Time Uncertainties


    You began, early in this thread, by claiming that my assertion that the Shively/Simpson encounter occurred at 10:40 was a recent invention of mine, and that until recently I had said the time of that event was 10:45. I researched the question and demonstrated that for more than the last two years I have used the 10:40 number, and I never used the 10:45 number as more than a vague indication of when that event had occurred. You now claim that the time of the encounter was not your point, but any AFOJS reader can simply look above in this thread and see the several and forceful places where you made that point. You are frankly beginning to look silly, Bob, and I'm not surprised that people don't take you much serious. Must be the cold up there does something to you brain, huh?

    "CAN'T KNOW THE TIME PRECISELY": Trying to deflect attention from you earlier gaff, you now assail me because I have specified the times of events in the crimes with a precision you will not find elsewhere. Such precision, you say, is impossible. Many others than you do not believe that it is possible to make such close estimates, either. Not because I think that you will understand, but for others who think the same, I will explain.

    First of all, not all of the time estimates in this crime are vague. I mention a few...

   STORFER: Ultimately his estimate of the time when he first heard the dog rests on the clock on a VCR that was set the previous spring. Some brief activities (putting his son to bed) occurred between the beginning of the barking of the dog, and his reference to the VCR, but these can be estimated closely. There was a known clock error that he applied. When all of these factors are accounted for, it is reasonable to estimate that his statement of the time is 10:20 +/- 1 minute.

   FENJVES: He was watching a TV program when he first noticed the dog's bark and by relating that event to the segment he was watching, he estimated the time. It is reasonable to assign an error of +/-3 minutes or so to a time estimated in this way.

    STEIN: She makes only a very rough estimate of the time, and her testimony is more useful for other reasons than the time. Although she estimates the same time as Fenjves, her estimate is based on knowing the length of time (half an hour) that she lay in bed without looking at a clock, waiting for her house mate to come home. I have put a ten minute tolerance on this.

    PILNAK: She is compulsively punctual, and is constantly aware of time. She consulted a clock, which itself is frequently calibrated, to give the time (10:21 to 10:24) she was talking to her friend on the porch. I estimate that the accuracy of the time of her observation is high, with an error of not more than a minute. That is, she and Telander came onto the porch at actual time of 10:20 to 10:22.

   SHIVELY: She related the time she saw Simpson at Bundy and San Vicente to a "little battery operated clock in her car." The relationship between the events and the clock time is very accurate - better than a minute, I should say. But the clock itself was in error by an ESTIMATED "eight to ten minutes." This estimate itself is subject to uncertainty. To deal with this situation, I first subtract from the clock time (10:50) of the event, the clock error (I'll use the figure of 9 minutes in this analysis to avoid argument.) But, to this, one must apply the observational error (less than a minute) and the uncertainty in the amount of the clock error (I figure 2 minutes). From this, the actual time that Simpson left the intersection is 10:41 plus or minus 3 minutes. He entered the intersection one minute earlier.

    HEIDSTRA: His estimate is not very precise. He knows when he left his apartment, and he estimates the time it took to get to the events in question, but those estimates are admittedly rough. (I am sorry now that when Rose and I walked this same route, I did not take the stopwatch with us. I would like to have an independent estimate of the time from his apartment to the alley.) However, I have traced and timed his actions after he started into the alley, and from that point, I have a confident idea of when things happened. I have put a tolerance of five minutes on the time he got to the alley.

    ELAPSED TIME, HEIDSTRA TO SHIVELY: In my recreation of Simpson's flight from Dorothy to San Vicente, I went at the same time of day, and day of the week that he did, and by the three most likely routes. My elapsed time on these was two minutes plus/minus 15 seconds, and I estimate that the error in those measurements is not more than 5 seconds, since they were done with a stopwatch in hand. Because these errors are small compared to other sources of error, I have neglected them, and say that the time between Dorothy and San Vicente is 2 minutes exactly, with no error -- a reasonable approximation, I think.

