SEARCHMASTER is a program designed to assist ground search and rescue personnel in finding lost persons. It provides brief but useful descriptions of various categories of lost person behaviour, including statistical probabilities on distances travelled from the point last seen; it assists with the difficult (and critical) decision-making process of selecting areas to search for the lost subject; and it helps with the planning of search tactics by providing quick advice on how many searchers to commit to an area, what spacing to use, and how large of an area to attempt to search with available manpower. Generally, SEARCHMASTER is a tool for maximizing the expertise of an experienced search director (and for minimizing the impact of mistakes); in no way can it substitute or compensate for a lack of expertise and experience. The author can therefore assume no responsibility for the outcome of any search in which SEARCHMASTER is used.
SEARCHMASTER is being offered to search and rescue colleagues on a Shareware basis. You may distribute the program, with the startup screen and this text file, freely. However, if you use SEARCHMASTER for any purpose (including training), you are asked to send $25.00 to the author, who will contribute the funds to the volunteer search and rescue team of which he is a member. Users will be provided with a tutorial on how best to use SEARCHMASTER, as well as notification of future updates.
What follows are descriptions of the three major components of the program.
SUBJECT INFORMATION
Each year, thousands of people become lost in the forests and wilderness areas of North America, and hundreds of ground search teams keep records on the outcomes of their efforts to find them. The assumption underlying SUBJECT INFORMATION is that there is much to learn from examining past incidences of lost person behaviour. Certain categories of lost people tend to suggest different patterns of responding to the experience of being lost. A small child may not answer callers; a ten-year-old may search randomly for a familiar landmark; an older person may be prone and difficult to detect; a hunter may travel an incredibly long distance to avoid the embarrassment of being "found". SUBJECT INFORMATION provides not only brief generalizations of the behaviour of past lost persons in different categories, but also indicates the respectives distances which they travelled from the point last seen, taking into account terrain and weather conditions. It suggests "probabilities" that represent statistical summaries of hundreds of past cases. For example, the assumption is made that, in the absence of clues to indicate otherwise, the area of highest probability of containing the lost subject is the median distance at which past lost persons of the same category were found from the point last seen.
The statistics refered to in SUBJECT INFORMATION are based on data collected by the late William Syrotuck (1976) from case histories of 229 lost persons, mostly in the States of New York and Washington. The Behaviour Profiles draw also from Syrotuck's work, as well as later research by Mitchell (as reported in LaValla et al, 1987), and the author's own review of hundreds of files of lost persons in Nova Scotia. The categories included are (1) Young Child , (2) Child, (3) Hiker, (4) Hunter, (5) Elderly, and (6) Other. The profiles offered in SUBJECT INFORMATION are necessarily brief, and represent only a fraction of available information on the behaviour of lost persons in each of the categories. The author has attempted to include those descriptions that should be most useful to a search director, such as those that pertain to the subject's motivation and how he or she might respond (or why they might not respond) to callers. When using SUBJECT INFORMATION, keep in mind that subjects within each category may vary a great deal, and many may not behave so predictably. A brief behaviour profile abstracted from case histories is no substitute for information about the lost subject gained through interviews with family and friends. However, it may prove especially useful at the beginning of the search, when the search director must make immediate decisions regarding hasty teams and confinement lines.
POA CALCULATOR
Whenever a search director decides to send a search team into a particular area, he or she has intuitively made some judgment about the probability that the subject is indeed in the area. However, when the subject is not found in that area, the director (again, intuitively) alters the probability judgement downwards, depending on how thorough the search was perceived to have been. If the search had been extremely thorough (e.g., a grid search with close spacing), the search director may entertain a very high level of confidence that the subject is not in the area. Finally, as the intuitive or subjective Probability of Area (POA) goes down in one area, it is logical that the POA of the remaining, unsearched areas should go up, in proportion to the size of their original probabilities.
It should be apparent from this discussion that searching for someone requires complex information processing of many variables in a systematic and mathematical fashion. However, research in psychology indicates that people tend not to think mathematically or even systematically when confronted with such a complicated task. Instead, they "simplify" the problem by applying various rules of thumb which they believe are relevant to the task at hand, or attempt to apply "solutions" which worked in the past on apparently similar problems. In particular, people seem to have considerable difficulty altering their subjective probabilities accordingly in the light of new evidence. For example, many people report having had an experience similar to the following. You always keep your wallet in the top drawer of your bureau. In the past, whenever you wanted it, you always found it in that very drawer. However, one day, as you're leaving for work, you look in the drawer and don't see the wallet there. You look somewhere else, with no luck. After five minutes of fruitless searching, you come back to the drawer, conduct a very thorough, "Type III" search, then look somewhere else again. A few minutes later you're back to the drawer, and so on. In the task of searching for a lost person, a common mistake (in the author's view) is to oversearch areas which the search director had held a high subjective probability of containing the subject. Rather than extending the search to new areas, the same area (or areas) receives one thorough search after another.
