The Psychological Record, 1975, 25, 255-264. RELIABILITY AND CONTROLLING EFFECTS OF THREE PROCEDURES FOR SELF-MONITORING SMOKING' LEE W. FREDERIKSEN, LEONARD H. EPSTEIN Veteran's Administration Center and University Medical Center Jackson, Mississippi and BERNARD P. KOSEVSKY Ohio University Measurement reliability and controlling effects of 3 procedures for self-monitoring smoking were examined. In Experiment I, subjects monitored their smoking rates using each recording procedure for 1 week. Results showed continuous recording to be the most reliable procedure. Self-reports indicated continuous recording to be most accurate, bothersome to use, and reactive. In Experiment 11 a betweensubject design evaluated the 3 procedures over a 5-week period. Continuous recording led to pronounced reductions in reported smoking rate and greater attrition. Reported smoking rates did not significantly differ at a 6-month follow-up. Results were related to demands imposed by recording procedures and differing interrecording intervals. Self-monitoring is a technique in which the subject discriminates discrete behavioral events and records their occurrence. Initially, self-monitoring was used to generate data to evaluate the effectiveness of behavior change strategies (Kanfer & Philips, 1970; Simkins, 1971). The reactive or controlling effects of self-monitoring were also noted and subsequently utilized as intervention procedures (see review by Kazdin, 1974; Thoresen & Mahoney, 1974). Thus the uses of self-monitoring include evaluation andjor treatment. When self-monitoring is used in treatment evaluation, measurement reliability is of critical importance. On the other hand, behavior control is of primary importance when self-monitoring serves as a treatment. Numerous techniques of self-monitoring have been developed. While these procedures vary on many dimensions, one important variable may be the schedule of monitoring. Investigations have employed continuous and intermittent schedules of recording. When continuous schedules are used (McFall, 1970), the subject is instructed to record each occurrence of the criterion response. Another self-monitoring strategy would be to assess continuously the ongoing behavior but record intermittently its estimated frequency. For example, recording could occur once per day (Levinson, Shapiro, , This study was done while the first author was at Ohio University. The authors thank Ms. L. Vella for her assistance in conducting this research. Reprints may be obtained from either thefirstauthoratthe Psychology Department, Veteran's Administration Center, Jackson, Mississippi 39216, or the secondauthor, who is now at the Department of Psychology, Haley Center, Auburn University, Auburn, Alabama 36830. 256 FREDERIKSEN, EPSTEIN & KOSEVSKY Schwartz, & Tursky, 1971) or less than once per week (Schmahl, Lichtenstein, & Harris, 1972). The intermittency or recording may affect both reliability (Simkins, 1971) and the control exerted by self-monitoring. Procedures that request recording at long intervals permit occurrence of numerous behaviors that can compete with the processes involved in self-observation. However, the recording interval may interact with such factors as response rate and day-today variability in determining reliability. For example, low-rate, easily discriminated responses, such as homosexual contacts, probably can be reliably recorded at relatively long interrecording intervals. However, responses occuring at high, variable rates, such as smiling or obsessive thoughts, probably require continuous recording or relatively short interrecording intervals. The control exerted by self-monitoring may be functionally related to the intermittency of recording. Mahoney, Moore, Wade, and Moura (1973) showed that continuous recording of correct responses du ring academic review increased study time relative to the recording of every third correct response. Further , recording at short interrecording intervals can be helpful when self-reward is used, as the opportunities for consequating occurrences of the recorded response are increased. The design of a self-monitoring program must also take into account the demands involved in recording. A recording procedure that is time consuming and interferes withconcurrent activities may not be consistently used by the subject. Such disruptions in selfmonitoring are unlikely to improve either reliability or behavioral control. Thus both intermittency of recording and the recording demands are important to the clinical utility of self-monitoring. This study was designed to assess the measurement reliability, controlling effects and subjective evaluations of three methods of self-recording. These procedures were selected to sampie continuous and intermittent (Le., daily and weekly) recording of cigarette smoking. Cigarette smoking was chosen because it is a relatively high-rate behavior that has been the target of numerous treatment interventions (Hunt & Matarazzo, 1973). Experiment I investigated measurement reliability, short-term control, and subjective evaluation of the three recording procedu:r:es utilizing a within-subject analysis. Based on the results of Experiment I, Experiment 11 was designed to evaluate controlling effects of the recording procedures during a 5-week treatment period and at a 6-month follow-up. A between-subject analysis was used. EXPERIMENT I Method Subjects Volunteers from introductory psychology classes (8 females, 7 males) received course credit for serving as subjects. The selection SELF-MONITORING SMOKING 257 criteria was a reported smoking rate ofatleast 15 cigarettes per day. The mean age of the subjects = 18.67 years, mean