The Affective Impact of a Clinical Social Worker’s Interviewing Style: A Series of Single-Case Experiments William R.

Nugent

University of Tennessee

A series of studies was conducted to investigate the immediate affective impact that a clinical social worker’s interviewing style has on clients. In Experiment 1, four simulated interviews were conducted using B1-B2-B1-B2 single-case designs. In Experiment 2, three further B1-B2B1-B2 single-case-design studies were conducted in order to demonstrate the replicability and generalizability of results from Experiment 1. In all of the interviews the outcome measures involved a combination of qualitative and quantitative assessments. The results illustrate that certain practitioner interviewing styles may cause immediate and profound affective responses among clients.

The

relationship between practitioner and client has long been viewed as fundamentally important in social work (Richmond, 1917). Numerous schools of counseling and psychotherapy also view the therapist-client relationship as being of critical importance to the helping process (Corey, 1977; Egan, 1975; Truax & Carkhuff, 1967). Recent social work texts have devoted much space to discussing both the importance and the development of the client-practitioner relationship (e.g., Compton & Galaway, 1984; Hepworth & Larsen, 1986). Research done by Alexander, Barton, Schaivo, and Parsons (1976) led them to conclude that relationship characteristics are not enough in and of themselves to promote change. More recent research seems to suggest that technical interventions may combine with relationship elements in an additive or interactive manner to affect outcome (Persons, 1989; Persons & Bums, 1985; Sweet, 1985). Sweet (1985), after a review of the research on the role of the practitioner-client relationship in treatment outcome, concluded that relationship variables combine with intervention techniques to mediate outcome. He concluded that relatively little is known about what Author’s Note:

Correspondence

may be addressed to William R.

Work, Henson Hall, University of Tennessee, Knoxville, TN 37996. Research 0 1992

6

on

Social Work Practice, Vol. 2 No. 1, January 1992 6-27

Sage Publications, Inc.

Nugent, College of Social

7

relationship variables are important and how specific practitioner behaviors affect either the client or eventual service outcome. He also called for research on, among other things, the impact of process-type interventions on both client and outcome. Northern (1969, pp. 53-58) wrote, &dquo;Relationship has been described as consisting ’primarily of emotional responses which ebb and flow from person to person as human behavior evokes different affective reactions. &dquo;’ Concerning the elements of the experience of being helped, Sweet (1985) notes that research suggests that the experiences that clients are more apt to remember are their own affective responses to the practitioner and to what the practitioner does. Given the saliency and impact that affective responses appear to have for human beings (see, e.g., Lazarus, 1982; Tataryn, Nadel, & Jacobs, 1989; Zajonc, 1980), practitioners may need to be aware of the affective impact that their behavior is likely to have on clients. Many social work texts describe specific verbal behaviors, or what Rosen and Proctor (1978) term &dquo;interventive responses,&dquo; that are recommended for use by practitioners (e.g., Hepworth & Larsen, 1986). Often these verbal behaviors are categorized into presumably undesirable and desirable categories. The undesirable category includes such verbal behaviors as premature advice giving and ordering, which are described as response types that may cause problems in the client-practitioner relationship and obstruct the helping process (Hepworth & Larsen, 1986). The desirable category includes such skills as active listening (Gordon, 1970; Rogers, 1966), which are described as beneficial to the practitioner-client relationship. However, although considerable research has been done on the impact of the practitioner-client relationship on service outcome, research has yet to be done on the immediate affective response that clients have to various practitioner verbal behaviors. Given the importance that these affective responses may have in both the process and outcome of service delivery (Sweet, 1985), it would seem that a productive line of research would be the investigation of the relationship between various social work practitioner interviewing styles and the affective responses of clients. This article describes a series of studies on the immediate affective impact that various verbal interventive responses have on clients. In Experiment 1, subjects observed interviews between a simulated client and a social worker using a B1-B2-B1-B2 single-case research design. In Experiment 2, three single-case studies of actual interviews were conducted to test the replicability of results from the initial simulation series. Each study employed a combination of qualitative and quantitative outcome measures, illustrating how these two methodologies may be fruitfully integrated. The following

8

research questions were investigated in these studies: (a) What is the immediate affective impact on clients of differing interviewing styles used by a clinical social worker? and (b) Can the systematic alteration of these interviewing styles produce demonstrably reliable changes in client affective reactions?

