Jump to content

Sign up for a free account!

Signing up for an account is fast and free. As a member you can join in the conversation, enter contests to win things like this Logitech Harmony Ultimate Remote and you won't get the popup ads that guests get. Click here to create your free account.

- - - - -

Need ideas for a statistics paper topic.

This topic has been archived. This means that you cannot reply to this topic.
19 replies to this topic

#1 of 20 Noah Gottula

Noah Gottula

    Stunt Coordinator

  • 96 posts
  • Join Date: Jan 20 2003

Posted October 20 2003 - 06:26 AM

Im taking an econometrics class in which we have to write a 5 page paper. In the paper I need to use various statistical methods such as regression. Past papers have been written on topics such as forecasting church attendence, modelling occupancy rates in residence halls, forecasting the demand for lift tickets at certain mountains, studying cable deregulation, forecasting box office revenues, and estimating the density of Douglas fir trees.

I need a good idea for a topic. Especially one thats easy for me to find the statistical data for (finding it on the web would be preferred). He's already stated he doesn't really like sports-related papers, so thats outta the question. But does anyone else have any interesting ideas?

#2 of 20 Mike Wladyka

Mike Wladyka

    Supporting Actor

  • 630 posts
  • Join Date: Jul 24 2003

Posted October 20 2003 - 06:40 AM

there has got to be plenty of traffic studies out there
Now they show you how detergents take out bloodstains, a pretty violent image there. I think if you've got a T-shirt with a bloodstain all over it, maybe laundry isn't your biggest problem. Maybe you should get rid of the body before you do the wash.

#3 of 20 Eric Brunton

Eric Brunton


  • 40 posts
  • Join Date: Jun 18 2003

Posted October 20 2003 - 06:50 AM

How about school attendance? It would be interesting if the data is available to see the trends over time.

#4 of 20 Ted Lee

Ted Lee

    Lead Actor

  • 8,398 posts
  • Join Date: May 08 2001

Posted October 20 2003 - 07:52 AM

how about the number of starbucks coffee cups being thrown away each day? i always wonder that everytime i buy from there. Posted Image

#5 of 20 David Baranyi

David Baranyi

    Stunt Coordinator

  • 81 posts
  • Join Date: Jul 17 2003

Posted October 20 2003 - 08:03 AM

I may suggest national crime statistics.

#6 of 20 AndyDC



  • 38 posts
  • Join Date: May 30 2003

Posted October 20 2003 - 08:35 AM

The last one is not a bad idea. I'm sure it's been done (I'm an economist, not someone who studies crime), but I once played around with some data on crime rates by smsa (standard metropolitan statistical area, I think). From the census, on line, you can get data on crime rates, density (pop/area), measures of average income, poverty rates, ethnicity, etc. and you could regress crime rates on these things. Check for "nonlinearities", i.g. whether say you need to put in income squared as a separate regressor. I think there are a few hundred of these smsas, and the online data I saw was in pdf format, so you have to type it in. But you could randomly pick 100 smsa (e.g. alphabetically) and use that as a sample and type in the data. If you want to get fancy, you could type in 10 more and see if your model works "out of sample". You could also test whether the first letter of the smsa matters for crime rates, testing your assumption that your sample of 100 is random.

#7 of 20 Lew Crippen

Lew Crippen

    Executive Producer

  • 12,060 posts
  • Join Date: May 19 2002

Posted October 20 2003 - 08:47 AM

Movie Theatre Attendance

Lots of data and lots of opportunities for different ways to look at the data: total tickets, dollars at the box office, adjusted dollars (for inflation, or CPI or whatever), domestic (U.S.), foreign, etc. You should have about 100 years of data available.

Factor in TV, VHS rentals, DVD sales and competing entertainment at various points in time as you consider trends and anomalies in the trends.

This could be a 50 page paper.
¡Time is not my master!

#8 of 20 Paul_Sjordal


    Supporting Actor

  • 831 posts
  • Join Date: May 29 2003

Posted October 20 2003 - 09:14 AM

Where do you work right now?

That is probably your easiest source of data (provided your employer agrees to let you do this), and more importantly, you'll likely understand all the nuances of what those numbers represent, which gives you an edge in interpreting the data.

