students (Harmon et al., 1994). The GOTs can be used to
help those students who are not prepared to select a major to understand their general
interests (Harmon et al., 1994). The themes and their relationship on the hexagon
provide a cognitive map for students who are trying to select and academic major from
what is often a confusing web of options (Harmon et al., 1994).

Research has demonstrated the benefits of helping students acquire a general
cognitive framework for exploring interests (Harmon etal., 1994). Shahnasarian and
Peterson (1988) showed a video that interpreted the six themes to a group of students to
help them identify alternatives for consideration. The study concluded that the cognitive
schema offered by the six themes sensitized the students to information in their
environment and helped them encode as well as retrieve information (Harmon et al.,
1994). J acoby, Rue, and Allen (1984) applied the idea of cognitive mapping literally.
Specifically, the authors created a map for each of the six types highlighting academic
programs, career possibilities, internship and volunteer possibilities, campus
organizations, and recreational activities (Harmon et al., 1994). J acoby et a1. (1984)
stated that such maps could be used successfully in recruitment, orientation programs,
academic advising, and career counseling (Harmon et al., 1994).

Research has also demonstrated the benefits of using a client’s GOT information
to inform the counseling process. A study by Kivlighan, Hageseth, Tipton, and
McGovern (1981) found that matching counseling approaches to an individual’s GOT
had a positive effect on career counseling outcomes (Hammer & Kummerow, 1997).

Realistic clients, for example, preferred counseling to be concrete and anchored to

33

examples in real life (Hammer & Kummerow, 1997). Investigative types on the other
hand preferred to understand the patterns and relationships among the scales and how the
tests they have taken relate to one another (Hammer & Kummerow, 1997).
Using Additicmal Sources of Infomtion

Beyond eliciting information regarding an individual’s interest patterns, the
authors of the Strong endorsed integrating information about an individual’s personality,
values, and beliefs via MBTI results to provide a deeper understanding regarding what
careers and work environments an individual is likely to find satisfying and rewarding
(Harmon et al., 1994). Authors of the Strong noted, “using the Strong and the Myers-
Briggs Type Indicator has been shown to be particularly useful” (Harmon et al., 1994, p.
238). From a P-E perspective, use of these two instruments together can, according to
Strong authors, provide deeper levels of self-understanding leading to improved
information-processing and critical thinking procedures and help confirm particular areas
that warrant further exploration (Harmon et al., 1994).
Myers-Briggs Type Indicator and Vocational Psychology

The MBTI has long been used as a tool in career counseling (Hammer, 1996). A
primary goal of Isabel Briggs Myers in creating the instrument was to help individuals
choose work that they would find meaningful and at which they could be productive
(Hammer, 1996). Over the course of the last fifty years, the MBTI has become, along
with the Strong, the most widely used instruments in college counseling centers,
community colleges, private career or educational counseling practices, and high schools

(Hammer, 1996).

34

The MBTI’s popularity as a career-counseling tool has been well established. In
1991, the National Research Council (NT C) conducted a survey of individuals who had
received an interpretation of the MBTI as part of a career development program
(Hammer, 1996). The survey was conducted five months after the participants completed
the career development program (Hammer, 1996). Among the findings reported by the
NTC, 97 percent of respondents remembered taking the MBTI and remembered their
type; 84 percent felt the results confirmed what they already knew about themselves; 61
percent said that it was the most helpful aspect of the program; a number of respondents
reported that the best way to improve the overall program was to provide more
information on the MBTI; the ratings for the MBTI were significantly higher than those
given to any other instrument used within the program (Hammer, 1996).

The MBTI has been used to explore a number of topics pertinent to vocational
psychology such as occupational selection, career decision-making, and job satisfaction.
As will be discussed, the following P-E model assumptions have been directly addressed
and tested: (a) people and work environments differ in reliable and meaningful ways; (b)
the greater the level of congruence between person and environment characteristics, the
greater the likelihood of occupational satisfaction and success; (c) person and
environment fit is a reciprocal process in which individuals shape the environment and
the environment shapes individuals (Chartrand, 1991).

Myers-Briggs Type Indicator—studying occupational selection

P-E theory states that people and work environments differ in reliable and

meaningful ways (Chartrand, 1991). This is not meant to imply that a single type of

person resides within a given occupational context but that important patterns exist and

35

can be better used to help people organize themselves within generally congruent
environments (Chartrand, 1991). If MBTI personality theory is correct and
commensurate with P-E theory, MBTI personality types should be distributed across
occupations in ways that are consistent with the characteristics of that work environment
(Hammer, 1996). The data to date indicates that MBTI types are distributed across
occupations in ways that are generally consistent with theoretical predictions (Hammer,
1996). In general, the more an occupation focuses on certain tasks, the less equally
distributed the types will be across occupations (Hammer, 1996). Descouzis (1989)
examined a sample of tax preparers--an occupation that requires careful attention to detail
in combination with an ability to recall a myriad of pertinent facts (Hammer, 1996).

Type theory would predict that individuals whose dominant function was Introverted
with Sensing (i.e., ISTJ and ISFJ) would predominate in this field (Hammer, 1996).
Descouzis (1989) found that 100 percent of the sample demonstrated a preference for
Sensing versus Intuitive (Hammer, 1996). ISTJ and ISFJ types comprised a statistically
significant 44 percent of the sample (Hammer, 1996). This pattern can also be observed
across occupations within a given organization. Steckroth, Slocum, and Sims (1980)
studied ninety-six business executives (Hammer, 1996). Personnel and sales positions
were related to a preference for Feeling whereas economics, operations research, and
operations management were related to preferences for Thinking (Hammer, 1996). A
preference for Perceiving was related to sales and operations management and preference
for Judging to plamring positions (Hammer, 1996). Differentiation by MBTI type is most

noticeable within clearly defined occupations. Conversely, broad occupations such as

36

teaching may provide a satisfactory environment for a large number of different types
depending upon the subject being taught (Hammer, 1996).

The pattern of distributions of types by occupations holds across different cultures
(Hammer, 1996). Using a Norwegian translation of the MBTI, Nordvik (1994) found
that pilots and industrial workers, two occupations that required workers to engage in
highly similar tasks, had a statistically significant similarity in personality types
(Hammer, 1996). Eighty-eight percent of the pilots were STJs as were 96 percent of the
industrial workers (Hammer, 1996). Hammer (1996) concluded that the evidence to
support the proposition that occupational choice is related to the preferences as measured
by the MBTI is “abundant and compelling” (p. 37).

The MBTI—studying career decision mm

Gibb (1991) used the MBTI to explore career indecision among a sample (N=139)
of White, married, middle-class women (Hammer, 1996). Results indicated that
Perceiving types tended to report higher levels of career indecision and lower levels of
confidence in decision-making skills (Hammer, 1996). Lynch and Hepp (1985) explored
career decision-making in college students and found that higher scores in Perceiving
were associated with self-reports of uncertainty and indecisiveness (Hammer, 1996).
Similarly, DiRusso, Carney, and Bryan (1995) reported that among undergraduates
within a sample classified as being least decided, Perceiving types in general and those
with SP and TP preferences specifically were significantly overrepresented (Hammer,
1996). lkenberry (1987) reported a connection between career indecision and the J -P as
well as E-I scales (Hammer, 1996). Conversely, Apostal (1988) reported a link between

career decidedness and high Thinking scores as well as linkages between S-N and T-F

37

dichotomies and expressed interests and perceived competencies (Hammer, 1996).
McKenna, Martin, and Schmidt (1990) studied the MBTI type of those considering
retirement (Hammer, 1996). These authors found that Judging Types, Extraverted and
Judging types in particular, were significantly overrepresented among those rated high on
retirement planning (Hammer, 1996). ESTJ types achieved the highest score on the
retirement planning measure (Hammer, 1996).

In summary of the data using the MBTI to explore career decision making
readiness, Judging types consistently demonstrated more career decisiveness than did
Perceiving types (Hammer, 1996). Less definitive linkages have been made between
career indecisiveness and the E-I, T-F, and S-N dichotomies (Hammer, 1996). The fact
that the MBTI distinguishes degrees of career decision and indecision by the Judging and
Perceiving personality dimension preferences potentially informs the counselor as to
unique personality needs within each of the following P-E information processing steps:
(a) perceiving and appraising the current situation, (b) establishing concrete and realistic
goals, (c) developing a plan, and ((1) taking action (Chartrand, 1991).

The MBTI—studying iob satisfaction

One assumption inherent within the use of the MBTI and commensurate with a P-
E fiamework is that knowledge of person and environment characteristics can be learned
and used to inform individuals about the probability of satisfaction within a given
occupation (Chartrand, 1991). A number of studies have explored the effect of MBTI
personality characteristics and job satisfaction, coworker satisfaction, and occupational
stress. Hammer and Macdaid (1992) reviewed fifteen studies on the relationship between

the MBTI and career satisfaction within the MBTI Career Report Manual. Among the

38

studies reviewed, five showed no relationship between type and occupation satisfaction
(Hammer, 1996). Within these five studies, the occupational populations explored only
dieticians and teachers (Hammer, 1996). In the other ten studies, the relationship
generally supported the hypothesis that the MBTI is related to job satisfaction in ways
consistent with theory (Hammer, 1996). Furthermore, the T-F scale in particular
appeared to be highly related to satisfaction with coworkers (Hammer, 1996). Fitzgerald
(1994) studied managerial skills, stress and coping, and job satisfaction in 386 managers
in a leadership development program (Hammer, 1996). Using the Managerial Job
Satisfaction Questionnm, Fitzgerald (1994) found that extraverted managers reported
significantly more job satisfaction than did introverted managers (Hammer, 1996). Using
the Occupational Stress Inventog, Fitzgerald (1994) found that managers preferring INP
personality dimensions reported more stress around role insufficiency and role ambiguity,
those preferring NTP personality dimensions reported more stress with role boundaries,
and Introversion was associated with more stress in the area of responsibility (Hammer,
1996). Perceiving types reported more vocational and physical strain (Hammer, 1996).
Introverts and Perceiving types reported using fewer coping responses (Hammer, 1996).
Managers with clear scores in either direction on the J -P and T-F scales reported less
stress than those with unclear scores on the two dichotomies (Hammer, 1996). Managers
with clear scores on the LP dichotomy also reported higher levels of job satisfaction

(Hammer, 1996). Johnson (1991) selected 100 pastors and administered the Pastoral

 

Satisfaction Survey (PSS). Among the 63 pastors who responded, a preference for
Thinking predicted satisfaction with administrative tasks whereas Feeling was a predictor

of satisfaction with professional interaction (Hammer, 1996).

39

Across findings within the literature, Introverts and Perceiving types reported
higher levels of job dissatisfaction than did Extravert and Judging types. The T-F scale
appeared to be important in predicting satisfaction with coworkers (Hammer, 1996).
Knowledge of the relationship between MBTI personality dimensions and work
satisfaction, coworker satisfaction, and work stress is purported to be useful within a P-E
framework (Hammer & Kummerow, 1997). Specifically, such data could be used to
inform the information-processing stages and reinforce solid and generalizable decision-
making and interpersonal skills based upon knowledge of self (Chartrand, 1991).

Another assumption inherent within the use of the MBTI in career counseling and
derived directly fi'om the P-E model states that those who choose occupations wherein
the tasks and the individuals are congruent with their personality preferences are more
likely to derive satisfaction from the work environment (Hammer, 1996). Furthermore,
the person and environment fit is believed to be a reciprocal process in which individuals
attempt to shape the environment and the environment influences the individual
(Hammer, 1996). A number of studies have tested this hypothesis using the MBTI.
Watson (1990) used the MBTI to study the relationship between personality and job
satisfaction among 63 agriculture teachers. Data revealed that the teacher participants
were primarily SJ types. Furthermore, those individuals with SP temperaments reported
the lowest levels of job satisfaction (Hammer, 1996). Buie (1988) reported that among a
sample of computer professionals, a profession primarily populated by Intuitive
preferences and not by Sensing preferences, those who preferred Intuition were more
satisfied than were the Sensing types (Hammer, 1996). Richard (1994) surveyed 1,220

lawyers and found a link between Extraversion, Thinking, and Judging types and work

40

satisfaction (Hammer, 1996). Preferences for Sensing or Intuition were related to choice
of specialty but not to satisfaction (Hammer, 1996). Scarbrough (1987) found that SF
accounting managers, who were the least frequently occurring types in a sample of
accounting managers, were significantly more likely to report an intention to leave the
job than were the majority NT types (Hammer, 1996). Shewchuk and O’Connor (1995)
found that a sample of 522 health care executives were largely TJ types and had a more
positive sense of well-being than did non-TJ types within the sample (Hammer, 1996).
Marcie, Aiuppa, and Watson (1989) also studied health care executives and found that
those executives who had preferences similar to the majority in the organization
possessed higher levels of self-esteem (Hammer, 1996). Roush (1989) compared the
MBTI types of those who voluntarily resigned from the United States Naval Academy to
those who persisted at the Academy (Hammer, 1996). Voluntary resignation was
correlated with preferences for Introversion, Intuition, Feeling and Perceiving types
(Hammer, 1996). Findings lend support for the P-E supposition that stress and
dissatisfaction are a result of an individual being in a work environment that does not
reward his or her natural preferences (Hammer, 1996). As P-E theory predicts, in
dissatisfying situations, individuals attempt to modify the environment to meet their
preference needs and if unable to adequately do so, often leave (Chartrand, 1991).

Research on the MBTI appears to have consistently supported the following P-E
assumptions: (a) people and work environments differ in reliable and meaningful ways;
(b) the greater the level of congruence between person and environment characteristics,
the greater the likelihood of occupational satisfaction and success; (c) person and

environment fit is a reciprocal process in which individuals shape the environment and

41

the environment shapes individuals (Chartrand, 1991). The MBTI manual states that the
instrument is useful toward increased self-awareness and knowledge applicable toward
improved information-processing and decision-making skills (Myers & McCaulley,
1985). The manual also recommends conjoint use of the Strong toward these ends
(Myers & McCaulley, 1985).

The Strong and the MBTI: Together in Researph

Hammer and Kummerow (1997) hypothesized that the Strong and MBTI have
achieved an extensive level of use because counselors and clients have found that, true to
P-E model objectives, both instruments enable people to better understand themselves
and their environments. Individuals are able then to apply this increased understanding
of self and environment toward general information-processing and decision-making
procedures. As a result, they are better able to confront daily problems and decisions
(Hammer & Kummerow, 1997). Yet as more career professionals are using both the
Strong and the MBTI, Hammer and Kummerow (1997) concluded that more research is
needed on the conjoint use of the Strong and MBTI so that a better understanding of the
relationship between the two instruments could be achieved.

Five known and published studies have compared the effectiveness of using the
instruments together (Katz, Joyner, & Seamen, 1999). Four of the studies did not
establish a statistically significant benefit for using the two instruments together (e. g.,
Hayslip, 1995; O’Neil, Price, & Tracey, 1979; Takai & Holland, 1979; Talbot & Birk,
1979). However, only one (Talbot & Birk, 1979) of these four studies used a control
group (Katz et. al., 1999). Furthermore, three were published in the 1970’s before

revisions were made to both the Strong and the MBTI (Katz et al., 1999). Additionally,

42

one study (Hayslip, 1995) used only 37 participants divided across three treatment
settings (Katz et al., 1999). The most recent study (Katz et al., 1999) to explore the topic,
however, established evidence suggesting the efficacy of joint usage of the Strong and
MBTI. Katz et al. (1999) found that the instruments, when used in tandem, resulted in
higher rates of career related change (i.e., change in career goal, change in specificity of
career goal, and increase in certainty of career goal) as compared to individuals who
received information from a single instrument. Specifically, 85 percent of participants in
the dual-usage group changed their career goal, changed the specificity of their career
goal, and/or changed the level of certainty with reference to their career goal as compared
to approximately 67 percent of participants in conditions involving the administration of
only one instrument or neither instrument (Katz et al., 1999). The results of this study,
according to Katz et al., (1999), supported the P-E model’s belief in the importance of
self-understanding toward effective career development. In short, greater self-
understanding was purported to encourage the widest possible consideration of careers
and to maximize the chances of reaching a satisfying choice (Katz et al., 1999).
Generalizability of the results is limited by the fact that the participants were all
undergraduate students at the same institution (Katz et al., 1999). Possible nesting effects
were therefore ignored. Furthermore, whole classes were assigned to treatment
conditions (Katz et al., 1999). It’s possible therefore that confounding variables were
introduced based on characteristics of students in various types of classes (Katz et al.,
1999). Finally, individual score interpretations were not provided within the study (Katz
et al., 1999). Although professional interpretation of Strong and MBTI results is not

overtly mandated within either instrument’s manuals, a test-and-tell model of instrument

43

administration and interpretation is widely discouraged (Chartrand, 1991). Counselors
are encouraged to help clients make use of their functions through practice and not
through understanding alone (Myers & McCaulley, 1985).

In light of the each instrument’s emphasis on interpretive issues, several articles
have addressed the interpretive uses of the Strong and MBTI from a conceptual
perspective. For example, Pinkney (1983) highlighted ways in which the MBTI may be
used as an alternative to the Strong. Specifically, Pinkney (1983) stated that the use of the
MBTI in the career counseling process might lead to psychological awareness on the part
of the client, which can augment vocational information received from a more traditional
career instrument (Buboltz et al., in press). Lowrnan (1991) discussed the Strong and the
MBTI as examples of interest and personality measures that may be used in conjunction
with information on abilities toward the creation of a comprehensive career assessment
approach. Lowrnan did not, however, directly address the interpretation of the MBTI and
Strong together (Hammer & Kummerow, 1997). Miller (1992) used a case study to
discuss how to synthesize results from the Strong and the MBTI. Miller (1992) stated
that the use of the MBTI in conjunction with the Strong may lead to better psychological
understanding on the part of an individual and can lead to individual’s understanding the
interaction of their preferences with different work situations (Buboltz, Thomas, &
Johnson, in press). Similarly, Peterson (1995) discussed the integration of the two
instruments through the provision of case examples. Neither Miller (1992) nor Peterson
(1995) however attempted to discuss the MBTI and Strong relationship in a

comprehensive manner (Hammer & Kummerow, 1997).

Beyond conceptual and qualitative work on the interpretation of the Strong and
the MBTI, Hammer and Kummerow (1997) stated that more empirical data was needed
regarding the strengths and limitations of using both instruments simultaneously.
Hammer and Kummerow (1997) suggested, “there are number of topics that seem to
offer a logical beginning for future research efforts” (p. 125). A suggestion offered was
the exploration of the correlational relationship between the Strong and MBTI using large
samples of adults and students (Hammer & Kummerow, 1997). Because the Strong
results are built upon the foundations of Holland’s (1985) theory of vocational interests
and the six interest themes therein reflected on the Strong General Occupational Themes
(GOT), a logical starting place for exploring the correlational relationship between the
Strong and the MBTI seemingly exists within the exploration of correlations between the
Strong GOTs and the MBTI personality dimensions.

In the last two decades, multiple studies have attempted to explore the
relationship between Holland interest themes contained within the Strong GOTs and
MBTI personality dimensions. Some studies have explored the relationship via
correlating the four MBTI personality dimensions (i.e., Extraversion-Introversion [E-I],
Sensing-Intuitive [S-N], Thinking-Feeling [T-F], and Judging-Perceiving [J -P]) with the
six interest themes (i.e., Realistic [R], Investigative [I], Artistic [A], Social [S],
Enterprising [E], Conventional [C]). Others have explored the relationship via grouping
MBTI personality dimensions by type (e. g., INFP, ESTJ, etc) and assessing the
relationship of these four-letter types with the six Strong interest themes. The findings of

each however are limited by important theoretical and methodological considerations.

