CO Estimated Harvest Stats are up

[ImBillT]

I actually don’t know on what level it’s randomized. I’m guessing it’s CID number or hunt code.

It’s likely not wildly inaccurate, for most hunts, regardless of how exactly CPW is doing it, but depending on what exactly they’re doing, there may be some simple changes that would greatly increase their statistical power without actually making any big chances to their process or costing them any money.

 

[elkduds]

Here is an excerpt from CPW Roundtable discussion of big game harvest survey methodology, 7/22/23:

The annual big game license setting timeline begins in winter when hunting seasons are still
ongoing. The harvest survey is sent out to 80,000 hunters at the end of the season. By March
and April, CPW biologists are receiving and processing data to begin population modeling on
which to base next year’s license quotas. In mid-April, a sneak peek is made publicly available
and in early May, big game draft quotas are presented to the PWC for public input and
finalization. In short, it is a quick timeline, with little room to release quotas earlier in the
year.
CPW would like to give hunters as much notice as possible to plan their hunts. However, due
to the compressed timeline, the current system provides the most notice that biologists and
managers can give. License numbers cannot be drafted until May. Mandatory reporting would
push the timeline later and CPW data analysts must work with the data before sending it to
the biology team. The process is mainly slowed by the timing of late season data.
Currently, 80% of hunters respond to the harvest survey. Moving to a mandatory survey would
not improve CPW’s data — for example, New Mexico receives an 85% response rate, but then
must try to assess non-responses.

 

[ImBillT]

Here is an excerpt from CPW Roundtable discussion of big game harvest survey methodology, 7/22/23:

The annual big game license setting timeline begins in winter when hunting seasons are still
ongoing. The harvest survey is sent out to 80,000 hunters at the end of the season. By March
and April, CPW biologists are receiving and processing data to begin population modeling on
which to base next year’s license quotas. In mid-April, a sneak peek is made publicly available
and in early May, big game draft quotas are presented to the PWC for public input and
finalization. In short, it is a quick timeline, with little room to release quotas earlier in the
year.
CPW would like to give hunters as much notice as possible to plan their hunts. However, due
to the compressed timeline, the current system provides the most notice that biologists and
managers can give. License numbers cannot be drafted until May. Mandatory reporting would
push the timeline later and CPW data analysts must work with the data before sending it to
the biology team. The process is mainly slowed by the timing of late season data.
Currently, 80% of hunters respond to the harvest survey. Moving to a mandatory survey would
not improve CPW’s data — for example, New Mexico receives an 85% response rate, but then
must try to assess non-responses.

It’s sad that they need more than one data analyst for this.

NM gets 85% of EVERYONE. CPW gets 80% OF THOSE SURVEYED. I’m not sure what percentage of hunters CPW surveys, but it’s not 100%, and 80% of whatever percentage they survey should not be compared to 85% of everyone. Obviously one of CPW’s supposed “data analysts” did not write that paragraph. Either that or they’re obviously not qualified.

In defense of CO’s harvest reporting system, it’s entirely possible to have harvest stats that are accurate enough for our purposes without surveying everyone, or using mandatory surveys. I personally do not concern myself with how accurate CPW’s harvest statistics are. I assume they’re good enough for my purposes. Additionally, comparing to NM, they give us harvest data a year late. Harvest data that you can see in NM when you apply for the 2024 season is from 2022. The primary reason is that NM’s mechanism of enforcement for harvest reporting is denial of draw privileges the following year. You can report your previous year’s harvest all the way up to the last minute of the draw deadline. I don’t know if NM looks at harvest reports that were submitted early when making decisions. They definitely could, and probably should, as they could infer quite a bit from that information. They could even publish estimated harvest data from preliminary harvest report data if they wanted to. Still, I’m not dissatisfied with CPW currently, at least based on what limited information we have. If they completely randomly sample, then there are some definite holes. If they tailor the percentage of hunters sampled to the number of tags available, and do so properly, then their estimates are probably well within reason. I may be too trusting that CPW “data analysts” are doing things properly. These are government jobs after all. It’s unlikely that they graduated at they top of their classes. Those people probably analyze data in Vegas. Then again, I completed the requirements for a minor in math, dropped out to get married, and do lawn work for a living. WTH do I know?

 

[stealthy_bowman]

I really don’t understand why states use a few surveys to estimate harvest rates. It’s an archaic method with huge error bars. It’s 2024 already and the technology to have mandatory reporting through a cell phone has been around for at least 10 years, probably longer.
Here in NJ, there is mandatory reporting required for every deer killed. All one needs to do is log into a simple system via cell phone, report the kill and bingo, you are issued a confirmation number. Takes a whole 30 seconds to do.

 

[5yearcoueshunter]

I think it all depends on the sample size for accuracy. In some of my amateur's analysis I've done for my job, rule of thumb is 400 as a min sample size. So obviously a lot of the samples on these hunts are way too small because they give so few tags. I'm assuming that's why if you look at a trending chart on GoHunt, it looks like a ping pong ball going up and down. haha This will be my first year applying for Elk in CO. I did use the harvest report as a factor in picking a unit, obviously not the sole factor or even the main factor. The units that had a more stable harvest %, I assumed were more accurate but who knows.