EXPERIMENT 1

Methods and Procedures

Participants The first simulated interview was conducted at an outpatient family therapy clinic. The participants were 10 family therapists who had agreed to participate during a workshop conducted by the author. Participants’ ages ranged from 24 to 57 years, with a mean (M) of 33.3 and a standard deviation (SD) of 9.4. Participants’ educational levels ranged from 16 to 25 years (M = 18.4, SD = 2.4). There were 6 men and 4 women. Eight of these participants were White Americans, and 2 were Latin Americans. Four of the participants were married, 4 were single and never married, and 2 were divorced. The simulated interview represented a marital therapy session. The second simulated interview was conducted in a residential program for adolescents. The participants were 11 members of the program staff, including cooks, child-care workers, residential counselors, and administrative staff. Participants’ ages ranged from 25 to 54 years (M = 37.2, SD = 8.8). Participants’ education levels ranged from 12 to 16 years (M = 14.1, SD = 1.6). Seven of the participants were women and 3 were men (1 person failed to indicate gender). Six participants were White Americans, and 3 were Black Americans (2 persons failed to indicate their race). Six of the participants were married, 2 were divorced, 2 were living with partners but not married, and 1 was single and never married. The simulated interview represented a

crisis intervention situation. The third simulated interview was conducted with volunteers who were attending a conference. The 12 participants included teachers, athletic coaches, bus drivers, and school administrators. Participants’ ages ranged from 22 to 56 years (M 32.1, SD 9). The range in education was from 12 to 18 years (M = 15.3, SD 1.8). There were 9 men and 3 women in this group, of whom 7 were White Americans and 5 were Black Americans. Four of the participants were married, 4 were single and never married, 2 were =

=

=

9

living with a partner but not married, 1 was separated, and The simulated interview represented a mediation session neighbors involved in a dispute.

1 was divorced. conducted with

The fourth simulated interview was conducted with 24 members of a graduate social work research class who agreed to participate. Participants’ ages ranged from 21 to 51 years (M = 35.2, SD = 9.6). The range of educational level was from 16 to 20 years (M = 17.6, SD = 1.2). There were 23 White Americans and 1 Black American. There were 18 women and 6 men in this group. Seven of the participants were married, 8 divorced, 6 single and never married, and 3 were living with a partner but not married. The simulation represented an individual counseling session with a woman abused by her live-in boyfriend.

Interviewing Styles (Independent Variable)

Type of interviewing style was the treatment variable in each of the simulation studies, Type 1 and Type 2. Type 1 interviewing style corresponded to what Strayhorn (1977) labeled &dquo;obstructive&dquo; and Hepworth and Larsen (1986) labeled &dquo;barriers to verbal communication.&dquo; Specifically, Type 1 verbal behaviors involved the social worker’s deliberately engaging in the following: blaming, reassuring, lecturing, leading questions, arguing, should statements, labeling, premature advice giving, warning, negative or psychological interpretations, and using put-down questions (Hepworth & Larsen, 1986; Strayhorn, 1977). Type 2 verbal behaviors corresponded to what Strayhorn (1977) labeled &dquo;facilitative&dquo; communication styles and involved active (or empathic) listening, open-ended questions, close-ended questions, summarizing statements, negotiating statements, positive reframing, and praise statements (Gordon, 1970; Hepworth & Larsen, 1986). Outcome Measure

The focus of these studies was upon the immediate affective response of observers to the social worker’s two interviewing styles. In order to capture self-reports of qualitatively different emotional reactions, affective responses were assessed using a variation of the thought-listing technique (Cacioppo & Petty, 1981; Goldberg & Shaw, 1989). This procedure allows the respondent to list thoughts and feelings that he or she has in response to specific environmental and social stimuli (Cacioppo & Petty, 1981). Cullen (1968) has compared the reliability of thought-listing techniques with the reliabilities of Likert-type and Thurstone scales. In Cullen’s (1968) study the average

10

split-half reliability of thought listings was .78 (compared to .83 for Likerttype scales and .55 for Thurstone scales), whereas the average test-retest reliability of thought listings was .64 (compared to .83 for Likert-type scales and .53 for Thurstone scales). Thus thought-listing procedures may have internal consistency reliabilities comparable with those of Likert-type selfreport scales. However, the reliability of the thought-listing procedure used in this study was not directly assessed. A number of studies have investigated the sensitivity of thought listings to detect changes due to interventions (e.g., Cacioppo & Petty, 1979; Cacioppo, Sandman, & Walker, 1978; Petty & Cacioppo, 1979; Petty, Wells, & Brock, 1976). The results of these investigations suggest that thought listings can detect response differences caused by an environmental manipulation (Cacioppo & Petty, 1981). The thought-listing method used in the studies reported here is described in detail below. Research Design