years smoked 3.68, and mean estimated baserate 22.68 cigarettes per day. = = Recording Procedures The procedures were chosen to sampie continuous and two schedules of intermittent recording. The continuous recording procedure was similar to that used by McFall (1970), the daily recording similar to the Levinson et al. (1971) procedure, and the weekly recording procedure approximates the follow-up recording procedures used by Schmahl et al. (1972). Continuous recording(CR). Subjects were given a supply of small cards cut to fit inside the cellophane wrapper on a cigarette package. Each morning the subjects were to select an unused card and place the date, their name, and their social security number on it. The recording procedure involved writing the time they started smoking each cigarette that day. Daily recording(DR). In this recording procedure subjects were instructed to re cord the total number of cigarettes smoked in a given day. The total for each day was recorded nightly on a small card wi tb space provided for name, date, and social security number. Weekly recording (WR). In this recording procedure subjects were contacted by telephone once a week and were asked to report the average number of cigarettes smoked per day during the preceding week. General Procedure All subjects attended four half-hour meetings held at 1-week intervals . During the first meeting subjects were informed they would be involved in a comparison of three procedures for recording smoking and were free to smoke as much or as little as they wished. They were instructed to buy all their cigarettes from the experimenter, smoke only the cigarettes that had been purchased in this manner, and bring the remaining cigarettes to the next meeting. Each subject was then randomly assigned to one of six conditions that counterbalanced possible orders in which the three recording procedures could be employed. Subjects purchased a week's supply of cigarettes (the subjects reported smoking rate plus about 25 %), and were instructed in the recording procedure for that week. During each of the next two meetings, subjects tallied unused cigarettes, purchased another supply of cigarettes, returned data cards (DR and CR methods), and were instructed in the monitoring procedure they were to use for the next week. At the final meeting the subjects completed a questionnaire, returned data cards, and tallied unused cigarettes. Results Reliability The reliability of each recording procedure was calculated by first obtaining the difference between each subject' s mean daily 258 FREDERIKSEN, EPSTEIN & KOSEVSKY actual smoking rate [(cigarettes purchased - unused cigarettes) /7] and his mean daily reported smoking rate (as recorded 'by the subject). This difference was then divided by the mean daily actual smoking rate. The resulting proportion was multiplied by 100 and expressed as percentage of agreement. An arcsine transformation was performed on the proportional data to satisfy the assumptions of analysis of variance (Kirk, 1968). Separate treatment X subject analyses of variance were performed with weeks as the within factor in one analysis and procedures the within factor in another analysis. Results indicated that reliabilities across weeks, independent of the procedures, were not significantly different (F < 1). However, the reliabilities of the procedures were significantlydifferent (F = 6.06, df= 2, 28, P <.01). Dunn's Multiple Comparison Test (p <.05) showed that CR was more reliable than either WR or DR, which did not significantly differ. The mean percentages of agreement for the three procedures were: CR=93.59%, DR=85.77%, WR=87.32%. Smoking Rate A treatment X subject analysis of variance, with weeks as the within factor, showed no significant week-to-week variation (F< 1) in mean smoking rate (based on cigarettes used). A separate treatment X subject analysis of variance was performed on reported smoking rate with recording procedures the within factor. This analysis showed that the reported smoking rates associated with each procedure (CR = 20.40, DR = 22.11, WR = 23.80) were not significantly different (F = 2.03, df =2,28, P >.05). Subjective Evaluation The results of the postexperiment questionnaire are shown in Table 1. The ratings in Questions 1 and 2 were subjected to separate TABLE 1 Results of Postexperiment Questionnaire Recording Procedure Question (1.) Rating of accuracy: 1 =Not accurate 7 = Very accurate (2.) Rating of "hassie": 1= No hassle 7 =Big hassle (3.) Avoided smoking because of procedure: (4.) Increased smoking because of procedure: (5.) Best overall procedure: WR DR eR 2.93 4.27 6.73 1.27 3.47 6.27 o o o o 11 o 12 aOne subject did not answer this question. treatment X subjectanalyses ofvariance, with recording procedures the within factor. There was a significant procedures effect in both Question 1 (F = 51.21, df= 2,28, P <.001) and Question 2 (F = 75.51, SELF-MONITORING SMOKING 259 d/=2,28,p <.001). An aposteriori analysis (Tukey's HSD Test) was performed on the ratings in Questions 1 and 2. The ratings ofthe three procedures were significantly different (p < .01) on Question 1 and Question 2. eR was rated as most accurate and involving the most "hassIe." Table 1 also indicates that on Question 3, 11 of the 15 subjects reported they avoided smoking because of the eR procedure, while no subjects reported avoiding smoking because of the DR or WR procedures. As indicated in Question 4, no subject reported increased smoking because of any recording procedure. Most subjects selected the DR procedure as the best overall recording procedure, with the eR procedure being the choice of 2 people (Question 5). Discussion The results of