45

Several studies have used correlational or similar techniques to analyze the
relationship between the four MBTI personality dimensions (i.e., E-I, S-N, T-F, LP) and
the six Strong interest themes (R-I-A-S-E—C). McCauley (1978) found statistically
significant but low correlations between the majority of Jungian and Holland types on the
two instruments. Using the updated version of the MBTI, Dillon and Wiessman (1987)
employed Pearson correlations and found significant relationships between several of the
MBTI poles and Strong interest themes. Specifically, the MBTI Extraversion pole was
found to “significantly agree” (p. 72)-—defined by the authors as an agreement over 50
percent (p<.05)—with Strong Social and Enterprising themes for men and women and
Strong Conventional themes for men (Dillon & Weissman, 1987). The MBTI
Introversion pole was not found to significantly agree with any Strong interest themes
(Dillon & Weissman, 1987). The MBTI Sensing pole was found to significantly agree
with Conventional Strong interest theme for men and women and Strong Realistic
interest theme for men only (Dillon & Weissman, 1987). The MBTI Intuition pole was
found to significantly agree with the Strong Artistic interest theme for men and women
and the Strong Investigative and Social themes for women only (Dillon & Weissman,
1987). The MBTI Thinking pole was found to significantly agree with the Strong
Realistic interest theme for men and the Strong Investigative and Enterprising interest
themes for women (Dillon & Weissman, 1987). The MBTI Feeling scale significantly
agreed with no Strong interest themes for women and with Artistic and Social interest
themes for men (Dillon & Weissman, 1987). The MBTI Judging pole was found to
significantly agree with only the Conventional Strong interest theme for both men and

women (Dillon & Weissman, 1987). The MBTI Perceiving pole significantly agreed

46

with the Artistic Strong interest theme for men and women and the Social interest theme
for women only (Dillon & Weissman, 1987). Dillon and Weissman suggested that the
meaning of GOTs could be enhanced with the integration of Jungian typology scales to
potentially increase an individual’s self-awareness, underlying decision-making style, and
understanding of preference style.

A few years following the analysis performed by Dillon and Weissman (1987),
Decola (1992) performed a similar analysis and concluded that the MBTI types are not
good predictors of the Strong interest themes but that consistent relationships between the
two instruments do exist (Buboltz et al., in press). Specifically, Decola (1992) found
relationships between MBTI Extraverted and Feeling types with Strong Social interest
theme, MBTI Introverted and Sensing types with Strong Conventional interest theme,
and MBTI Intuitive types with Strong Enterprising and Conventional interest themes
(Buboltz et al., in press). Corresponding findings between Decola (1992) and Dillon and
Weissman (1987) were achieved with both purporting significant correlations between
MBTI Extraversion and Strong Social interest theme, MBTI Sensing and Strong
Conventional interest theme, and MBTI Feeling and Strong Social interest theme interest
theme. All other purported significant correlations between the two studies, however,
failed to correspond.

In 1997, Hammer and Kummerow compiled correlations of the MBTI personality
dimensions with the Strong interest themes for 757 adults and college students. Within
their large sample, Hammer and Kummerow (1997) reported “moderate” (p. 41)
correlations between MBTI Extraversion and Strong Social and Enterprising interest

themes, MBTI Intuition and Strong Artistic interest theme, MBTI Sensing and Strong

47

Conventional interest theme, MBTI Thinking and Strong Realistic interest theme, MBTI
Feeling and Strong Artistic and Social interest themes, MBTI Perceiving and Strong
Artistic interest theme, and MBTI Judging and Strong Conventional interest theme.
Hammer and Kummerow (1997) concluded that because the correlations of the Strong
GOTs with the MBTI personality dimensions were symmetrical across the Holland
hexagon, the pattern of correlations attained from the data offered evidence of convergent
and divergent validity. In short, opposite preferences on MBTI scales tended to correlate
with interest themes diametrically opposed on the Holland hexagon (Hammer &
Kummerow, 1997). In comparison to previous studies, Hammer and Kummerow (1997)
were the only researchers to emerge with such broad and theoretically supported
correlation patterns. In comparison to the findings of Dillon and Weissman (1987) and
Decola (1992), additional evidence was asserted for the existence of the following
correlations: MBTI Extraversion and Strong Social interest theme, MBTI Sensing with
Strong Conventional interest theme, and MBTI Feeling with Strong Social interest theme.
Within the last year, Buboltz, Thomas, and Johnson (in press) revisited the
relationship between MBT I personality dimensions and Strong interest themes but used
MANOVA analysis as opposed to Pearson correlation to do so. The dominant Holland
theme served as the independent variable and the Extroversion-Introversion, Judgment-
Perception, Thinking-Feeling, and Sensing-Intuition continuous scores of the MBTI
served as the dependent variables. Significant main effects based on the Holland code
were found for three of the four dependent MBTI variables. Specifically, Buboltz et al.
(in press) ascertained significant relationships between MBTI Extroversion and Strong

Enterprising interest theme, MBTI Feeling and Strong Social as well as Artistic interest

48

themes, and MBTI Intuitive and Strong Artistic interest theme. Data from Buboltz et
al.(in press) corresponded with Dillon and Weissman (1987) and Hammer and
Kummerow (1997) for the following relationships: MBTI Intuition and Strong Artistic
interest theme, MBTI Extraversion and Strong Enterprising theme, and MBTI Feeling
and Strong Artistic interest theme. Data from Buboltz et al. (in print) corresponded only
with Decola (1992) on the purported relationship between MBTI Feeling and the Strong
Social interest theme. Universal agreement among Buboltz et al. (in press), Hammer and
Kummerow (1997), Decola (1992), and Dillon & Weissman (1987) was obtained only for
a reported correlation between MBTI Feeling and the Strong Social interest theme.
Within the Dillon and Weissman (1987) analysis however, the correlation between MBTI
Feeling and Strong Social interest theme was significant only for males. Buboltz et al.
(in press) concluded from their analysis that the majority of the differences within their
data occurred for individuals with either Artistic or Social Strong interest themes. These
differences in personality, according to Buboltz et al. (in press), were in expected
directions. For example, commensurate with Dillon and Weissman (1987) and Hammer
and Kummerow (1997) findings, Artistic individuals demonstrated higher levels of MBTI
Intuition than all other codes. Commensurate with Dillon and Weissman (1987), Decola
(1992), and Hammer and Kummerow (1997) findings, individuals with a Strong Social
interest theme demonstrated higher levels of MBTI Feeling than all other Strong interest
themes (Buboltz et al., in press). Taken together, these results indicated with regard to
the MBTI scales, both Artistic and Social interest themes might possess unique

personality profiles (Buboltz et al., in press).

49

Above and beyond the fact that questionable levels of consistency existed across
research findings, the validity of the reported findings by McCauley (1978), Dillon and
Weissman (1987), Decola (1992), Hammer and Kummerow (1997), and Buboltz et al.,
(in press) is limited by at least one important consideration: bivariate correlational
analysis and/or MANOVA have limited validity when applied to MBTI personality
dimensions and Strong interest themes (Frank, personal communication, May 2001). A
bivariate correlational design or MANOVA design fails to capture the full depth and
breadth of relationships occurring among and between the MBTI personality dimensions
and the Strong interest typologies (Frank, personal communication, May 2001). The
bivariate correlational analysis such as that employed by, for example, Dillon &
Weissman (1987) and Hammer and Kummerow (1996) simply correlated isolated MBTI
personality dimensions by independent Strong interest themes and in turn, failed to
consider the interactive nature of each. The use of MANOVA, as employed by Buboltz
et al., (in press), created a similar problem in that it permitted only one Strong interest
theme to serve as a predictor variable at any one time and in turn, failed to fully consider
interest themes vis-a-vis other interest themes. It also treated the MBTI variables as
categorical, in comparison to continuous, and therefore reduced variability.

Evidence abounds from the development of both the MBTI personality
dimensions and the Strong interest themes speaking to the interactive and dynamic nature
inherent within both descriptive systems. To illustrate the interactive nature of the Strong
interest theme system, Holland (1985) noted in a revision of his theory that a
classification system comprised of only six independent themes was insufficient to firlly

represent human and occupational workplace diversity (Harmon et al., 1994). Holland

50

therefore (1985) expanded his vocational interest classification system to include
combinations of the six themes such as Realistic-Artistic and Social-Enterprising. Based
upon the evolution of Holland’s theory, the General Occupational Themes on the current
version of the Strong are ipsative. Practically stated, the six GOTs are essentially ordinal
in nature and the continuous score on any theme is relevant only when compared to the
scores of the other 5 themes (Buboltz et al., in press).

To illustrate the multivariate nature of the MBTI personality dimensions, Myers
and McCaulley (1998) discussed the fact that in developing personality theory,

Jung, Myers, and Briggs were constantly attentive to the dynamic
character of the personality types they were describing. The richness,
depth, and breadth of their descriptive systems result from the dynamic
interplay of the functions and attitudes inherent in each type. It is
important to bear this in mind in applying MBTI theory and interpreting
the MBTI personality inventory (p. 23).

MBTI theory asserts that the interactions of MBTI personality dimensions result
in behaviors that cannot be predicted from exploration of each preference separately
(Myers & McCaulley, 1998). Because each preference in a given type interacts with all
of the other preferences in that type, MBTI personality theory asserts that any
combination of two or more preferences will yield behaviors that are more than the sum
of their parts (Myers & McCaulley, 1998).

The study of the refinement of Jung’s personality theory demonstrates the
importance he appeared to place on the interactive nature of his evolving personality
dimensions. Initially, Jung’s attempt to explain individual differences in personality

stemmed from his observation that there were two types of people: extraverts and

introverts (Myers & McCaulley, 1998). However, ten years after his initial observations,

51

Jung noted that these two types did not provide a complete picture (Myers & McCaulley,
1998). Jung stated,

What struck me now was the undeniable fact that while people may be classified

as introverts or extraverts, this does not account for the tremendous differences

between individuals in either class. So great, indeed, are these differences that I

was forced to doubt whether I had observed correctly in the first place. It took

nearly ten years of observation and comparison to clear up this doubt (Jung,

1921/1971, p. 535; Myers & McCaulley, 1998, p. 29).

Jung ensuingly subdivided his initial extravert and introvert types into six types
by identifying Sensing versus Intuitive type and Thinking versus Feeling type (Myers &
McCaulley, 1998). Myers and Briggs extended Jung’s model by adding the Judging
versus Perceiving dichotomy (Myers & McCaulley, 1998). According to MBTI
personality theory therefore, there are unique relations among the personality dimensions
that suggest characteristics or behaviors that may not be apparent at a univariate or
bivariate level of analysis (Myers & McCaulley, 1998). Myers and McCaulley (1985), for
example, interpreted a high score on Thinking differently when it was paired with
Sensing than when it was paired with Intuition. Hammer (1996) therefore stated,
“researchers are implored to make more use of (MBTI) preference combinations or
(MBTI) whole types in their research” (p. 50).

When introducing the review of existent MBTI personality and Strong interest
theme research, it was stated within the present manuscript that multiple studies have
attempted to explore the relationship between the two. To this point, studies exploring the
relationship between the four MBTI personality dimensions (i.e., Extraversion-
Introversion [E-I], Sensing-Intuitive [S-N], Thinking-Feeling [T-F], and Judging-

Perceiving [J -P]) with the six interest themes (i.e., Realistic [R], Investigative [I], Artistic

52

[A], Social [S], Enterprising [E], Conventional [C]) via Pearson correlation or MANOVA
have been reviewed. Cognizant of dynamic and interactive relationships within MBTI
personality dimensions and within Strong interest themes, other researchers have
explored the relationship between MBTI personality dimensions and Strong interest
themes via grouping MBTI personality dimensions by type (e. g., INFP, ESTJ, etc) and
assessing the relationship of these four-letter types with the six Strong interest themes. It
is upon these studies that attention will now be focused.

Based upon a sample of 468 college students, Hammer and Kummerow (1997)
calculated the three MBTI four-letter personality types with the highest mean score
within each of the six GOTs and ranked them in descending order. According to this
methodology, the following results emerged: (a) GOT Realistic was correlated with
MBTI types ISTP, ENTP, and INTP; (b) GOT Investigative was correlated with MBTI
types INTJ, INTP, and ISFP; (c) GOT Artistic type was correlated with MBTI types
INFP, ENFP, and ENFJ; (d) GOT Social type was correlated with MBTI types ENFJ,
ENFP, INFJ; (e) GOT Enterprising type was correlated with MBTI types ESTJ, ISFJ, and
ENFP; (f) GOT Conventional Type was correlated with MBTI types ESTJ, ISFJ, and
INTJ. Social and Enterprising Strong interest types were correlated with eleven of the
sixteen types within the data cited, corroborating earlier findings that college students
generally have higher mean scores on the Social and Enterprising Strong scales than do
adult samples (Hammer & Kummerow, 1997). Yet the finding also presented some
theoretical conundrums. First, Strong interest theme Conventional was the highest theme
code for the MBTI type INTJ yet the dominant MBTI function within the INT J type,

Intuition, was theoretically associated with Strong interest theme Artistic. Bivariate

53

correlational analyses by Dillon (1987) and Hammer and Kummerow (1997) as well as
MAN OVA analysis by Buboltz et al. (in print) also documented moderately significant
correlations between MBTI Intuition and the Strong Artistic interest theme. Second, the
Strong Social interest theme was the highest theme for MBTI type INTP (Hammer &
Kummerow, 1997). Yet the dominant function of Thinking and the dominant Attitude of
Introversion theoretically, according to the authors, aligned the INTP type more with
Strong interest type Conventional. In contrast to the authors’ theoretical prediction
associating INTP and Conventional, previously discussed bivariate analyses uniformly
stated that MBTI Introversion did not significantly correlate with any Strong interest
themes (Dillon & Weissman, 1987; Decola, 1992; Hammer & Kummerow, 1997;
Buboltz et al., in press) and MBTI Thinking correlated with the Realistic Strong interest
theme (Hammer & Kummerow, 1997). According to these results therefore, INTP could
theoretically also be aligned with the Strong Realistic interest theme if any theme at all.
Although bivariate correlation analysis and the present typology analysis
contained several inconsistencies across findings, some consistencies also existed.
Bivariate analyses displayed moderate but significant correlations between MBTI
Intuition (Dillon, 1987; Hammer & Kummerow, 1997; Buboltz et al., in press), MBTI
Feeling (Dillon, 1987; Hammer & Kummerow, 1997; Buboltz et al., in press), MBTI
Perceiving (HMer & Kummerow, 1997) and the Artistic Strong interest theme. Within
the present Hammer and Kummerow (1997) analysis, two of the three four-letter MBTI
types associated with the Artistic Strong interest theme were INFP and ENFP. The third

type identified was ENFJ (Hammer & Kummerow, 1997).

54

Based upon data derived from 426 college students, Buboltz, Thomas, and
Johnson (in press) employed Chi-square analysis to determine significant relationships
among MBTI four-letter personality types and Strong interest themes. Buboltz et al. (in
press) Chi-square analysis revealed that the sixteen MBTI types and the Strong GOT
types were not independent (Buboltz et al., in press). Several of the MBTI types appeared
to cluster within one or two of the Strong interest themes. Specifically, individuals with
the ENTJ personality type appeared to cluster in the Strong Artistic and Enterprising
interest themes (Buboltz et al., in press). The INFP type tended to cluster in the Strong
Artistic and Social interest themes (Buboltz et al., in press). The ESFJ type appeared to
cluster with the Strong Social interest theme (Buboltz et al., in press). Significant
differences between the six Strong interest types by the sixteen MBTI personality types
were difficult to discern among the following personality types: (a) INTJ, (b) ISTP, (c)
ESTP, (d) ESFP, (e) and ENTJ (Buboltz et al., in press). Interestingly, no four-letter
MBTI type placements by Strong interest themes corresponded between Buboltz et al. (in
press) and Hammer and Kummerow (1997). Regarding other findings of note between
the two analyses, both placed only MBTI Introverts within Strong Investigative.
However none of the previously discussed bivariate or MAN OVA analysis found any
correlation neither between MBTI Introversion and the Strong Interest themes nor
between Strong Investigative and the MBTI personality dimensions. Corresponding with
bivariate and MANOVA analyses findings, both placed MBTI Introversion exclusively
within the Strong Artistic interest theme and MBTI Feeling exclusively within the Strong

Social interest theme.

55

In addition to the questionable levels of consistency between the findings of
Hammer and Kummerow (1997) and Buboltz et al. (in press), the validity of the reported
findings of these two typology studies is limited by at least one important consideration:
by taking a pure typology approach within the statistical design, investigators have made
MBTI personality dimensions and Strong interest types discrete as opposed to continuous
and have therefore overlooked the subtle complexity of the potential relationships among
MBTI personality dimensions and Strong interest themes. For example, individuals
scoring at the 5th and 35th percentiles of the Thinking-Feeling MBTI scale is treated the
same (Healy, 2000). In contrast, interpreting the score as continuous assumes that people
with higher scores are more likely to show a given personality quality than people with
lower scores regardless of whether they are in the same or different category (Healy,
2000). Pertaining to the MBTI, Healy (2000) explored the validity of categorical (i.e.,
type) analysis versus continuous (i.e., personality dimensions) analysis. The analysis
revealed that type, or categorical, MBTI score interpretation was consistent with two of
the four expected relationships between the MBTI personality dimensions and Strong
interest type scores for women but not for any of the relationships for men within the
sample (Healy, 2000). Healy (2000) concluded that because his statistical design may
have capitalized on sampling error and the findings for male participants did not support
categorical interpretation, replication would be appropriate.

Strong interest themes and MBTI personality dimensions—a call for new research

Researchers have attempted to gain a better understanding of the relationships
between MBTI personality dimensions and Strong interest themes via bivariate

correlational analysis, typology analysis, chi square analysis, and profile analysis. By

56

virtue of not adequately exploring or accounting for potential relationships within each
classification system and minimizing variability therein through the use of categorical
variables or bivariate analyses, firller understandings of the potential relationships
between the MBTI personality dimensions and the Strong interest types have not been
reached. In short, bivariate correlational analyses and the use of categorical data are, in
the opinion of this author, not fully appropriate toward the development of deeper
understandings of the relationship between MBTI personality dimensions and Strong
interest themes. The existence of a gap in the field’s understanding of the relationship
between the MBT I personality dimensions and the Strong interest types is therefore being
asserted. The relevance of exploring this gap hinges upon the fact that the Strong and the
MBTI are the two most frequently used vocational assessment instruments across a
variety of settings (Hammer & Kummerow, 1997). Furthermore, both instruments
promote conjoint usage.

The current proposed study seeks to distinguish itself from previous research
through exploring the relationship between and among MBTI personality dimensions and
Strong interest themes fiom a multivariate perspective. The present analysis will reveal
which MBTI and Strong variables most strongly covary. Practically therefore, results
will highlight those variables that “go together” most frequently. To do so, canonical
correlation will be employed to address the following research question: “to what extent
and in what ways do MBTI personality dimensions overlap with Strong Interest themes?”

The question will be addressed in three steps. First, canonical correlation will
treat the four MBTI dimensions (i.e., E-I, S-N, T-F, J-P) as one set X and will treat the

six Holland codes as one set Y. The treatment of each as a set, as Opposed to independent

57

constructs, is theoretically sound. Second, one new variable will be created for set X
(MBTI dimensions) and will be called the function of X {f(X)}. It will be comprised of
those MBTI personality dimensions that most strongly correlate with one another in
relation to Strong interest themes. One new variable will also created for set Y (Holland
themes) and will be called the function of Y {f(Y)}. It will be comprised of those Strong
GOTs that most strongly correlate with one another in relation to MBTI personality
dimensions. Third, the degree of correlation between the two functions will be
computed.

