A reported score is a measurement made under particular conditions. Its precision depends on the assessment and the evidence available for it. There is no single validated margin of error that can be applied to every online IQ test.
Spread and uncertainty are different
The standard deviation of a group describes differences between its scores. The standard error of measurement concerns uncertainty associated with individual measured scores. These quantities have related statistical roles, but they are not interchangeable.
A wide distribution of results does not automatically mean that each result is imprecise. Similarly, a tightly clustered group does not prove high precision for every individual.
An illustrative calculation
In a common classical-test framework, an overall standard error can be expressed as score standard deviation multiplied by the square root of one minus reliability. If a hypothetical test has standard deviation 15 and reliability 0.84, the calculation is 15 × √0.16 = 6.
Those are invented inputs for explaining the formula. They are not measured values for starp. A provider needs appropriate reliability evidence before substituting numbers and presenting an uncertainty estimate.
Read confidence intervals in context
An assessment may report an interval alongside the score using a documented method and confidence level. Consult that method rather than assuming any interval is simply the score plus or minus a standard number of points.
Do not interpret overlapping intervals as a complete statistical test of a difference. Comparisons can require additional information, particularly when two results come from the same person or different instruments.
What this means for online results
Avoid attaching excessive meaning to a one-point difference. Check whether the provider gives a supported precision estimate and whether it applies to the score you are reading. Where no validated estimate is available, say so instead of borrowing one from another test.
For practice, focus on the actual answers and the reasoning behind them. That evidence is more concrete than pretending a numerical score has precision its method has not established.