See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/227147781

The Therapeutic Factors Inventory: Development of a

Scale

Article  in  Group · December 2000

DOI: 10.1023/A:1026616626780

CITATIONS READS
96 7,346

2 authors, including:

Rebecca R. MacNair-Semands
University of North Carolina at Charlotte
33 PUBLICATIONS   882 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Ethical issues related to suicide risk assessment in group therapy View project

All content following this page was uploaded by Rebecca R. MacNair-Semands on 16 September 2015.

The user has requested enhancement of the downloaded file.

P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

Group, Vol. 24, No. 4, 2000

The Therapeutic Factors Inventory:

Development of a Scale
Karen P. Lese1,3 and Rebecca R. MacNair-Semands2

Although Yalom’s (1995) framework of the “therapeutic factors” facilitating out-

come in group has been accepted by group specialists, no empirically based in-
strument assesses all of these factors. The Therapeutic Factors Inventory (TFI),
with 11 scales based on the therapeutic factors, has been designed to fill this gap.
This article summarizes the development and preliminary reliability testing of the
TFI. Each scale of the instrument demonstrated high internal consistency; how-
ever, one scale obtained unacceptably low test-retest reliability. Further validity
testing is needed. Implications of these findings are discussed.
KEY WORDS: therapeutic factors; group therapy; psychological assessment; psychometrics; instru-
ment development.

Yalom’s (1970, 1975, 1985, 1995) classification of the therapeutic factors

in therapy groups has been the most widely adopted version of this popular con-
cept. His articulation of the therapeutic factors includes eleven elements thought
to beneficially affect client growth. The factors include instillation of hope, uni-
versality, imparting information, altruism, the corrective recapitulation of the pri-
mary family group, development of socializing techniques, imitative behavior,
interpersonal learning, group cohesiveness, catharsis, and existential learning.
As described in an early article examining the factors (Rorhbaugh & Bartels,
1975, p. 430), “The therapeutic factors are . . . provisional, interdependent, and not
mutually exclusive. Some refer to mechanisms of change (e.g., catharsis) while
others seem to be conditions of change (e.g., group cohesiveness, instillation of
hope).”

1 Counseling Center, University of San Diego.

2 Counseling Center, University of North Carolina at Charlotte.
3 Correspondence should be directed to Karen Lese, University of San Diego Counseling Center, 5998
Alcala Park, San Diego, CA 92110-2492; e-mail: internet to klese@acusd.edu.

303
0362-4021/00/1200-0303$18.00/1 °
C 2000 Eastern Group Psychotherapy Society
P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

304 Lese and MacNair-Semands

These concepts have great utility for group therapists. The importance of
the factors has been discussed for many types of clients, including those strug-
gling with learning disabilities (Brown, Hedinger, & Mieling, 1995), incest expe-
riences (Randall, 1995), alcohol addiction (Lovett & Lovett, 1991), hearing im-
pairment (Card & Schmider, 1995), obsessive-compulsive disorder (Kobak, Rock,
& Greist, 1995), and grief (Price, Dinas, Dunn, & Winterowd, 1995). It has been
suggested that what is therapeutic in a given group may vary by client cultural
factors (DeLucia-Waack, 1996) or age (McLeod & Ryan, 1993) and that within-
group cultural differences may mediate any therapist’s ability to foster therapeutic
factors (Johnson, Torres, Coleman, & Smith, 1995). Treatment setting (Fuhriman,
Drescher, Hanson, Henrie, & Rybicki, 1986), group format (Goldberg, McNiel,
& Binder, 1988; Kellerman, 1987), and time-limits (Flowers, 1987; Marcovitz
& Smith, 1983) may also cause variability in the influence of the therapeutic
factors. Furthermore, the stage of group development affects what therapeutic
factors are significant (Yalom, 1995) and how interventions work (Clark, 1993).
Given such conceptual understanding, it makes sense that the therapist’s ability
to target different therapeutic factors for different presenting problems, cultural
groups, or stages of group development could improve the effectiveness of any
given group (see Crouch, Bloch, & Wanlass, 1994, for discussion). However,
tailoring of interventions to therapeutic factors within specific groups is extraordi-
narily difficult to accomplish without reliable and valid methods of assessing the
factors.
Measurement of the therapeutic factors has been similarly difficult (Bednar
& Kaul, 1994). Two basic approaches, characterized by Bloch and Crouch (1985)
as “direct” and “indirect,” have been utilized. The indirect approach to measuring
therapeutic factors in group therapy has also been called the “critical incident” ap-
proach (see MacKenzie, 1987, for a discussion of this approach). Research using
this approach asks participants to describe the most significant event that occurred
in group. Descriptions are then rated qualitatively. The format for this approach
varies from study to study, as does the degree to which the underlying factor
structure is validated (Kivlighan, Multon, & Brossart, 1996). Because its format is
unstructured, the indirect approach biases the respondent less than direct inquiry
(Bloch & Crouch, 1985; Crouch, Bloch, & Wanlass, 1994). However, because
methods vary, the results may not generalize across samples. Furthermore, the
critical incidents selected by group members might or might not provide compre-
hensive descriptions of the therapeutic factors. Thus the technique has limitations
that hamper its use in research designed to compare therapeutic factors.
Yalom’s (1970, 1995) Q-sort is prototypical of the direct approach, which
measures factors by client response to specific questions about the group experi-
ence. Sixty statements assess 12 therapeutic factors. The factor constellation of the
instrument differs from that described in Yalom’s theory. Interpersonal learning
is divided into two factors (input and output), development of socializing tech-
niques is not included, and an additional factor labeled self-understanding (insight
P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