Applied to the present study therefore, at least three questions are being asked and
directly addressed within the larger research question. First, which MBTI personality
dimensions overlap (i.e., covary) with one another in relation to Strong interest types?
Second, which Strong interest themes overlap (i.e., covary) with one another in relation
to MBTI personality dimensions? Third, how strongly do the MBTI dimensions and
Strong dimensions overlap? These questions have not been, to the best of the author’s
knowledge, previously addressed within the literature.

Hypothesizing regarding potential answers to research questions one and two is
theoretically laden and fraught with errant empirical foundations. First, predicting which
MBTI personality dimensions overlap with one another, aside from which overlap within
the context of Strong interest themes, presents a conundrum. Jung’s concept of the
personality dimensions grew from his “empirical observations over many years” (Myers
& McCaulley, 1998. p. 23) from which Jung reportedly concluded, “I distinguish these
functions from one another because they cannot be related to or reduced to one another”

(Myers & McCaulley, 1998, p. 23). Second, the prediction of which Strong interest types

58

overlap with one another in the context of MBTI personality dimensions has not been
previously explored in the literature. From a bivariate perspective, the prediction of
which GOTs are related to one another has a clearer empirical basis for hypothesis
formulation. For example, according to Holland’s hexagonal model and principal of
consistency, Realistic and Investigative as well as Enterprising and Social should share
the highest correlation coefficients. Conversely, Conventional and Artistic should be
weakly correlated (Holland, 1985). All other dimensional relationships remaining should
share a moderate, at best, degree of correlation (Holland, 1985). However, as was
previously stated, the effect of the MBTI dimensions on these relationships is unknown at
present. Third, one may look to previous research for guidance but findings are based on
bivariate or categorical analyses of the two constructs. As has been discussed, bivariate
research has commonly reported the following moderate relationships between the MBTI
personality dimensions and the Strong interest themes: MBTI Extravert and Strong
Social, MBTI Intuitive and Strong Artistic, MBTI Feeling and Strong Artistic, MBTI
Feeling and Strong Social, and MBTI Sensing and Strong Conventional. These
relationships should be expected to be valid and reliable in their entirety only if the MBTI
personality dimensions and Strong interest themes do not significantly correlate from a
multivariate perspective (i.e., if MBTI personality dimensions should be considered as
isolated poles and Strong interest themes as isolated types).

The third question addressed via canonical correlation pertains to the extent of
overlap between correlated NHBTI personality dimensions and Strong interest themes.
Hammer and Kummerow (1997) reported bivariate correlations between MBTI

preferences and Strong interest themes ranging 0.00 to .48. Specifically, Strong Realistic

59

was moderately correlated with MBTI Thinking whereas its opposite, Strong Social, was
moderately correlated with MBTI Feeling and Extraversion (Myers & McCaulley, 1998).
Strong Artistic was moderately correlated with MBTI Intuition, Perceiving, and Feeling
whereas its opposite—Strong Conventional—was moderately correlated with MBTI
opposites Sensing and Judging (Myers & McCaulley, 1998). Strong Enterprising was
moderately correlated with MBTI Extraversion (Myers & McCaulley, 1998). Formally
stated, the research hypothesis asserts that the correlation between the two sets (i.e.,
MBTI personality dimensions and Strong interest themes) and the percent of variance
accounted for within the first canonical correlation will be considerably higher than any
percentage of variance accounted for by a bivariate regression and correlation analysis
procedures.

The relevance of the present study is as follows: it is the first known study to
explore the multidimensional complexity of the relationships between MBTI personality
dimensions and Strong interest themes from the perspective of (a) the ways in which each
of the two interest and personality sets represented on the Strong and MBTI overlap (i.e.,
covary) and (b) the extent of the overlap between the two sets.

Practically, counselors can gain a theoretically valid understanding as to what
MBTI personality dimensions and Strong interest themes tend to go together.
Correlations between MBTI personality dimensions and Strong interest types have been
used, among other things, in the past to suggest counseling strategy and treatment
planning. Using bivariate correlation analysis data, Hammer and Kummerow (1997
identified typical versus unusual MBTI and Strong combinations. Based upon this data,

the authors suggested that when a client’s Strong and MBTI results indicated an unusual

60

combination (i.e., in the opposite direction of the correlations), more personal exploration
around identity issues was suggested (Hammer and Kummerow, 1997). Hammer and
Kummerow (1997) also argued that the joint use of the Strong and the MBTI provided
information regarding counseling strategies. For example, with clients whose MBTI type
and Holland codes were congruent, counseling strategies were purported to be clearer
than with those whose results were incongruent (Hammer & Kummerow, 1997). An
ESTJ client with GOT dominant themes Conventional and Enterprising was purported to
appreciate a structured and detailed career search strategy whereas an INTJ client with an
Investigative and Realistic GOT dominant theme was hypothesized to prefer more
individual work encouraging research and reflection (Hammer & Kummerow, 1997).
However, the examples drawn by Hammer and Kummerow (1997) are based upon
correlation research drawn from analysis of MBTI poles as opposed to dimension
combinations. More information is therefore needed highlighting patterns of covariance

between interest and personality.

61

CHAPTER THREE
METHODS

Participants

The sample consisted of 423 undergraduate state university students within the
western region of the United States. Forty-five percent (192) of the individuals in the
sample were males, 54% (231) were female and the remainder of the sample (n=3), who
were excluded from the analysis and were not counted toward the total, did not indicate
gender on their assessment materials. The racial and ethnic demographics of the sample
were unknown. However, general setting characteristics suggest the sample was
primarily White. The university within which the sample was derived announced a year
2000 enrollment of approximately 26,000 students. Students were enrolled at the
university representing all 50 states and 102 foreign countries. The international student
population in the year 2000 was reported to be approximately 1,600. The majority (i.e.,
approximately 18,000) of students however were reported to be in-state students
representing all counties within the state. The state, according to year 2000 estimates,
reported population of approximately 2,000,000 people. Individuals identifying as White
comprised approximately 1.9 million of the population totals. Individuals identifying as
Asian (apx. 37,000) or as American Indian (apx. 30,000) were the next most frequently

occurring population groups.

62

College students were selected for two reasons. First, the design and hypotheses
of the study are an outflow of previous research largely employing college students as the
sample. Second, the demand for vocational counseling services appears to be especially
high on college campuses and therefore particularly relevant to this population (e.g.,
Slaney & MacKinnon-Slaney, 1993 in Hammer & Kummerow, 1997).

Procedures

Data for this study were obtained from undergraduate students taking a career
development and planning class at a large state university in the western region of the
United States. All students who participated in the class were required to complete the
Strong Interest Inventory (Harmon, Hansen, Borgen, & Hammer, 1994) and the Myers-
Briggs Type Indicator (Myers & McCaulley, 1985). Human Subjects approval was
obtained to use the data, which was stored on a computer in the testing center of the
university. Once the data was downloaded from the computer, the instruments were
matched for each individual. After the matching was completed, the scores on the MBTI
indexes were converted to continuous scores as per the procedure outlined in the Manual
(Myers & McCaulley, 1985). Within the manualized transformation, a midpoint was set
at 100 (Myers & McCaulley, 1985). Preference scores for E, S, T, and J were subtracted
from 100. Preference scores for I, N, F, and P were added to 100. The same procedure
was employed for the three other polar dimensions of the MBTI. Once the
transformations were completed for the MBTI, results from both instruments were then
entered in a data file for purposes of statistical analyses. Although continuous variables
will be used for the present proposed study, MBTI type categories can be determined

based on these continuous scores. For example, on the E-I dimension a score below 100

63

would be typed as Extravert (E), while scores above 100 would be typed Introvert (I)
(Myers & McCaulley, 1985).

The sample was diversified in terms of both personality types and vocational
interests with fairly adequate representation of most types and interests. With regards to
the MBTI dimensionality, the sample was found to be 47.2% Extravert, 54.2% judging,
46.2% sensing, and 44.1% thinking. Similarly, the continuous scores for these four MBTI
scales were 102.5 (SD = 27.4), 100.2 (SD = 27.4), 102.0 (SD = 27.1, and 100.6 (S_ =
23.8) respectively. Noting once again that a continuous score of 100 indicates the
dividing point between Extravert/introvert, judging/perceiving, sensing/intuition, and
thinking/feeling, the sample was almost evenly split on all four dimensions of the MBTI.
Measures
Strong Interest Inventory (Harmon, Hansen, Borgen, & Hammer, 1994).

The Strong contains 317 items that assess interests in occupations, interests in
different types of people and environments, and interests in leisure activities. Results of
the Strong are organized into four areas: Six General Occupational Themes (GOT), 23
Basic Interest Scales, 162 Occupational Scales, and 4 Personal Style Scales (Harmon,
Hansen, Borgen, & Hammer, 1994). For the purpose of the present study only the six
GOTs were explored (Harmon et al., 1994).

The six General Occupational Themes (GOTs) are based on Holland’s (1985)
theory of vocational types and reflect a person’s overall occupational orientation. The six
types are as follows: (a) realistic — people who like action, are concrete, adventurous and
who have areas of interest which include mechanical, construction and repair activities;

(b) investigative - people who prefer to work alone, have a strong scientific orientation

and enjoy discovering new ideas and facts; (c) artistic — people who value aesthetic
qualities, like self-expression, and enjoy being spectators or observers; ((1) social — peOple
who enjoy working with others, are caring and nurturing, and solve problems through
discussion; (e) enterprising — pe0ple who are verbally oriented and enjoy selling, seeking
positions of power and leadership; and finally (f) conventional - people who like
activities that require organization and attention to detail and accuracy (Buboltz, Thomas,
& Johnson, 01).

The GOT scales were normed on a general reference sample consisting of 18,951
employed adults selected on the basis of job satisfaction, job experience, typicality of job
description, and age (Harmon et al., 1994). The general reference sample was purported
to be randomly constructed and balanced for gender as well as being equally
representative of the fifty new occupations added to the Strong (Harmon et al., 1994).
Sixty-two percent of women in the sample reported being “very satisfied” in their
occupation; the percentage of men reporting the same was 58.7 (Harmon et al., 1994).
The remainder of the sample, both male and female, reported being “somewhat satisfied”
(Harmon et al., 1994). Women in the sample averaged 13.8 years of experience in their
occupation and men averaged 18.2 years of experience (Harmon et al., 1994). The
average age of women in the sample was 40.5 years and average age of male participants
was 44.6 years (Harmon et al., 1994). Within the total general reference sample, 18,789
reported their ethnicity. Eight percent (n=l,424) were non-Caucasian, two percent
(n=3 78) were African-American, two percent (n=363) were Hispanic/Latino American,
less than one percent (n=80) was Native American/American Indian, and approximately

1.4 percent (n=254) was of other ethnicity (Harmon et al., 1994). Across the fifty

65

occupations represented by the reference sample, 26 had five to ten percent ethnic
minority representation, 9 occupations had more than ten percent ethnic minority
representation, and 15 occupation had less than five percent ethnic minority
representation (Harmon et al., 1994). The highest ethnic minority representations were
for translators (26.7%), childcare providers (20.3%), and flight attendants (19.5%). The
lowest minority representations were for farmers (2.1%), veterinarians (2.3%) and
accountants (2.3%) (Harmon et al., 1994).

The GOT scales were not appreciably changed for several years until the 1994
revision (Harmon et al., 1994). For the 1994 Strong revision, the new GOT scales were
constructed by deleting items weakly correlated with the existing scales and adding items
that were more highly correlated (Harmon et al., 1994). On the new GOTs, Cronbach’s
alpha was increased to .90 or better for all six scales, a level previously reached by only
two scales (Harmon et al., 1994). Test-retest reliabilities were also improved. When all
individuals in the 1994 general reference sample were scored on the 1985 and the 1994
GOT scales, the test-retest correlations were .95 for females and .96 for males (Harmon
et al., 1994). The alpha reliabilities for the general occupational themes range fi'om .90
for the Social scale to .94 for the Artistic scale (Harmon et al., 1994). Test-retest
reliabilities were examined over 3 and 6-month time intervals and ranged from .84 for
Enterprising to .92 for Realistic (Harmon et al.). Several research studies have examined
the validity of the GOTs and consistently have concluded that the General Occupational
Themes of the Strong provide the best measure of the Holland RIASEC hexagon (Tracey
& Rounds, 1993). Futhermore, Harmon et al. (1994) stated that the GOTs appear to

separate occupations in accordance with theoretical expectations. Donnay and Borgen

66

(1996) stated that “it is possible to accurately predict exact occupational group
membership fiom the Strong. . .people in diverse occupations show large differences in
likes and dislikes, whether in terms of vocational interests or in terms of personal styles.
The Strong provides valid and comprehensive measures of these differences” (p. 290).

The General Occupational Themes produced on the Strong are ipsative in nature,
which is to say that the continuous score on any of the six GOTs are relevant only when
compared to scores from the other 5 themes (Buboltz et al., in press). Within the present
analysis therefore, individual scores on the six GOTs will be treated as continuous
variables to maximize variability.

Myers-Briggs Type Indicator (Myers & McCaulley, 1985).

The MBTI is a 126 item forced-choice instrument. Results fi'om the MBTI
classify individuals along four interlocking dimensions, each dimension having two polar
preferences. The Extroversion-Introversion index (E-I) indicates whether a person directs
his/her perception and judgment upon the environment (E) or upon the inner world of
ideas (I). The Sensation-Intuition index (S-N) indicates whether a person relies on
sensing (S) or intuition (N) processes in perceiving and dealing with the world. The
Thinking-Feeling (T-F) dichotomous dimension indicates whether a person relies on
thinking (T) or feeling (F) processes in making judgments. The Judging-Perceiving (J -P)
index, indicates whether a person uses judging (J) or a perceptive (P) attitude in dealing
with the environment. With the use of four indices, 16 types can be generated and the
individuals in these types are presumed to differ in preferences, personality
characteristics, value judgments, interests, and numerous surface traits (Buboltz, Thomas,

& Johnson, in press).

67

For each index, a series of dichotomous items exists. The first dimension is a
general attitude toward the world either extraverted (E), which reflects the individual’s
preference for an interactive outer world, or introverted (I), where attention and energy
are directed inward, focusing on internal representations of the world.

The second dichotomy that Jung defined is Sensation-Intuition (S-N), which
reflects two different modes of perceiving. The Sensation type prefers perception of the
observable environment directly through the senses and orients cognitively to facts and
details. Intuition, on the other hand, refers to going beyond the information that is
provided by the senses and looks for meaning and possibilities within the data received
and processed. The Sensation dimension is reflected in practical and realistic
characteristics of the environment whereas the Intuition dimension is associated with the
preference for creative and imaginative processes (Buboltz et al., in press).

The Thinking-Feeling (T-F) dichotomy reflects two different ways of judging the
world. Once information is received, it is assumed to be processed in either a Thinking
(T) or Feeling (F) style. The Thinking style prefers a reliance on reasoning, logic, and
impersonal methods to make judgments about the world. Conversely, the Feeling (F)
style arranges the contents of consciousness based on relevance to needs for affiliation,
harmony, and warmth (Buboltz et al., in press).

The final dimension, J udgment-Perception (J -P), reflects a preference for judging
or perceiving functions. This dimension was proposed by Myers (1962) and serves a
dual purpose in the interpretation of the MBTI. The Judging (J) theme prefers situations
to be planned and orderly, to be completed, and for issues to be resolved. Individuals who

prefer the Perception (P) theme prefer to live a more flexible style, in a reactive manner,

68

tending to keep their options open while gathering more information. Individuals are
viewed as needing both of these attitudes, but with one of the styles being dominant over
the other style. Another purpose of the J -P dimension is to determine dominant versus
auxiliary functions (Buboltz et al., in press).

Although the MBTI items were written within a forced-choice format, the
instrument is not purported to be as aversive to most as other forced-choice instruments
(Myers & McCaulley, 1985). This claim is based upon the fact that all items reflect two
opposing rather than two competing choices. Construction of the current version of the
MBTI entailed the following process: (1) determining items to reflect preferences
described by Jung for all four dichotomies; (2) identifying the dominant and auxiliary
functions via the development of the J -P dichotomy; (3) writing, testing, weighting, and
selecting items that would achieve the greatest separation and least amount of overlap
between the dichotomies; (4) establishing precision in the center of the scale so that
persons with indeterminate preferences would be more likely classified according to their
true preferences; (5) establishing an objective check on the division points for each
dichotomy (Myers & McCaulley, 1985).

The original research for the MBTI measured the responses of a variety of
populations ranging from fourth grade students to older adults (Myers & McCaulley,
1985). From 1942 to 1944, a large number of MBTI questions were written and
evaluated based upon a criterion group consisting of approximately twenty of the authors’
relatives and friends (Myers & McCaulley, 1985). Items created fi‘om this initial
validation trial became Form A. Form B was a rearrangement of the same items (Myers

& McCaulley, 1985). Form C reflected an advance in validity testing wherein items that

69

were highly intercorrelated were omitted (Myers & McCaulley, 1985). Form C3
represented yet another statistical advance, incorporating item weights to better
differentiate individuals scoring toward the middle of the poles (Myers & McCaulley,
1985). Form D was published in 1962. Revisions to Form D focused upon item delivery.
Specifically, key words or phrases within items were thought to trigger preference
differentiation. Therefore, Form D introduced word—pairing that consisted of key words
only and eliminated potentially obfuscating or unnecessary item content (Myers &
McCaulley, 1985). Ensuing Forms E and F evolved from sex and type evaluation of
Form D items (Myers & McCaulley, 1985). Form G, originating in 1977 and revised in
1985, was normed on 1,114 males and 1,111 females in grades 4-12 within three
Maryland public schools and four private schools in Pennsylvania (Myers & McCaulley,
1985). Form G reflects several improvements to Form F. Forty items were omitted for
validity reasons. Nine of the items were rewritten to avoid ambiguity. The items were
rearranged so that the scored items were presented prior to the unscored items. The
respondent therefore must provide feedback on the first 95 items for the instrument to be
scored (Myers & McCaulley, 1985). The additional thirty-one items were unscored and
experimental.

The internal consistency of the four MBTI scales was quite high in all samples
available to date, whether computed using logical split-half, consecutive item split-half,
or coefficient alpha (Myers & McCaulley, 1998). A summary of Form G data suggests
that sample characteristics related to type development may create reliability variation
across groups. Specifically, groups composed of people who are believed to have less

command over perception and judgment (e. g., chronologically younger participants)

70

should exhibit lower reliabilities (Myers & McCaulley, 1998). The quality of perception
and judgment has also been found to be influenced by an individual’s level of
achievement (Myers & McCaulley, 1998). It is therefore expected that within samples
consisting of comparable age levels, those individuals with higher achievement levels
will report their preferences more consistently (Myers & McCaulley, 1998). College and
university samples possessed higher reliabilities than did high school samples (Myers &
McCaulley, 1998). Reliabilities were also found to be higher in groups with higher
average intelligence as measured by standard intelligence tests (Myers & McCaulley,
1998). Nonetheless, internal consistency coefficients based on coefficient alpha have
been deemed to be acceptable for adults and adequate for younger samples alike (Myers
& McCaulley, 1985). Based upon a data bank consisting of 32,671 individuals, split half
correlations for Form G were .82 (BI), .84 (S-N), .83 (T -F), and .86 (J-P) (Myers &
McCaulley, 1998).