For instance if you work for a pizzeria, you could track the rate of consumption of 50-lb bags of flour. Or if you worked for a surveillance company you could track the rate at which surveillance reports come in from the field. You'd be amazed at what such innocuous numbers can tell you about what's going on in the business in question (I was once able to use such data to identify the dates of a particular manager's vacation). When looking for numbers, look for something the business tracks on a daily basis, and try to get numbers going back at least 3 to 5 years.

If that doesn't work out, look to the bureaucrats at your school. You'd be surprised what kinds of things they keep track of.
“It’s great to be known, but it’s even better to be known as strange.” —Takeshi Kaga

#9 of 20 Noah Gottula

Noah Gottula

    Stunt Coordinator

  • 96 posts
  • Join Date: Jan 20 2003

Posted October 22 2003 - 06:26 AM

Thanks for all the help everyone

I work at the university archives, so there isn't much there that I can think to use for a report.

So far im thinking the easiest would be the crime rate suggestion. Movie theater attendance might be a little more entertaining but its getting a little close to someone else's papers which was forcasting box office revenues.

I haven't totally decided yet though, so im still open for suggestions.

#10 of 20 Mark Dubbelboer

Mark Dubbelboer


  • 1,011 posts
  • Join Date: Oct 06 1999

Posted October 22 2003 - 09:20 AM

how about sporting attendance and seeing if there's a trend with winning percentages of the franchises.
both sets of data would be very easy to obtain on the internet, and if you looked hard enough you could probably find it in spreadsheet form somewhere as well.

#11 of 20 Ashley Seymour

Ashley Seymour

    Supporting Actor

  • 945 posts
  • Join Date: Jun 29 2000

Posted October 22 2003 - 10:39 AM

For all the data available on crime rates, the age make-up of the population should have a high statistical level of confidence.

If a city had a median age of its population of 28, I would assume it would have higher crime levels than a population with a median or average age of 32.
Go Orange, Go Blue,
Fight Fight Fight BSU!

It's not whether you win or loose that counts, but whether you win.

#12 of 20 Eric_L



  • 1,991 posts
  • Join Date: Nov 02 2002

Posted October 22 2003 - 11:55 AM

How about a comparative statistic on taxes vs poverty.

Do states (or counties, cities or municipalities) with higher tax rates have more or less poverty than those with lower rates?

Compare the budgets of the high and low tax specimens and see which allocate a larger portion to social programs - both as a percentage of the budget as well as per capita.

Also worth considering is where the poverty is 'deeper' (lower average income below the poverty level) high or low tax areas?

I personally would be facinated with your results.

#13 of 20 AndyDC



  • 38 posts
  • Join Date: May 30 2003

Posted October 23 2003 - 05:13 AM

The last two suggestions make for an interesting comparison, from the point of view of feasibility of the project.

The first (age and crime) is pretty easily testable, and the data are fairly straightforward. Also, not too much risk that you are picking up reverse causality (from crime to age), though one could maybe tell a story along those lines.

The second (tax rates and poverty) is very interesting, but much trickier. The suggestion is (sensibly) about looking at correlations. But for a regression (which he should probably do for his paper, I gather) he would have to take a stand on causality.

First measurement. State tax systems vary a lot, so how do you characterize overall low and high rates (e.g. if the business income tax rate is high but the personal income tax rate is low)? You can't use revenues/population or something because that will of course depend on poverty itself and related characteristics. Then, causality. One could easily tell stories either way about how poverty is a function of tax rates (say through effects on business climate or work effort) but also how tax rates are a function of poverty (e.g. states with lots of poor try to raise more revenue to redistribute it to the poor). If all these things are going on at once to varying degrees, it will be hard to make much of your results.

If you go the poverty/tax route, some potential problems are not so much reverse causality as spurious correlation and could be dealt with by including more variables. For example, it would be important to control for the level and change in GDP, since maybe a recession in the state causes both more poverty and higher tax rates (due to balanced budget requirements). Also, the demographic characteristics of the state may determine both poverty rates but also political allegiance and hence attidutes towards taxes, so you could include those too.

