Baseline Data
When developing an intervention plan for a student, we begin by taking baseline data. Baseline data should be stable prior to implementing the intervention. Baseline data that is stable typically would have a trend line with zero celeration. In other words, the trend line that represents the data would be parallel to the abscissa or x-axis.
However, when we are working with a behavior that is decreasing (decelerating trend line) and we want that behavior to be increasing (accelerating trend line), it is appropriate to intervene on the behavior that is decreasing in ordinate value.
Likewise, when we are working with a behavior that is increasing (accelerating trend line) and we want that behavior to decreasing (decelerating trend line), it is appropriate to intervene on the behavior that is increasing in ordinate value.
In some cases, the data may be highly variable and is it is difficult to determine the trend. In these cases we can calculate the stability of the data.

• Follow these steps to calculate stability:
  Find the mean of your data. To do this, add the values and divide by the number of data points.

For example, you have the data points of 2, 7, 4, 7. Adding these together equals 20. Then divide this total by the number of data points (in this case 4). 20 divided by 4 gives you a mean of 5.

Calculate 50% of the mean. To do this you divide the mean by 2.

In the preceding data the mean was 5. Dividing this number by 2 equals 2.5.

Add the number that represents 50% of the mean to the mean.

In this example that would be 5 (the mean) plus 2.5 equaling 7.5.

Subtract the number that represents 50% of the mean from the mean.

In this example that would be 5 (the mean) minus 2.5 equaling 2.5.

You now have determined the range where all of your data points should fall to determine stability.

In this example the established range is 2.5-7.5.

Going back to the original data points (2, 7, 4, 7) you notice that 2 falls outside this range and thus your data points do not represent stable baseline data. The solution is to continue to collect baseline data until you can determine stability.