Therapeutic Factors Inventory 305

achieved during the group) is assessed. Group members rank each statement, as-
sessing which factors are most helpful to them in group. Each statement is rated
with a seven-point, Likert-type scale ranging from most to least helpful. Because
the factors are compared to one another, this assessment model presupposes that
each factor is present to some degree within each group.
While the Q-sort assesses the therapeutic factors more comprehensively
than any other available instrument, it also has several limitations. Unfortunately,
Yalom’s (1985, 1995) texts provide no psychometric information about the Q-sort.
Consequently, it is difficult to judge the reliability and validity of the instrument
in spite of the extensive base of clinical experience the author brought to its con-
struction. The discrepancies between the model assessed by the Q-sort and the
model described in Yalom’s texts make the Q-sort a less than perfect test for re-
searchers who want to do a pure test of his theoretical formulation. Researchers
have observed that the ranking process is difficult for clients to complete (Corder,
Whiteside, & Haizlip, 1981). Early factor analysis research examining the Q-sort
items’ relationship to the identified factors found some support for the therapeutic
factor categories, but the Q-sort items did not necessarily correspond to their as-
signed categories (Rorhbaugh & Bartels, 1975). Furthermore, Weiner (1974) has
argued that the Q-sort is biased toward interpersonal learning, because this scale
has twice the number of items as the other factors. Additionally, the use of a Q-sort
methodology presumes that the goal of a study utilizing an instrument is to com-
pare relative utility of the various therapeutic factors. Researchers or practitioners
with the goal of assessing the degree to which the therapeutic factors are present in
a group would, therefore, find the instrument less helpful than would those doing
factor comparison research. Overall, the Q-sort has limitations including lack of
psychometric support, difficulty of administration, and structural concerns.
Other available instruments have similar problems. A review of the literature
found many instruments assessing specific factors (e.g., Attraction Scale: Stockton,
Rohde, & Haughey, 1990; Harvard Community Health Plan Group Cohesiveness
Scale: Budman et al., 1987). Other studies reanalyzed the factor constellation
(Kellerman, 1987; Murillo, Shaffer, & Michael, 1981; Stone, Lewis, & Beck,
1994), examined single session impact from different perspectives in the literature
(e.g., Group Counseling Helpful Impacts Scale: Kivlighan, Multon, & Brossart,
1996), or examined group climate as a whole (e.g., Group Atmosphere Scale:
Silbergeld, Koenig, Manderscheid, Meeker, & Hornung, 1975; Group Climate
Questionnaire—Short Form: MacKenzie, 1983). However, there is no established
instrument with psychometric support that assesses the presence of all of Yalom’s
therapeutic factors in a given group (see Crouch, Bloch, & Wanlass, 1994, and
Fuhriman & Barlow, 1994, for reviews).
The lack of a unified scale measuring all therapeutic factors at one time
provides one explanation for the decades of research on therapeutic factors that
have provided little association between these factors and outcome. Bednar and
Kaul (1994) observed that far too many studies in the group therapy literature
P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

306 Lese and MacNair-Semands

still investigate whether or not group therapy works, rather than how it works.
They observe, however, that greater precision in definition and measurement of
group process elements is necessary for higher quality studies (Bednar & Kaul,
1994). As early as 1981, Bloch, Crouch, and Reibstein (1981) decried the “lack
of suitable measures” of the therapeutic factors (p. 525). Bednar and Kaul (1994)
made essentially the same statement thirteen years later:
[Yalom’s] observations are outstanding candidates for the conceptual and measurement
refinement needed to develop the conceptual tools and technology required for the next
developmental step in group research and clinical practice. We simply must establish more
consistency and depth in multidimensional conceptions and measurement of central curative
factors before their effects can be tested and compared. (p. 644)

