That simply isn’t true, and therein lies the source of your misunderstanding.

The required sample size depends on what question is being asked, and on the nature of the data.

For example, let’s consider a hypothetical breed, the Andean Shepherd Hound (ASH).
40 years ago, ASH had a median longevity of 7.0 years.
15 years ago, ASH had a median longevity of 10.0 years.

Now lets consider another hypothetical breed, the Tundra Hairless (TH).
40 years ago, TH had a median longevity of 9.5 years.
15 years ago, TH had a median longevity of 10.0 years.

What sample sizes for ASH and TH populations are required to determine these trends?

The required sample sizes for the ASH populations will be smaller than for the TH populations, because the longevity change for the ASH was larger than for the TH.

Beyond that, the plus/minus longevity spread in each population also affects required sample sizes.

I once did a statistical comparison of two dog breed populations, and it only required sample sizes of 30 each to reach the commonly accepted p less than 0.05 threshold for statistical significance. Increasing sample sizes beyond 30 each reduced the p value as expected, but it did not change the calculated difference between the two populations to any meaningful extent. This happened despite the fact that the calculated difference between these two breed populations was relatively small.

There is no simple answer to what the required sample size is for a dog breed longevity study. To use your phrase, it isn’t “statistically valid” to dismiss a study that sampled “tens” rather than hundreds or thousands.

That simply isn’t true, and therein lies the source of your misunderstanding.

The required sample size depends on what question is being asked and on the nature of the data.

For example, let’s consider a hypothetical breed, the Andean Shepherd Hound (ASH).
40 years ago, ASH had a median longevity of 7.0 years.
15 years ago, ASH had a median longevity of 10.0 years.

Now lets consider another hypothetical breed, the Tundra Hairless (TH).
40 years ago, TH had a median longevity of 9.5 years.
15 years ago, TH had a median longevity of 10.0 years.

What sample sizes for ASH and TH populations are required to determine these trends?

The required sample sizes for the ASH populations will be smaller than for the TH populations, because the longevity change for the ASH was larger than for the TH.

Beyond that, the +/- longevity spread in each population also affects required sample sizes.

I once did a statistical comparison of two dog breed populations, and it only required sample sizes of 30 each to reach the commonly accepted p<0.05 threshold for statistical significance. Increasing sample sizes beyond 30 each reduced the p value as expected, but it did not change the calculated difference between the two populations to any meaningful extent. This happened despite the fact that the calculated difference between these two breed populations was relatively small.

There is no simple answer to what the required sample size is for a dog breed longevity study. To use your phrase, it isn't "statistically valid" to dismiss a study that sampled "tens" rather than hundreds or thousands.

But sample sizes in the tens, are just laughable.

And a sample size of one, is just trying to make it look like you have more results than you really have.

Other breeds for which the “studies” had sample sizes under three digits. (I have included a few where two studies combined may be over three digits, but none where any single study was over three digits, or where there were three or more studies)

Leonberger 47 dogs (and another with 78 dogs)

Russian Black Terrier 4 dogs

Hovawart 19 dogs

Chinese Crested 14 dogs

French bulldog 71 dogs

Schipperke 36 dogs

Tibetan Terrier 95 dogs (an another with 7 dogs)

Affenpinscher 21 dogs

Brussels Griffon 71 dogs

English Toy Spaniel 22 dogs

Havanese 3 dogs

Japanese Chin 38 dogs

Lowchen 9 dogs

Maltese 42 dogs

Pomeranian 29 dogs

Silky Terrier 3 dogs

Toy Manchester Terrier 19 dogs

Toy Poodle 20 dogs (and another with 11 dogs)

Yorkshire Terrier 63 dogs

Basenji 46 dogs

Pharaoh Hound 15 dogs

Basset Fauve de Bretagne 15 dogs

Basset Griffon Vendeens (all sizes) 76 dogs (with a note saying see Petit and Grand)

Grand Bleu de Gacogne 6 dogs

Hamiltonstovare 6 dogs

Miniature Dachshund (all coat types) 21 dogs ( with note saying See also Dachshunds (all varieties))

Otterhound 41 dogs

Petit Basset Griffon Vendeen “N probably 45”

Smooth-haired Dachshund 75 dogs (See also Dachshunds (all varieties))

Wire-haired Dachshund 46 dogs (See also Dachshunds (all varieties))

Borzoi 87 dogs

Rhodesian Ridgeback 10 dogs

Sloughi 1 dog

Anatolian Shepherd Dog 23 dogs

Estrela Mountain Dog 3 dogs

Giant Schnauzer 39 dogs

Komondor 10 dogs

Maremma Sheepdog 19 dogs

Australian Cattle Dog 11 dogs

Australian Shepherd Dog 22 dogs

Briard 71 dogs

Canaan Dog 2 dogs

Cardigan Welsh Corgi 53 dogs

Polish Lowland Sheepdog 11 dogs (and another with 19 dogs)

