This can be avoided by pinpointing the target behavior precisely enough so that you can give a simple yes or no answer: Yes, Melody is talking on the phone; or no, she isn't. Yes, Otto made a negative comment about the program; or no, he didn't. Suppose Otto says, "I think we should reconsider each step of this program in light of the new developments in the market." Is this to be counted as a negative comment or not? Any answer besides a definite yes or no should alert us to an incompletely defined target behavior. Georgia would reduce the effectiveness of her team and would limit Otto as well if she employed a change program that decreased the frequency of Otto's problem-solving comments along with his negative ones. Before Georgia can continue collecting baseline data, she needs to return to the first step of specifying the problem behavior. When she can clearly define a negative comment about the program, she is ready to resume collecting the baseline data.
There are times when it is not necessary to observe directly, because the information has already been collected. Institutional sources of behavioral data include payroll records for absenteeism, time cards for tardiness, and log records for speed of response (such as in police calls). Whenever possible, utilize such records-it saves time.
Some natural questions are, "How do I count?" and "What tool do I use?" There are a number of methods, the most obvious and traditional of which is the tally. Each time you observe the target behavior, make a check on a piece of paper or on a 3" x 5" index card. At the end of the day or of the meeting, count up the checks and transfer this data onto your graph. A second tool you can try is the golf counter. The best is the type you wear on your wrist-it looks like a watch. These are particularly effective, because you can use them without interrupting your activity or diverting your attention. The point is to count. The method you use doesn't matter as long as you get accurate data, so devise a method that is easy and that works for you.
The best approach is to have the subordinates in question count and record their own behavior. For example, chronically late employees can record their time of arrival each day, packers on an assembly line can count the number of boxes correctly packed, a secretary can record the length of each personal phone call. An advantage of self-recording is that it gives employees immediate feedback on their behavior and involves them in all stages of the change process. It is important that you reinforce employees for collecting data on their own behavior.
There is no set length of time for collecting baseline data; you need to keep counting until the data is stable-that is, until you can see a consistent trend. If the target behavior occurs several times a day, then most likely a week-long baseline would suffice. When there is a lot of variation from day to day, a longer baseline period will be needed. The purpose is to establish how often the behavior occurs. This counting should continue until you feel you understand the pattern of the target behavior.
If the target behavior occurs regularly, the actual time involved gathering baseline data can be shortened considerably by using a time sample. A time sample is similar to an opinion poll: Just as those conducting the poll cannot interview everyone in the country, you probably cannot count every occurrence of the behavior. Opinion polls solve the problem by selecting a sample of people and talking to them. With a time sample, you count the occurrence of the behavior in a few short time periods each day. For example, Georgia might select two five-minute time periods in each meeting to count Otto's comments.
How the sample is selected is also important. If an opinion poll were to base its conclusions solely on the opinions of Californians, the results might not be representative of the country as a whole. The time sample, too, must represent the whole time period covered. This is best done by randomly selecting the time periods. Georgia could assign a number to each five-minute period in the meeting, put them into the proverbial hat, and select two at random to use as her sample.
Behavior modification techniques are not crisis techniques; they are foundation-building techniques that require careful analysis before using them. Grounded in the scientific method, behavior modification is an approach in which you become an applied behavioral scientist solving a problem. Each behavior change program is a unique challenge, because no two people are alike. What one person finds reinforcing, another does not. What provokes one person has no impact on another. To be an effective behavior manager, you need to understand and respond to each person as an individual. In part, you can accomplish this by observing, analyzing, and counting. By understanding each supervisee's uniqueness you can develop a change program suited to his or her individual needs. By continuing to collect data during the intervention and through comparison with the baseline, you can obtain an immediate indication of the effectiveness of the change program. And such data can be used as concrete evidence of your effectiveness as a manager. If comparisons reveal that the data are moving in the desired direction (if the frequency of undesired behaviors decreases and that of desired behaviors increases), you can rely on your program. If the trend begins to plateau or to reverse, this signals that it is time for troubleshooting. Finally, baseline comparisons are a good indicator of how long to continue in the intervention phase and of when to move to the maintenance phase of the change program. When the frequency of the target behavior reaches a predetermined goal, it is a signal to implement techniques that will maintain the behavior at that frequency-in other words, it signals that the intervention phase is complete.
