Most of the problem with is with the language. Some is with using IT.
First of all, an 'error' is a mistake and we don't use that word. If you've made a mistake, you do it again. So we don't use 'error bars' we mark the range of uncertainty.
So what is uncertainty? Well, it's as simple as "I know it's on my desk, not sure where, but it IS on my desk". (Much easier to understand if you've ever seen my desk.) We know where the thing we are looking for is, within bounds. We are uncertain as to its exact position, but we can say within what parameters it will lie. In this case my desk.
A stage further - numbers. "There are 200 children, give or take 10." We all use the term "give or take", so we know that the number of children must lie somewhere between 190 and 210.
The number of children is 200 + or - 10, not a huge step and neither is using the '+' over the '-' sign that I have no idea how to type in here.
When we put this number on a graph, as we do not know the exact number, but we do know the bounds in which it will lie, so we draw a line and somewhere along it is the actual number.
So far so good?
Let's add another dimension.
We are measuring the number of children every hour, but our time-keeper could be anywhere between 5 minutes early and five minutes late. Our measurement of children is such-and-such and hour give or take 5 minutes. All the above applies.
But what if we are putting both (uncertain within bounds) numbers on a graph? For example the number of children against hours.
Well, we mark the line somewhere along which the actual number lies at the hour. But the hour is "give or take 5 minutes" so we mark a line horizontally to show that uncertainty.
Those two lines mark out a box. Somewhere within that box is the actual value. It could be far bottom left - 5 minute to the hour and 190 kids. It could be far right - 5 minutes after the hour and 210 kids. But it is SOMEWHERE in that box - we can be sure of that.
So why is IT unhelpful? Students put in numbers without actually thinking about what they are doing with them. They are two steps away from the data - especially if the software then produces a line for them. The students use the information the computer has calculated with no appreciation of what it has done for them. It is like thinking that a computer that reads FOR you makes you better at reading. Or an automatic car makes you better at changing gear in a car with a gear stick.
We might give students the ability to make pretty graphs, but we take their understanding of them. Are we really sure we want this?