THE THERAPEUTIC FACTORS INVENTORY

The objectives of the current project were to create a questionnaire assessing

Yalom’s therapeutic factors and subject it to preliminary reliability testing. The
intent was to develop an instrument that is easy to utilize and useful for practi-
tioners as well as researchers. Furthermore, the goals included the investigation
of preliminary psychometric support (e.g., item analysis, internal consistency, and
test-retest reliability) for the instrument.
The Therapeutic Factors Inventory (TFI) was designed to provide a com-
prehensive, empirically-based measure to determine the presence or absence of
therapeutic factors in a particular group. The TFI assesses group member percep-
tions of the degree to which the therapeutic factors described by Yalom (1995) are
present in a given group. Scales of the TFI include Instillation of Hope, Univer-
sality, Imparting Information, Altruism, Corrective Reenactment of the Primary
Family Group, Development of Socializing Techniques, Imitative Behavior, Inter-
personal Learning, Cohesiveness, Catharsis, and Existential Factors.

METHOD

Review of Available Instruments

All relevant, available, and psychometrically adequate scales assessing ther-

apeutic factors were collected and reviewed. Yalom (October 21, 1996; personal
communication) and MacKenzie (October 21, 1996; personal communication)
were consulted during this stage. Criteria for review included psychometric sup-
port, comprehensiveness of the assessment of Yalom’s therapeutic factors, and
availability of the instrument. Following the review, several conclusions were
reached. First, no psychometrically validated instrument that assesses all ther-
apeutic factors was found. Second, only Yalom’s Q-sort assessed these factors
comprehensively. Third, many of the scales reported in the literature (e.g., the
P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

Therapeutic Factors Inventory 307

Group Atmosphere Scale: Silbergeld, Manderscheid, Meeker, & Hornung, 1975;

Scale for the Evaluation of Group Counseling Experiences: Murillo, Shaffer, &
Michael, 1981) were not investigated after their initial study. Such scales are not
accessible to researchers, illustrating both a lack of replication and the waxing and
waning of interest in specific scales (Bloch, Crouch, & Reibstein, 1981) that have
taken place in the group measurement literature.

Item Development

Therapeutic factors were defined using Yalom’s (1995) descriptions, and

items were generated based on this formulation. The authors of the scale were two
doctoral-level psychologists with graduate level specialties in group psychother-
apy. Each of the authors had over nine years of experience leading, researching,
and studying groups. The authors independently generated items. They then cri-
tiqued and revised items based on lack of clarity, lack of correspondence with
the factor definition, and redundancy. Following this critique, problematic items
were revised or eliminated. A total of 174 items were kept for analysis. Each scale
comprised 15 to 19 items. Items were assessed along a seven-point Likert-type
scale rated from 1 (strongly disagree) to 7 (strongly agree). A few items in each
scale were reverse-coded to reduce response bias (for example, one item originally
phrased “I feel more positive about the future as a result of being in group” was
changed to “I feel less positive about the future as a result of being in group.”)
Items were then placed in random order. Table I provides sample items from each
scale.

Participants

Undergraduate and graduate-level college students participating in counsel-

ing and support groups at university counseling centers in three major universities
in the Northeast, the Southeast, and the Southwest, were recruited. In order to
maximize generalizability and number of participants, participation was requested
in groups of various formats including open-ended therapy groups, structured
groups, and support groups. In addition, subjects were recruited at different stages
of group development in order to increase variance and, consequently, generaliz-
ability. The largest proportion of participants came from process-oriented therapy
groups.
Seventy-seven undergraduate and graduate college students agreed to partic-
ipate. Approximately three-quarters of the sample (59) were female. Seventy of
the participants defined themselves as Caucasian; one was African American, four
were Hispanic/Latino, and one reported as Other. Age ranged from 19 to 58. The
mean age was 25.50, with a standard deviation of 8.34.
P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

308 Lese and MacNair-Semands

Table I. Sample Items for TFI Scales

Scale Sample items

Altruism It has impressed me that people in my group can be so kind and giving to
one another.
Helping others in group makes me feel better about myself.
Catharsis I can “let it all out” in my group.
I can’t express my feelings here.
Cohesiveness We cooperate and work together in group.
Even though we have differences, our group feels secure to me.
Corrective reenactment I have found myself playing the same role in the group that I played in
of primary family my family at times.
group In group I’ve really seen the social impact my family has had on my life.
Development of Group doesn’t teach me anything about how to have good relationships.
socializing Group helps me learn how to be more clear and direct with other people.
techniques
Existential factors In group I have learned that I am responsible for my own improvement.
This group helps empower me to make a difference in my own life.
Imitative behavior I learn how other people act in group and imitate them when it is
appropriate.
I keep my eye on what other people do in group so I can try different
things.
Imparting information In group I get “how-to’s” on improving my life situation.
We share ideas about resources in group.
Instillation of hope I don’t think group helps me feel any better about the future.
Seeing others change in my group gives me hope for myself.
Interpersonal learning I learn in group by interacting with the other group members.
Expressing myself in group has not freed me to express myself better in
my outside life.
Universality We have little in common in my group.
In group I have a sense that we all share similar feelings.

