Syrotuck's main study is his 1976, with N=242. But he gives much more detail about distance travelled in his 1975 paper, breaking distance down every 0.2 miles. Unfortunately, he only reports probabilities, not numbers, and doesn't even report total N. We know he got more data between 1975 and 1976, but didn't know how much. Is the 1975 breakdown representative of the 1976 data? Unfortunately, no one has Syrotuck's original data. But we re-created it. (Spreadsheets available!)
Originally 2 Apr 2007, updated 4 Jul 2008; Posted on the new site 23 Mar 2012. Updated links 8 Apr 2023.
Noting that Syrotuck's probabilities were quantized (as they should be, since we know that lost people come in units!), Bob Koester & I were able to work out the quantum for each table. (Think of it as SAR's version of Millikan's oil drop experiment). From that, we were able to figure out the N for the various tables, and even replace the quantized probabilities with counts, re-creating the original data.
For the 1975 distance data, N_tot=100.
- 38 on flat terrain (including those going "beyond 3.0").
- 62 on hilly terrain (including 8 with no known distance; maybe beyond, maybe missing).
The attached spreadsheet (or google sheet) shows the reasoning, and re-creates the data. I'm quite confident. Figuring out the best explanation for the typos took a little while, but they make a very satisfying confirmation. (The very round total of 100 suggests maybe Syrotuck selected, or stopped at, a pleasing number, but it's not an artifact of this analysis: I worked on the individual tables, getting 38 and then 62.)
So what's the relationship to the 1976 data, N=242. Did he really have 242 cases with distance, or was 242 just the total set, with a lot of missing data (like we have), meaning he had mostly the same data as in 1975?
I think it's genuine: in 1975 he had 100 with distance data, maybe some more without. In 1976 he had about 242. Several reasons. First, the medians differ in the 1976 report. Second, in 1975 he had 11 children and kept them as a single group. In 1976, he broke children down into 2 groups. He doesn't subdivide unnecessarily: the 1976 forward says he tested NY vs WA medians for similarity before combining.
So he likely had about twice as much data in 1976. Which got me thinking. That data split about evenly between NY (95 cases) and WA (117). So I reread the 1976 Acknowledgements with this in mind.
I wish to express my gratitude to Mr. James Lord... for providing a large number of case histories from the state of New York. This information broadens the scope of this paper, which otherwise could have been considered regional to the Pacific Northwest. ....
I think that clinches it. Further, it shows that his 100 cases in 1975 were all Pacific NW cases. Maybe all from WA (117 by 1976), maybe including some of the 17 from {Idaho, Oregon, California, Alaska, New Mexico, Wyoming, and Tennessee}, at least 3 of which are Pacific NW.
Now, in the 1976 Forward, he says the WA & NY medians were either exactly the same, or differed by less than 0.2 miles. But the 1975 data have two categories that differ by more: hikers on flat terrain (1.6 vs 2.0 in '76), and hunters on flat terrain (1.0 vs. 1.6 in '76). So unless he made a mistake, he got more Western data by 1976, changing the medians to those reported in 1976 for the combined dataset, which we can take to be the same for both WA and NY. (The 95% zones of course would be different, as Mitchell showed so nicely in his massive NASAR study.)
So, I feel pretty happy saying that the attached spreadsheets reconstruct at least 80% of his 1976 WA distance data as well. The 1975 book has other summary info: concealed, survival, etc. We've been building a Bayesian model that tries to squeeze all the information out of the 1976 report. Knowing the separate values for the WA vs NY data will help, since we do know from the 1976 report that the survivability differed radically between the two states.
If you know Bayes nets, several are in syrotuck_nets.zip.
