Sunday, November 22, 2009
s_pooled Revisited
Thursday, September 24, 2009
Spreadsheet Form Page
I thought I had fixed all of that about two weeks ago and that this continued problem was something we'd have to live with. Today I discovered I made a really silly mistake. And, of course, now it's been corrected! So starting today onwards those of you using XLSX files made from the newest version of Excel will be able to send me your work from the Spreadsheet form page.
Sunday, September 20, 2009
DevSq is the Word
Several of you have been posting questions about small rounding off errors. There are a few ways to handle this including: (1) willing to live with small rounding off errors when hand calculating, (2) using Raylene's super-smooth calculation trick (email/phone her), or (3) using your new found spreadsheet skills.
In the end, the real direction you want to go is to become comfortable enough with spreadsheets that you use them for all your basic calculation needs. (The notable exception is that you'll want to use dedicated statistical analysis software for big and complicated analyses.)
1. Spreadsheet Setup
You can easily calculate all of the homework Problem Set 5 answers using your current spreadsheet skills. However, I'm going to make it even easier! First, set up your spreadsheet for the calculations. You should have something like what you see at the top of this post. Note that I've filled in some cells, but left others empty for you to figure out. (All the shown calculations you've already calculated in previous problem sets.)
2. New Spreadsheet Command
The new spreadsheet command that makes life easier for you: DEVSQ. This is a command to figure out the Sum of Squares (or deviations squared) for a specified array of numbers. With this command you do not need to create Deviation and Deviation Squared columns.
Notice that you have three possible DEVSQ commands you'd want to use with the data presented above: SS for the female group, SS for the male group, and SS for the total sample.
3. Calculate
All that is left is to do the calculations. SS Total is very simple since it treats all the data as belonging to one group. SS Within is also straightforward since it's the same thing as adding up the SS for each of your groups. So the one place where you'll really be practicing your formula-creation-chops is figuring out SS Between. That said the formula for this is much simpler than the one used for s_pooled. Finally, eta^2 is easy to calculate once you have SS Between.
4. Spreadsheet Concepts Revisited
There's a conceptual difference between what you do by hand, and how a spreadsheet operates. When you round off then you end up using those rounded-off-numbers in subsequent calculations. In this way it's easy for hand calculations (even if done correctly) to vary a small amount from the actual correct answer. (Of course if your calculations vary a large amount then something went askew!) However, spreadsheets work differently. In essence, spreadsheets always use the full number in all calculations. However, with a spreadsheet you can choose to have only part of the full number (e.g. two decimal places) displayed on the screen. So with spreadsheets you get the best of both worlds: all calculations done with the full number, but your eyes only need to look at the nicely formatted two-decimal-place rounded off number.
Thursday, September 17, 2009
Surveys Open
It's probably easiest to answer all 5 at the same time. There will be some repetitious questions, but the data ends up going to different groups so please be understanding.
I have erased all submittals posted before today. There have been various small tweaks made to the surveys and they simply weren't ready for real submittals until this morning.
I have linked the five surveys. That means when you successfully complete Survey 1 you'll be taken to Survey 2 and so on. After completing Survey 5 you'll be taken to a general confirmation notice page. In total you can probably respond to all 5 surveys in less than 5 minutes. What a deal!
Please be honest when you are responding. I won't know the names of individual respondents, but I will have running counts of how many people from a few different courses have responded.