The majority of participants (n = 33) had participated in group for more

than eight sessions. Twenty-seven participants had been in group for five to eight
sessions, and 17 had been group members for two to four sessions. No participant
had been in group for less than two sessions.

Procedures

Group leaders explained the project and its voluntary nature to members and
distributed an informed consent form, the TFI and a brief information sheet re-
questing demographic information and length of participation in group. Group
members of two counseling centers completed forms at the end of a group session
and returned their packets to front office staff. Group facilitators at the third coun-
seling center requested that the forms be completed out of session. At that center,
participants were asked to return their packets, whether complete or incomplete,
to an anonymous drop box. A MANOVA was performed on the three settings
to ensure that these procedural differences did not skew results. No differences
P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

Therapeutic Factors Inventory 309

between the three counseling center settings were found on mean factor scores,
F(22, 96) = 1.65, p > .05, Hotellings = .76.

Item Analysis

Results were compiled and item analysis completed. Although no method of

item analysis is universally accepted for all purposes, Anastasi (1988) recommends
that researchers employ a combination of methods including statistical and judg-
mental procedures. The judgmental procedures in the current study included using
preliminary item reviews by licensed psychologists specializing in group ther-
apy. Following the judgmental procedures, statistical procedures were followed to
measure internal consistency.
Rejecting items that had low correlations with the total scale score provided a
means of purifying and homogenizing the instrument. Homogeneous instruments
are preferable because their scores permit fairly unambiguous interpretation. The
items with the highest correlations were retained. This process increases test va-
lidity, as each scale measures a single trait (Anastasi, 1988).

RESULTS

Internal Consistency

With the goal of balancing the need for a reliable and comprehensive instru-
ment with the need for a manageable length, items with the lowest correlations
between that item and the relevant factor scale score were removed. Because
Yalom’s Q-sort was critiqued for having more items in one scale than the others
(Weiner, 1974), it was judged important to have an equal number of items per
scale. Thus, different cut-off levels were used for each scale. Following analysis
of the item correlations, seventy-five items were deleted. The resulting instrument
had nine items per scale, with a total of 99 items. In some cases, the procedure
lowered the coefficient alpha slightly; in other cases, the alphas increased slightly.
Coefficient alphas (Cronbach, 1951) were computed for each scale after items
with the lowest scale correlations were deleted. The final coefficient alphas were
as follows: Altruism = .88; Catharsis = .83; Cohesiveness = .94; Existential
Factors = .83; Instillation of Hope = .93; Imitative Behavior = .88; Imparting
Information = .85; Interpersonal Learning = .88; Corrective Reenactment of the
Primary Family Group = .82; Development of Socializing Techniques = .92; and
Universality = .86. All reflected acceptable levels of reliability. Mean scores on
the scales ranged from 41.07 (Corrective Reenactment of Primary Family Group)
to 54.91 (Cohesiveness), with standard deviations ranging from 7.51 (Catharsis)
to 9.65 (Corrective Reenactment of Primary Family Group).
P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

310 Lese and MacNair-Semands

Test-Retest Reliability

In a second stage of analysis, test-retest reliabilities on the scales of the

99-item TFI were performed on four groups (N = 31) in two geographically dif-
ferent settings. Groups used in this stage were process oriented experiential and
supervision groups. These groups were recruited from a peer advising class, a
group supervision class, and a practicum in group therapy. These different formats
were tested to expand heterogeneity of the sample and generalizability of results.
In order to ensure that therapeutic factors did not naturally increase in strength
due to additional group sessions between administrations, the TFIs were given
one week apart. Pearson Product Moment correlations were run to evaluate test-
retest reliabilities, and were found to be as follows: Altruism, r = .87, p < .001;
Catharsis, r = .89, p < .001; Cohesiveness, r = .93, p < .001; Existential Fac-
tors, r = .64, p < .001; Instillation of Hope, r = .88, p < .001; Imitative Behav-
ior, r = .78, p < .001; Imparting Information, r = .84, p < .001; Interpersonal
Learning, r = .74, p < .001; Corrective Reenactment of Primary Family Group,
r = .28, p > .05; Development of Socializing Techniques, r = .72, p < .001;
and Universality, r = .85, p < .001.