Rough Collie 51 dogs

Swedish Vallhund 7 dogs

Welsh Corgis (both breeds combined) 13 dogs (See also Cardigan Welsh Corgi and Pembroke Welsh Corgi)

Bullmastiff 96 dogs (and another with 7 dogs)

Dogue de Bordeaux 5 dogs

Greater Swiss Mountain Dog 71 dogs

Mastiff 94 dogs

Tibetan Mastiff 10 dogs

Chow Chow 7 dogs (and another with 31 dogs)

Finnish Lapphund 5 dogs

Finnish Spitz 42 dogs

German Spitzes (all varieties) 43 dogs

Norwegian Buhund 17 dogs

Shiba Inu 3 dogs

Greenland Dog 2 dogs

Braco Italiano 1 dog

German Long-haired Pointer 2 dogs

German Wire-haired Pointer 41 dogs

Italian Spinone 47 dogs

Large Munsterlander 69 dogs

Small Munsterlander 31 dogs

Wire-haired Vizsla 15 dogs

Chesapeake Bay Retriever 45 dogs

Curly-coated Retriever 40 dogs

Portuguese Water Dog 6 dogs

Field Spaniel 68 dogs (and another with 71 dogs)

Irish Water Spaniel 95 dogs

Kooikerhondje 7 dogs

Sussex Spaniel 42 dogs

Miniature Bull Terrier 20 dogs (and another with 53 dogs)

Bedlington Terrier 48 dogs (another with 6 dogs)

Cesky Terrier 9 dogs

Dandie Dinmont Terrier 62 dogs

Fox Terriers (all varieties) 44 dogs

German Pinscher 24 dogs

Glen of Imaal Terrier 6 dogs

Russell Terriers (all varieties) 75 dogs (See also Jack Russell Terrier and Parson Russell Terrier)

Sealyham Terrier 12 dogs

Skye Terrier 37 dogs

The sad truth is you can’t extract any meaningful information from these tiny studies. Such small samples could be distorting the results in either direction. They, literally, don’t tell us anything.

You know next to nothing about the people who did most of the studies. It might come as a surprise to you what kind of people serve on breed health committees. It’s not unusual for these committees to include people with medical or scientific backgrounds. People like Denise Wall, PhD, who has served on her breed’s health committee.

Furthermore, based on the comments you’ve made in this thread it’s obvious that you haven’t bothered to read the studies or very much of the content on Dr. Cassidy’s site. Yet you feel qualified to attack their work nevertheless. That’s not how science works.

There is considerable evidence that many breeds have a shorter median or average longevity than one would expect given the typical size of the dogs in those breeds.

There is evidence that breeds for which longitudinal data are available are not living any longer in recent years than they were in the past. As Dr. Cassidy said:

“Even with consideration of possible methodological problems with some of the studies, the data are very discouraging. There have been dramatic improvements in veterinary medicine, particularly in treatments of the most common killers of dogs (cancer and heart disease). If breed health is staying about the same, then breed life spans would be expected to be increasing because of improved veterinary care. If breed life spans are staying about the same, then improved veterinary care may only be compensating for declining breed health.”

Laura, you are missing the rather central point that I don’t have “a conclusion.” We really have no idea whether dogs, either as an overall average, or within particular breeds except the few for which there have been genuinely good studies, are living longer lives, shorter lives, or lives of similar length compared to forty or fifty years ago.

And surveys conducted by breed clubs without the collaboration of professionals are, whatever you choose to believe, far more likely than not to be utter crap, because the average person has no idea how to design a survey so that it produces meaningful, useful results.

Finally, the fact that Dr. Cassidy includes (and especially, includes without caveat of any kind) a ‘study’ in which the sample size was fourteen dogs, and includes another “study,” the Papillon breeder survey, with a report that Dr. Cassidy admits was extremely badly written and which someone who attempted to fill out that survey reports was extremely confusing and badly designed, really does undermine the credibility of the rest of the “information.”

If that isn’t clear enough for you: Obvious statistical invalidity and an extremely poorly-written report that ought to have cause a serious person to ask if the study design was any better, were not regarded as reasons for excluding a ‘study’ from the compilation. Therefore, the fact that a study was included, does not mean we should take it seriously or can extrapolate anything at all from it. Inclusion of obvious bad data means that we can’t rely on any of the data, except in cases where we can personally look at the actual surveys and their results—and we can’t extrapolate from those to any other breeds than the ones they specifically studied.

You have no basis for making that sweeping generalization. You are the one in this thread who has been twisting evidence to support “your conclusion”.