In each simulated interview, participants were trained to use a type of thought listing, what might be termed feeling listing. Participants were taught to observe a role-played interview in a manner such that they put themselves into the &dquo;shoes&dquo; of the person in the client role. They observed the role-played interview as if they were in the client’s place, experiencing the client’s problem, saying the things that the client was saying to the social worker, and hearing the things said by the social worker. As they participated in this manner, the observers wrote down in

one

column of

a

two-column data sheet

specific things said by the person in the social worker role and then noted in a second column the affective response that they had to these verbal behaviors of the social worker. The left-hand column of the data sheet was headed, &dquo;When you said...,&dquo; and the right-hand column was headed, &dquo;My feeling reaction was....&dquo; Thus, for example, suppose the social worker said, &dquo;I am afraid that you will be hurt by him if you don’t leave.&dquo; Further suppose that one observer, as she participates and hears this statement by the social worker, responds by feeling afraid. Thus the observer would write in the left-hand column, &dquo;I am afraid that you will be hurt by him if you don’t leave,&dquo; and in the right-hand column, &dquo;afraid.&dquo; Participants were trained in this procedure until they could observe a role-played interview and record social worker statements and the affective responses they had to them. Then a role-played interview was conducted in which a person in a client role discussed a problem with a person in the social worker role. Each observer had four of the two-column data sheets. Prior to

11

the role play the client described the problem and how he or she was feeling and thinking. Time was spent helping observers develop a sense of how they might think and feel if they were in the situation described by the client. The client and social worker then role-played an interview in which they talked about the client’s problem. For the first 5 min (labeled a Bl, or obstructive, phase) the social worker would use only Type 1 (obstructive) verbal responses. Observers recorded their affective responses to social worker verbal behavior on the first data sheet. During the second 5 min (labeled a B2, or facilitative, phase) the social worker used only Type 2 (facilitative) verbal responses, and observers recorded their affective responses on the second data sheet. During the third 5 min (a second Bl, or obstructive, phase) the social worker used only Type 1 verbal responses, and observers recorded their affective responses on a third data sheet. Finally, during the fourth 5 min (a second B2, or facilitative, phase) the social worker used only Type 2 verbal behaviors, and the observers recorded their affective responses on a fourth data sheet. Thus the data-collection process conformed to the parameters of a Bl-B2B1-B2 single-case design. In all phases the outcome measure was the recorded affective responses of the observers to the specific verbal behaviors of the person in the social worker role. Results

The results of the four simulated interviews included in Experiment 1 are shown in Table 1 and in Figures 1 through 3. In Table 1 the feeling responses listed by observers in response to the social worker’s verbal behaviors during the first interview are listed by phase. As can be seen in Table 1, there is a clear difference between obstructive and facilitative phases as to the type of affective response reported by the observers. The obstructive phases, during which only Type 1 verbal communications were used by the social worker, are primarily characterized by negative or unpleasant feelings. In contrast, facilitative phases, during which only the Type 2 verbal behaviors were used, are characterized by what might be classified as positive feelings. Although the qualitative differences between phases are clear through visual inspection, the feeling responses in each phase in Table 1 have been grouped by the class (negative or positive) to which they appear to belong. The feeling responses listed in each phase of the other three simulation studies were similarly grouped and the results shown in Figures 1 through 3. These groupings were used to compute a X2 statistic for the results of each simulated interview. For the first interview, X2 = 53.47 (3 df, p < .001 [two

12

TABLE 1:

Feelings and Thoughts

Elicited

During the

First Interview of

- Experiment 1

tailed]). For the second interview, X2 81.01 (3 df, p < .001 [two tailed]), and for the third interview X2 58.79 (3 df, p < .001 [two tailed]). For the fourth interview, x2 = 160.1 (3 df, p < .001[two tailed]). The critical a level for each test was set at .0025 in order to provide a familywise Type 1 error rate of .01 for the set of four tables. As can be seen, all four x2 values are statistically significant at this .0025 level. If the type of interviewing style is used to predict the category of feeling response (positive or negative) elicited, then for the data in the simulation studies, type of interviewing style (social worker’s verbal behavior) explains about 89% of the variability of the feelings that were reported. Qualitative and quantitative changes in type of feeling response clearly follow the changes in the social worker’s interviewing style, suggesting that the type of verbal behavior used by the social worker may have caused the =

=

0 ..c::

. M

Q)

L_

O

C +r

3 a

1j Q.) c

~5

15

0

― (D -.0 c

.~e ~ u ae

1’. ;(c.Q)(f )

w&dquo;&dquo;a. OC O (f)

~ ~

t j §f &dquo; v -~~-a) v

.5~ ’t:j1j 1’1.1 Q.)