Between-Scale Correlations

Pearson Product Moment correlations between scales were performed to

study the relationship between scales. Because the chance of familywise error
increases unacceptably when many correlations are run simultaneously, a Bonfer-
roni adjustment was made. In this process, a more conservative level of significance
for individual p levels was adopted to control for familywise error. In order to de-
rive the more conservative test of significance, the significance level was divided
by number of variables (.05 divided by 11 = .004).
Table II summarizes the intercorrelations between scales. Altruism, Cathar-
sis, Cohesiveness, Existential Factors, Instillation of Hope, Imitative Behavior,
Interpersonal Learning, Socializing Techniques, and Universality all correlated
significantly with one another ( p < .004). In addition, Imparting Information
correlated significantly with Altruism, Catharsis, and Universality ( p < .004).
Corrective Reenactment of Primary Family Group correlated significantly with
Existential Factors, Instillation of Hope, Imitative Behavior, Interpersonal Learn-
ing, Socializing Techniques, and Universality ( p < .004). Other correlations were
not significant.

DISCUSSION

The purposes of the current study were to empirically develop and test a mea-
sure assessing Yalom’s (1995) therapeutic factors. This study provided preliminary
 P1: GAR/FTT
Group [group]
P2: FPX
PH010-292464

Table II. Intercorrelations of TFI Scales

Therapeutic Factors Inventory

Altru Cathar Cohes Exis Hope Imitat Impart Interp Reena Soc Univ

Altruism 1.00 .61∗ .69∗ .68∗ .74∗ .70∗ .36∗ .79∗ .32 .71∗ .64∗
Catharsis — 1.00 .63∗ .49∗ .55∗ .43∗ .42∗ .56∗ .33 .50∗ .55∗
Cohesiveness — — 1.00 .42∗ .54∗ .44∗ .22 .50∗ .18 .37∗ .67∗
November 30, 2000

Existential factors — — — 1.00 .82∗ .57∗ .21 .77∗ .44∗ .71∗ .68∗
Hope — — — — 1.00 .67∗ .24 .80∗ .45∗ .72∗ .71∗
Imitative behavior — — — — — 1.00 .30 .74∗ .40∗ .70∗ .59∗
9:26

Imparting information — — — — — — 1.00 .27 −.01 .20 .14

Interpersonal learning — — — — — — — 1.00 .59∗ .87∗ .63∗
Corrective reenactment of — — — — — — — — 1.00 .51∗ .39∗
primary family group
Socializing techniques — — — — — — — — — 1.00 .59∗
Universality — — — — — — — — — — 1.00
∗p < .004.
Style file version Nov. 19th, 1999

311
P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

312 Lese and MacNair-Semands

and qualified support for the TFI. This new instrument demonstrated strong inter-
item consistency, demonstrating its homogeneity as a form of reliability. Test-retest
reliability was adequate with the exception of one scale. Corrective Reenactment of
the Primary Family Group obtained low test-retest reliability (r = .28, p > .05).
The low test-retest reliability obtained for this scale may indicate that it does not
reliably measure this factor over time, or that this factor is highly variable within
short time frames. However, the low test-retest reliability for this scale might also
have been obtained due to the fact that the groups used were all connected with
counseling classes. It is less likely that specific attention would be paid in this
context to family roles, than would occur in therapy groups. In future research, it
may be necessary to reassess test-retest reliability on the Corrective Reenactment
of the Primary Family Group scale.

Between-Scale Relationships

Many of the therapeutic factor scales correlated significantly with one an-
other. The scales that had the weakest correlations with other scales were Imparting
Information and Corrective Reenactment of the Primary Family Group. Interest-
ingly, these more independent variables are historically less valued by leaders and
members than other variables (Crouch, Bloch, & Wanlass, 1994; Yalom, 1995).
Several possible interpretations exist for the intercorrelations of the scales.
One possibility is that the therapeutic factors as defined by this project could be
meaningless concepts. Despite the fact that the definitions of the factors were
derived from Yalom’s (1995) work, validity testing would be necessary before
it could be reasonably argued that Yalom’s conceptualization of the therapeu-
tic factors is similarly meaningless. Better definitions might be accomplished
by a broader, overarching category system, possibly divided into family reen-
actment, imparting information, and several factors encompassing the remaining
nine.
Alternatively, all or some of the factors could be seen as so overly inclusive that
significant differences between them are negated. Bloch and Crouch (1985) have
observed that many therapeutic factor categories subsume a number of different
concepts, and that the research in general has been plagued by imprecise definitions.
They argued for greater precision in therapeutic factor category systems and have
developed their own category system for therapeutic factors. A similar argument
has been made by Fuhriman, Drescher, Hanson, Henrie, and Rybicki (1986), who
argued that interpersonal learning is an overly inclusive concept and that, rather
than measuring such learning separately, group researchers should assume that all
therapeutic activity in a group takes place in an interpersonal context.
If theoretical flaws did not contribute to the intercorrelations, measurement
errors could explain the strong between-scale relationships. The correlated scales
could be measuring the same variable due to flaws in scale construction, rather
than flaws in conceptualization of the factors. If the items did not accurately
P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