Experiment I indicate that the three recording procedures differ on reliability and subjective evaluations. The measurement reliability was higher for continuous recording than for either daily or weekly recording. While the differences between reliability estimates were significant, the coefficients obtained were all above 85 %, which is adequate for the most measurement purposes. Although subjects were never so informed, they may have been aware that the reliability of their recording was being assessed. This may lead to reliabilities that are somewhat higher than those obtained when subjects are unaware of reliability checks (Lipinski & Nelson, 1974). On the questionnaire eR was evaluated as being most accurate, most demanding, and also exerting the greatest control on smoking behavior. Whenasked to take all factors into account (e.g. accuracy, time required, etc.), the subjects tended to see DR as the best overall procedure. It is also interesting to note that no subject reported increased smoking because of a recording procedure. The similarity in the short-term controlling effect of the three recording procedures was unexpected. This may mean that all of the procedures are equally good, or bad, controllers of responding. In either case the clinical utility, in terms of behavioral control, must be evaluated over a longer duration. A second experiment was thus designed to evaluate control of smoking over a 5-week treatment period and maintenance at a 6-month follow-up. EXPERIMENT 11 Method Subjects Volunteers (23 males, 13 females) were recruited from introductory psychology classes to participate in the experiment for course credit. The selection criteria was the same as in Experiment I. The mean age of the subjects = 19.20 years, mean years smoked 3.07, mean estimated baserate 23.31 cigarettes per day. = = 260 FREDERIKSEN, EPSTEIN & KOSEVSKY Recording Procedures The recording procedures (CR, DR, WR) were the same as in Experiment I. General Procedure Subjects attended one half-hour meeting. They were told that they would be involved in a 5-week comparison of three recording procedures and were free to smoke as much or as little as they wished. Subjects were randomly divided into three groups (CR, DR, WR). One-way analyses of variance performed on the variables of age, number of years smoked, number of attempts to quit smoking, and duration of longest abstinence indicated that the groups were not significantly different (p> .05). Each group received recording instructions identical to those used in Experiment I. Subjects in the DR and CR groups returned completed data cards and obtained unused cards at a convenient location. Six months after completion of the 5-week self-monitoring procedure, all available subjects were contacted by telephone. An estimate of their current smoking rate was obtained and served as a follow-up measure. Results Attrition Attrition was defined as terminating participation in the experiment prior to its completion (excluding follow-up). No subjects from either the WR or DR group failed to complete the experimental procedures. However, three subjects from the CR group dropped out of the experiment-one during the first week and two during the final week.' When contacted by telephone, all three subjects cited dissatisfaction with the demands of the continuous recording procedure as their reason for quitting. Smoking Rate The mean reported smoking rate for all three groups is shown in Figure 1. A two-factor mixed analysis of variance performed on the estimated baserate and self-monitoring data indicated a significant groups x trials interaction (F= 5.66, df=10, 160, P<.OOI). Computation of simple main effects (Kirk, 1968) showed that mean reported smoking rates for the groups were not significantly different (p< .05) at baserate or Week 1 of self-recording. There were significant group differences in mean reported smoking rate during Weeks 2-5. Tukey's HSD test indicated that the mean reported smoking rate for the CR group was significantly (p < .05) lower than the mean reported smoking rate for WR group during Weeks 2-5 and significantly lower than the mean reported smoking rate for the DR 'The smoking rates ofthe two subjects not reporting during Week 5 were estimated using a least-squares procedure (Kirk, 1968). The alternative of dropping the two subjects from the analysis leads to more pronounced experimental effects than those reported. SELF-MONITORING SMOKING 261 group during Weeks 3-5. The mean reported rates for the WR and DR groups were significantly different during Week 4 only. Six out of the nine subjects who completed the five weeks of CR ~WR1 ·OR ~RJ - - r _ _ - - r - - _ ---~ '-t~ ..- - t - - _ 2 ~<}..... WEEKS OF SELF·MONITORING ,~yo. Figure 1. Mean reported smoking rate for the three experimental groups during treatment andatthe6-monthfollow-up.Baserate and follow-up data were obtained by having subjects estimate their smoking rate. reported rates of less than 10 cigarettes per day du ring the final week ofrecording. Ofthese six subjects, three reported smoking less than one cigarette per day, and one reported smoking between one and two cigarettes per day. In contrast, no subjects in either the WR or DR group reported similar reductions during the final week of recording. Further , only one of these subjects (DR group) reported smoking less than 10 cigarettes per day (8.57 cigarettes per day). Follow-uP At the six month follow-up the mean reported smoking rates for the three groups (Figure 1) were not significantly different (F <1). Of the nine sUbjects contacted from the CR group, no subject reported smoking less than 10 cigarettes per day. Two out of nine subjects