=NO .- ....

O

c en Q c: ’.J~

1’1.1 (f) ’s~ ’&dquo; Q.)

-5r E~ ~ 8 ..
-

.a &dquo;0

Q) c:

i15 0

. 0) ― D

- = Ë CO u c: ES

.- en

C~’-en l.c:QJ1o0 ~0enaQ) 0 0

s3, v

QJOJ ’? s

L

QJ~0 10 Q) Q)

ë’E~ CC ~a e ’&dquo;-

X !~ 0

’1:1 OJ 100 C

E-.ê 4. O

~ ~

-5 ~

H

M 8 .. < N

100..... ~O

¡ZZ 14

o .c

:5

enQ) -5 B 0) c

Y $ >-

.0

1T1 &dquo;0 Q) m

:E o

~o

_.0 en c::

~C

ES ï:

£m

en

’S.S

xaNc ’-0-

Q

en

3 °~

.~ 0> ~ae is -a S ’o Q) i~ S a) ri5õ

.so> C~

iF

O N ~

’3~ ’IJ~

3 ~, E a

=~ ~ Eu0 .. < ~ ..

W =15)0

E z 15

16

TABLE 2:

Feelings and Thoughts Listed as Responses to Active-Listening Statements During All Four Interviews of Experiment 1

Respondent noted that feeling was due to reliving an event in memory that the active-listening response triggered. b. Respondent noted that feeling response arose from active-listening statement whose wording did not match the intensity of feelings felt by respondent. a.

different types of feelings reported by observers. However, these results must be considered within the context of the methodological limitations in these simulated interview studies. These limitations will be discussed below. The two Type 2 verbal behaviors most frequently used in all of the facilitative phases were active listening and open-ended questions. The feelings listed by observers as their responses to active-listening statements are shown in Table 2. About 37% of the feeling words listed as responses to activelistening statements were the word understood If words implying feeling understood are considered, then about 50% of all responses to active-listening statements were feelings of being understood. About 91% of the listed feelings were positive, whereas only about 9% of the listed feeling responses to active-listening statements were what can be classified as negative. Table 3 shows the feelings listed as responses to open-ended questions across all four interviews. No clear theme of feeling responses was apparent in this listing. However, about 89% of the affective responses were positive, whereas about 11 % of the affective responses listed for open-ended questions could be categorized as negative.

17

TABLE 3:

Feelings and Thoughts Listed Questions

as

Responses to Open-Ended

The most frequently used Type 1 verbal behaviors were premature giving of advice, the use of warnings, should statements, negative interpretations and labeling, and reassurance. The large majority of feeling responses to these interviewing tactics can be classified as negative. Anumber of the listed feeling responses to a warning statement were positive. These responses were listed for a warning given about the possibility of physical harm in the simulation involving a client abused by her live-in boyfriend. A number of the feeling responses listed for reassurance statements concerned, interestingly, abandonment. No other common feeling themes could be discerned in the feeling response listings for the Type 1 interviewing style. The recorded responses of several participants contained explanations noting that the observer’s feeling responses to the verbal behaviors of the social worker were caused, at least in part, by the switch from the first facilitative phase to the second obstructive phase and from the second obstructive phase to the second facilitative phase. Some of the observers who gave these explanations noted that the switches from Type 2 back to Type 1 verbal communication patterns evoked such feelings as abandonment, hurt, confusion, and anger. Several respondents noted that their negative feeling responses to active-listening statements were due to wordings that failed to match the intensity of the feelings they had. Others noted that their negative feeling responses were brought on by an active-listening statement that elicited the memory of an unpleasant event.