Test-retest reliabilities of the MBTI demonstrate consistency over time with levels
of agreement much greater than chance (Myers & McCaulley, 1998). From 3 meta-
analytic sample consisting of 1,139 individuals, Form G demonstrated test-retest
reliabilities across a time period of less than nine months of .84 (El), .81 (SN), .77 (T-
F), and .82 (J -P) (Myers & McCaulley, 1998). Across a time period greater than nine
months, test-retest reliabilities were .70 (BI), .68 (S-N), .59 (T-F), and .63 (LP) (Myers
& McCaulley, 1998). The reliability coefficient for the T-F scale is the lowest of the four
scales in F orm G (Myers & McCaulley, 1998). When subjects reported a change in type,
it was most likely to occur in only one preference, and in scales where the original

preference clarity was low (Myers & McCaulley, 1998). Changes in mood were not

71

found to significantly affect test-retest correlations (Myers & McCaulley, 1998). Isabel
Myers administered one of the original forms of the indicator to the 87 members of a high
school class graduating in 1943. At the 50th reunion of the class, Katherine Myers
administered Form G to 39 of these same people. Over this fifty-year period, eight
people (21%) had the same type, 13 (33%) had changed one letter, 16 (41%) had changed
two letters and two (5%) had changed three letters (Myers & McCaulley, 1998). No one
in the group changed all four letters. Over a fifty-year interval therefore, 54 percent
changed either none or one letter whereas the level of agreement expected by chance
would be approximately 6 percent (Myers & McCaulley, 1998).

Hammer (1996) concluded that the reliability of the MBTI continuous preference
scores appeared to be quite good. For individuals whose preference scores lie outside of
the uncertainty zone surrounding type cutoffs, type assignment appeared to be reasonably
stable over time (Hammer, 1996). Yet test-retest reliability has been referred to as the
Achilles heel of the MBTI among its critics (Hammer, 1996). Hammer (1996)
highlighted that because cutoff scores are used within each of the four dichotomies, the
size of acceptable standard error around the cutoff score is decreased. Furthermore,
under the current scoring system research indicates that many individuals receive low
preference scores (Hammer, 1996). Therefore, forcing such individuals to fit into a two-
category structure can produce a high rate of erroneous classifications (Hammer, 1996).
Most of the type-switches within test-retest analyses occur not surprisingly in individuals
with low preference clarity (Hammer, 1996). Hammer (1996) suggested efforts to

improve the measurement precision of the MBTI at the cutoff point via the addition of

72

test items or the use of Item Response Theory-based scoring to decrease the number of
individuals scoring close to the cutoffs.

Exploratory and Factor analytic results have offered strong support for the
construct validity of the MBTI (Myers & McCaulley, 1998). Several exploratory studies
employing large samples have produced factor analytic structures that almost exactly
matched hypothesized patterns of loadings (Myers & McCaulley, 1998). Myers and
McCaulley (1998) recently concluded, “there is no question that the results of the factor
analytic studies reported over the last ten year period have been very supportive of the
validity of the four-scale structure of the MBTI” (p. 173). Hammer (1996) concluded
large amounts of empirical data indicated that MBTI scores demonstrated good
convergent, discriminant, and predictive validity.

The MBTI scales have been correlated with a number of personality and interest
scales across multiple studies (Myers & McCaulley, 1998). Results of the analyses are
presented within the Myers-Briggs Type Indicator Manual (Myers & McCaulley, 1998).
MBTI Extraversion has been found related to extraversion as measured by the 16
Personality Factors, Millon Index of Personality Styles, NEO-PI, and the Jungian Type
Survey (Myers & McCaulley, 1998). Constructs related to MBTI Extraversion included:
(a) sense of comfort in the environment which is demonstrated in measures named self-
acceptance, well-being, autonomy, self-confidence, and stability, (b) action on the
environment as shown in correlations with dominance, assertiveness, social boldness,
capacity for status, leadership, change, aggression, and exhibitionism; (c) sociability and
relatedness to others as shown in scales such as outgoing, sociability, social interaction,

social relations, valuing of community, empathy, high interpersonal needs, and peer

73

cohesion; (d) responsiveness to the environment shown in measures such as liveliness; (e)
ways of coping by the use of social and emotional resources; and (f) career interests that
relate to general social and enterprising interests (Myers & McCaulley, 1998).

MBTI Introversion type has been correlated with introversion as measured by the
16 Personality Factors, Millon Index of Personality Styles, NEO-PI, and the Jungian
Type Survey (Myers & McCaulley, 1998). Constructs related to MBTI Introversion
included: (a) lack of comfort with the environment in measures such as anxiety,
hesitating, retiring, abasement, deference, yielding, and accommodating; (b) need to
control external stimulation in scales such as privateness and preserving; (c)
independence from the environment in measures such as self-reliance and self-control;
(d) tendency to be stressed by environmental demands as evidenced by correlations with
burnout scales such as exhaustion and depersonalization; (e) career interests in the
investigative and scientific areas (Myers & McCaulley, 1998).

MBTI Sensing type has been correlated with sensing as measured by the Millon
Index of Personality Styles and the Jungian Type Survey (Myers & McCaulley, 1998).
Constructs related to MBTI Sensing included: (a) taking a practical approach such as in
correlations with tough-mindedness; (b) managing reality as demonstrated by correlations
with scales such as control and order; (c) accepting reality as seen in relationships with
measures of rule consciousness, accommodating, retiring, self-control, conforming,
yielding, socialization, endurance, achievement via conformity, norm favoring, deference
and a valuing of home activities, economic security, and physical comfort; and (d) career
interests including conventional, athletics, mathematics, law, organizational management,

data management, and office services (Myers & McCaulley, 1998).

74

MBTI Intuitive was correlated with intuitiveness as measured by the Millon Index
of Personality Styles, and the Jungian Type Survey (Myers & McCaulley, 1998).
Constructs related to MBTI Intuitive included: (a) openness to the environment as
demonstrated by correlations with variables such as modifying, sensitivity, openness to
change, risk-taking, flexibility, and variety; (b) thinking abstractly in relationships with
measures of reasoning, abstractedness, intellectual efficiency, and valuing of an academic
learning environment; (c) independence as seen in correlations with scales measuring
achievement via independence, free child, and autonomy; and (d) career interests such as
music, drama, art, applied arts, and writing (Myers & McCaulley, 1998).

MBTI Thinking type has been correlated to thinking as measured by the Millon
Index of Personality Styles and the Jungian Type Survey (Myers & McCaulley, 1998).
Constructs related to MBTI Thinking included: (a) logical analysis as demonstrated by
correlations with scales such as asserting, dissenting, complaining, tough-mindedness,
expressed control, and order; (b) behaviors related to thinking and acting independently
as with measures of independence, autonomy, achievement, endurance, and risk-taking;
and (c) career interests in the realistic and investigative areas and in mechanical activities,
science, mathematics, law, and data management (Myers & McCaulley, 1998).

MBTI Feeling type has been found to be related to feeling as measured by the
Millon Index of Personality Styles and the Jungian Type Survey (Myers & McCaulley,
1998). Constructs related to MBTI Feeling included: (a) concern for others as
demonstrated by correlations with measures of nurturing, warmth, and affection; (b)
adaptability to others as shown in variables such as accommodating, agreeing, sensitivity,

deference, and desire for peer cohesion; and (c) career interests in the social and artistic

75

areas and activities such as music, drama, art, culinary art, social service, and people-
oriented workstyle (Myers & McCaulley, 1998).

MBTI Judging type has been correlated with (a) attempt to create order in the
environment as shown in correlations with scales such as systematizing, perfectionism,
and control; (b) respect for social mores as shown in relationships with scales such as
order, conforming, rule consciousness, responsibility, socialization, self-control,
endurance, and valuing economic security; and (c) career interests in the conventional
area and in data management and office services activities (Myers & McCaulley, 1998).

Constructs relating to MBTI Perceiving type include (a) openness to environment
as shown by relationships with measures of innovation, openness to change,
abstractedness, flexibility, variety, creativity, change, and risk-taking; (b) career interests
in the artistic area and interests in activities related to nature, music, drama, art, and
writing (Myers & McCaulley, 1998).

Myers and McCaulley (1998) asserted at least two warnings regarding
interpreting correlation results. First, several of the correlation analyses performed
employed MBTI preference scores as opposed to MBTI continuous scores. The
conventional notation therefore that positive correlation associated with, for example, I,
N, F, and P connote negative correlation with E, S, T, and J are not applicable. On the
other hand, correlations of MBTI continuous scores have their limitations as evidence for
construct validity. They report only the four preference scores one at a time and do not
show the sixteen types as dynamic entities. Correlations also have the problem of clarity

of preference.

76

The principal criticism to the validity of the MBTI is the use of bipolar types.
Jungian theory posits the existence of dichotomies yet some have argued that evidence of
bimodal distributions of preference scores is nonexistent. If Jung was correct about
dichotomous populations and if the MBTI correctly separates the two populations on
each preference scale then discontinuity around the midpoint of each scale is expected
(Myers & McCaulley, 1998). In order to constitute a discontinuity, there must be a
significantly greater rate of change at the point of discontinuity than at either side of the
point (Myers & McCaulley, 1998). According to Myers and McCaulley (1998) recent
research has indicated that evidence for the dichotomous nature of the scales was
established in plots of preference scores against external variables such as intelligence,
vocabulary skill, and academic major. Analyses of these plots demonstrated that the only
significant differences between successive groups of scores were exactly at the midpoint
of the scales. (Myers & McCaulley, 1998).

Design

Canonical correlation will be employed to explore the following question: to
what extent and in what ways do MBTI personality dimensions overlap with Strong
interest themes? This method of data analysis will address the research question in three
stages. First, canonical correlation will treat the eight MBTI dimensions as one set X and
will treat the six Holland themes as one set Y. The treatment of each as a set is
theoretically valid and allows for the exploration of relationships within each set (i.e., X
and Y). Second, one new variable will be created for set X (i.e., MBTI poles) and will be
called the function of X {f(X)}. One new variable will also created for set Y (i.e.,

Holland types) and will be called the function of Y {f(Y)}. Third, the correlation between

77

f(X) and f(Y) will be analyzed. Analysis of the correlation between firnctions (f)
facilitates a multivariate exploration of the relationship between sets MBTI and Strong.

The composition and analysis of Eigen values and Eigen vectors serves as the
foundation for canonical correlation. First, correlation within and between variable sets
X and Y yields Eigen values. Second, Eigen values mathematically yield Eigen vectors.
Third, Eigen vectors provide the recipe for combining set X and set Y into firnctions.
Specifically, Ei gen vectors always collapse set X and set Y in such a way as to yield the
best possible correlation. Fourth, after sets X and Y have been combined into a function,
a canonical correlation value is given indicating the level to which the canonical
functions are correlated. Finally, the canonical correlation value squared indicates the
proportion of variation accounted for in set Y by set X within a particular function. The
first function is typically the only function that is analyzed because it represents the best
fit between the sets. However, dimension reduction analysis can be employed to
statistically determine the amount of functions to explore.

Canonical correlation is not limited by unequal cell size considerations (K. Frank,
personal communication, February 2001). Given the fact that the proposed study is
attempting to estimate at least 16+6+l parameters, 200 observations were needed to
achieve fairly stable results (K. Frank, personal communication, February, 2001). A

sample size of 423 was therefore comfortably sufficient to perform the analysis.

78

CHAPTER F OUR
RESULTS

Archival data were collected from 426 undergraduate students fi‘om a state
university located in the western region of the United States. The data of three
individuals were excluded due to missing information, reducing the sample to 423. Each
of the 423 individuals had taken both the Myers-Briggs Type Indicator (1986, Form G)
and the Strong Interest Inventory (1994). Scores on each of the four polar dimensions of
the MBTI indexes were converted to continuous scores creating the following pairs: (1)
Extravert/Introvert (E1), (2) Sensing/Intuitive (SN), (3) Thinking/Feeling (T F), and (4)
J udging/Perceiving (JP). The data also included individuals scores on the following six
General Occupational Themes (GOT) within the Strong: (1) Realistic, (2) Investigative,
(3) Artistic, (4) Social, (5) Enterprising, (6) Conventional.
Sample Demogaphics and Distribution

Frequency data was generated to determine if basic distributional assumptions
underlying multivariate analysis were satisfied. Frequency analysis demonstrated that the
sample consisted of 192 (45 percent) males and 231 (55 percent) females. Frequency
data regarding MBTI personality dimensions revealed the following sample distribution:
47 percent were classified as Extravert (n=200) and 53 percent as Introvert (n=223); 46
percent were Sensing (n=l96) and 54 percent (n=227) were Intuitive; 44 percent (n=187)
were Thinking and 56 percent (n=236) were Feeling; 54% (n=228) were Judging and 46
percent (n=195) were Perceiving. The mean EI continuous score was 102.5 (SD = 27.4),
the mean SN score was 100.2 (SD = 27 .4), the mean TF score was 102.0 (SD = 27.1),

and the mean JP continuous score was 100.6 (SD = 23.8). Probability plots for all four

79

continuous dimensions demonstrated normal distributions of scores around the mean of
100.

Frequency and descriptive data on the six General Occupational Themes (GOT)
revealed that each were normally distributed around the mean of fifty. The mean
Realistic score was 47.09 (SD = 9.59), the mean Investigative score was 46.86 (SD =
9.93), the mean Artistic score was 51.57 (SD = 9.89), the mean Social score was 51.58
(SD = 10.07), the mean Enterprising score was 52.29 (SD = 9.36), and the mean
Conventional score was 49.90 (SD = 9.30). Probability plots for each of the six GOTs
corroborated the fact that the data was normally distributed.

Bivariate Relationships between MBTI dimensions apd GOTs: Full Dgagtiet

Similar to research previously performed and cited exploring the relationship
between MBTI poles and Strong interest themes, Pearson correlations were calculated to
determine what bivariate relationships existed within the sample (a) among the MBTI
dimensions, (b) among the six GOT interest themes, and (c) between the four personality
dimensions of the MBTI and the six GOTs. These analyses were performed for to satisfy
two purposes. First, the analyses permitted the testing of the hypothesis that the percent
of variance accounted for within the first canonical correlation for the full data set would
be higher than any percentage of variance accounted for by bivariate analysis procedures.
Second, because canonical correlation describes the relationships between canonical
variates but does not describe relationships between the variables on the same side of the
equation (e. g., the direction of the relationship between variables within the Strong
variate), bivariate analyses can be used to hypothesize direction of relationships within

canonical variates.

80

As highlighted on Table 1 (see below), bivariate analysis among the MBTI

dimensions revealed the following statistically and practically significant (p<.05, r<.3)

relationship: SN and JP were positively correlated (r = .407). This finding is consistent

with previously cited bivariate analyses. Results are also confirmed Jung’s theory that

the personality dimensions are, from a bivariate perspective, largely independent of one

another.

Table 1 - Bivariate Correlations Among MBTI Personality Dimensions: Full Data Set

 

 

 

 

 

 

Extravert- Sensing- Thinking- J udging-
Introvert value iNtuition value Feeling value Perceiving
value
Extravert- 1.000 -0.1163 -.0613 -. 1043
Introvert value
Sensing- -.l 163 1.000 .0813 .4071*
iNtuition value
Thinking- -.0613 .0813 1.000 .0289
Feeling value
Judging- -.1043 .4071 * .0289 1.000
Perceiving
value

 

 

 

 

 

Note: asterisk (*) indicates statistically significant relationship at the .05 level

Bivariate analysis among the six GOTs revealed numerous statistically and

practically significant (p<.05, r>.3) relationships. As Table 2 indicates (see next page),

the strongest relationship existed between Realistic and Investigative. This finding was

commensurate with Holland’s theory of consistency. The strength of relationships

between Enterprising and Conventional, Enterprising and Social, and Artistic and Social

supported Holland’s hexagonal arrangement of RIASEC according to the theory of

consistency as well. Perhaps of mild surprise vis-a-vis theory was the strength of the

relationship found between Investigative and Conventional (see Table 2 next page).

Although the fact that a significant relationship between the two dimensions does not

81

 

contradict Holland’s (1985) theory of consistency, the strength of the relationship was

more significant than other relationships, such as Enterprising and Social, predicted by

Holland’s (1985) consistency theory to be stronger. Unique sample characteristics may

have been at play, suggesting the need for retesting before reliable generalizations of the

present findings are made.

Table 2 - Bivariate Correlations Among General Occupational Themes: Full Data Set

 

 

 

 

 

 

 

 

Realistic Investigative Artistic Social Enterprising Conventional
Realistic 1.000 5083* .2044 .1428 .1012 .2077
Investigative .5083* 1.000 .2456 .1525 .0457 3487*
Artistic .2044 .2456 1.000 3173* .2228 .0061
Social .1428 .1525 .3173* 1.000 .3264“ .2631
Enterprising .1012 .0457 .2228 3264* 1.000 .4515*
Conventional .2077 .3487* .0061 .2631 .4515* 1.000

 

 

 

 

 

 

 

Note: asterisk (*) indicates statistically significant relationship at the .05 level (r>.3)

Bivariate correlations were also calculated between the four MBTI personality

dimensions and the six interest themes (see Table 3 below). A positive correlation

indicated a relationship between higher MBTI dimension scores and the respective GOT.

Conversely, a negative correlation indicated a relationship between the lower MBTI

dimension scores and the respective GOT.

Table 3 - Bivariate Correlations Between the MBTI Dimensions and GOT: Full Data Set

 

 

 

Strong/MBTI Extravert- Sensing- Thinking- Judging-
Introvert iNtuition Feeling Perceiving
Realistic -.02077 -.03664 -.11468* .18965*

 

 

 

 

 

82

 

 

Table 3 (cont’d).

 

 

 

 

 

 

Investigative .05906 .08469 -.10108* .01416
Artistic -.07976 .050426* .11882* .13828*
Social -.14242* -.00313 .38351* -.O9575*
Enterprising -.29950* -.041 85 -.07846 -.06842
Conventional -.01267 -.25047* -.05 895 -.21647*

 

 

 

 

 

Mg: asterisk (*) indicates statistically significant relationship at the .05 level
Statistically and practically significant (p<.05, r>.3) relationships were found
between MBTI personality dimension SN and Strong GOT Artistic (p<.0001; r=.504),
MBTI personality dimension TF and Strong GOT Social (p<.0001; r=.3 84), and between
MBTI personality dimension EI and Strong GOT Enterprising (p<.0001; r= -.299). Weak
yet statistically significant relationships (p<.05, r<.3) were found between the following
nine MBTI personality dimensions and Strong GOT interest themes: MBTI personality
dimension SN and GOT Conventional (p<.0001; r= -.250), MBTI personality dimension
JP and GOT Conventional (p<.0001; 1': -.2l6), MBTI personality dimension JP and GOT
Realistic (p<.0001; F .190), MBTI personality dimension EI and GOT Social (p=.0033;
r= -.142), MBTI personality dimension JP and GOT Artistic (p=.0044; r=.138), MBTI
personality dimension TF and GOT Artistic (p=.0145; r=.119), MBTI personality
dimension TF and GOT Realistic (p=.0183; r= -.115), MBTI personality dimension TF
and GOT Investigative (p=.03 77; r= -.101), and MBTI personality dimension JP with
GOT Social (p=.049l; r= -.096). Table 4 (see next page) summarizes these relationships

in descending order by absolute value of the correlation.

