[macintologist]

I managed a paper-based survey at my university so I have about two or three hundred little pieces of paper with the answers that each person gave.

I was wondering if someone here could point me in the right direction as to how to enter all this information into some software and then be able to analyze it multiple ways.

Let me give you an example:

There are multiple questions on the survey, the first one asks, what class year are you? The second question asks what gender are you?

So what I want to be able to do is, once I've entered all the information, how can I manipulate the data so if I wanted to find out how many of those who answered "Freshmen" on the first question answered "Male" on the second question, rather than just tallying up all the answers into one big spreadsheet and not being able to analyze the data from multiple angles.

Another example of what I want to be able to do: Of the total 300 respondents, 40% answered that they were "very satisfied" with their classes. However, of all the Freshmen who took the survey, only 5% answered that they were "very satisfied" with their classes.

That's something that you can't do with just tallying all the answers into Excel, am I right?


I hope this is making sense.... I'm not asking anyone to hold my hand, I just need some pointers as to what software to use, I can figure out the rest on my own, I just have no idea how to start.

Thanks all!

-Mac

[RAILhead]

Yes, Excel can do all that.

[macintologist]

If I were to google around for a tutorial on this... what search terms do you recommend?

[turtle777]

Pivot tables will be a good Excel tool to analyze the data.

But first, you'd need to enter the information in Excel. Do it in a structured way, this will help you to sort and analyze it later on.

-t

[RAILhead]

Pivot table, autosort, all kinds of stuff. There's not really a tutorial -- it's just how you choose to pull the data.

[turtle777]

You could use 3rd party apps.

http://www.surveyminer.com/

The demo version actually might be good enough for you.

-t

[Doofy]

I'd have to resort to good old Access for stuff like this. But then I am teh lose at survey stuff.

[mduell]

A database is the right way to do it. Excel might be enough for your purposes.

[klb5090]

depending on the type of analysis your going to do, for most of my economic statistical analysis courses we use STADA, extremely simple but more powerful than most people will ever need.

[SVass]

Any flat file database is easier to use than a spreadsheet.
The simplest one is part of AppleWorks and is directly accessible from its spreadsheet.

The free beta of StarOffice also seems to have a data base component.
sam

[BRussell]

A real stats program will be a lot better than excel at those crosstabs (and a database would be worst of all). Your university probably has something like SPSS. I'd go to a social science-type and ask them. They'd be able to show you how to analyze it in like 10 minutes.

[macintologist]

Crosstabs.... now we're getting closer to the kind of jargon that identifies exactly what I'm looking for

[torsoboy]

Originally Posted by BRussell 

A real stats program will be a lot better than excel at those crosstabs (and a database would be worst of all). Your university probably has something like SPSS. I'd go to a social science-type and ask them. They'd be able to show you how to analyze it in like 10 minutes.

Do you even know what a database is?? Your "real stats program" almost assuredly uses a database when it is creating its numbers and graphs.

[BRussell]

Originally Posted by torsoboy 

Do you even know what a database is?? Your "real stats program" almost assuredly uses a database when it is creating its numbers and graphs.

"Using a database" and "being a database" makes all the difference in the world.

[turtle777]

Originally Posted by BRussell 

"Using a database" and "being a database" makes all the difference in the world.

Seconded.

The learning curve for typical database apps (like Access) is much steeper than for Excel or some statistical analysis app.

If you can make it "less databasy", it'll help many users to get the results quicker. Doesn't really matter what the app is doing in the background.

-t

[philm]

I do a lot of this sort of thing, and my view is that an Excel Pivot Table is by far the best option. They are easy to set up, highly customisable and very dynamic. Once set up, you can drag fields into, out of and within the table to get precisely the sort of information you are looking for.

You need to ensure that your data are set up correctly. All you need to do is to ensure that your table has a single row for the data titles, not two or more rows like you sometimes see. Simplicity is the key here. Then, each respondent needs to occupy only one row. Something like this:



Then, you select the columns of data and start the Pivot Table, using Data...Pivot Table Report in Excel. It takes you through a couple of steps and then you are presented with an empty Pivot Table:



and a small dialog with your fields:



Drop the fields into the table and you are ready to go.

You need to get the fields in the position you want, and then you can move them around to get exactly the data you need, as below:











In this table, I have the data variable ('Satisfaction') set to present as a percentage of each row. You can have the raw count instead, or another of other options. These are accessed by hitting the little yellowy square on the Pivot Table dialog box.

Enjoy.

(Note: you'll see that I have the three 'years' as first, second and third. This is how we do it in the UK. I know that in the US it's all very complicated with Freshmen year, junior years, something about sorority, all sorts of pseudo-Greek letter things (beta alpha kappa) and a frat house. Basically, my understanding comes from watching Animal House in about 1985. Anyway, please convert our simple 1-2-3 method to your approach).

[brassplayersrock²]

Thank you very much philm. I'm about to do some database things as well, so thank you very much for your very detailed post.

[philm]

Originally Posted by brassplayersrock² 

Thank you very much philm. I'm about to do some database things as well, so thank you very much for your very detailed post.

Much appreciated! Cheers.