Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: Collapsing data to daily data
From
Brigham Whitman <[email protected]>
To
[email protected]
Subject
Re: st: Collapsing data to daily data
Date
Wed, 2 Mar 2011 14:53:01 -0500
Previous research has indicated that a snow depth over of 38 cm (15
in) plays a role in increasing deer mortality and prompting deer
migrations to winter yards. How the deer themselves measure the snow
depth, I do not know.
On Wed, Mar 2, 2011 at 2:27 PM, Nick Cox <[email protected]> wrote:
> Interesting.
>
> Not pertinent really, but still intriguing: 38 cm (well, 381 mm) looks to me like 15 inches in disguise. Whether deer use metric or non-metric units in reacting to snow is no doubt a subject for future research. Perhaps this means that there is some device that takes up to 15 inches' worth of snow, after which it is deemed to overflow.
>
> Nick
> [email protected]
>
> Brigham Whitman
>
> Yes, at first I wasn't going to collapse my data for a survival
> analysis. It is a huge data set (68 individual white-tailed deer with
> a total of 72,000 data points with up to 4 data points per day per
> deer), but Stata and R could handle it fine. But my advisors for my
> thesis said I should collapse the data, and another man I contacted
> who has done the same analysis suggested I should collapse it, so I
> did. That man had mentioned something about how I might inadvertently
> be increasing the amount of time each animal is at risk if I use
> multiple locations per animal per day. I can't say that I totally
> understand why that is. It does make more sense to me now to look at
> daily data, and I've incorporated a 'Cumulative days of snow depth
> over 38 cm' variable and running averages for certain variables over
> 14 and 28 day moving windows, which I am not sure how I would do if I
> hadn't collapsed the data. But yes, I should have a better idea of
> why I would "coarsen" my data and lose some of that accuracy. Thank
> you for the input.
>
>
> On Tue, Mar 1, 2011 at 9:44 AM, Nick Cox <[email protected]> wrote:
>
>> I don't see that you need to coarsen your data for a survival analysis -- unless the sheer size of the dataset is problematic.
>>
>> It could even be that time of day data has a secondary bearing on survival...
>>
>> Nick
>> [email protected]
>>
>> Brigham Whitman
>>
>> Yes, these options worked out for me, thank you.
>>
>> I am collapsing the data set to prepare it for a Cox proportional
>> hazard model. Each data point already has a few variables (distance
>> to cover, distance traveled, snow depth encountered) with measurements
>> for each point that I can average together for each day before I make
>> the model.
>
> *
> * For searches and help try:
> * http://www.stata.com/help.cgi?search
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/