83

 

Table 4 - Significant Bivariate Correlations Between the MBTI Dimensions and GOT
Listed by Absolute Correlation Value: Full Data Set

 

 

 

 

 

 

 

 

 

 

 

 

 

MBTI GOT Correlation (r)
SN Artistic .504
TF Social .384
E1 Enterprising -.300
SI Conventional -.250
JP Conventional -.216
JP Realistic .190
El Social -. 142
JP Artistic .138
TF Artistic .1 19
TF Realistic -.1 15
TF Investigative -.101
JP Social -.096

 

 

 

 

 

The most variation accounted for via bivariate analysis occurred between MBTI
personality dimension SN and Strong GOT Artistic (r=.503). It may therefore be stated
that Strong Artistic accounted for 25 percent of the variance in SN scores within the
present sample.

Canonical Correlation Analysis: Full Data Set
Unique to the present analysis was the use of canonical correlation to identify

latent relationships between the four MBTI personality dimensions and the six

84

occupational interest themes. The specific question being addressed is as follows: to

what extent and in what ways do the MBTI personality dimensions and the Strong GOTs

overlap? Theoretically and statistically, canonical correlation represents the soundest data

analytic design to explore the relationship among and between these variables. Through

the simultaneous analysis of correlations within sets X (i.e., MBTI personality

dimensions) and Y (i.e., Strong GOT) and between sets X and Y, canonical correlation

creates a statistical procedure for combining sets X and Y into a function. A function

represents the best possible correlation between variable sets X and Y and yields a value

that indicates the level to which the variable sets are correlated. The canonical

correlation value squared indicates the proportion of variation accounted for in set Y by

set X within a particular function. The first firnction is typically the only function that is

explored because it represents the best fit between the variable sets (K. Frank, personal

communication, 11/99).

To What Extent Do the MBTI Personality Dimensions and the Strong GOTs Overlap:

Full Data Set

Canonical correlation analysis on the full data set'yielded four statistically

significant canonical functions (see Table 5 below).

Table 5 - Canonical Correlation Analysis: Full Data Set

 

 

 

 

 

Canonical Function P-value Canonical Squared Canonical
Correlation Correlation

1 <.001 .600 .360

2 <.001 .497 .247

3 <.001 .339 .1 15

 

 

 

 

85

 

Table 5 (cont’d).

 

 

4 <.001 .289 .084

 

 

 

 

Although all four fimctions differed significantly (p<.05) from zero, the practical
significance of functions three and four was questionable. Specifically, a danger with a
sample size as large as the one employed in the present analysis was that several
functions might appear significant (Grunden, personal communication, February 2002).
What is practically significant, as determined by the magnitude of the relationship,
therefore becomes an important consideration (Grunden, personal communication,
February 2002). The first two functions accounted for 36 percent and 24 percent
respectively of the proportion of shared variance between the four MBTI personality
dimensions and the six Strong occupational interest themes. Functions one and two
accounted for 60 percent of total shared variance. In comparison, function three
accounted for only 12 percent of shared variance and function four accounted for only 8
percent of shared variance. Statistically, redundancy analysis results may be explored to
determine the practical significance of a given function (Grunden, personal
communication, February 2002). The redundancy index (see Table 6 next page) on the
present canonical analysis revealed significant decreases from functions one (proportion
index = .0942) and two (proportion index = .0623) to fimctions three (proportion index =
.0287) and four (proportion index = .0199). The analysis therefore proceeded with the

interpretation of the first two significant canonical functions.

86

 

Table 6 - Canonical Redundancy Index: Full Data Set

 

 

 

 

 

 

Canonical Function Canonical R-Square Proportion Index
1 .3604 .0942
2 .2470 .0623
3 .1152 .0287
4 .0837 .0199

 

 

 

In Whflays Do the MBTI PersonalityDimensions and the Strong GOTs Overlap: Full

 

Data Set.

A general principle of canonical interpretation is to include within a function any
variable with a standardized coefficient greater than or equal to +/—.25 (Grunden, personal
communication, February 2002; K. Frank, personal communication, March 2002).
Typically, if the standardized coefficient value closely approximates .25 but its validity
for inclusion within a function remains in question, confirmation of the most important
variables can be attained via exploration of values yielded as a result of correlations with
the canonical variate and correlations with the Opposite canonical variate (Grunden,
personal communication, February 2002; K. Frank, personal communication, March
2002). By this method, a variable may reliably be included within a canonical function if
at least two of the three methods discussed above indicated (Gnmden, personal
communication, February 2002; K. Frank, personal communication, March 2002).

Using the inclusion criteria described above, SN was the most important
(standardized coefficient = 1.0340) within the MBTI variate on the first canonical

function and Artistic (.9195), Realistic (-.3651), and Conventional (-.3056) were the most

87

 

important variables within the GOT set on the first canonical function (see Table 7

below).

Table 7 - Inclusion Criteria for First Canonical Function: Full Data Set

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

T” Canonical Standardized Correlation w/ Correlation w/
Function Coefficient Canonical Variate Opposite Canonical
Variate

MBTI l MBTI I GOT l
Extravert-Introvert .l 17 .005 .003
Sensing-iNtuition* 1 .034 .962 .578
Thinking-Feeling .184 .255 .153
Judging-Perceiving -.180 .234 .140

GOT I GOT 1 MBTI 1
Realistic“ -.365 -. 159 -.096
Investigative .200 .122 .073
Artistic“ .920 .848 .509
Social -.023 .113 .068
Enterprising -.l66 -.134 -.081
Conventional -.306 -.387 -.232

 

 

Nata asterisk (*) indicates inclusion in the first canonical function, full data set

The complete relationship determined by the first canonical firnction was as
follows: MBTI = .1173 E1 + 1.0340 SN + .1835 TF - .1801 JP; GOT = -.3651 R + .1997
I + .9195 A - .0228 S -.1656 E -.3056 C. This relationship may be summarized as
follows: SN (n) Artistic — 1/3 (Realistic + Conventional) =.60. The equation
symbolizes the fact that the maximum correlation between the four MBTI personality
dimensions and the six Strong GOTs was obtained when an individual’s SN score was
correlated with the linear combination of their Artistic score minus one-third the sum of
the Realistic and Conventional scores.

The correlation being interpreted fi'om canonical correlation is that which exists
between the two canonical variates or sets of variable. Based upon canonical correlation

findings, it may therefore validly be stated that higher levels of a variable on one side of

88

the canonical equation are correlated with higher or lower, depending upon the factor
loading direction, values of a variable on the other side of the equation. Within canonical
correlation however, inferences regarding the relationship between two variables on the
same side of the function based upon the function alone cannot be validly drawn. One
must look at the simple bivariate correlations for that information (Grunden, personal
communication, May 2002).

Applied to the present data, 36 percent of the variation within an individual’s SN
score was accounted for by their Artistic, Realistic, and Conventional scores. Simple
bivariate correlations suggested that within the Strong variate, Artistic and Realistic were
positively and significantly (weak) correlated (r=.20), Artistic and Conventional were
uncorrelated (r=.00), and Realistic and Conventional were positively and significantly
(weak) correlated (r=.21).

Analysis revealed that the second canonical function was composed of MBTI
variables TF and JP and GOT variables Social, Realistic, and Enterprising (see Table 8
below).

Table 8 - Inclusion Criteria for Second Canonical Function: Full Data Set

 

 

 

 

 

 

 

 

 

 

 

 

2'“l Canonical Standardized Correlations w/ Correlations w/
Function Coefficients Canonical Variate Opposite Canonical
Variate

MBTI 2 MBTI 2 GOT 2
Extravert-Introvert -.044 -.044 -.022
Sensing-iNtuition -.155 -.223 -.1 l l
Thinking-Feeling* .916 .896 .445
Judging-Perceiving* -.362 -.394 -.196

GOT 2 GOT 2 MBTI 2
Realistic“ -.337 -.336 -.167
Investigative .224 -.228 -.1 13
Artistic -.l37 -.O32 -.016
Social“ .993 .790 .393

 

 

 

 

89

 

Table 8 (cont’d).

 

Entemising -.392 - .056 -.028

 

 

 

 

 

Conventional .193 .128 .064

 

No_te: asterisk (*) indicates inclusion in the second canonical function, full data set

The complete relationship determined by the second canonical function was as
follows: MBTI = -.0436 E1 - .1548 SN + .9162 TF - .3623 JP; GOT = - .3367 R -.2242 I
- .1369 A + .9931 S - .3918 E + .1928 C. The relationship determined by the second
canonical function is summarized as follows: TF — 1/3 JP (“9 Social — 1/3 Realistic =
.497. The equation signifies that the second largest correlation occurs when the linear
combination of an individual’s TF score minus one-third of their JP score is correlated
with their Social score minus one-third their Realistic score. Although the method of
standardized coefficients suggests that Enterprising might have had some importance, the
other two methods disagreed with this result (see Table 8 above). Practically, the second
canonical ftmction indicated that 24 percent of the variation within an individual’s TF
and JP scores was accounted for by their Social and Realistic scores. Bivariate
correlations suggested that TF and JP were positively but not significantly correlated
(r=.01). Social and Realistic were positively and significantly (weak) correlated (r=.14).

A sensitivity analysis validated the results obtained within the first and second
canonical functions. Three of the variables in the GOT set were selected and removed
one at a time and canonical correlations were repeated. Coefficients and correlations did
not vary significantly upon the removal of one of the variables from the GOT set

(Grunden, personal communication, 11/01).

90

 

Gender Considerations: Male Demoggaphics and Distributions

In addition to the use of canonical correlation, a second unique aspect of the
present study was the consideration of the effect of gender on the latent relationships
between the four MBTI personality dimensions and the six occupational interest themes.
Because differences are purported to exist between males and females on the GOT
dimensions, the data set was divided according to gender. Bivariate correlations between
the four MBTI personality dimensions and the six GOTs were computed as were
canonical correlations on each gender set.

Frequency data indicated that the male sample consisting of 192 individuals was
normally distributed around the mean of fifty on the six GOT scales. The mean Realistic
score was 51.06 (SD = 9.93), the mean Investigative continuous score was 48.44 (SD =
10.19), the mean Artistic continuous score was 48.83 (SD = 9.98), the mean Social
continuous score was 50.07 (SD = 10.19), the mean continuous Enterprising score was
52.47 (SD = 9.83), and the mean Conventional continuous score was 49.19 (SD = 8.96).
Males were also normally distributed around the mean of 100 along each of the four
MBTI dimensions. The mean EI value was 103 (SD=25.59), the mean SN value was 99
(SD=27.19), the mean TF value was 95 (SD=22.61), and the mean JP value was 104
(SD=27 . l 7).

Bivariate Correlations Between MBTI Dimensions and GOTs: Male Data Set

Simple bivariate correlations were calculated to explore two-dimensional
relationships within the male sample between the four personality dimensions of the
MBTI and the six GOTs. As highlighted on Table 9 (see next page), bivariate analysis

revealed statistically significant (p<.05) relationships between MBTI personality

91

dimension SN and Strong GOT Artistic (r=.481) and between MBTI personality

dimension TF and Strong GOT Social (r=.361). Statistically significant but weak

relationships were found between MBTI personality dimension EI and Strong GOT

Enterprising (r= -.284), between MBTI personality dimension SN and GOT Conventional

(r= -.241), between MBTI personality dimension JP and GOT Conventional (r= -.257),

between MBTI personality dimension JP and GOT Artistic (r=. 148), and between MBTI

personality dimension JP with GOT Social (r= -.187).

Table 9 - Bivariate Correlations Between the MBTI Dimensions and GOT: Male Data Set

 

 

 

 

 

 

 

 

Extravert- Sensing- Thinking- Judging-

Introvert iNtuition Feeling Perceiving
Realistic -.07674 -.13825 -.00515 .06353
Investigative -.001 84 .01845 .09765 -.09396
Artistic -.01038 .48139* .12742 .14847“
Social -.11821 .02940 .36088* -.18706*
Enterprising -.28375* -.01746 .03211 -.12528
Conventional -.04432 -.24108* .04719 -.25728*

 

 

 

 

 

Note: asterisk(*) indicates statistical significance at the .05 level.

Differences regarding strength of relationship between bivariate correlations

performed on the full data set in comparison to bivariate correlations performed on the

male data set are outlined on Table 10 (see next page).

92

 

Table 10 - Comparison of Significant Bivariate Correlations Between the MBTI
Dimensions and GOTs Listed By Absolute Correlation Value: Full Data Set and Male

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Data Set

MBTI GOT Full Data Set (r) Male (r)
Sensing-iNtuitive Artistic .504 .481
Thinking-Feeling Social .384 .361
Extravert-Introvert Enterprising -.300 -.284
Sensing-iNtuitive Conventional -.250 -.241
Judging-Perceiving Conventional ‘-.216 -.257
Judging-Perceiving Realistic ' .190

Extravert-Introvert Social -.142

J udging-Perceiving Artistic .138 .148
Thinking-Feeling Artistic .1 19

Thinking-Feeling Realistic -.115

Thinking-Feeling Investigative -.101

Judging-Perceiving Social -.096 -.187

 

 

 

 

The five largest correlations for the full data set remained significant for the male
gender set. However, non-statistically significant differences in the strength of the
relationships emerged. For example, correlations within the male sample were slightly
stronger for JP and Conventional, JP and Artistic, and JP and Social relationships.
Correlations were slightly weaker for SN and Artistic, TF and Social, EI and
Enterprising, and SN and Conventional. The most variation accounted for in the full data

set was 25 percent, which was between MBTI SN and GOT Artistic. In comparison, the

93

 

most variation accounted for within the male data set was 23 percent, also occurring
between MBTI SN and GOT Artistic.
Gender Considerations: Female Demographics and Distributions

Frequency data indicated that the female sample consisting of 231 individuals was
normally distributed around the mean of fifty on the six GOT scales although, true to
Strong Interest Inventory theory, the Realistic scale produced the lowest average female
scores. The mean Realistic score was 43.79 (SD = 9.93), the mean Investigative
continuous score was 45.54 (SD = 9.52), the mean Artistic continuous score was 53.84
(SD = 9.24), the mean Social continuous score was 52.82 (SD = 9.81), the mean
Enterprising continuous score was 52.14 (SD = 8.96), and the mean Conventional
continuous score was 49.00 (SD = 9.59). The fact that females reflected lower levels of
interest in Realistic and Investigative domains vis-a-vis Artistic and Social domains
highlights perhaps the effects of gender socialization within the culture from which the
sample was drawn. Females were normally distributed around the mean of 100 along
each of the four MBTI dimensions. The mean EI value was 102 (SD=28.54), the mean
SN value was 104 (SD=26.91), the mean TF value was 105 (SD=23.84), and the mean JP
value was 97.11 (SD=27. 12).
Bivariate Correlations Between MBTI Dimensions and GOT§: Female Data Set

Simple bivariate correlations were calculated to determine what two-dimensional
relationships existed within the female sample between the four personality dimensions
of the MBTI and the six GOTs. Bivariate analysis revealed statistically significant
(p<.05) relationships between MBTI personality dimension SN and Strong GOT Artistic

(r=.517), between MBTI personality dimension TF and Strong GOT Social (r=.373),

94

between MBTI personality dimension EI and Strong GOT Enterprising (r= -.317),
between MBTI personality dimension SN and GOT Conventional (F -.258), between
MBTI personality dimension JP and GOT Conventional (r= -.l91), between MBTI
personality dimension JP and GOT Realistic (r=.247), between MBTI personality
dimension JP with GOT Artistic (r= .207), and between MBTI personality dimension EI
with GOT Social (r=-.156).

The five largest correlations for the full and male data sets remained significant
for the female data set (see Table 11 below). No statistically significant differences
emerged between full, male, and female data sets across the five strongest correlations.
Beyond the five largest correlations, significant yet weak (r<.3) bivariate correlations
were found within the female data set that did not exist within the full or male data sets.
These relationships included the following: SN and Investigative, TF and Enterprising,
SN and Realistic, and TF and Conventional (see Table 11 below). Only one pair of
correlated variables showed a significant (p<.05) difference between males and females.
Specifically, the relationship between TF and Investigative differed significantly between
males (r= .098) and females (r=—.220). At a less restrictive level of significance (p<.10),
SN and Realistic significantly differed between males (r= -.138) and females (r=.138).

Table 11 - Comparison of Significant Bivariate Correlations Between the MBTI

dimensions and GOTs Listed by Absolute Correlation Value: Full Data Set, Male Data
Set, Female Data Set

 

 

 

 

MBTI GOT Full Data Set ® Male ® Female ®
Sensing- Artistic .504 .481 .517
iNtuitive

Thinking— Social .384 .361 .373
Feeling

 

 

 

 

 

95

 

Table 11 (cont’d).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Extravert- Enterprising -.300 -.284 -.3 1 7
Introvert

Sensing- Conventional -.250 -.241 -.258
iNtutive

Judging- Conventional -.216 -.257 —.191
Perceiving

Judging- Realistic .190 .247
Perceiving

Exravert- Social -.142 -.156
Introvert

Judging- Artistic .138 .148 .207
Perceiving

Thinking- Artistic .1 19

Feeling

Thinking- Realistic -.1 15

Feeling

Thinking- Investigative -.101 -.220
Feeling

Judging- Social -.096 -.187

Perceiving

Sensing- Investigative .171
iNtuitive

Thinking- Enterprising -.l7l
Feeling

Sensing- Realistic .138
iNtutive

Thinking- Conventional -.135
Feeling

 

 

 

 

 

 

To What Extent Do the MBTI Personality Dimensions and the Strong GOTs Overlap:
Male Data Set

Canonical correlation analysis on the male set produced three statistically
significant functions (see Table 12 next page). Function one accounted for 39 percent of
the variance, function two accounted for 24 percent of the variance, function three
accounted for 10 percent of the variance, and function four accounted for 4 percent of the

variance. Function four was not statistically significant (p<.05).

96

Table 12 - Canonical Correlation Analysis: Male data set

 

 

 

 

 

 

Canonical Function P-value Canonical Squared Canonical
Correlation Correlation

l <.001 .624 .389

2 <.001 .494 .244

3 <.001 .324 .105

4 <.001 .194 .038

 

 

 

 

Based on reasoning employed within the analysis of full-data set canonical

functions, only canonical functions one and two were deemed practically significant.

Redundancy analysis confirmed the practical significance of functions one and two

versus three and four (see Table 13 below).

Table 13 - Canonical Redundancy Index: Male Data Set

 

 

 

 

 

 

Canonical Function Canonical R-Square Proportion Index
1 .3893 .1058
2 .2443 .0598
3 .1052 .0258
4 .0377 .0090

 

 

 

In What Ways Do the MBTI Personality Dimensions and the Strong GOTs Overlap:

Male Data Set

MBTI variable SN was the most important (standardized coefficient = 1.0039)

within the first canonical firnction and the Strong GOT variables Artistic (.9495),

 

 

Realistic (-.4692), and Conventional (-.3100) were the most important within the first

canonical function (see Table 14 below).

