A student started a new reading intervention three weeks ago. You have four scores. The line looks flat. A colleague says, "Time to change it." Another says, "Give it a month." Both want what is best, and neither has a number to point to.
This guide collects the numbers the sources do give, and explains why the answer is a range and not one figure.
Different jobs, different minimums
"How many data points" is really several questions. Each has a different answer.
| Job | What the source says | Source |
|---|---|---|
| Academic baseline | Three progress monitoring probes, then median or average (or a screening score) | NCII goal-setting guide |
| Behavior baseline | At least five data points, ideally stable | NCII behavior guide |
| A goal from the student's own growth | Six to nine data points to index the slope | NCII goal-setting guide |
| Four-point decision rule | 3 weeks of instruction and 6 data points, then the 4 latest | NCII decision rules |
| Trend line or median of three | 4 weeks of instruction and 8 data points | NCII decision rules |
| Review of recent points | After six to nine data points, review the latest four | MTSS Center article |
The numbers are minimums for starting to apply a rule. They are not targets to stop at. NCII also says to assess whether data are stable. For behavior, it says highly variable baseline data may mean you need more data and a closer look at how the behavior is defined.
Frequency sets the wait
The rules count weeks and points. How often you collect decides how soon you reach both.
| Collected | 6 points | 8 points | Both conditions met at |
|---|---|---|---|
| Twice a week | 3 weeks | 4 weeks | 3 weeks for four-point, 4 for trend |
| Once a week | 6 weeks | 8 weeks | 6 weeks for four-point, 8 for trend |
| Every other week | 12 weeks | 16 weeks | 12 weeks for four-point, 16 for trend |
The numbers come from simple division. The weeks-of-instruction condition is met well before the data condition in the last two rows, so the data points are what you wait for. In a school year of roughly 36 weeks, a student monitored every other week would not reach the eight-point minimum until about week 16. That is a reason to choose a schedule on purpose.
The MTSS Center's chart of progress monitoring features lists at least weekly monitoring at tier 3 and at least biweekly at tier 2. NCII's behavior guide says a goal should be monitored often enough to determine progress and make timely decisions. Utah's goals guide says it is best to collect behavior data every day.
What to do before you reach the minimum
Waiting is allowed. But waiting should not mean ignoring the graph.
- Keep collecting the same way. Changing the measure resets the clock.
- Mark events. Absences, schedule changes, and new materials belong on the graph.
- Set a review date. The MTSS Center's article suggests reviewing data on a regular schedule, such as every two weeks, so that waiting has an end.
- Check fidelity now. Missed sessions or a changed lesson plan explain a flat line better than any rule.
When you cannot wait
NCII's behavior guide says to collect at least five baseline points "unless there is an ethical reason to begin immediate intervention." Harm to the student or others is that kind of reason. In that case, start the intervention, keep recording, and compare later data with whatever baseline you have, noting its limits.
For academic skills, there is rarely an emergency of that kind. A student who is clearly not growing after the minimum has passed is a reason to act. Fewer points than the minimum are a reason to look closer, not a reason to decide.
Why not decide on two or three
Two points make a line, but the line is not reliable. A single low score can be an off day. NCII's handout lists a caution for every rule: outliers in the last four points can mislead the four-point rule, a trend line needs enough points for a reliable trend, and the median of the last three can miss a trend when scores vary widely. More points make each of those problems smaller, but they do not make them vanish. Pair the numbers with a look at the student and the lessons.
A calendar example
Suppose instruction on a new goal starts on Monday, October 5, and you give the probe each Friday. The first probe is Friday, October 9. Counting forward, the sixth lands on November 13 and the eighth on November 27. That Friday is the day after Thanksgiving in 2026, and many schools are closed. So the eighth point slips to the next school day, and a week of instruction that did not happen does not count toward NCII's weeks of instruction either.
The lesson is to count on a real calendar before you promise a review date. List the weeks with breaks, testing days, or field trips, and push the review back to match. A team that says "we will decide at eight points" without looking at the calendar often finds the date has passed or has not yet arrived.
Counting slips
- Counting the baseline as monitoring data. Baseline points sit before the intervention begins. Count the points after it starts.
- Counting duplicates. Two probes on the same day with different forms are a check on the probe, not two weeks of growth.
- Mixing tools. Points from two different measures cannot share one line.
- Dropping the dull points. If a score looks unremarkable, it is still data. Keep it.
A schedule you can adapt
See a weekly data routine for a caseload for fitting this into a busy week, and decision rules: when to change an intervention for what to do once you have the points.
What the sources do not say
None of the sources says a single number works for every student, measure, or goal. District procedures and tool publishers may require more. Where they do, follow them and say so in your notes. Terms are in the glossary.