I need help with data!!

#1
Hi! I am new to the board. I sure wish I had found this at the beginning of the summer....I am in the last week of my internship, and I need to finalize the analysis of a bunch of data.

Here's the scoop:

I have been cycle testing scissors. What we're looking to do is compare data from the first cut vs data from the last cut (which may be after 2,500 - 10,000 cycles). Testing is done via a machine, so it is consistant. Also, we would like to compare data from one pair of scissors to another (ie US made vs identical model made in China).

The data: the force to cut paper is measured, then plotted (in Excel) via the machine controller.

I have only taken 1 semester of Stats, so I'm not exactly sure what to do here. So far, I have toyed with the following----paired t-test, and ANOVA single factor. Keep in mind, the only tool I have for analysis is the Data Analysis Toolpak in Excel.

I'll try and attach an Excel file.

Does anyone have any suggestions??? It would be of great help!! :)

Thanks,
Jason
 

JohnM

TS Contributor
#2
What does your supervisor say would be the most effective way to analyze and present the data?

In my opinion, your graph says a lot more than what you could produce with the outputs of statistical tests.
 
#3
JohnM- Thanks for your post.

Well, nobody really has an opinion, except for one engineer asked me to compare the area under each plot.

The positive slope to the plots is due to the mechanical advantage of scissors...takes more effort to cut at the tips. Simply showing these plots to marketing/sales may not be effictive, as they likely won't understand the concept, etc.

I'm curious is there is a good way to evaluate the "spread" between the first and last cut data.

The key data they want is the comparison of US -vs- Chinese made (same model, just made at new supplier in China...moved mfg overseas). My thought was:

1) subtract the 1st cut data from the last cut.
2) compare US/China via t-test or ANOVA....saying that there is or is not a statistical difference (small p-value).

sound confusing? on the right track? or too much?
 

JohnM

TS Contributor
#4
I think you're on the right track. Try your t-test idea.

To measure the spread between the plots, use Excel to compute the area under each, and compute the difference - just like in calculus where you would use a computational method (i.e., trapezoidal).

However, if you're going to be successful in the working world, you'll need to figure out a way to communicate technical concepts to non-technical people.

I would at least include 1 graph in your analysis - I'll bet they'll understand it if you explain it the way you explained it to me - I mean, who hasn't used scissors? Of course it makes sense that it's easier at the beginning of the cut and then becomes "harder" as you approach the tips.