J. Behav. Ther. & Exp. Psychiat. Vol. 5, pp. 7-12. Pergamon Press, 1974. Printed in Great Britain.
A
MULTIPLE
BASELINE ANALYSIS CONTROL
OF
COVERANT
LEONARD H. EPSTEIN* and MICHEL HERSEN Veterans Administration Center and University of Mississippi Medical Center, Jackson, Mississippi Summary--Coverant procedures, assessed in a multiple baseline design, were used to modify three high rate, target behavior problems (finger picking, lip biting, cigarette smoking) in a psychiatric patient. An additional target behavior 0nood) remained untreated to serve as a control. Results indicated successful control and maintenance by coverant techniques; however, instructional or self-monitoring effects could not be discounted. EXPERIMENTAL control is important in the evaluation of self-management procedures. As Mahoney (1970) has pointed out, demand expectancies, instructional sets, and selfmonitoring are inherent components of selfmanagement, and may influence behavior idependently of the programmed treatment. Therefore, an experimental design must provide controls for these effects in addition to evaluating treatment. This paper presents the use of a multiple baseline in evaluating coverant control, a popular self-management procedure. Coverant control procedures were initially discussed by Homme (1965). He outlined techniques for modifying rates of coverants (covert operants) using "Premack" reinforcers. The Premack principle states that a high probability response can be used to reinforce a lower probability free operant response (Premack, 1965). In applying the Premack principle to coverant control techniques, Homme proposes that the frequency of coverants can be changed by the subject programming high probability behaviors contingent on the emission of the coverant. Subsequent single case reports indicated that coverant control procedures were effective in modifying inter-
personal problems (Johnson, 1971), depression (Mahoney, 1971; Todd, 1972), negative selfevaluation responses (Todd, 1972), and obsessive thinking (Mahoney, 1971). Group studies have also been used to evaluate both the outcome and the process of coverant control. Keutzer (1968) found that coverant control was equal in effectiveness to breath holding and negative practice for modifying smoking. Tyler and Straughan (1970) showed that coverant control was as effective as breath holding and relaxation training in the treatment of obesity. In a process study of coverant control, Horan and Johnson (1972) found mean weight loss for obese subjects greatest for those in the coverant control group in which coverants were reinforced by contingent "Premack" reinforcers. However, mean weight loss for subjects given dietary information and for those emitting coverants on a predetermined schedule in absence of Premack reinforcement was not statistically different from the coverant control group. The withdrawal and greup designs have several disadvantages for the study of selfmanagement which are obviated by use of the multiple baseline design. In the withdrawal design, overt experimental control for engaging
*Requests for reprints should be addressed to Leonard H. Epstein, Psychology Service, Veterans Administration Center, 1500 Woodrow Wilson, Jackson, Mississippi 39216.
8
LEONARD
H. EPSTEIN
in self-management can be removed, but it is difficult to ensure that the client does not then resort to covert self-management. Also, some of the changes observed may be irreversible. That is, the targeted behavior may be influenced by naturally occurring environmental events, therefore resulting in no change when treatment is withdrawn. Group designs, as Sidman (1960) indicates, introduce inter-subject variability. Such variability may obscure the specific effects of treatment for a given subject as only the “average” subject is described. A further difficulty involved in group comparison designs is the problem of matching subjects. If matching procedures are to be done accurately, a large pool of subjects is required. This may prove to be extremely difficult when the investigator is intent on studying the more esoteric disorders. In addition, the number of subjects needed is directly a function of the number of experimental controls required. In the present study coverant control was evaluated using a multiple baseline design (Baer, Wolf and Risley, 1968; Barlow and Hersen, 1973). This experimental strategy provides for within subject measurement of several concurrent behaviors. After an initial baseline is established for all concurrent behaviors, treatment is introduced for one target ‘behavior. Treatment is then introduced sequentially for each of the remaining concurrent behaviors, providing an extended baseline for each succeeding behavior, The multiple baseline averts the problems noted for withdrawal and group designs by providing within-subject controls for such variables in self-management as ,time in treatment, therapist attention, self-monitoring, and instructions. Control for self-monitoring is of particular importance as it can effect behavior change (e.g. Johnson and White, 1971; McFall, 1970). In ,the present multiple baseline evaluation, concurrent self-report measurements were taken on four behaviors. Three were selected as target behaviors (nail picking, lip biting, and cigarette smoking), while the fourth (mood) remained untreated and served as a control. However,
