The Last 10%

There was an interesting conversation a few days ago on the YesBiscuit! blog about whether the 90% live release rate standard for shelters isn’t enough and shelters should be saving a higher percentage. The conversation got me wondering whether the statistics for the 90%+ shelters listed by this blog could shed some light on the question. With almost 200 communities listed at this point, we have a pretty good sample size to work with. Quite a few communities have live release rates in the 95% to 100% range. So one question is whether there are characteristics of communities that affect how difficult it is to go beyond 90% to 95% or higher.

The first thing I checked was whether climate has any effect on the number of shelters at 90%-94% versus 95%-100%.  Over the years that I’ve been researching shelters I’ve gotten an impression that shelters in colder climates, on average, have higher live release rates than shelters in warmer climates. Even so, I was surprised at how great the discrepancy was when I ran the numbers. I found a chart that ranks each U.S. state by average temperature, then counted how many communities in the blog’s right sidebar are located in the warmest 25 states versus the coldest 25 states. Out of the 184 communities currently listed, fully 131 (71%) were in the 25 colder states, while only 53 (29%) were in the 25 warmer states. The 20 coldest states had a breakdown of 30 shelters that were at 95% to 100% versus 9 that were at 90% to 94%, whereas the 20 warmest states had 12 shelters at 95% to 100% versus 9 shelters at 90% to 94%. I don’t think these numbers are being distorted by any differential in north-south population distribution: check out this map, which shows a rather even population distribution between north and south. So, if you are running a shelter in a warmer state, you have your work cut out for you.

Another impression I had was that there were fewer large jurisdictions with 95% to 100% live release rates. We have to be careful in evaluating large versus small jurisdictions, though, because most jurisdictions are small (median size of 25,000 people). So, I looked at live release rates for communities under 25,000 in population versus communities over 25,000. I found a huge discrepancy in live release rates based on community size. The shelters listed by this blog that serve a population of under 25,000 people overwhelmingly have live release rates of 95% to 100% (21 to 2). Shelters in communities of over 25,000 people, by contrast, were almost evenly divided, with 23 having live release rates in the 90% to 94% range and 22 having rates of 95% to 100%. Overall, we have a striking pattern where smaller jurisdictions are more successful in getting to 95% and above.  (Interestingly, a previous statistical study I did showed minimal effect of population size on the ability of a shelter to get to the 90% threshold; thus, the effect of population size seems to pertain only to live release rates of 95% to 100%.)

One of the issues brought up in the YesBiscuit! discussion was whether it’s helpful for advocates to criticize shelters that have live release rates in the 90-95% range, or whether such criticism is setting the bar too high and actually hurting the shelter reform movement. Austin was brought up as an example of a community that has taken severe criticism from advocates for killing some animals that advocates believed were savable. Austin is in the deep south, and it is the largest city in the U.S. that has achieved a 90% or more save rate.  Austin’s live release rate was 91% in 2011. I’ve not been able to find full 2012 calendar year statistics for the Austin Animal Center, but Austin Pets Alive! reported that the city had close to a 95% live release rate in fiscal year 2011-2012. Given that Austin is a big city in the deep South, the city’s attainment of a live release rate of close to 95% seems to me like a pretty awesome accomplishment. In fact, it’s unprecedented. I can’t see how criticizing Austin does the movement any good.

Going beyond the head-to-head comparisons of shelter live release rates, there are several strategic considerations that are illuminated by the 184 successful communities identified thus far. One of those considerations is that the the U.S. population living in identified 90%+ communities is about 6.8 million, which is only a little more than 2% of the total U.S. population. We can probably safely double that number due to the fact that not all of the 90%+ communities have been identified, but that still leaves about 96% of the U.S. population served by shelters saving less than 90%. In these circumstances, if advocates claim that a standard that only about 4% of the population has attained isn’t good enough, it might discourage more communities than it encourages.

Another consideration is that raising the bar to 95%+ could serve as an incentive for shelters to report their data in ways selected to maximize their live release rates. As of now, there is no standardized reporting system for U.S. shelters. Shelters are free to choose how they report animals in foster care, cats who have received TNR, neonatal dogs and cats, etc. A shelter that wants to make its live release rate look as good as possible can do so without being dishonest or fraudulent, simply because of the lack of any set reporting standards. One concern I have is that if unrealistic standards are set for shelter performance, shelters may start to change their reporting standards to present a rosier scenario, simply out of fear that they will be dragged through the mud by local advocates if they don’t.

One final strategic consideration is that shelters that have highly effective pet retention programs will show somewhat lower live release rates than shelters that do not have such programs. This is because the animals that are retained by their owners (and thus not counted at all in shelter statistics) are ones that likely would not have been euthanized by the shelter and would have been live releases. Thus, the shelter is left with a population of animals that on average will need more rehabilitation and be harder to adopt, thereby having a higher euthanasia rate. If we insist on live release rates of 95% or more, it may have the unintended consequence of discouraging pet retention programs.

I draw two conclusions from the factors outlined above. First, there are several strategic reasons having to do with performance incentives to stick with the 90% standard rather than moving to a 95% to 100% standard. Second, it’s unfair to judge all shelters by the same standard. The fact that rural shelters in the upper peninsula of Michigan can frequently put up numbers in the 95-100% range does not mean that city shelters in the warmer parts of the country can also do so, at least under the current state of things.

Sticking with the 90% standard for now wouldn’t mean forever — once it becomes routine and expected for shelters to achieve 90%, then expectations could be increased. By that time we will have a much better professional infrastructure to help shelters, and communities will have been able to build large coalitions that will help them save the last 10%.


  1. Thank you for this very insightful analysis. It would be interesting to see if the larger no-kill communities slightly lower kill rates are due to lower adoption rates, shorter stays (due to smaller capacity), or reduced owner redemption. I noticed a pattern of lower owner redemptions in New Jersey for community shelters covering larger areas. However, this was not a formal analysis and did not involve no kill communities.

    • I think you mean slightly lower “save” rates, but the point you bring up about owner redemptions is something that’s extremely important. It makes sense that the bigger the area, the more difficulty people are going to have redeeming their animals, because the bigger the area the more organizations tend to be involved and the more ground there is for owners to cover. You’ve hit upon a key factor, because if it is true that redemptions are lower in larger jurisdictions, then that alone could be the explanation for why larger communities, although there is no difference in their ability to get to 90%, seem to have a harder time getting to 95%. When I get time I’ll check and see if I have enough data to do a statistical analysis of that question.

Leave a Reply

Your email address will not be published. Required fields are marked *