Therapeutic Factors Inventory 313

measure differences between the factors, the scales would be expected to correlate
significantly.
Finally, and perhaps most likely, the therapeutic factors could be distinct en-
tities, but could correlate significantly because they are definitionally interrelated.
Given that Yalom believes that cohesiveness is a prerequisite for all other factors,
for example, it is no surprise that in this study this factor correlated significantly
with most other factors. Similarly, Fuhriman et al. (1986) argue that several of the
factors “nest” within one another. Yalom (1995) observes that the therapeutic fac-
tors are interrelated by definition: “Universality, like the other therapeutic factors,
does not have sharp borders; it merges with other therapeutic factors” (p. 7). Given
that the overlap between factors is an aspect of Yalom’s theory, if these scales do
measure Yalom’s concepts, the intercorrelations may reflect an aspect of Yalom’s
definitions that are inseparable from his work. Some scholars (e.g., Crouch, Bloch,
& Wanlass, 1994) who have advocated for more discrete factor categories view
definitional overlap as a problem in the theory and, consequently, in theory-driven
instruments such as the TFI.

Utility of the Theory

The logical question that follows is whether or not the factors, and the scales
this instrument measures, are meaningful and useful despite the intercorrelations.
Given the systematic observation Yalom’s work has undergone over decades of
clinical and research experience (Bednar & Kaul, 1994), as well as the practical
utility of Yalom’s system, most group practitioners believe that his system has
great value. It has previously been observed that group research often defines one
aspect of process in isolation, leading to an effort that teaches us little about the
interrelatedness of variables (Fuhriman & Burlingame, 1994). Perhaps recognizing
that variables are related, and measuring them in a way that reflects this, is not
inherently problematic. Rather, studying the complex and interrelated nature of
factors impacting group development across different developmental stages may
provide us with the most meaningful results (Kivlighan & Lilly, 1997; McGrath,
1997). If so, it would follow that measuring the factors together rather than looking
at specific variables in isolation would be most meaningful. For example, is it
possible that altruism in a therapy group cannot be meaningfully viewed in isolation
from most other therapeutic factors? The intercorrelation of therapeutic factors
may not necessarily imply that the concepts are meaningless, but that they are
meaningless in isolation, reflecting the complexity inherent in therapy groups.

Limitations of the Current Study and Future Directions for Research

The development of the TFI has a strong beginning, but several limitations
should also be considered. Having a third item generator or external judge of the
P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

314 Lese and MacNair-Semands

original items would have provided a divergent perspective for item development.
Statistical limitations of the current research begin with sample size, which was
low given the number of items. This limited power and generalizability, and also
limited the types of statistical analysis that could be performed. In addition, despite
recruiting subjects from three university campuses, the sample was skewed, with
only six participants identifying as non-Caucasian. Future research would do well
to expand and diversify the participant pool.
This investigation provided only information about reliability. Although some
aspects of construct validity, focusing on the role of psychological theory in test
construction, were considered in the development of the TFI, the current project
includes no external criterion. The reader is directed to another research report
(MacNair-Semands & Lese, 2000) which provided preliminary support for the
TFI’s construct validity. Further assessment of construct and criteria-related valid-
ity is the next logical step in evaluating the utility of this instrument. Such validity
testing is essential before any conclusions can be drawn regarding the applica-
bility of the TFI to Yalom’s theory. Furthermore, confirmatory factor analysis is
necessary.
The interrelations between the scales provide important implications for re-
search. For example, researchers applying the TFI will need to exercise caution in
making conclusive statements about therapeutic factors in isolation. Difficulties
in conducting valid outcome studies may also follow. Research indicating that
therapeutic factors are in fact therapeutic has been sorely needed in the literature
(Bednar & Kaul, 1994; Dies, 1993). If the therapeutic factors cannot be reliably
viewed as separate entities using this instrument, they also cannot be indepen-
dently connected to outcome. Thus, if we are to continue to use this instrument to
measure Yalom’s concepts, we must keep in mind that the limitations of Yalom’s
theory will also be reflected in the limitations of the instrument.

Applications and Implications for Treatment

If the TFI holds up to further validity and reliability testing, the instrument
could be extremely useful for group therapists. Crouch, Bloch, and Wanlass (1994)
have suggested that therapeutic factor classifications should assist therapists in
tailoring interventions to groups. Formal assessment of the therapeutic factors
could provide a reliable method of doing so. For example, if group members
rate a group low on cohesiveness, leaders are likely to prioritize fostering this
factor in group, given that cohesiveness is thought to be a precondition for ther-
apeutic change in groups (Yalom, 1995). Results of this study indicate that al-
truism may be an additional precondition for change. Assessment may also be
useful for individual group members. If one member’s TFI scores are signifi-
cantly different from the rest of the group, this may reflect alienation from the
group. The member may need increased time to process his or her differences
P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