#14 of 20 Mark Dubbelboer

Mark Dubbelboer


  • 1,011 posts
  • Join Date: Oct 06 1999

Posted January 21 2004 - 05:37 PM

I got a similar paper this semester. how did yours go noah? any suggestions?

and AndyDC I think i accidentally sent you a pm i was trying to send to Noah. whoops, sorry about that.

#15 of 20 AndyDC



  • 38 posts
  • Join Date: May 30 2003

Posted January 22 2004 - 02:18 AM

OK, here's an idea tht goes with the forum. Why don't you see if you can estimate the correlation between different critics ratings' and box-office success. Use various critics' ratings (they would need to have some sort of star or grading system) prior to opening and look at box office after say 12 weeks. You could

= ask whether on average "good movies" sell more tickets
= ask about which critics seem to predict sales best, which would mean either (1) they have the most influence or (2) they are best at predicting what people will like/have the most common taste.
= You could try to distinguish among the two theories above by asking whether say sales in Washington DC are relatively high (compared to the national average) for movies which the local critic likes especially.
= run a more general equation of:
sales = f(marketing expenditure, critics ratings, box office gross of last movie by main star, etc etc.)

Doing all of this would be overkill for your paper, that's for sure. But I think in general the statistics would be fairly easy to process and the data collection would be a bit onerous but not too bad (and maybe intersting, if you're into movies).

This suggestion is free, but if you take it, I want to see the final paper. :-)

#16 of 20 Mark Dubbelboer

Mark Dubbelboer


  • 1,011 posts
  • Join Date: Oct 06 1999

Posted January 22 2004 - 06:14 AM

Wow. that's good. And i've got like 3months to get this together so that would be about perfect.
You're gonna have to e-mail me your real name so I can cite you now.Posted Image
I never actually read critics reviews, does anyone have links to "well-respected" critics?
I also have no idea where to find marketing expenditure...or even box office information.

What have I been doing on the internet all these years anyway?!?!

#17 of 20 Grant B

Grant B


  • 3,213 posts
  • Join Date: Mar 29 2000

Posted January 22 2004 - 07:14 AM

My brother-In-Law use to work for the California Prisons as a statistian. Every year he would have to show how crime went up in the state so the prisons could ask for more money even though crime was decreasing. He would find some way and they loved him for it. He couldn't take it after a while and quit. An interesting subject."How to fudge the facts to keep your job and lose your soul".
"Whatever it is, I'm against it!" G. Marx

Sony CDP-CX270 CDP-CX355 CDP-CX270 CDP-CX355
Sony S-TS 700ES XPR32450Pioneer Elite CLD 79 KEFs (5) Q75s & 20B Subwoofer

#18 of 20 Noah Gottula

Noah Gottula

    Stunt Coordinator

  • 96 posts
  • Join Date: Jan 20 2003

Posted January 22 2004 - 07:57 AM


After taking some of the advice here, my topic ending up being based on home values. Basically what I did was created a regression that predicted average home values of a MSA as suggested above. I used around 9 or so variables including crime rates, median age, median income, population density, etc.

Doing one on movies sounded like a good idea to me, but someone in a class before had done something very similar to that, so I didn't wanna come off as copying them.

I can send you links to mine and other papers if you want some examples.

#19 of 20 Mark Dubbelboer

Mark Dubbelboer


  • 1,011 posts
  • Join Date: Oct 06 1999

Posted January 22 2004 - 01:09 PM

Noah, can I see those links.
Plus if that classmate of yours that did the box office paper has any recommendations that would also be greatly appreciated. thanks

#20 of 20 AndyDC



  • 38 posts
  • Join Date: May 30 2003

Posted January 23 2004 - 02:14 AM


No need to cite. I'm sure you'll find by the time you're done that what you've done is somewhat different from these suggestions. There will always be unanticipated difficulties (or opportunities) that cause you to change course a bit.

I don't know much about data sources, but I guess I'd start with www.imdb.com which might have some of the sales info etc., and www.mrqe.com has a list (comprehensive?) of review articles available online for any movie you type in. Cool site.