**Date:** Tue, 31 Aug 2010 12:13:49

**Author:** Donald Smith

**Subject:** Re: Sonic Ranger

**Post:**

Greetings,

On Fri, Aug 27, 2010 at 11:09 AM, Paul Nord wrote:

> It would be much better to do a curve fit of 0.5*a*t^2 + bt + c. Curve fitting is a little beyond the introductory lab students. However, it would give you clear graphical confirmation that your model fit the data. A curve fit uses all of the data at once rather than just pairs of points.

This is what I do. Then they can clearly see that the data points are

*not* following a parabola. What's amazing is even though it's

obviously *not* a parabola, many of them are so convinced it *should*

be a parabola that they ignore what they can see and insist it's

accelerating downward at a constant 9.8 m/s/s. Some of them do

realize that air drag is important, but many don't. It provides a

good lesson in how prone we are to seeing what we expect, and how

important it is to let the data take priority over our expectations.

I don't ask them to understand what the curve fit does at that point

in the curriculum. I hate to use the computer as a black box, but in

this case I think the benefit is worth it.

Yours,

--

Donald Smith

Guilford College Physics Department

http://www.guilford.edu/physics/dasmith

From tap-l-owner@lists.ncsu.edu Tue Aug 31 12:19:04 2010