Calling all math/statistic geeks.... I NEED HELP!!!!

[oogieboogie]

Can someone explain to me what standard deviation is and it's applications?

My coworker is working on a project and her boss wants her to do a standard deviation from number of accounts and their corresponding balances. She came to me cause she was stumped and me being a polysci major in college am just as stumped.

I've read some of the definitions and they're just confusing me even more. So, if any of you could explain what it is in layman's terms, that would be so so wonderful.

Thanks

 

[caitycaity]

standard deviation measures the variability of the data. it helps you to "see" what the data looks like.

on a "normal" distribution, 68% of the data will be within one standard deviation of the mean, and 95% will be within 2 standard deviations of the mean. you can compare the standard deviation of your data against these benchmarks to see what the distribution of the data is like.

does that help?

 

[Bob Slydell]

What caity said -- that's about as clear of an explanation as there is.

 

[cynsaun]

I'm sorry, Oogie. You lost me after you said math!


Sure, Bob, it's clear as mud!

 

[Miss Jasmine]

Ahhhh statistics. That was a great explanation Caity Caity!

 

[winniedapooh]

Warning LONG!


OK let's hope I get this right. The square root of the variance. I am going to copy a simple example from an old text so that you can cut and paste and give it to your friend.

8 steps- example is the distibution gives the number of years of employment of 6 employees

years of employment agency-1,2,4,8,10,11

this can be done in column form

step 1 list values of distribution
ex:
1 - 6 = -1 1
2 - 6 = -1 1
4 - 6 = 0 0
8 - 6 = 0 0
10 - 6 = 1 1
11 - 6 = 1 1

0 4 4/6=.67 Square root of .67=.72 (Sd)

step 2 compute the mean of the values from step 1
step 3 list the mean in column b (that is the number 6 above)
step 4 subtract the mean from each value in column a and place this value in column c
step 5 square each value in column c and place this value in column d
step 6 compute the sum of the squares in column d
step 7 divide the sum of the squares in column d by the total number of values minus one (for simple data) or simply the total number of values (for population data) in column a
step 8 compute the square root of the number computed in step 7 (called the variance) this results in the standard deviation fo the values from column a

It really is easy and can be fun to do.....hope this helps! If she has too much data may want to see if she has a stats package available at work eg SPSS

 

[caitycaity]

Originally posted by Miss Jasmine
Ahhhh statistics. That was a great explanation Caity Caity!

thanks. it helps that i just took 2 semesters of Ph.D. level statistics.

ETA, if you don't have a statistics package, you can run basic statistics in excel.

 

[mickeyboat]

Huh?





Denae

 

[DisneyVillain]

A small standard deviation tells you that the numbers are bunched up around the mean (average). A larger standard deviation means the numbers are more spread out.

Standard deviation = usual difference (from the mean)

 

[DisneyVillain]

See! I CAN make a post with no satire or sarcasm. (I just don't enjoy it as much.)

 

[Cindy B]

I'm impressed caity caity! I do remember the definition,but couldn't word it as well as you did.

 

[cardaway]

Caity's is right on but this site's graphics and examples may be of help.

http://infotech.indiatimes.com/articleshow/452423.cms

 

[oogieboogie]

Wow, you guys are awesome. Maybe it helps if I show what numbers we're working on:

Background: I work on customer data for a bank. We have a free checking product. We're in the process of tweaking said product. We know how many accounts are in the product. We know what the balances are of those accounts.
What her boss wants is a standard deviation graph that will tell him how many people are on the edges of the bell curve because that will be the ones he wants to target for the tweak.

In this case, lets say I have 20,000 accounts. Can I break these accounts by 1000's and run a standard deviation on the number of accounts? Then do the same for the balances? So that I would end up with 2 graphs in Excel?

Is this the case? I totally suck at math and stats, and I haven't really touched it since college (almost 20 yrs ago).

Caity, that explanation was great. I'm just confused on how to apply it to my problem.

Thanks

 

[DISJAZZ]

I'm not sure that you can run a standard deviation on the number of accounts. If you have 20,000 accounts, then you have 20,000 accounts. What would be the mean for 20,000? I would focus on the standard deviation for the amount of money in the account. This sounds like an interesting application of statistics. Good luck!

 

[Bob Slydell]

DISJAZZ is right -- doing a Std Dev on merely the number of accounts isn't going to show you anything. You could, however do one (or both) of two things -- run a Std Dev on the balances in all the accounts (which would give you an average balance a standard deviation and then the number of accounts within one or two Std. Dev. from the mean (average).

The other thing you could do would be group the balances into classifications ($0-$500, $501-$1000, etc.) and then do some sort of bar chart showing the distribution of accounts by level of their account balances.

Maybe that will help? Good luck.

 

[tcwitt]

You guys are scaring me. I have to take statistics this coming year and it sounds very confusing. Please tell me that it's not as confusing as you guys are making it seem. I may just cry.
I'm usually very good at math but this just seems worse.

Thanks,
Christine

 

[oogieboogie]

Update:

I told my co-worker what you guys said and we've come to the conclusion that a standard deviation project is not feasible for the type of report we're trying to do.

So.. no more banging my head on the wall!

 

[DISJAZZ]

It sounds like the boss is in the process of making a decision. If you're trying to go above and beyond the call of duty, try providing this additional information. Talk about the total number of accounts. If I remember right, this is "the population". Discuss the Range of the population (lowest number - highest number). State what the Mean is. Ask the boss what his/her hypothesis is that he/she is gathering data for. I wish my bosses would try to use statistics. They all "just know" the answers (even when they're wrong). Good luck!

 

[Miss Jasmine]

Originally posted by caitycaity
thanks. it helps that i just took 2 semesters of Ph.D. level statistics.

ETA, if you don't have a statistics package, you can run basic statistics in excel.

YIKES. I was happy to make it through my undergrad statistics class and my Public Relations research class (where we had to use statistics of course).

tcwitt, I am not great at math but I did great in these classes. Go to class, pay attention and do the work. Don't worry about it.