and MICHEL
HJXRSEN
to provide controls for both instructional effects and self-monitoring, the subject was informed that mood was #being modified and he was instructed to record this “behavior” throughout treatment. Following baseline measurement, coverant control for nail picking and lip biting was introduced while the baseline for cigarette smoking was extended. Treatment was then simultaneously withdrawn for nail picking and lip biting to assess continued control in ‘the absence of overt manipulations. In the following phase, coverant control was introduced for cigarette smoking while concurrent measures of pipe smoking were taken. Finally, treatment for cigarette smoking was withdrawn and follow-up data were obtained. Thus, experimental effects of coverant control were assessed via the multiple baseline strategy; withdrawal of treatment was used to evaluate duration of change. METHOD Subject The subject was a 28-yr-old, white, male who was admitted to the Psychiatry Service of the Veterans Administration Center, Jackson, Mississippi, with a previous diagnosis of schizophrenia, paranoid type. Throughout his hospitalization he was administered Thorazine, 100 mg., q.i.d.; Stelazine, 2 mg., t.i.d.; and Cogentin, 1 mg., b.i.d. A behavioral analysis revealed a marked deficit in appropriate independent responses that appeared to be maintained by contingencies at home. Subsequent to his previous hospitalization the subject’s mother had failed to implement a behavioral program designed to structure his activities. Given this history, the need for teaching the subject self-control was apparent. Moreover, it was expected that at the conclusion of his treatment, data obtained would be used to motivate the sublect to change additional problem behaviors through the use of self-control procedures (Epstein and Peterson, 1973 ; Goldiamond, 1965). The subject identified several discrete target behaviors that he thought warranted modifi-
A MULTIPLE BASELINE ANALYSIS
OF COVERANT
CONTROL
9
cation (finger picking, lip biting and cigarette smoking). Effects of these behaviors were observable: bleeding around the cuticles, chapped and split lips, and excessive coughing and wheezing. Observations also indicated that cigarette smoking and coffee drinking were two high probability behaviors that might be used as “Premack” reinforcers (Premack, 1965) to control coverant rates.
times daily (upon arising, breakfast, lunch, dinner, prior to retiring). Ratings were obtained by having the subject record a “+” if he was in a “good” mood, and a “-” if he was in a “bad” mood. Data used in analyses were number of “-” ratings per day. Data for the other target behaviors were also placed on this index card and were collected by the senior author each morning during ward rounds.
Recording Target behaviors were monitored daily by the subject on a wrist recorder that included three channels. The recorder consisted of a leather band with six pipe cleaners arranged horizontally, with nine beads on each pipe cleaner (see Fig. 1). This arrangement allowed for recording of 99 occurrences of a target behavior per day. The two pipe cleaners per channel were used to count 10’s and 1’s. Thus, the left channel in Fig. 1 indicates 0 occurrences: the middle channel, 4 occurrences; and the right channel, 17 occurrences. No systematic reliability checks were obtained on the accuracy of the subject’s recording behaviors. However, he was observed by the senior author and a medical student at various times during the day and by a social worker in group psychotherapy four times a week. He was specifically observed to have recorded cigarette smoking and lip biting each time it occurred. Finger picking was not observed in the presence of these staff members.
Coverant control procedures During a pretreatment interview the subject was informed that an attempt would be made to modify his mood. As no contingencies were arranged to control mood, this procedure served to evaluate instructional effects. He was also instructed that he would be expected to record a number of target behaviors. Prior to treatment the subject was asked to construct his own wrist counter in Occupational Therapy r~_~.. a model provided by the senior author. rrom This pretreatment requirement was used to assess motivation to participate in selfmanagement. The subject was also asked to generate a list of positive statements, a number of which were used to prompt coverants. After baseline measurement, 10 of these statements were selected to be used in the treatment phase for finger picking and lip biting. Two were directly related to the target behaviors selected for modification (“I can stop picking my fingers”; “I can stop biting my lips”); while the remaining eight were unrelated (e.g. “I can drive well”). Unrelated statements were added to facilitate the evaluation of instructional effects. Without their inclusion the credibility of this procedure would have been decreasal, particularly if the subject was informed to alter his mood while being given statements related to the specific target behaviors. The procedure for coveranr control involved a chain in which the subject was instructed to: (1) read to himself a card from among the 10 in his breast pocket; (2) take a cigarette; and (3) then record a smoking response on his wrist recorder before actually smoking the cigarette