Table 14 - Inclusion Criteriafor First Canonical Function: Male Data Set

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

7"? Canonical Standardized Correlation w/ Correlation w/
Function Coefficient Canonical Variate Opposite Canonical
Variate

MBTI l MBTI I GOT l
Extraversion- .233 .153 .096
Introversion
Sensing-iNtuitive* 1 .004 .963 .601
Thinking-Feeling .135 .159 .099
Judging-Perceiving -.O73 .333 .208

GOT I GOT 1 MBTI 1
Realistic“ -.469 -.260 -.162
Investigative . l 59 .061 .038
Artistic“ .950 .781 .487
Social -.054 .103 .064
Enterprising -. 174 -.1 13 -.070
Conventional* -.310 -.364 -.227

 

Note: asterisk (*) indicates inclusion in the first canonical function, male data set

The complete relationship determined by the first canonical function was as

follows: NIBTI = .2333 El + 1.0039 SN + .1354 TF - .0728 JP; GOT = -.4692 R + .1591

1+ .9495 A - .0544 S -.1742 E -.3100 C. Based upon the standardized coefficient cutoff

value of +/- .25, the relationship determined by the first canonical function may be

summarized as follows: SN ("9 Artistic — 1/2 Realistic — 1/3 Conventional = .624. This

symbolizes the fact that for males, the maximum correlation occurs when SN was

correlated with the linear combination of the Artistic score minus one-half the sum of the

Realistic and one-third the sum of the Conventional scores. Across the three methods of

inclusion, disagreement existed regarding the importance of MBTI E1. E1 therefore was

not included within the first function. Practically, it may be concluded that 39 percent of

the variation within a male’s SN score was accounted for by their Artistic, Realistic, and

 

Conventional scores (see Table 12). Male bivariate correlation data suggested that a
positive, significant, and moderate correlation exists between Artistic and Realistic
(r=.25); a positive and nonsignificant relationship existed between Artistic and
Conventional (r=.09); and a positive, significant, and moderate relationship existed
between Realistic and Conventional (r=.25).

As indicated on Table 15 (see below), the second canonical function was as
follows: MBTI = -.0161 E1 - .1194 SN + .7652 TF - .7022 JP; GOT = - .4822 R -.2077 I
- .2396 A + 1.0394 S - .2340 E + .2318 C. TF and JP were the most important variables
in the MBTI variate. Social was the most important variable in the GOT variate followed
by Conventional and Investigative. GOT Realistic appeared important when considering
standardized coefficients, however, the other two inclusion methods minimized its
importance. Realistic was therefore excluded from the second function.

Table 15 - Inclusion Criteria for Second Canonical Function: Male Data Set

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2'“I Canonical Standardized Correlations w/ Correlations w/
Function Coefficients Canonical Variate Opposite Canonical
Variate

MBTI 2 MBTI 2 GOT 2
Extraversion- —.016 -.056 -.028
Introversion
Sensing-iNtuitive .119 -.134 -.066
Thinking-Feeling* .765 .750 .371
Judging-Perceiving -.702 -.628 -.311

GOT 2 GOT 2 MBT12
Realistic -.482 -. 129 -.064
Investigative“ .208 .289 .143
Artistic -.240 .103 .051
Social“ 1.039 .835 .413
Entegrrising -.234 .233 .l 15
Conventional“ .232 .382 .1 89

 

 

 

 

Note: asterisk (*) indicates inclusion in the second canonical function, male data set

99

 

The second function may be smnmarized as follows: TF — JP (--) Social + 1/5
(Investigative + Conventional) = .494. This symbolizes the fact that the second largest
correlation occurred when the difference between an individual’s TF and JP scores was
correlated with their Social score plus one-fifth the sum of Investigative and
Conventional scores. It may also be concluded that 24 percent of the variation within an
individual’s TF and JP scores was accounted for by Social, Investigative, and
Conventional scores (see Table 12). Bivariate correlation data suggests that a positive
and nonsignificant relationship existed between TF and JP within the MBTI variate
(r=.03). Within the Strong variate, bivariate data indicated that a positive, significant,
and moderate relationship existed between Social and Investigative (r=.30), Social and
Conventional (r=.30), and Investigative and Conventional (r=.45).

To mt Extent Do the MBTI Personality Dimensions and the Strong GOTs Overlap:
Female Data Set

Canonical correlation analysis on the female set produced four statistically

significant functions (see Table 16 below).

Table 16 - Canonical Correlation Analysis: Female Data Set

 

 

 

 

 

 

Canonical Function P-value Canonical Squared Canonical
Correlation Correlation

1 <.001 .601 .362

2 <.001 .5 19 .270

3 <.001 .378 .143

4 <.001 .303 .092

 

 

 

 

100

 

Function one accounted for 36 percent of the variance, function two accounted for
27 percent of the variance, firnction three accounted for 14 percent of the variance, and
firnction four accounted for 9 percent of the variance. Based on redundancy analysis,

only canonical functions one and two were deemed practically significant (see Table 17

 

 

 

 

 

 

below).

Table 17 - Canonical Redundancy Index: Female Data Set
Canonical Function Canonical R-Square Proportion Index
1 .361 7 .0935
2 .2691 .0746
3 .1429 .0368
4 .0916 .0189

 

 

 

In What Ways Do the MBTI Pergrality Dimens_iop_s_ and the Strong GOTs Overlap:
anale Dag Set

SN was the most important MBTI variable (standardized coefficient = .9934)
followed by TF (standardized coefficient = -.3642) within the first female canonical
function (see Table 18 next page). For the GOT variate, the most important variables
were Artistic (standardized coefficient = .8255), Social (standardized coefficient = -
.4310), Investigative (standardized coefficient = .3833), and Conventional (standardized

coefficient = -.2330).

101

 

Table 18 - Inclusion Criteria for First Canonical Function: Female Data Set

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1St Canonical Standardized Correlation w/ Correlation w/
Function Coefficient Canonical Variate Opposite Canonical
Variate

MBTI 1 MBTI I GOT 1
E1 .126 .002 .001
SN* .993 .919 .553
TF“ -.364 -.293 -.l76
JP -.061 .323 .194

GOT I GOT l MBTI 1
Realistic -.075 .254 .153
Investigative* .383 .430 .259
Artistic* .826 .791 .476
Social“ -.431 -.350 -.210
Enterprising -.004 -.062 -.038
Conventional“ -.233 -.323 -.195

 

 

 

 

 

m: asterisk (*) indicates inclusion in the first canonical function, female data set

The relationship may be summarized as follows: SN — 1/3 TF (--) Artistic + 1/2
(Investigative - Social) - 1/4 Conventional = .601. This symbolizes the fact that for
females, the maximum correlation occurred when the SN score minus one-third of TF
score was correlated with the Artistic score plus half the difference of Investigative minus
Social less one-fourth of Conventional scores. It may also be stated that 36 percent of the
variation within a female’s SN and TF scores was explained by their Artistic,
Investigative, Social, and Conventional scores (see Table 16). Female bivariate data
suggested that SN and TF were positively yet weakly correlated (F.08). Within the
Strong variate, bivariate data suggested that Artistic and Investigative were moderately
and positively correlated (F.27), Artistic and Social were weakly and positively
correlated (F. 17), Artistic and Conventional were weakly and negatively correlated (F-

.06), Investigative and Social were positively and weakly correlated (F.06), Investigative

102

and Conventional were positively and moderately correlated (F27), and Social and

Conventional were positively and weakly correlated (F.24).

For the second canonical function generated by the female data, TF was the most

important variable in the MBTI variate followed by SN (see Table 19 below). The most

important variable in the GOT variate was Social, with all other GOT variables excluding

Realistic having some importance (see Table 19 below).

Table 19 - Inclusion Criteria for Second Canonical Function: Female Data Set

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1st Canonical Standardized Correlation w/ Correlation w/
Function Coefficient Canonical Variate Opposite Canonical
Variate

MBTI 1 MBTI I GOT l
Extraversion- -.0844 -.1565 -.0812
Introversion
Sensing—iNtuitive* .3001 .3798 .1970
Thinking-Feeling* .9232 .9491 .4924
Judging-Perceiving -.0172 .1965 .1019

GOT I GOT 1 MB’I'I 1
Realistic -.1288 -.0787 -.O408
Investigative“ —.3350 -.3116 -.1617
Artistic“ .4301 .3611 .1874
Social“ .7790 .6576 .341 l
Enterprising“ -.4128 -.2872 -. 1490
Conventional“ -.2577 -.3854 —. 1999

 

 

 

 

Note: asterisk (*) indicates inclusion in the second canonical function, female data set.

The second function may be summarized as follows: TF + 1/3 SN(--)Social +

1/2 (Artistic - Enterprising) — 2/5 Investigative — 1/3 Conventional = .519. This

symbolizes the fact that the second largest correlation occurred when the linear

combination of TF plus one-third of SN was correlated with Social plus one-half the

difference of Artistic minus Enterprising minus two-fifths of Investigative minus one-

third of Conventional. Practically, it may be concluded that 27 percent of variation within

a female’s TF and SN scores was accounted for by Social, Artistic, Enterprising,

 

Investigative, and Conventional scores (see Table 16). Bivariate data suggested that TF
and SN were positively but weakly correlated (F.08). Within the Strong variate, female
bivariate data suggested that Social and Artistic were weakly and positively correlated
(F. l 7), Social and Enterprising were positively and weakly correlated (F.23), Social and
Investigative were positively and weakly correlated (F.06), Social and Conventional
were positively and weakly correlated (F.24), Artistic and Enterprising were positively
and weakly correlated (F.18), Artistic and Investigative were positively and moderately
correlated (F.26), Artistic and Conventional were negatively and weakly correlated
(F.06), Enterprising and Investigative were positively and weakly correlated (F.02),
Enterprising and Conventional were positively and strongly correlated (F.43), and
Investigative and Conventional were positively and moderately correlated (F.27).
To What Extent Do the MBTI Personality Dimensions and Strong Interest Themes
Overlap: Summary

Canonical correlation analysis accounted for more variance explained than did
bivariate correlation analysis. The first and second canonical functions together
accounted for 61 percent (full data set), 63 percent (male data set), and 63 percent (male
data set) of the variance between MBTI personality dimensions and GOTs. In
comparison, the strongest bivariate correlation across all three data sets, SN and Artistic,
accounted for 25 percent, 23 percent, and 27 percent of the variance between MBTI

personality dimensions and GOTs within full, male, and female data sets respectively.

104

In What Way_s_ Do the MBTI Personalifl Dimensions and Strong Interest Themes

Overlap: First Canonical Function Summary

 

Table 20 (see below) highlights canonical correlation analysis findings across all

 

 

 

three data sets.
Table 20 - Canonical Correlation Summary
First Canonical Function Second Canonical Function
Full Data Set SN ‘I'Artistic - T F -1/3 JP¢3 Social — 1/3
1/3(Realistic + Conventional) Realistic
936% variance accounted for 925% variance accounted for
Male Data Set SN 4:. Artistic — 1/2 TF — JPm Social + 1/5
Realistic - 1/3 Conventional (Investigative + Conventional)
939 % variance accounted 924% variance accounted for

 

 

Female Data Set SN - 1/3 TFm Artistic + ‘/2 TF + 1/3 SN mSocial + V2
(Investigative — Social) — 'A (Artistic-Enterprising) — 2/5
Conventional Investigative - 1/3 Conventional
936% variance accounted for 927% variance accounted for

 

 

 

The first canonical function represents the best (i.e., most highly correlated)
combination of the two variable sets in question. Analysis of the first canonical function
on the full data set revealed that 36 percent of the variation within an individual’s SN
score was explained by their Artistic, Realistic, and Conventional Scores. The relative
significance of each of the GOTs was summarized as follows: SN 69 Artistic — 1/3
(Realistic + Conventional). The first canonical function on the male data set revealed
that 39 percent of the variation within a male’s SN score was explained by their Artistic,
Realistic, and Conventional GOT scores. The relative significance of the GOTs in
relation to each other was summarized as follows: SN (-9 Artistic — 1/2 Realistic — 1/3
Conventional. The first canonical function on the female data set revealed that 36

percent of the variation within a female’s SN and TF scores was explained by their

105

 

Artistic, Investigative, Social, and Conventional scores. The relative significance of
these variables in relation to each other was summarized as follows: SN - 1/3 TF 69
Artistic + V2 (Investigative — Social) —- V4 Conventional. There appears to be very little
difference between these functions across the male and full data sets; however, the
relationship for female data set appears to be more complicated. Specifically, the first
canonical function for the female data set included, in addition those variables shared
with the other two data sets, MBTI TF and GOT Artistic. Across all three data sets,
varying combinations of Strong interest themes explained a significant amount of

variance in MBTI SN scores.

In WLIEIIWQLS. Do the MBTI Personalig Dimensions and Strong Interest Themes

 

OverlQ: Second Canonical Function Summa_ry

Typically, only the first fimction within canonical correlation is explored.
However, because the second canonical functions generated by the full, male, and female
data sets were both statistically and practically significant, the analysis included
consideration of the second function. The second canonical function represented the best
(i.e., highest correlated) combination of the two variable sets after accounting for
relationships outlined in the first canonical function. Analysis of the second canonical
function on the full data set revealed that 25 percent of the variation within an
individual’s TF and JP scores was explained by their Social and Realistic scores. The
relative significance of the GOTs in relation to each other was indicated by the value of
their respective standardized coefficients and may be summarized as follows: TF — 1/3 JP
69 Social — 1/3 Realistic. The second canonical function on the male data set revealed

that 24 percent of the variation within a male’s TF and JP scores was explained by their

106

Social, Investigative, and Conventional scores. The relative significance of the GOTs in
relation to each other was indicated by the value of their respective standardized
coefficients and may be summarized as follows: TF -- JP (-9 Social + 1/5 (Investigative
+ Conventional). The second canonical function on the female data set revealed that 27
percent of the variation within a female’s TF and SN scores was explained by their
Social, Artistic, Enterprising, Investigative, and Conventional Scores. The relative
significance of each of the GOTs in relation to each other was indicated by the value of
their respective standardized coefficients and may be summarized as follows: TF + 1/3
SN (-9 Social + 1/z (Artistic -— Enterprising) — 2/5 Investigative — 1/3 Conventional.
Across the second canonical ftmctions (i.e., full, male, and female data sets), the
consistently important variables were MBTI TF and GOT Social. Beyond these, the
important variables differed across the three data sets. Realistic appears to be important
for the full data set but not for the male and female sets. Investigative and Conventional
appear to be important with positive coefficients for the male data set but with negative
coefficients for the females. The variables Artistic and Enterprising appear to be

important for only the female data set.

107

CHAPTER FIVE
DISCUSSION

Most researchers would agree that compelling evidence exists in support of a
strong positive correlation between work satisfaction and mental health (Blustein &
Spengler, 1995). Studies highlight the reciprocal relationships between unemployment
(Osipow & Fitzgerald, 1993), job stress (Osipow, 1979) and work related problems (Herr,
1989 in Blustein & Spengler, 1995). Similar links have been found between career
satisfaction and factors such as anxiety, depression, somatic difficulties, and self-esteem
(Blustein & Spengler, 1995). In response to these observations, intervention models have
been proposed for infusing the career counseling processes with greater attention toward
increased intrapersonal and interpersonal awareness (Blustein & Spengler, 1995). .

The Person-Environment theory (P-E) represents a widely accepted theoretical
approach intended to promote individual adjustment to vocational and avocational
contextual factors as well as the ability to modify one’s vocational and avocational
environments via self-understanding, information processing skills, and goal-setting skills
(Chartrand, 1991) . In order to facilitate these objectives, the P-E process has traditionally
included assessment instruments (Buboltz, Thomas, & Johnson, in press). Two of the
most frequently used (Hammer & Kummerow, 1997) instruments are the Strong Interest
Inventory (Harmon, Hansen, Borgen, & Hammer, 1994) and the Myers Briggs Type
Indicator (Myers & McCaulley, 1985).

Both the Strong and MBTI manuals recommend using the two instruments
together (Healy, 2000). Proponents of using the two instruments together have suggested

that MBTI results can assist an individual toward a better understanding of their Strong

108

results (Healy, 2000). Furthermore, Hammer and Kummerow (1997) argued that the
MBTI and the Strong presented unique but complimentary information. Yet questions
remain regarding the integration of the two instruments (Hammer & Kummerow, 1997).
Five studies, for example, have compared with varying findings the effectiveness of using
the instruments together (Katz, Joyner, & Seamen, 1999). Four of the aforementioned
studies did not establish a statistically significant benefit for using the two instruments
together (e.g., Hayslip, 1995; O’Neil, Price, & Tracey, 1979; Takai & Holland, 1979;
Talbot & Birk, 1979) whereas the most recent study to explore the question established
evidence supporting the use of both instruments in the career counseling process (Katz et
al., 1999). Specifically, Katz et al. (1999) found that the instruments, when used in
tandem, resulted in higher rates of career related change (i.e., change in career goal,
change in specificity of career goal, and increase in certainty of career goal) as compared
to individuals who received information from a single instrument.

A parallel body of research has attempted to gain a better understanding of the
integration of the two instruments via explorations of the relationships between Holland
and Jungian personality theories. Researchers have historically attempted to do so via
bivariate correlation analysis, MAN OVA, chi square analysis, and profile mean score
analysis. By virtue of not fully exploring or accounting for potential relationships within
each classification system or by minimizing variability therein through the use of
categorical variables or bivariate analyses, fuller understandings of the potential
relationships between the MBTI personality dimensions and the Strong interest themes

have not been reached. Bivariate correlational analyses and the use of categorical data

109

are not, therefore, fully appropriate toward the development of deeper understandings of
the relationship between MBTI personality dimensions and Strong interest themes.

The current study distinguished itself from previous research through exploring
the relationship between and among MBTI personality dimensions and Strong interest
themes from a multivariate perspective. Specifically, the research design uniquely
positioned the analysis to simultaneously explore relationships among all four MBTI
personality dimensions and all Strong interest types as well as significant relationships
between MBTI personality dimensions and Strong interest themes. Canonical correlation
addressed the following research question: to what extent and in what ways do MBTI
personality dimensions overlap with Strong Interest typology? Because the Strong
Interest Inventory (Harmon et al., 1994) is normed differently for males and females, the
analysis included canonical correlation computations for the full data set as well as by
gender. These questions have not been, to the author’s awareness, simultaneously
addressed using canonical correlation in the literature thus far. Results from canonical
correlation demonstrated that (a) multivariate analysis of the relationship between MBTI
personality dimensions and Strong interest themes resulted in more significant amounts
of variance explained between the two variable sets than did bivariate correlation, (b)
multivariate analysis of the relationship between MBTI personality dimensions and
Strong interest themes resulted in a more comprehensive explanation of the relationships
between the two variable sets than did bivariate correlation, and (c) gender affects the

relationship between the two sets.

110

To What Extent Do the MBTI Personalig Dimensions and Strong Interest Themes

 

Overlap?

Canonical correlation analysis on the full data set generated four statistically
significant canonical fimctions. However, only the first two functions were considered
practically significant, together accounting for about 61 percent of the variability between
the two sets of variables. The first two canonical functions on the male data set
accounted for 63 percent of the variance between the two data sets. The first two
canonical functions on the female data set also accounted for 63 percent of the variance
between the data sets. In comparison, Hammer and Kummerow (1997), in their review
of the literature, reported bivariate correlations between MBTI preferences and Strong
interest themes ranging 0.00 to .48. This translates to, at best, bivariate analysis
accounting for 23 percent of the variance between the two data sets. Within the present
analysis, bivariate correlation accounted for, at best, 25 percent (r = .50) of the variance
between the two sets. The percentage of variance accounted for by canonical correlation
was therefore, as predicted, considerably higher than any percentage of variance
accounted for by a bivariate regression and/or correlation analysis procedures.

Support of this finding is likely attributable to the fact that MBTI personality
dimensions and Strong interest themes are indeed, as theory suggests, interactive and
dynamic in nature and should not be isolated to bivariate or categorical analysis. As has
been discussed, Holland (1985) noted in a revision of his theory that a classification
system comprised of only six independent themes was insufficient to fully represent
human and occupational workplace diversity (Harmon et al., 1994). Holland therefore

(1985) expanded his vocational interest classification system to include combinations of

111

the six themes. MBTI theory asserts that the interactions of MBTI personality
dimensions result in behaviors that cannot be predicted from exploration of each
preference separately (Myers & McCaulley, 1998). Because each preference interacts
with all other preferences, MBT I personality theory asserts that any combination of two
or more preferences will yield behaviors that are more than the sum of their parts (Myers
& McCaulley, 1998). Unique relationships among the MBTI personality dimensions
therefore exist that may suggest characteristics or behaviors that not apparent at a
bivariate level of analysis (Myers & McCaulley, 1998).

