2
$\begingroup$

I am sure that this has been addressed in other threads, but after half an hour or so of looking, I thought I'd just ask. My apologies in advance.

I am currently looking at suicide rates/100,000 pop by county in a state. The county pops ranges from almost 1 million to ~1,500. As you can imagine, the rates in the smaller counties are highly variable. 3 suicides may result in a rate of 106/100000. I am comparing these rates across counties.

A) Should I weight my data to address this volatility? B) What would be the best way of going about this?

Data are: 36 observations of suicide rate/100,000; 36 observations of county population

edit: This is part of a larger project creating a time-series animation of rates for counties over a 15 year span. Basically, each slide will be a heatmap of suicide rate of each county in the state. I'm trying to create a graphical representation of where rates have been high over time. Would smoothing be useful in creating a more faithful representation?

edit: Here's a version of the time series graphic I'm working on. I would like to remove the volatility of the smaller counties - i.e. make it look less jumpy with the extremes brought more into line. I am relatively inexpert when it come to hard modelling in r, so something simple perhaps if it can retain the overall integrity of the data?

Rate of Suicide by Firearm, Oregon, 2001-2015

Improve this question
$\endgroup$
4
  • 2
    $\begingroup$ Because "looking at" is a vague statement of your objectives, could you elaborate on what you're trying to accomplish? $\endgroup$
    – whuber ♦
    Jan 5, 2017 at 20:56
  • $\begingroup$ This is part of a larger project creating a time-series animation of rates for counties over a 15 year span. Basically, each slide will be a heatmap of suicide rate of each county in the state. I'm trying to create a graphical representation of where rates have been high over time. $\endgroup$
    – NFP
    Jan 5, 2017 at 20:59
  • $\begingroup$ Please include that important information within the post itself: it opens up the possibility of many more solutions, such as smoothing the series over time. $\endgroup$
    – whuber ♦
    Jan 5, 2017 at 21:02
  • $\begingroup$ Please register &/or merge your accounts (you can find information on how to do this in the My Account section of our help center), then you will be able to edit & comment on your own question. $\endgroup$
    – gung - Reinstate Monica
    Jan 6, 2017 at 17:31

1 Answer 1

Reset to default
0
$\begingroup$

Realized that I just needed to normalize my data. Utilized the answer from here. That did the trick! Stats skills are definitely in need of a refresher...

Improve this answer
$\endgroup$
2
  • 1
    $\begingroup$ Unfortunately, normalizing the data does not solve the problem you asked: the counties with smaller populations will continue to exhibit greater volatility. $\endgroup$
    – whuber ♦
    Jan 8, 2017 at 18:30
  • $\begingroup$ Was weighting a better solution, then? $\endgroup$
    – NFP
    Jan 8, 2017 at 19:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.