Therapeutic Factors Inventory 315

in group, or may require additional training about how to make best use of
group.
If construct validity is established, empirical evidence for Yalom’s factors
could be obtained by measuring client outcome in group as it relates to the client’s
perspective of the strength of certain therapeutic factors. If clients improve more
in groups that members rate as having strong therapeutic factors than in groups
that rate lower on the factors, this pattern may suggest an association of factors
and outcome above and beyond client variables.
From a pragmatic viewpoint, the most valuable research involving the TFI
might investigate the applied utility of the instrument. What implications would
testing at different stages of group development have? How would testing at inter-
vals improve leaders’ abilities to assess the status of the group and make modifica-
tions in strategy? Does the effectiveness of structured groups and therapy groups
depend upon different therapeutic factors? Alternatively, researchers may wish to
consider the relative importance of different factors at different stages of group
development.
This investigation provided a sound base for future work with the TFI. Solid
methods of item construction and selection were utilized. Definitions of the factors
and their anticipated interrelationships were clarified, and evidence for internal
consistency was obtained. Although additional research is needed, the TFI shows
promise for future application.

ACKNOWLEDGMENTS

We thank Alex Goncalves and the staffs of the Center for Counseling and
Psychological Services at Penn State University, the Counseling Center of Texas
Tech University, and the Counseling Center of the University of North Carolina at
Charlotte for their assistance in data collection.
Portions of this article appeared in a poster session presented at the 105th
Annual Convention of the American Psychological Association, Chicago, 1997.

REFERENCES

Anastasi, A. (1988). Psychological testing (4th ed.). New York: MacMillan Publishing Company.
Bednar, R. L., & Kaul, T. J. (1994). Experiential group research: Can the cannon fire? In S. L. Garfield
& A. E. Bergin (Eds.), Handbook of psychotherapy and behavior change (pp. 631–663). New
York: J. Wiley.
Bloch, S., & Crouch, E. (1985). Therapeutic factors in group psychotherapy. Oxford: Oxford University
Press.
Bloch, S., Crouch, E., & Reibstein, J. (1981). Therapeutic factors in group psychotherapy: A review.
Archives of General Psychiatry, 38, 519–526.
Bloch, S., Reibstein, J., Crouch, E., Holroyd, P., & Themen, J. (1979). A method for the study of
therapeutic factors in group psychotherapy. British Journal of Psychiatry, 30, 63–76.
P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

316 Lese and MacNair-Semands

Brown, B. P., Hedinger, T., & Mieling, G. (1995). A homogenous group approach to social skills
training for individuals with learning disabilities. The Journal for Specialists in Group Work, 20,
98–107.
Budman, S. H., Demby, A., Feldstein, M., Redondo, J., Scherz, B., Bennett, M. J., Koppenaal, G.,
Daley, B. S., Hunter, M., & Ellis, J. (1987). Preliminary findings on a new instrument to measure
cohesion in group psychotherapy. International Journal of Group Psychotherapy, 37, 75–93.
Card, K. J., & Schmider, L. (1995). Group work with members who have hearing impairments. The
Journal for Specialists in Group Work, 20, 83–90.
Clark, A. J. (1993). Interpretation in group counseling: Theoretical and operational issues. The Journal
for Specialists in Group Work, 18, 174–181.
Corder, B. F., Whiteside, L., & Haizlip, T. M. (1981). A study of curative factors in group psychotherapy
with adolescents. International Journal of Group Psychotherapy, 31, 345–354.
Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 297–334.
Crouch, E. C., Bloch, S., & Wanlass, J. (1994). Therapeutic factors: Interpersonal and intrapersonal
mechanisms. In A. Fuhriman & G. Burlingame (Eds.), Handbook of Group Psychotherapy (pp.
269–315). New York: Wiley.
DeLucia-Waack, J., (1996). Multiculturalism is inherent in all group work. The Journal for Specialists
in Group Work, 21, 218–223.
Dies, R. R. (1993). Research on group psychotherapy: Overview and clinical applications. In A.
Alonso & H. I. Swiller (Eds.), Group therapy in clinical practice (pp. 473–518). Washington, D.
C.: American Psychiatric Press.
Flowers, J. V. (1987). Client outcome as a function of agreement or disagreement with the modal
group perception of curative factors in short-term, structured group psychotherapy. International
Journal of Group Psychotherapy, 37, 113–117.
Fuhriman, A., & Barlow, S. H. (1994). Interaction Analysis: Instrumentation and issues. In A. Fuhriman
& G. Burlingame, (Eds.), Handbook of group psychotherapy (pp. 191–222). New York: Wiley.
Fuhriman, A., & Burlingame, G. M. (1994). Measuring small group process: A methodological appli-
cation of chaos theory. Small Group Research, 25, 502–519.
Fuhriman, A., Drescher, S., Hanson, E., Henrie, R., & Rybicki, W. (1986). Refining the measurement
of curativeness: An empirical approach. Small Group Behavior, 17, 186–201.
Goldberg, F. S., McNiel, D. E., & Binder, R. L. (1988). Therapeutic factors in two forms of inpatient
group psychotherapy: Music therapy and verbal therapy. Group, 12, 145–156.
Johnson, I. H., Torres, J. S., Coleman, V. D., & Smith, M. C. (1995). Issues and strategies in leading
culturally diverse counseling groups. The Journal for Specialists in Group Work, 20, 143–150.
Kellerman, P. F. (1987). Psychodrama participants’ perception of therapeutic factors. Small Group
Behavior, 18, 408–419.
Kobak, K. A., Rock, A. L., & Greist, J. H. (1995). Group behavior therapy for Obsessive-Compulsive
Disorder. The Journal for Specialists in Group Work, 20, 26–32.
Kivlighan, D. M., & Lilly, R. L. (1997). Developmental changes in group climate as they relate to
therapeutic gain. Group Dynamics: Theory, Research, and Practice, 1, 208–221.
Kivlighan, D. M., Multon, K. D., & Brossart, D. F. (1996). Helpful impacts in group counseling:
Development of a multidimensional rating system. Journal of Counseling Psychology, 43, 347–
355.
Lovett, L., & Lovett, J. (1991). Group therapeutic factors on an alcohol in-patient unit. British Journal
of Psychiatry, 159, 365–370.
MacKenzie, K. R. (1983). The clinical application of a group climate measure. In R. R. Dies & K. R.
MacKenzie, (Eds.), Advances in group psychotherapy: Integrating research and practice. New
York: International Universities Press.
MacKenzie, K. R. (1987). Therapeutic factors in group psychotherapy: A contemporary view. Group,
11, 26–34.
MacNair-Semands, R. R., & Lese, K. P. (2000). Interpersonal problems and the perception of therapeutic
factors in group therapy. Small Group Research, 31, 158–174.
Marcovitz, R. J., & Smith, J. E. (1983). Patients’ perceptions of curative factors in short-term group
psychotherapy. International Journal of Group Psychotherapy, 33, 21–39.
McGrath, J. E. (1997). Small group research, that once and future field: An interpretation of the past
with an eye to the future. Group Dynamics: Theory, Research, Practice, 1, 7–27.
 P1: GAR/FTT P2: FPX
Group [group] PH010-292464 November 30, 2000 9:26 Style file version Nov. 19th, 1999