Target behaviors For purposes of recording, finger picking and lip biting were operationally defined as picking at cuticles of fingers and gnawing of lips for at least 5 sec. This time duration was included to facilitate self-observation of each behavior. A second occurrence of each behavior was not recorded until the subject -had discontinued his previous response for 5 sec. Rate of cigarettes per hour was calculated by dividing number of cigarettes smoked per day by total number of hours the subject was awake. The subject’s mood was recorded on a 3 X 5 index card five
10
LEONARD
H. EPSTEIN and MICHEL HERSEN
or drinking a cup of coffee. This chain was practised during a pre-experimental session. Discriminative stimuli for initiation of the chain were those associated with both the target behavior and the reinforcer. Coverants were programmed immediately prior to occurrence of high frequency reinforcers (cigarette smoking and coffee drinking) in order to increase frequency of their emission. Johnson (1971) reported a similar procedure for increasing rate of coverants. The cards were removed from the ,subject following treatment for finger picking and lip biting and he was instructed to continue to record these behaviors. On Day 18, he was given feedback as to the “true” nature of the experiment and with respect to his performance. He also was questioned as to whether he now wished to modify his rate of cigarette smoking He agreed, and three anti-cigarette, pro-pipe smoking statements were placed on cards and substituted for the initial finger picking and lip biting items and an unrelated item. The three statements were: “Pipe smoking is better than cigarette smoking’*. Girls like pipe smokers.” “Pipe tobacco tastes better than cigarette tobacco.” No other instructions regarding pipe smoking were provided. Despite the fact that the subject had a pipe in his possession at the time of admission, he had not been observed, nor did he report smoking it prior to initiation of smoking treatment. After treatment for smoking, the cards were again taken from the subject, and additional inpatient follow-up data were collected. The results were then discussed with him in order to provide reinforcement for his performance and to motivate future attempts at using selfcontrol procedures. Post-discharge behavior was recorded by the patient using a checklist of 11 items that were desired by his family and were essential to self-maintenance. Examples of these included straightening his room, attending to his job, and saving money. These behaviors were recorded either daily or weekly, depending on their rate of occurrence. Self-reports were con-
firmed at designated intervals by his aunt with whom he lives. RESULTS The effects of coverant procedures on the two sets of target behaviors and the control behavior (mood) are presented in Figure 2. Rates of both finger picking and lip biting
Iarsllne
Covert Control
WithdnwrlC
Followup
FIG. 2. Rates of target and control behaviors during baseliie, coverant control (labeled covert control iu Figure), and withdrawal conditions.
decreased during treatment. Subsequent withdrawal of treatment did not result ,iu an increase in rate. The mean rates during baseline, treatment, and follow-up phases were 6.0, 1.35, and O-55 for finger picking and 4.0, 1.83, and 0.45 for lip biting. Rates of cigarette smoking were stable during the ‘baseline and treatment phases for finger picking and lip biting. Introduction of treatment for cigarette smoking resulted in a decrease in rate of cigarettes smoked per hour. Removal of treatment for cigarette smoking did not result in a rate increase. Rates for cigarette smoking during
A MULTIPLE BASELINE ANALYSIS OF COVERANT CONTROL
baseline, treatment, and follow-up phases were 4.25, 2.57, and 2.58 per hour. As shown in Fig. 2, the reduction in cigarette smoking appeared related to ,increased pipe smoking. From a pretreatment baseline of 0.0, the subject smoked an average of 6.83 pipes per day during coverant control and 7.25 pipes per day after termination of treatment. Scores on the five-point ‘mood” scale decreased gradually during the experiment. The means for the baseline, first coverant control phase, extended baseline, second coverant control phase. and withdrawal conditions were 2.83, 2.83, 2.33. 