 

[ImBillT]

Here is an excerpt from CPW Roundtable discussion of big game harvest survey methodology, 7/22/23:

The annual big game license setting timeline begins in winter when hunting seasons are still
ongoing. The harvest survey is sent out to 80,000 hunters at the end of the season. By March
and April, CPW biologists are receiving and processing data to begin population modeling on
which to base next year’s license quotas. In mid-April, a sneak peek is made publicly available
and in early May, big game draft quotas are presented to the PWC for public input and
finalization. In short, it is a quick timeline, with little room to release quotas earlier in the
year.
CPW would like to give hunters as much notice as possible to plan their hunts. However, due
to the compressed timeline, the current system provides the most notice that biologists and
managers can give. License numbers cannot be drafted until May. Mandatory reporting would
push the timeline later and CPW data analysts must work with the data before sending it to
the biology team. The process is mainly slowed by the timing of late season data.
Currently, 80% of hunters respond to the harvest survey. Moving to a mandatory survey would
not improve CPW’s data — for example, New Mexico receives an 85% response rate, but then
must try to assess non-responses.

Sooo. I actually just clicked the elk harvest report. The excerpt you quoted conflicts substantially with what is in the report.

On the first page of the report that isn’t an index or a map, it says that CPW surveyed the holders of 117,531 out of about 178,745 licenses sold. That’s a pretty high sample rate! They also claim a 49% response rate. That’s quite a ways from the 80% you found mentioned in an excerpt, but they still have a lot of data. They even list 95% confidence intervals on the DAU level. At a glance the intervals are not excessively wide. Just skimming the very first table in the report it seems to me that the numbers CPW is estimating on the DAU level are good enough that I wouldn’t feel like more accurate data would change where I decided to hunt.

Addressing the individual hunt code issue, again on the first real page, the report states that they used a stratified random sampling of licenses based on hunt code, residency, and age. While they don’t give us any details, they do say that the sampling isn’t completely random, but is stratified at the hunt code, and further divided by residency and age. This suggests to me that hunt codes with fewer tags get a higher sample rate, and that even if the residents don’t need a higher sample rate, they will increase sampling of non-residents or youth if the number of non-resident or youth tags indicates a need for higher sampling.

Page three, the first real page of the harvest report, gives me enough confidence in CPW’s survey results that I do not find them meaningfully inferior to any other states.

 

[ImBillT]

I think it all depends on the sample size for accuracy. In some of my amateur's analysis I've done for my job, rule of thumb is 400 as a min sample size. So obviously a lot of the samples on these hunts are way too small because they give so few tags. I'm assuming that's why if you look at a trending chart on GoHunt, it looks like a ping pong ball going up and down. haha This will be my first year applying for Elk in CO. I did use the harvest report as a factor in picking a unit, obviously not the sole factor or even the main factor. The units that had a more stable harvest %, I assumed were more accurate but who knows.

The required sample size to reach a certain level of precision. Obviously smaller data sets require lower sample sizes, but they actually require a higher percentage of the data to be sampled. I don’t remember any of that from my stats class, but I think that from what CPW has published it is reasonable to assume they’ve done things close enough to right that we shouldn’t be concerned.

My personal opinion is that CPW harvest data is plenty good for me, and I hope that anyone who thinks it’s inaccurate goes in some other direction from myself.

 

[Hyhunter]

Here's something that always confuses me on harvest stats each year that maybe someone can explain or tell me I'm wrong. I'm not a turkey hunter. Turkey harvest stats indicate Lake County to be one of the far best counties in the state for harvest, despite its size. Turkey distribution map shows there's barely any there. I've certainly never seen a turkey on mt Elbert or massive

 

[Smtn10pt]

My experience with CO harvest numbers is limited to looking up a unit that I had one of 8 deer tags in. I hunted with two other tag holders and all 3 of us tagged deer. When the numbers came out it only listed the unit as having 2 deer killed by non residents during that hunt.

 

[5yearcoueshunter]

The required sample size to reach a certain level of precision. Obviously smaller data sets require lower sample sizes, but they actually require a higher percentage of the data to be sampled. I don’t remember any of that from my stats class, but I think that from what CPW has published it is reasonable to assume they’ve done things close enough to right that we shouldn’t be concerned.

My personal opinion is that CPW harvest data is plenty good for me, and I hope that anyone who thinks it’s inaccurate goes in some other direction from myself.

Makes sense. Some of the YoY trends just look kind of wonky, but maybe they're accurate based on weather or other factors. Either way, hope I picked a decent unit and point creep/lower NR quota won't bite me.

 

[ImBillT]

Makes sense. Some of the YoY trends just look kind of wonky, but maybe they're accurate based on weather or other factors. Either way, hope I picked a decent unit and point creep/lower NR quota won't bite me.

Another thing to remember is that harvest stats can be correct while painting an inaccurate picture. Smaller numbers of tag holders not only require a higher percentage of the tag holders to be sampled, perhaps 100%, but the actual harvest can also swing wildly year to year. 3-4 guys with access to hunt or drive through private might draw one year and really bump up the harvest stats. A group could have an illness or a break down and not even hunt or barely hunt. A weather system could push animals down onto public early. Migration hunts often have more variation than hunts focused primarily on resident deer or elk. I’m sure there are other reasons.

There’s a wilderness unit in NM that consistently has very low harvest success, and good draw odds but one year a big snow storm came early and pushed all the elk down low and into the open. That unit, for that week, had one of the best harvest successes of any hunt code in the state. The draw odds plummeted to 1/3 of what they had been the year before those harvest stats were published. A) People didn’t look at previous year’s harvest stats or understand the unit. B) Because NM gets harvest reports just before the draw deadline, those published harvest stats weren’t even from the prior year. They were two years old. The following season, which took two years to get data on, had one of the lowest harvest successes for a number of years, probably because a large percentage of tag holders had no idea what they were getting into and expected an easy hunt since it had one of the highest harvests in NM the year they checked.

 

[elkmo]

Assuming responses are honest. Why inflate success rate when it can hurt future draw odds?? Some of the uninformed take success rates as gospel.