Therapeutic Factors Inventory 317

McLeod, J., & Ryan, A. (1993). Therapeutic factors experienced by members of an outpatient therapy
group for older women. British Journal of Guidance & Counselling, 21, 64–72.
Murillo, N., Shaffer, P., & Michael, W. B. (1981). The development and validation of a preliminary
measure for student evaluation of group counseling experiences. Educational and Psychological
Measurement, 41, 463–472.
Price, G. E., Dinas, P., Dunn, C., & Winterowd, C. (1995). Group work with clients experiencing
grieving: Moving from theory to practice. The Journal for Specialists in Group Work, 20, 159–
167.
Randall, D. A. (1995). Curative factor rankings for female incest survivor groups: A summary of three
studies. The Journal for Specialists in Group Work, 20, 232–239.
Rohrbaugh, M., & Bartels, B. D. (1975). Participants’ perceptions of “curative factors” in therapy and
growth groups. Small Group Behavior, 6, 430–456.
Scheidlinger, S. (1997). Group dynamics and group psychotherapy revisited: Four decades later. Inter-
national Journal of Group Psychotherapy, 47, 141–159.
Silbergeld, S., Manderscheid, R. W., Meeker, B. F., & Hornung, C. A. (1975). Assessment of
environment-therapy systems: The Group Atmosphere Scale. Journal of Consulting and Clin-
ical Psychology, 43, 460–469.
Stockton, R., Rohde, R., & Haughey, J. (1990). Effects of structure on group cohesion, engagement,
avoidance, and conflict. Paper presented at the 98th Annual Convention of the American Psycho-
logical Association, Boston.
Stone, M. H., Lewis, C. M., & Beck, A. P. (1994). The structure of Yalom’s Curative Factors Scale.
International Journal of Group Psychotherapy, 44, 239–245.
Weiner, M. F. (1974). Genetic vs. interpersonal insight. International Journal of Group Psychotherapy,
24, 230–237.
Yalom, I.D. (1970). The theory and practice of group psychotherapy. New York: Basic Books.
Yalom, I. D. (1975). The theory and practice of group psychotherapy (2nd ed.). New York: Basic
Books.
Yalom, I. D. (1985). The theory and practice of group psychotherapy (3rd ed.). New York: Basic Books.
Yalom, I. D. (1995). The theory and practice of group psychotherapy (4th ed.). New York: Basic Books.

View publication stats