from the DR group reported smoking less than 10 cigarettes per day with one ofthe two reporting a smoking rate of O. In the WR group five out of ten subjects reported smoking less than 10 cigarettes per day, with three of the five reporting complete abstinence. GENERAL DlSCUSSION The results of Experiment 11 show that continuous recording resulted in a greater reported smoking reduction than the intermittent procedures. This finding is consistent with the greater controlling effects of continuous monitoring that were observed by Mahoney et al. (1973). It is important to note that continuous recording did not produce effects significantly different from the intermittent procedures until Week 3. These delayed controlling effects indicate the importance of an extended trial when self-recording is used in the baseline phase of a self-management program. There were no significant differences between groups at the sixmonth follow-up. Thus follow-up data do not indicate that the different 262 FREDERIKSEN, EPSTEIN & KOSEVSKY recording histories produced differential changes in the contingencies governing smoking during the six-month posttreatment interval. Self-monitoring does not automatically produce lasting changes in the environmental events that affect smoking. There are several possible explanations for the superior controlling effects of continuous recording. One possible explanation is related to the demands of the recording procedure. The demands of recording are important because the recording procedure is imposed on ongoing behavior. It is expected that as the time and effort demanded by a recording procedure increase, concurrent behaviors will be disrupted. This disruption is likely to reduce the opportunities for a vailable reinforcement ofthe ongoing behavior, thus resulting in respons~ costduring each recording episode. The subject can avoid this response cost by decreasing the frequency of the response to be recorded-in this case smoking. The fact that the subjects evaluated continuous recording to be more demanding in Experiment land the finding that continuous recording exerted more control in Experiment 11 are consistent with this possibility. An alternative avoidance tactic would be for the subjects to discontinue recording. This alternative is supported by the greater attrition rate associated with continuous recording in Experiment 11. Another possible explanation for the superior controlling effects of continuous recording is related to the increased possibility of "unprogrammed" consequation by the subject. Continuous recording increases the number of opportunities for consequation of the recorded response. This property makes continuous recording especially useful in self-management programs employing selfreward. Although a self-reward component was not programmed in conjunction with self-recording in the current study, the subjects may have nevertheless reinforced decreases in smoking. In addition to the noted superiority in controlling effects, continuous recording was also associated with the highest reliability. Reliability may be related to the interrecording interval. As the interval between the occurrence and recording of the response is decreased, there is less opportunity for the occurrence of interfering events. Although this question has not been researched, short interrecording intervals would seem to be more critical when a highrate behavior, such as smoking, is monitored. However, intermittent recording procedures may be appropriate for accurate monitoring of low-rate behaviors. In summary, the choice ofa self-monitoring procedure must take into account both measurement reliability and controlling, or reactive, effects. In the present investigation, continuous recording was both the most reliable and exerted the most control. Thus continuous recording, as it was the most controlling, should be used with caution when a nonreactive measure is required. On the other hand, continuous recording may be the monitoring procedure of choice in clinical self-management, as it provides both reliable da ta and maximal opportunities for self-reinforcement. SELF -MONITORING SMOKING 263 REFERENCES HUNT, W. A., &MATARAZZO, J. D. 1973. Three years later: Recent developments in the experimental modification of smoking behavior. Journal of Abnormal Psychology, 81, 107-114. KANFER, F., & PIDLLIPS, J. 1970. Learning foundations of behavior therapy. New York: Wiley. KAZDIN, A. E. 1974. Self-monitoring and behavior change. In M. J. Mahoney & C. E. Thoresen (Eds.), Self-control: Power to the person. Monterey: Brooks/Cole. KIRK, R. E.1968. Experimental design: Procedures for the behavioral sciences. Belmont: Brooks/Cole. LEVINSON, B. L., SHAPffiO, D., SCHWARTZ, G. E., & TURSKY, B. 1971. Smoking elimination by gradual reduction. Behavior Therapy, 2, 477-487. LIPINSKI, D., & NELSON, R. 1974. The reactivity and unreliability of self-recording. Journal of Consulting and Clinical Psychology, 42, 118-123. MAHONEY, M. J., MOORE, B. S., WADE, T. C., & MOURA, N. G. M. 1973. Effects of continuous and intermittent self-monitoring on academic behavior. Journal of Consulting and Clinical Psychology, 41, 65~9. McFALL, R. M. 1970. Effects of self-monitoring on normal smoking behavior. Journal of Consulting and Clinical Psychology. 35, 135-142. SCHMAHL, D. P., LICHTENSTEIN, E., & HARRIS, D. E. 1972. Successful treatment of habitual smokers with warm, smoky air and rapid smoking. Journal of Consulting and Clinical Psychology, 38, 105-111. SIMKINS, L. 1971. The reliability of self-recorded behaviors. Behavior Therapy, 2, 83~7. THORESEN, C. E., & MAHONEY, M. J. 1974. Behavioral self-control. New York: Holt, Rinehart & Winston.