Students are often called upon to draw lines of best fit by hand, without the process being well
defined. For example, the line might be linear, but with no information as to the size or direction
of any error bars, or the line might be curved but in an unfamiliar fashion. A survey of students
showed good agreement amongst the sample as to the best straight line for noisy data, although it
differed from the expected value. When asked to draw an exponential curve (as for a half-life
experiment) there was a significant difference between results when a point was included or omitted
on the y -axis. This may inform what we ask students to do in experimental work.