Practically and generally, these results suggest the importance of practitioners and
researchers alike attending to the multivariate nature of the relationship between the
MBTI personality dimensions and the Strong interest themes because the multivariate
results are (a) more descriptive and (b) account for a greater degree of overlap between
the two sets. Furthermore, gender appears to be an important moderator within the
relationship between MBTI personality dimensions and Strong interest themes.

In erat Ways Do the MBTI Personality Dimensions and Strong Interest Themes
Overlap?

As has been discussed above, bivariate analysis demonstrated that the strongest
relationship between the two sets occurred between MBTI Intuitive and GOT Artistic
(F.50). In comparison, canonical correlation revealed that although MBTI SN and GOT
Artistic were indeed correlated, the maximum correlation between the two sets was
obtained when MBTI SN was correlated with Strong Artistic, Realistic, and Conventional
theme scores. Specifically, the canonical formula generated on the first function (i.e.,

SN(—9Artistic -l/3(Realistic + Conventional» implied that higher levels of SN (i.e.,

112

more Intuitive) were correlated with higher levels of Artistic preferences and lower levels
of Realistic and Conventional preferences. The data therefore indicated that the strongest
correlation between the two constructs existed between the level of which an individual
processes information through facts and details, the present and what is (Sensing), versus

through the big picture, the future and what can be (Intuitive); with the level of
their interests in creating and enjoying the arts (Artistic), with engaging in hands on
activities and the outdoors activities (Realistic), and with processing and organizing
activities (Conventional).

It is important to note that these findings do not imply that Artistic preferences
were negatively correlated with Realistic and Conventional interest themes. To the
contrary, this conclusion is false based upon bivariate analyses findings. What these
findings do imply is that higher levels of SN preferences covary with higher levels of
Artistic interests and then lower levels of Realistic and Conventional interests. Canonical
correlation does not address what underlying factor(s) unify these variables. A factor
analytic design would likely be more appropriate to address this question directly. Given
this caveat, it may be intuitively and theoretically rationalized that a correlation exists
between preferences toward taking in and processing information from a “gut-level” and
future-oriented perspective (i.e., Intuitive) and interests in activities commensurate with
the Artistic theme such as the performing, visual, and musical arts. The same may be
said for reasoning a correlation between taking-in and processing information via the five
senses, prioritizing facts and details, and focusing on that which is “known” in the here
and now (i.e., Sensing) with interests in activities commensurate with the Realistic theme

such as athletics and the military and Conventional theme such as data processing and

113

accounting. The use of continuous scores in canonical correlation highlights the fact
however that a purely dichotomous (i.e., typology) conceptualization of the relationship
may not be wholly appropriate. For example, an individual may have a median SN score
(e. g. 95-105) that may be correlated with mildly to moderately elevated Artistic interests
(e. g. 55-60), average to above average Realistic interests (e. g., 50-55), and average to
above average Conventional interests (e. g., 50-55).

In conclusion of the first canonical function, a fact that is clearly supported by the
data was as follows: among the potential MBTI and GOT combinations, MBTI SN and
Strong GOT Artistic, Realistic, and Conventional covaried, or went-together, most
frequently. In practical language, it may therefore be stated that a common pattern
between the MBTI personality dimensions and the GOTs was as follows: higher SN
scores covaried with higher Artistic scores and then lower Realistic and Conventional
scores. Stated another way, canonical correlation revealed that when MBTI SN was
elevated, it was a good probability Strong Artistic was elevated and then Strong Realistic
and Conventional were de-elevated.

Canonical analysis demonstrated that the second strongest correlation relationship
between the two sets occurred when MBTI TF and JP were correlated with Strong Social
and Realistic theme scores. Specifically, the canonical formula generated on the second
function (i.e., TF - 1/3 JP 69 Social -l/3 Realistic) implied that higher levels of TF
(i.e., more Feeling) and lower levels of JP (i.e., more Judging) were correlated with
higher levels of Social interests and lower levels of Realistic interests. The data therefore
indicates that the second strongest correlation between the two constructs exists between

the level of which an individual makes decisions using facts and depersonalized rationale

114

(i.e., Thinking) versus through placing themselves within a situation and measuring the
affective impact of a decision on self and others (i.e., Feeling) and the way in which they
deal with or present to their external world; whether it be it be in an organized manner
(Judging) or in a spontaneous manner (Perceiving), with the intensity of their preferences
for helping, instructing, and working with others (Social) and for hands-on activities
(Realistic).

As was applicable to the interpretation of the fist canonical function, these
findings do not imply that the MBTI personality dimensions TF and JP were negatively
correlated or that the Strong interest themes Social and Realistic were negatively
correlated. Based upon bivariate analyses findings within the present sample, TF and JP
were positively yet quite weakly correlated (F02) and Social and Realistic interest
themes were positively and weakly yet statistically significantly correlated (F.14). What
these findings do imply is that increased levels of TF preferences and decreased levels of
JP preferences are correlated with higher levels of Social interests and lower levels of
Realistic interests. Although canonical correlation does not address what underlying
factor(s) unifies these variables, results from the second canonical function can be
theoretically rationalized. A correlation between higher levels of TF (i.e., more Feeling)
and higher levels of Social interests is somewhat intuitive. It stands to reason that
individuals who tend to engage in decision-making processes by considering the
emotional impact of a decision on self and others are also interested in helping and
serving others through teaching, counseling, or the ministry. However, the correlation
between higher levels of TF (i.e., more Feeling) and lower levels of JP (i.e., more

Judging) with higher Social interests and lower Realistic interests is somewhat

115

counterintuitive. Specifically, it may reasonably be hypothesized that individuals who
prefer flexibility, ambiguity, and spontaneity (i.e., MBTI Perceiving) may be linked more
closely with higher Social interests and lower Realistic interests. Instead, the second
canonical fimction revealed that individuals’ who prefer organization and a systematic
approach to the external world (i.e., MBTI Judging) are correlated with higher Social
interests and lower Realistic interests. On the other hand, this finding may come as no
surprise to many counselors and teachers who prefer planfullness and orderliness, task
completion, and issue resolution. Conversely, it is theoretically possible that many
individuals with Thinking preferences and Realistic interests prefer spontaneity and
flexibility (i.e., more MBTI Perceiving). Consider for example the military troop (Strong
Realistic) who relishes ambiguous and complicated battle scenarios that require ongoing
flexibility (MBTI Perceiving). When engaged in such situations, the soldier enjoys the
process of gathering as much information as possible; analyzing potential courses of
action vis-a-vis standard operating procedures, operational instruction manual guidelines,
and experience (MBTI Thinking); selecting and implementing a course of action; and
reinitiating the process based upon changing environmental data

In conclusion of the second canonical function, a fact that is clearly supported by
the data is as follows: after accounting for the relationship within the first canonical
function, MBTI TF and JP and Strong GOT Social and Realistic covaried, or went-
together, most fiequently. In practical language, it may therefore be stated that the
second most common pattern between the MBTI personality dimensions and the GOTs
was as follows: higher TF and lower JP scores covaried with higher Social and then

lower Realistic scores. Stated another way, canonical correlation revealed that when

116

MBTI TF scores were elevated and JP scores were de-elevated, it was a good probability
that Strong Social scores were elevated and then Strong Realistic scores were de-

elevated.

Implications to the PmatitionaraQanonipal Correlation Findings, Full Data Set

Authors of the Strong endorsed integrating information about an individual’s
personality, values, and beliefs via MBTI results to provide a deeper understanding
regarding what careers and work environments an individual is likely to find satisfying
and rewarding (Harmon et al., 1994). Furthermore, Hammer and Kummerow (1997)
argued that the MBTI and the Strong presented both unique and complimentary
information. Results from the first canonical correlation suggested that the search for
statistically complimentary information starts with an exploration of the relationship
between MBTI SN and Strong Artistic, Conventional, and Realistic scores. Specifically,
the data indicates that the most significant correlation between the MBTI personality
dimensions and the Strong interest themes existed between the level in which an
individual processes information through the facts and what is (Sensing) versus through
consideration of the future and what may be (Intuitive) and the level of their interests in
Artistic, Realistic, and Conventional activities. The data suggested that increased levels
of SN preferences are correlated primarily with increased Artistic interests but also with
lower Realistic and Conventional interests. Based upon second canonical function
findings, the practitioner is next challenged to consider the level to which an individual
prefers to make decisions by obtaining facts (Thinking) versus by employing a subjective

values-oriented approach (Feeling) and the level to which an individual prefers to

117

approach the external world, whether it be in an organized and procedural manner
(Judging) or in a spontaneous and flexible manner (Perceiving). These preferences are
directly correlated primarily to the level of an individual’s interests for Social activities
such as teaching and counseling but also with lower levels of interest in Realistic
activities such as construction, engineering, or military service. Unique information such
as MBTI E1 preferences and level of interest in Strong Investigative and Enterprising
theme domains may be used to complete the explanatory picture created for an individual
seeking to increase self-understanding, promote decision-making facilitation toward
career preferences and interests, and to ultimately achieve career satisfaction.

A number of valid factors may create a scenario wherein individuals do not fit the
profile highlighted. For such individuals, the career counselor should consider the fact
that the generalizability of the present study is limited in several important ways. Given
this caveat however, Holland (1985) stated that individuals with similar preference
patterns tend to locate themselves within similar occupations. Furthermore, Holland
(1985) stated that when a mismatch between individual interests and the occupational
environment occurred, the individual tended to either adjust their interests to meet the
demands of the environment or to adjust the environment to adapt to their interests and
preferences. If unsuccessful with either, the individual tended to pursue other
occupational options or environments. The career counselor therefore would benefit by
being aware of common patterns between MBTI and Strong data and by appropriately
preparing individuals who do not fit within these common patterns for potential inter and

intrapersonal hurdles they may face in pursuit of their interests and preferences.

118

In What Ways Do the MBTI Personaliy Dimensions and Strong Interest Themes

 

Overlap: The Effect of Gender

Based upon the data, researchers and practitioners alike are also encouraged to
consider gender as an important moderator variable when interpreting similarities
between MBTI personality dimensions and Strong Interest themes. Canonical correlation
results differed across males and females within the present sample. Gender results also
differed at times from results generated from the firll data set.

The first canonical function for males included the same variables as the first
canonical fimction for the full data set (see Table 20). Canonical correlation revealed that
the maximum correlation among males between the two sets was obtained when MBTI
SN was correlated with Strong Artistic, Realistic, and Conventional theme scores. The
canonical formula generated on first function (i.e., SN69Artistic -1/2 Realistic — 1/3
Conventional) implied that, among the potential MBTI and GOT combinations, MBTI
SN and Strong GOT Artistic, Realistic, and Conventional covaried, or went-together,
most frequently within the male sample. In practical language, it may therefore be stated
that the most common pattern between the MBTI personality dimensions and the GOTs
for males was as follows: higher SN scores covaried with higher Artistic scores and then
lower Realistic and Conventional scores. Stated another way, canonical correlation
revealed that when MBTI SN was elevated, it was a good probability that Strong Artistic
was elevated and then Strong Realistic and Conventional were de-elevated. The data
therefore indicated that the strongest correlation among males between the two variable
sets existed between the level of which a male processes information through facts and

details, the present and what is (Sensing), versus through the big picture, the future and

119

what can be (Intuitive); with the intensity of his interests in creating and enjoying the arts
(Artistic), with engaging in hands on activities and the outdoors activities (Realistic), and
with processing and organizing activities (Conventional).

It is notable that although the first canonical function for males appeared similar
to the first canonical function derived from the full data set, Strong interest theme
Realistic played a more significant role in terms of its contribution to the canonical
equation (see Table 20). The fact that the Realistic interest theme exacted a more
significant role with males is supported by Strong Interest Inventory normative data.
Furthermore, the Realistic theme elicited the second highest mean for males within the
present sample (x=51.1), second only to Enterprising (x=52.47). However, the
significance of the correlation between SN and Realistic in connection with Artistic and
Conventional interest levels is new data.

Canonical correlation revealed that the second most significant correlation
between the two male variable sets was obtained when MBTI TF and JP were correlated
with Strong Social, Investigative, and Conventional theme scores. Specifically, the
canonical formula generated on the second fimction (i.e., TF-JP69 Social
+l/5 {Investigative + Conventional}) implied that, after accounting for the relationship
within the first canonical function, MBTI TF and JP and Strong GOT Social,
Investigative, and Conventional covaried, or went-together, most frequently. In practical
language, it may therefore be stated that the second most common pattern within males
between the MBTI personality dimensions and the GOTs was as follows: higher TF and
lower JP scores covaried with higher Social, Investigative, and Conventional scores.

Stated another way, canonical correlation revealed that when MBTI TF scores were

120

elevated and JP scores were de-elevated, it was a good probability that Strong Social,
Investigative, and Conventional scores were elevated. The data therefore indicated that
the second strongest correlation between MBTI personality dimensions and Strong
interest themes among males existed between the level of which a male makes decisions
via facts and detached rationale (Thinking) versus through placing himself within a
situation and assessing the affective impact of options on others and self (Feeling), and
his attitude toward the outside world; whether he prefer spontaneity and flexibility
(Perceiving) or order and structure (Judging); with the intensity of his interests in creating
and enjoying the arts (Artistic), with engaging in hands on activities and the outdoors
activities (Realistic), and with processing and organizing activities (Conventional).
Canonical correlation revealed that the maximum correlation among females
between MBTI personality dimensions and Strong interest themes was obtained when
MBTI SN and TF were correlated with Strong Artistic, Investigative, Social, and
Conventional theme scores. Specifically, the canonical formula generated on first
function (i.e., SN — 1/3 TF69Artistic +1/2 {Investigative — Social} —- ‘/4 Conventional)
implied that MBTI SN and then JP and Strong GOT Artistic, Investigative, Social, and
Conventional covaried, or went-together, most frequently among females within the
sample. In practical language, it may therefore be stated that the most common pattern
between the MBTI personality dimensions and the GOTs for females was as follows:
higher SN and then lower TF scores covaried with higher Artistic and Investigative
scores and then lower Social and Conventional scores. Stated another way, canonical
correlation revealed that among females in the sample, when MBTI SN scores were

elevated and TF scores were de-elevated, it was a good probability that Strong Artistic

121

and Investigative scores were elevated and then Strong Social and Conventional scores
were de-elevated. The data therefore indicated that the strongest correlation among
females between the two variable sets existed between the level of which a female
processes information through facts and details, the present and what is (Sensing), versus
through the big picture, the future and what can be (Intuitive) and the level to which a
female goes about making decisions based upon objective and rational methodology
(Thinking) versus through a more personal, subjective, and values-oriented approach
(Feeling) with the level of her Artistic, Investigative, Social, and Conventional interests.
The first female canonical function highlighted the fact that Investigative and
Social are both moderately and nearly equally important after Artistic (standardized
coefficients = .38 and -.43 respectively) but are related differently to MBTI SN and TF.
This finding paralleled Holland’s (1985) theory that Investigative and Social were highly
dissimilar. Bivariate analysis within the present sample noted a correlation between the
two only slightly above zero (i.e., .06). Interestingly, the only variables not included
within the first female canonical function were Realistic and Enterprising, two interest
themes commonly associated with males. It should not be implied however from this
data alone that females’ necessarily possess low levels of Realistic and Enterprising
interests. Data indicates that females within the sample did possess lower mean levels of
Realistic interests (x=43) but above average mean Enterprising scores (x=52). What the
data does indicate is that Realistic and Enterprising interests did not significantly
contribute to significant percent of variance explained within the MBTI variate. Interests
in these domains likely should be considered as contributing unique information toward

self-understanding in addition to El and JP preferences.

122

 

Canonical correlation revealed that the second strongest correlation between the
two variable sets occurred when MBTI TF and SN were correlated with levels of interest
in Social, Artistic, Enterprising, Investigative, and Conventional activities. The canonical
formula generated for the second female function, TF + 1/3 SN 69 Social +1/2 (Artistic
- Enterprising) - 2/5 Investigative — 1/3 Conventional, suggested that after accounting for
the relationship within the first canonical firnction, MBTI TF and SN and Strong GOT
Social, Artistic, Enterprising, Investigative, and Conventional covaried, or went-together,
most frequently. In practical language, it may therefore be stated that the second most
common pattern between the MBTI personality dimensions and the GOTs for females
was as follows: higher TF and SN scores covaried with higher Social and Artistic Scores
and then lower Enterprising, Investigative, and Conventional scores. Stated another way,
canonical correlation revealed that when MBTI TF and SN scores were elevated, it was a
good probability that Strong Social and Artistic scores were elevated and then Strong
Enterprising, Investigative, and Conventional scores were de-elevated.. The data
therefore indicated that a significant correlation existed between (a) the level to which a
female processes information through facts and details, the present and what is (Sensing),
versus through the big picture, the future and what can be (Intuitive); (b) the level to
which a female goes about making decisions based upon objective and rational
methodology (Thinking) versus through a more personal, subjective, and values-oriented
approach (Feeling); (c) and the level of her Social, Artistic, Enterprising, Investigative,
and Conventional interests.

It is notable in the second canonical function that all Strong interest themes with

the exception of Realistic played a significant role toward the percent of variance

123

explained within TF and SN scores. According to Holland’s (1985) theory, this could be
due to females’ within the sample possessing undifferentiated (aside from Realistic)
interest preferences. According to Holland (1985), differentiation occurs when an
individual clearly expresses a preference for a Holland personality type(s) in relation to
other types. Theoretically, differentiation will be evidenced only when an individual’s
self-identity is well developed. Instead of labeling a client with an identity issue
however, canonical correlation may provide a stimulus for processing undifferentiated
vocational interests. The practical implications of the present data are as follows: if a
counselor is concerned about level of differentiation with individual clients, she is
encouraged to explore the client’s clarity of preference and dominant functions within TF
and SN domains. According to the first and second canonical functions, if preferences
are somewhat clear within these domains and Sensing or Intuition form the dominant
functions, the counselor might consider the exploration of Artistic interests. If the client
has higher levels of Intuitive preferences, a counselor might explore, based upon
canonical correlation findings, career options that are primarily Artistic but also blend
aspects of Investigative activities. If the client has higher levels of Sensing preferences,
the counselor may consider exploring Artistic, Social, and Conventional interests,
prioritizing a balance between Artistic and Social. If the client’s dominant function is
TF, higher scores (i.e., more Feeling) may suggest the consideration of primarily Social
occupations with a blend of Artistic activities. If scores are more toward MBTI
Thinking, a counselor may start with the exploration Artistic and Enterprising interests
mixed with levels of interest in Investigative and Conventional interests. As this

discussion and examination of the canonical functions has highlighted, a female’s level of

124

interest in Artistic and Social activities has an effect on the relationship between MBTI
SN and TF. It is important to note however that other Strong interest themes impact this
relationship as well. The specifics of this are missed when performing bivariate analysis
only.

In What Ways Do the MBTI Personality Dimensions and Strong Interest Themes

 

Overlap: A Summm

Hammer and Kummerow (1997) stated that more information was needed
regarding unique versus complimentary contributions of the MBTI personality
dimensions and the Strong interest themes. The present data suggested that (a) the two
constructs do offer both unique and complimentary information and (b) the specifics
therein are moderated by gender.