POA CALCULATOR keeps track of the changing probabilities of all areas in the search locale as specific areas are searched and the subject is not found. Moreover, when recomputing the changing probabilities, it takes into account the degree of thoroughness by which each area is searched. For example, if one area is searched very thoroughly, the new POA for that area will drop significantly, and the POA for the remaining areas will increase more than if the area had been searched less thoroughly. The measure of thoroughness is the Probability of Detection (POD), that is, the probability that the subject would have been found--if he had been in the area--given the width of spacing between searchers. Closer spacing makes for a more thorough (though slower) search, with a higher POD. POA CALCULATOR makes use of a mathematical equation called Bayes Theorum, a formula first applied to ground searching by Syrotuck (1975). To use POA CALCULATOR, the search director, preferably in consultation with other experienced search managers, should do the following:
(1) Divide the search locale into identifiable segments or areas, between 3 and
16 in number. Ideally, each area should be searchable by one team during one
shift or "slog". In practice, however, some areas may have to be smaller
or larger than this. Make use of natural barriers as much as possible to define
the boundaries of the areas (e.g., lakes, trails, streams) to minimize confusion
by field personnel. Each of the areas should have some non-trivial probability
of containing the subject, in the opinion of the search director or search manage-
ment team. The sum of the probabilities for all of the areas combined should be
over 90%, but less than 100% (there is always some chance the subject is
not in the search locale). POA CALCULATOR "Info" provides further information.
(2) In consultation with experienced searchers, decide on the specific probabilities
for each of the areas. Each area must be assigned a "prior probability" that
is some number between 1 and 50 (e.g. "50" = 50%). Again, the sum of all
areas should be as close to 99 as possible.
You are now ready to use POA CALCULATOR. Select "Run" and a window should appear, asking for the total number of areas selected (unless you have already started POA CALCULATOR and have temporarily been using some other part of SEARCHMASTER, in which case you will be returned to where you left off). After selecting the number of areas, you will be asked to enter the prior probabilities for each area. You can enter each number separately, type "S" to start over, or type "Q" to quit. Also, you can type "A" at any time during entry to average or distribute evenly the remaining probabilities to the unassigned areas. If you select the latter option, make sure you are satisfied with the POA assigned to "Outside Locale" before continuing, as the probability for the unsearched area can only get higher as the search continues.
Click the "Continue" box (or simply hit the Return key) if you want to use the assigned probabilities appearing on the screen. At that point, a new screen will appear, asking for (1) the number of the area you have searched, and (2) the Probability of Detection for that search. The latter figure is estimated with the formula POD = 100 - (S X .5), where S = the spacing (in feet) between searchers (this formula is based on studies conducted in Washington State by Wartes, as reported by Syrotuck, 1975; see "Info" for further discussion). Thus, for example, a spacing of 20 feet yields a POD of 90%, while a spacing of 100 feet leads to a POD of 50%. If you're not too good with mental arithmetic, you can either compute the POD's with pencil and paper (or calculator), or select "What is the POD?" from the menu under PLANNER. Unfortunately, no POD formulas are available for non-grid types of searching, such as hasty teams, search dogs, or helicopters. In these instances the search director will have to provide his or her own estimates.
Once the Probability of Detection is entered, another screen appears containing the revised probabilities for each of the areas, as well as for that area outside the search locale. Again, you can continue, start over, or quit to the main menu of SEARCHMASTER (quitting does not lose the current POA's; only starting over or quitting SEARCHMASTER altogether does that). If you chose to continue, you will be returned to the screen asking for the number of the area searched and the POD of the search. These latter two screens will then alternate until the search is terminated.
PLANNER
This part of SEARCHMASTER provides the search director with tools he or she needs to utilize human resources in the most efficient way possible. As its name suggests, PLANNER helps "plan" the ground search. PLANNER will instantly calculate the number of searchers necessary to cover an area, how much time the search will take, what spacing the searchers should use, and how much of an area to attempt to search with available manpower. (An important assumption underlying these calculations--also based on research with actual search teams in motion--is that it takes a team an average of 3.5 hours to travel 1 mile.) Using PLANNER, the search director can play "What If...?" with different figures before deciding to commit search teams to their respective areas. He or she can entertain realistic expectations about how long a team should be in the field, and can make informed decisions as to whether or not there is enough manpower to get the job done.
In addition, PLANNER will compute distances between two points when provided with U. T. M.
map coordinates (i.e., eastings and northings), and based on these bearings will compute both map and compass directions between the points. In the latter case, the user will have to enter the local declination (SEARCHMASTER "defaults" to the declination for central Nova Scotia), or select "skip" to ignore the compass direction computation. This part of PLANNER should prove useful for double-checking bearings that are provided to search teams, where accuracy is critical.
Finally, the computations done by PLANNER can be accomplished in either feet/miles or meters/kilometers, whichever the user selects from the menu.
KENNETH HILL, PH. D., the designer and programmer of SEARCHMASTER, is External Training Officer for the WAVERLEY GROUND SEARCH AND RESCUE TEAM, an organization of volunteers serving greater Halifax County, Nova Scotia. He would like to thank Ron Marlowe, Don Hirtle, and Don Bower for encouragement and advice in the development of the program.
SEARCHMASTER is dedicated to the memory of Bill Syrotuck.
REFERENCES
LaValla, R., Stoffel, S., & Wade, B. (1987). Search is an emergency (2nd Ed.). Olympia, WA:
The Emergency Response Institute.
Syrotuck, W. G. (1976). Analysis of lost person behavior. Westmoreland, NY: Arner
Publications.
Syrotuck, W. G. (1975). An introduction to land search probabilities and calculations.