18

Discussion of Experiment 1

There are a number of methodological limitations that should be considered when interpreting these results. First, there is no demonstration of the reliability of the observers’ reports of feeling responses in the positive and negative categories. Second, there was no demonstration that the classification by the author of the observers’ feeling statements into these categories was a reliable one. It could be claimed that some feelings listed as positive are, in fact, negative (and vice versa). Changes in categorizations would have an impact on the x2 statistics reported above. However, a careful examination of the feeling responses shown in Table 1 will clearly show the qualitative

differences between feelings elicited during obstructive and facilitative phases in the first simulation. The feeling responses listed in the other three simulated interviews show the same clear qualitative distinctions among

phases. Another limitation concerns the lack of evidence supporting the integrity of the treatment variable, the two interviewing styles. Evidence supporting the reliability of changes in verbal behaviors from Type 1 to Type 2 between phases, as well as evidence showing the consistent use of a specific interviewing style within phases, would further clarify the results discussed above. Future studies using the methods described here should provide such evidence. Although there is a clear difference between the feeling responses listed during obstructive and facilitative phases, the abrupt shift from the consistent use of Type 1 behaviors to use of Type 2 behaviors may have, in some manner, played a role in the types of feeling reactions elicited. That is, the characteristics of the B1-B2-B1-B2 type design may, in and of themselves, have had an effect on the affective responses of clients. Future replications of these studies might employ alternating-treatments single-case designs, with the use of Type 1 and Type 2 interviewing styles randomly determined. In this manner any effect due to the abrupt shift may be eliminated. These simulated interviews are, in a sense, two levels removed from the reality of a practitioner-client relationship. The responses listed by &dquo;clients&dquo; were to the reactions they felt to observing an interaction between two players in a role-played interview. Thus, rather than representing true client feeling reactions, they represent reactions to role-played interactions. The singlecase-design studies described below represented an effort to overcome this limitation.

19

EXPERIMENT 2

Methods and Procedures

Clients

The first single-case study involved a 25-year-old White female who worked as a secretary/receptionist. This woman was attending a university and working on her undergraduate degree. During the experimental interview she discussed an event that had occurred a couple of hours earlier that had left her angry. The woman completed the State Anger Scale (SAS; see below) at the end of each phase of the single-case study. The second single-case study involved a 30-year-old crisis intervention counselor with a master’s degree. During the interview she talked about whether or not she should go back to school and get her doctorate. The third single-case interview involved a 37-year-old woman who worked as an administrator in a human service organization. This woman, who held a master’s degree in criminology, talked about an ongoing major problem with her best friend. Interventions

The same intervention was used in these three single-case-study interviews as was used in the four simulated interviews constituting Experiment 1. Outcome Measures

The same feeling-listing procedure was used as an outcome measure in these three single-case studies as was used in Experiment 1. The SAS was also used in the first single-case study interview of Experiment 2 as an outcome measure (Corcoran & Fischer, 1987; Spielberger, Jacobs, Russel, & Crane 1983). Scores on the SAS can range from 15 to 60, with higher scores indicative of higher levels of anger at the moment. Corcoran and Fischer (1987) report a coefficient a estimate of reliability of the SAS to be .93. Spielberger et al. (1983) review the positive evidence regarding the validity of this scale.

20

TABLE 4: -

Feelings and Thoughts Elicited During First Interview in Experiment 2

Research Design

Three single-case-design studies were conducted in order to examine the replicability and generalizability of results obtained during simulated interviews to actual interviews with people discussing real problems. In each of these studies a &dquo;client&dquo; was trained in use of the feeling-listing technique. The client then discussed a personal problem or issue of concern with the

clinical social worker. After each discrete interaction with the social worker, the client noted her affective responses to what the social worker had said. In these studies the client did not write down the social worker’s comments during the interviews, as was done with the simulated interviews in Experiment 1. Either tape recordings or notes were made by the social worker, so that client affective responses could be matched up with the practitioner’s verbal behavior. As in the simulations, there were four phases in each study, with the Type 1 interviewing style used during obstructive phases and the Type 2 interviewing style used during facilitative phases. Thus each singlecase design followed a B1-B2-B1-B2 format. After the interview was completed, the client was interviewed about her experience during each phase of the single-case-design study. She was also given a further opportunity to discuss her problem or concern in a nonresearch interview. Results

The results of these three

single-case studies are shown in Table 4

and

Figures 4 through 5. Table 4 contains a listing by phase (interviewing style) of the affective responses described by the client during the first single-case study. Figures 4 and 5 show graphs of the numbers of positive and negative feelings listed by the clients during the second and third single-case studies.

0

s M 0) L

~ 0)

.r 5

a ’o

a~

. c ~ m

C’E L~ E o

’= -a

~

0

~~ ’SS

-~~

3 Q.

XC ~ ~

M en E

Q ’0

eo ~ ~

-e

~ o

~s’

C_

~C

0)

0 E

~ ~ O =d

~~ 0 .a

w

tL 0 E z 21

0

s 0 0

t_ ~

0

53>~ -0

c

di

0 E ~

~)