2.17, and l-75, respectively. Subsequent to treatment the subject reported greater confidence in his ability to control his behavior. He also reported that his cuticles had stopped bleeding and that they had begun to heal; this was confirmed by observation. In addition, a 1Zweek follow-up of weekly checklist self-reports indicated a high rate of the self-maintenance behaviors. Three months after follow-up an unsolicited report indicated that the subject had obtained high grades in a local junior college and was placed on the Dean’s List. DISCUSSION The results of this experimental ana.Jysis demonstrate the effectiveness of coverant procedures in controlling high rate overt target (problem) behaviors. The target behaviors modified in this study (finger picking, lip biting, smoking) add to the increasing list (overeating, obsessive thoughts, depression, interpersonal problems) successfully managed through coverant control (Horan and Johnson, 1972; Johnson, 1971; Mahoney, 1971; Todd, 1972, respectively). The multiple baseline used in this experiment allowed for the systematic replication of coverant procedures in two independent sets of target behaviors while maintaining an untreated target behavior for purposes of control. However, the use of mood as the control was not ideal because the subject did not identify mood as a problem and, therefore, may have been less
11
motivated to change that behavior. Also, the patient’s mood gradually changed during the experiment. Within ‘behavior withdrawals of treatment were used to evaluate durability cf change. The fact that rates for modified target behaviors did not change after coverant control procedures were removed may be interpreted in two ways. First, even though external prompts (coverant cards) were removed, the subject continued to covertly emit these statements. Second, reduced rates may have come under control of naturally occurring environmental events. McFall (1970) points out that self-monitoring may at&t independent behaviors differently and, ,in this study, self-monitoring may have affected finger picking and lip biting differently than cigarette smoking. However, due to decreasing trends in baseline rates, it is difficult to infer the effects of self-monitoring or of the coverant procedure on finger picking or lip biting. However, the effects of self-monitoring on cigarette smoking were controlled by the extended stable baseline measurement of cigarette smoking. There are several reasons why coverant control appears to ‘be a useful therapeutic technique. First, it can be very easily engineered for an apparently large variety of target behavior problems. Second, the treatment requires relatively little therapist time and in many cases may be implemented by a well-trained para-professional. Third, the technique can then be taught to a patient for use in the selfcontrol of other, even future, problem ‘behaviors. The subject’s continued application of these procedures in controlling problem behaviors will depend both on environmental maintenance of more adaptive behaviors and the relative aversiveness of engaging in the former problem behavior (Epstein and Peterson, 1973: Kanfer and Phillips, 1970). The long range control of problem behaviors may be facilitated by programming responses that have a high probability of being reinforced by the environment. In this study, pipe smoking served as a new adaptive response.
12
L E O N A R D H. EPSTEIN and M I C H E L H E R S E N
Acknowledgement--The authors t h a n k Linda M. Lyall for her assistance in preparing the figures. REFERENCES BAER D. M.. WOLF M. M. and R1SLEY T. R. (1968) Some current dimensions of applied behavior analysis, J. appl. Behav. Anal. l, 91--97. BARLOW D. H. and HERSEN M. (1973) Single case experimental designs: Uses in applied clinical research, Archs gen. Psychiat. 29, 319-325. EPSTEIN L. H. and PETERSON G. L. (1973) The control of undesired behavior by self-imposed contingencies, Behav. Therapy 4, 91-95. GOLDIAMOND I. (1965) Self control procedures in personal behavior problems, Psychol. Rep. 17, 851868. HOMME L. E. (1965)Perspectives in psychology: XXIV: Control of coverants, the ogerants of the mind, Psychol. Ree. 15, 501-511. HORAN J. J. and JOHNSON R. G. (1971) Coverant conditioning through a self-management application of the Premack principle: Its effect on weight reduction, J. Behav. Ther. & Exp. Psychiat. 2, 243-249. JOHNSON W. G. (1971) Some applications of H o m m e ' s coverant control therapy: Two case reports, Behav. Ther. 2, 240-248. JOHNSON S. M. and WroTE G. (1971) Self observation as an agent of behavior change, Behav. Therapy 2, 488-497.
KANFER F. H. and PHILLIPS ]'. S. (1970) Learning Therapy Foundations o[ Behavior Therapy, Wiley, New York. KEVTZER C. S. (1968) Behavior modification of smoking: The experimental investigation of diverse techniques, Behav. Res. & Therapy 6, 137-157. MAHONEY M. J. (1970) Toward an experimental analysis of coverant control, Behav. Therapy 1, 510-521. MAHONEY M. J. (1971) The self-management of covert behaviors: A case study, Behav. Therapy 2, 575 579. MAttONEY M. J. (1972) Research issues in self management, Behav. Therapy 3, 45-64. McFALL R. M. (1970) Effects of self-monitoring on normal smoking behavior, J. corr. olin. Psychol. 35, 135-147. PREMACK D. (1965) Reinforcement theory, Nebraska Symposium on Motivation (Edited by LEWNE D.), University of Nebraska Press, Lincoln. SIDMAN M. (1960) Tactics of Seienti]ie Research, Basic Books. New York. TODD F. J. (1972) Coverant control of self-evaluation responses in the treatment of depression: A new use for an old principle, Behav. Therapy 4, 91-94. TYLER V. O. and STRAUC,~AN J. H. (1970) Coverant control and breath holding as techniques for the treatment of obesity, Psychol. Rec. 20, 473 478.