Data 6pm the first canonical function, full data set, revealed that Artistic,
Realistic, and Conventional scores covaried with MBTI SN scores. Among males and
females combined therefore, the most common pattern between the two instruments was
as follows: when SN scores were elevated, it was a strong probability that GOT Artistic
scores would be elevated and then Realistic and Conventional scores would be de-
elevated. The same most common relationship was true for males; however, the intensity
of interests for Realistic tasks was more important toward the explanation of variance in
SN scores. Data from the first canonical function, female data set, revealed that Artistic,
Investigative, Social, and Conventional scores covaried with MBTI SN and TF scores.
When females were evaluated in isolation therefore, the most common pattern between

the two instruments was as follows: when SN scores were elevated and JP scores de-

125

elevated, a strong probability existed that Artistic and Investigative scores were elevated
and then Social and Conventional scores were de-elevated.

Data fiom the second canonical firnction, full data set, revealed that Social and
Realistic scores were correlated with MBTI TF and JP scores. The second most common
pattern therefore between the two instruments was as follows: when TF scores are
elevated and JP scores are de-elevated, it is a good probability that GOT Social will be
elevated and then GOP Realistic scores will be de-elevated. Data from the second
canonical ftmction, male data set, revealed that Social, Investigative, and Conventional
scores were complimentary toward variance explained in TE and JP scores. For males
therefore, the second most common pattern between the two instruments was as follows:
when TF scores are elevated and JP scores are de-elevated, it is a good probability that
GOT Social, Investigative, and Conventional scores will be elevated The second
canonical firnction for the female data set revealed that several Strong interest themes
(i.e., Social, Artistic, Enterprising, Investigative, and Conventional) were significant
toward proportion of variance explained in MBTI TF and SN scores. For females
therefore, data from the first and second canonical functions revealed that knowing where
women are on MBTI SN and TF scales is most important toward revealing overlap
between the two instruments.

Contrary to Jungian theory and one of the research hypotheses within the present
analysis, MBTI personality dimensions were found to correlate with one another within
four of six canonical functions generated. This is likely not because Jung was incorrect
in asserting the independence of his dimensions but more likely attributable to the fact

that when analyzed in context of the Strong interest themes, MBTI dimensions may

126

indeed be complimentary while others remain unique in their contribution. The
practitioner is encouraged to bear this in mind when interpreting results of the two
instruments. For example, the practitioner should consider that data indicates a
significant proportion of variance within SN and TF scores are explained by a female’s
level of interest in Artistic, Investigative, Social, and Conventional interest themes. The
practitioner is therefore encouraged, based upon the data, to consider and to discuss the
ways in which a female’s interests in performing arts or drama (Artistic), science
(Investigative), education (Social) and data processing (Conventional) may be influenced
by the ways in which they gather information (SN) and make decisions (TF). Based upon
these. findings, the researcher is encouraged to move beyond bivariate analysis with
MBTI dimensions and to use multivariate strategies.

Findings across canonical functions were largely commensurate with Holland’s
(1985) theory and with research hypotheses. On the first canonical function across full,
male, and female data sets, Artistic was the consistently important variable. On the
second canonical function across full, male, and female data sets, Social was the most
important variable. These findings correspond to previous research. Buboltz et al. (in
press), for example, concluded from their analysis that the majority of the differences
within their data occurred for individuals with either Artistic or Social Strong interest
themes. Commensurate with Dillon and Weissman (1987) and Hammer and Kummerow
(1997) findings, Artistic individuals demonstrated higher levels of MBTI Intuition than
all other codes. Commensurate with Dillon and Weissman (1987), Decola (1992), and
Hammer and Kummerow (1997) findings, individuals with a Strong Social interest theme

demonstrated higher levels of MBTI Feeling than all other Strong interest themes

127

(Buboltz et al., in press). Taken together, these results indicated with regard to the MBTI
scales, both Artistic and Social interest themes might possess unique personality profiles
(Buboltz et al., in press). What previous research missed however was the role of other
Strong interest themes within the context of MBTI personality dimensions. Specifically,
although Artistic was the most important variable within the first canonical function,
Realistic and Conventional played statistically significant roles in the full and male data
sets. Investigative, Social, and Conventional played statistically significant roles in the
female data set. The roles that these themes played were consistent with Holland’s
(1985) theory. In the full and male data sets, an individual’s scores on the Realistic and
Conventional themes was considered vis-a-vis their level of interest in Artistic activities
and contributed to the proportion of variance explained in MBTI SN. Realistic and
Conventional are located adj acently along Holland’s hexagon and therefore proposed to
be similar. Realistic and Artistic are located in close proximity along the hexagon and
therefore prOposed to be moderately to weakly correlated. Conventional and Artistic are
located diametrically opposite of one another along the hexagon and therefore are
proposed to be highly dissimilar. The canonical function for the full data set captured
these dynamics by adding Realistic and Conventional scores together then subtracting the
total by one-third from the Artistic score. This combination was able to account for 36
percent of the variance in SN scores whereas the Artistic-SN bivariate connection only
accounted for 25 percent of the variance. Canonical function for the male data set
captured the same Holland dynamics but also accounted for the fact that Realistic tends to
be stronger in males and therefore have a greater impact on Artistic interests. A greater

measure of the Realistic score was therefore subtracted from the Artistic score. Within

128

the female data set, Investigative and Conventional scores were subtracted from one
another and the resultant sum added to Artistic. Investigative and Conventional are
located opposite one another on the hexagon and therefore considered to be dissimilar.
Yet they both are located adjacent to the Artistic theme and therefore considered to be
congruent with this theme. Conventional, located opposite of Artistic and only weakly
correlated with Investigative and Social, was subtracted as well. Taken together, this
combination of scores accounted for 36 percent of the variance in MBTI SN and TF.
Given this information, practitioners and researchers are encouraged to consider the
dynamic nature of the two constructs in relation to one another and to look beyond
bivariate associations.
Conclusion

Assumptions underlying the use of the Strong Interest Inventory (Strong) and the
Myers-Briggs Type Indicator (MBTI) are consistent with the P-E model (Hammer &
Kummerow, 1997). Authors of both the Strong and the MBTI recognized a need to go
beyond a simplistic matching of people with jobs, that people can act as change agents in
shaping their own careers, and that instrument users should learn a career exploration
process based upon a self-exploration rather than simply attain a particular career goal
(e. g., choosing a vocation) (Hammer & Kummerow, 1997). Authors of both instruments
also believed that the knowledge of individual characteristics can be used to not only
evaluate the match with a possible job but also to inform the counseling approach used
with clients. Authors of the Strong noted, “using the Strong and the Myers-Briggs Type
Indicator has been shown to be particularly useful” (Harmon et al., 1994, p. 238). From a

P-E perspective, use of these two instruments together can, according to Strong authors,

129

provide deeper levels of self-understanding leading to improved information-processing
and critical thinking procedures and help confirm particular areas that warrant further
exploration (Harmon et al., 1994). The current study calls into question the validity of
treatment planning and implementation based upon bivariate research and suggests the
need for further research matching treatment planning with multivariate findings.
Specifically, does a client’s and a counselor’s knowledge and understanding of
multivariate MBTI-Strong relationships significantly enhance the facilitation of Person-
Environment goals including increased self-understanding, increased decision-making
process skills, increased clarity of occupational selection, and increased occupational
satisfaction. Present findings suggest but do not confirm that P-E process goals will be
better facilitated using multivariate analysis considerations versus bivariate and/or
categorical analysis.

Hammer and Kummerow (1997) stated that more information was needed
regarding unique versus complimentary contributions of the MBTI personality
dimensions and the Strong interest themes. The present data suggests that (a) the two
constructs do offer both unique and complimentary information and (b) the specifics
therein are moderated by gender. Further research is needed to test the reliability of the
present findings, bearing in mind that test-retest reliability has been referred to as the
Achilles heel of the MBTI among its critics (Hammer, 1996). Hammer (1996)
highlighted that because cutoff scores are used within each of the four dichotomies, the
size of acceptable standard error around the cutoff score is decreased. F urtherrnore,
under the current scoring system research indicates that many individuals receive low

preference scores (Hammer, 1996). Therefore, forcing such individuals to fit into a two-

130

category structure can produce a high rate of erroneous classifications (Hammer, 1996).
Most of the type-switches within test-retest analyses occur not surprisingly in individuals
with low preference clarity (Hammer, 1996). The present study sought to resolve this
issue in part via the use of continuous scores as opposed to typology status. Continuous
scores are still susceptible, however, to decreased reliability due to ambiguous clarity of
preference on the MBTI dimensions.

The generalizability of the data is limited by several important considerations.
The fact that only college students were used within the sample limits generalization to
other groups. College students were used for at least two important reasons. First,
Strong interest theme data is purported to be particularly useful when counseling or
advising beginning college students (Harmon et al., 1994). The GOTs can be used to
help those students who are not prepared to select a major to understand their general
interests (Harmon et al., 1994). The themes and their relationship on the hexagon are
purported to provide a cognitive map for students who are trying to select and academic
major from what is often a confirsing web of options (Harmon et al., 1994). Second,
previous research primarily employed a college student population. This is largely
attributable to the fact that the most readily available and applicable population from
which to sample is college students who are typically confronting important occupational
choices. Aside from the homogenous nature of the sample, other factors relevant to the
college sample have been found to impact reliability. A summary of MBTI Form G data
suggested that sample characteristics may create reliability variation across groups.
Specifically, groups composed of people who are believed to have less command over

perception and judgment (e. g., chronologically younger participants) should exhibit lower

131

reliabilities (Myers & McCaulley, 1998). An individual’s level of achievement has also
been found to affect the quality of perception and judgment (Myers & McCaulley, 1998).
It is therefore expected that within samples consisting of comparable age levels, those
individuals with higher achievement levels will report their preferences more consistently
(Myers & McCaulley, 1998). College and university samples possessed higher
reliabilities than did high school samples (Myers & McCaulley, 1998). Reliabilities were
also found to be higher in groups with higher average intelligence as measured by
standard intelligence tests (Myers & McCaulley, 1998). A sample comprised of differing
age, achievement, and intelligence levels is therefore suggested toward increased
generalizability of findings.

A second limitation to generalizability is the racial/ethnic homogeneity of the
sample. Although the exact racial and ethnic demographics of the sample are unknown,
general setting characteristics suggest the sample was primarily White. Individuals
identifying as Asian or as American Indian were the next most fiequently occurring
population groups within the university from which the sample was derived. As has been
discussed, Holland’s theory in particular has been scrutinized for cross-cultural
applicability. In the early to middle 1990’s, the belief emerged that perhaps confidence in
the structure of Holland’s model and its applicability to the culturally diverse was
inaccurate. Gati (1991) claimed that Holland’s order did not adequately account for the
relationships among the six types (Rounds & Tracey, 1996). Rather, according to Gati
(1991) a three-group partition [(RI) (AS) (EC)] demonstrated a better fit with cross-
cultural populations (Rounds & Tracey, 1996). According to Gati’s (1991) model,

correlations between types belonging to the same category (e.g., R, I) were greater than

132

correlation’s between pairs of types outside the category (e. g., R, A) (Rounds & Tracey,
1996). Both Holland’s and Gati’s models have since received criticism for the
relationship ascribed between the A and S types (Rounds & Tracey, 1996). Several
researchers have noted the cross-cultural discontinuity between Artistic (A) and other
Holland types. Rounds and Tracey (1996) attempted to address this discrepancy through
adjusting Gati’s (1991) model so that the A type existed within its own category and the
S type was added to the EC category [i.e., (RI) (A) (SEC)]. To test their hypothesis,
Rounds and Tracey (1996) conducted a structural meta-analysis to evaluate the cross-
cultural validity of the three models. Results indicated that Gati’s (1991) model and
Rounds and Tracey’s (1996) model achieved the best fit to the data. Rounds and Tracey
(1996) suggested that the results might be used to (a) counterpoint previous conclusions
supporting Holland structure across cultures and to (b) warn publishers, authors, and
assessors that the knowledge pertaining to consistency within Holland’s model may not
be applicable to cross-cultural groups. In 1998, however, Day and Rounds seemed to
offer contradictory evidence. Day and Rounds (1998) asserted that people of different
ethnicity and sexes hold the same cognitive map of the workworld when the structures of
preferences are explored. Furthermore, Day and Rounds (1998) stated that the “RIASEC
arrangement of general occupational themes prevails legitimately. Previous research
suggesting otherwise probably suffered multiple perils of . . . small sample size,
unrepresentative sampling, selective attrition. . .and attenuated categorization” (p. 735).
Due to the mixed results regarding the cross-cultural validity of Holland’s model, ensuing
studies are encouraged to pursue racial diversity within their sample or sample racial and

ethnic groups in addition to White/Caucasian.

133

The use of canonical correlation limits the findings in three important ways. First,
canonical correlation is designed to examine the relationships between two groups of
variables. It therefore does not answer questions regarding specific relationships within
each variate outside of the context of the opposite variate. For example, canonical
correlation pernrits the conclusion that higher levels of SN are correlated with higher
levels of Artistic and lower levels of Realistic and Conventional interests. This does not
imply that outside of the context of the relationship with SN, Artistic and Realistic and
Artistic and Conventional are necessarily negatively correlated. Bivariate analysis is
therefore useful toward the explanation of relationships within canonical variates.

Second, canonical correlation is an explanatory tool and not a predictive tool. Future
research should consider employing predictive data analytic models to the present
findings. Finally, as the well-known statistical principle states, correlation does not equal
causation. Although canonical correlation uncovered important correlations between
MBTI personality dimensions and Strong interest themes, it does not address potential
underlying factors that unify the relationships discovered and which could validly be said
to “cause” the relationship. Future research studies should consider a data analytic
method that allows for the exploration of, specifically, underlying factors unifying
correlated relationships between the MBTI personality dimensions and the Strong Interest

themes.

134

REFERENCES

Blustein, D. & Spengler, M. (1995). Personal adjustment: Career counseling and
psychotherapy. In Walsh, B.W., & Osipow, S., (Eds), Handbook of Vocational

Psychology

Buboltz, W., Jr., Thomas, A., & Johnson, P., (in press). The relationship between
the MBTI and S11: Implications for Conjunctive Use in Counseling. In Columbus, F .,
(Ed.), Advances in Paycholagy Research.

Buboltz, W., Jr., Johnson, P, Cassiday, S., & Miller, M.A. (1999). The
relationship between Strong personal style scales and MBTI poles. Journal of Career
Assessment, 8, 131-146.

Chartrand, J.M. (1991). The evolution of trait-and-factor career counseling: A
person x environment fit approach.” J om] of Counseling apd Development, 6_9_, 518-
524.

Day, S.X., & Rounds, J. (1998). Universality of vocational interest structure
among racial and ethnic minorities. American Psychologist, 53, 728-736.

DeCola, L. A. (1992). The relationship between the Myers-Briggs Type Indicator
and Strong-Campbell Interest Inventory. (Doctoral Dissertation, University of Akron,
Akron, Ohio).

Dillon, M., & Weissman, S. (1987). Relationship between personality types on
the Strong-Campbell and Myers-Briggs Instruments.” Measurement and Evaluation in
Counseling and Developmeat, 2Q, 68-79.

DiTiberio, J .K., & Hammer, A.L., (1993). Manual: Introduction to m in
college. Palo Alto, CA: Consulting Psychologists Press, Inc.

Dowd, ET. (1998). Counseling psychology. In R.J. Corsini & A.J. Auerbach
(Eds), Concise encyclopedia of psychology (pp 178-179). New York: John Wiley and
Sons

Graff, R.W., Larrimore, M., Whitehead, G.I., & Hopson, NW. (1991). Career
counseling practices: A survey of college/university counseling centers. Poster
presented at the annual meeting of the American Psychological Association, San
Francisco, CA.

Hackett, G., & Lent, R.W., (1992). Theoretical advances and current inquiry in

counseling psychology. In S.D. Brown & R.W. Lent (Eds), Handbook of counseling
psychology (pp. 419-452).

135

Hall, C.S., & Lindzey, G., (1970). Theories ofpersonality (2 ed). New York,
NY: John Wiley and Sons, Inc.

Hammer, A.L., (1996). MBTI applications: A decade of research on the Myers-
Briggs Type Indicator. Palo Alto, CA: Consulting Psychologists Press, Inc.

Hammer, A.L., & Kummerow, J .M. (1997). Strong and MBTI career
development guide - revised. Palo Alto, CA: Consulting Psychologists Press, Inc.

Harmon, L.W., Hansen, J .C., Borgen, F.H., & Hammer, AL. (1994). Strong
interest inventory applications and technical gpide. Stanford, CA: Stanford University
Press.

Healy, CC, (2000). Interpreting the Myers-Briggs Type Indicator to help clients
in understanding their Strong Interest Inventory. J oml of Career Development, 26,
295-308.

Holland, J .L. (1985). Making vocational choices (2 ed.). Englewood Cliffs, NJ:
Prentice-Hall.

J accard, J ., & Becker, M.A. (1990). Statistics for the behavioral sciences (2 ed.)
Behnont, CA: Wadsworth Publishing Company.

Jones, A.J., Stefflre, B., & Stewart, NR. (1970). Principles of Guidance (6‘h ed.).
New York: McGraw-Hill.

Katz, L., Joyner, J.W., & Seaman, N., (1999). Effects of joint interpretation of
the Strong Interest Inventory and the Myers-Briggs Type Indicator in career choice.
Journal of Career Assessment, 7. 281-297.

Lowrnan, R.L. (1987). Occupational choice as a moderator of psychotherapeutic
approach. Psychotherapy, _2_4_, 801-808.

Lowrnan, R.L. (1991). The clinical practice of career assessment: Interests},
abilitiesaand personalig. Washington, DC: American Psychological Association.

Miller, M.A., Buboltz, W.C., (1999, April). The relationship between MBTI
attitude types and SH Holland interest types. Poster presented at the annual meeting of
the Great Lakes Conference, Ohio State University, Columbus, OH.

Miller, M.J. (1992). Synthesizing results from an interest inventory and a

personality inventory to improve career decision-making. J oumal of Employment
Counseling, _2_2, 50-59.

136

Myers, I.B., & McCaulley, M.H. (1998). Manual: A guide to the developmenj
and use of the Myers-Briggs Type Indicator. Palo Alto, CA: Consulting Psychologists
Press, Inc.

Myers, I.B., & McCaulley, M.H. (1985). Manual: A guide to the development
and use of the Myers-Briggs Type Indicator. Palo Alto, CA: Consulting Psychologists
Press, Inc.

Pinkney, J .W. (1983). The Myers-Briggs Type Indicator as an alternative in
career counseling. Personnel and Guidance Journal, _6_l, 173-177.

Rounds, J. & Tracey, TJ. (1990). From trait and factor to person-environment fit
counseling: Theory and process. In W.B. Walsh & S. H. Osipow, (Eds), Career
Counseling. Hillsdale, NJ :Erlbaum.

Rounds, J. & Tracey, T]. (1996). Cross-cultural structural equivalence of
RIASEC models and measures. Journal of Counseling Psychology. 43. 310-329.

Strong, E.K. (1943). Vrgtional interests of men and women. Stanford, CA:
Stanford University Press.

Zunker, V.G. (1998). Career counseling. Applied concppts of life planning. (5
ed.) Pacific Grove, CA: Brooks/Cole Publishing.

Zunker, V.G. (1994). Career counseling. Applied concppts of life planning. (4
ed.) Pacific Grove, CA: Brooks/Cole Publishing.

137