The power of this analysis, and the thing that most people overlook, is that independent observations, each with their errors and uncertainties, nonetheless have to match, if they are caused by the same event.

    For example, Stein's poor estimate of the time the dog began to bark amounts to "10:05 to 10:25." But, Fenjves estimate is "10:15 to 10:20," and so these two estimates reinforce each other, and we know that Stien's experience was really at "10:15 to 10:20," even though she could not estimate it that close herself.

    On the other hand, Storfer heard the dog begin to bark at "10:19 to 10:21" (accounting for observational errors) and continue for "several minutes, at least." But, Pilnack thought she was in a position to begin hearing from "10:20 to 10:22" and did not hear. So, if the dog barked for at least three minutes (which it seems to have according to Storfer) there is not enough error in their time estimates to account for the fact that one heard the dog, and the other one did not. The attorneys resolved this conflict by convincing the public that "someone was lying," and the public -- unable to decide which -- just disregarded the testimony entirely. But there is another explanation, which I believe. There are locations from which the dog could have been barking so that Fenjves, Stein, and Storfer would hear, but Pilnak would not. Nobody was lying. And, Pilnak's testimony is valuable for telling us where the dog's location was when the others heard it.

   HEIDSTRA/SHIVELY OBSERVATIONS: The foregoing shows the power of combining time estimates from different observers as applied to the early timeline. When done with the late timeline it results in sharpening the precision with which we know the time beyond the accuracy of the individual estimates. In the particular case of Shively/Heidstra it does this because most of the Shively error is to make the actual time earlier than the raw observation, and with Heidstra it is to make the actual time later than nominal. It is a happy circumstance that there is only a small overlap in the possibilities. Consider...

    According to Shively's data (using a 9-minute clock error) the time when Simpson left the Bundy/San Vicente intersection was 10:41 +/ 2 minutes (see above). He entered the intersection one minute earlier, at 10:40 +/- 2 minutes. He left Bundy/Dorothy two minutes before that, at 10:38 +/- 2 minutes.

    Now, working from Heidstra's end, he left home at 10:15, and arrived at the alley about 15 minutes later. But both his exact time of leaving and his elapsed time of getting to the alley have some uncertainty. Heidstra testified that he entered the alley at "10:30 to 10:35," but on cross-examination conceded that it could have been as early as 10:29. This gives a nominal time of 10:32 with a tolerance of +/- 3 minutes. Following events can be estimated rather closely and produce a time of the "Hey, hey, hey" occurring 3 minutes after entering the alley (at 10:35), and a time of seeing the "white SUV" as 5 minutes after entering the alley (at 10:37). From this, we are led by Heidstra's testimony to think that he saw the SUV at 10:37 +/- 3 minutes. That is, between 10:34 and 10:40.

    But this has to match with Shively's observation which puts the "white SUV" at Bundy and Dorothy at 10:38 +/- 2 minutes, or between 10:36 and 10:40. To account for other factors (Heidstra's apparent friendliness to the defense, which wanted his observations as late as possible, and later events at Rockingham, I have taken my estimate as the middle (10:37) of his range, even though that puts the same number at the low extreme of Shively's range. Nonetheless, I have said that the actual time might be moved a couple of minutes to account for errors in the witness estimates of time. (To be strictly accurate, I think the Heidstra sighting of the "white SUV" occurred at 10:37 +3/-1 minutes.)

    In this way, Bob, I hope that you can see that the time of inter-related and independently observed events can actually be known more accurately than any one observation gives it.

    PROFESSIONAL APOLOGY: Now, if there are any statisticians out there, I must quickly say that this is a casual analysis, done for a layman's contemplation. In a thorough analysis, the actual distribution of errors for the individual sources would be estimated, and the overlap computed, then the resulting distribution in the times of a commonly observed chain of events would be reported. In this way, one does not simply consider a fixed error, but the likelihood of particular possible errors. Although it is more accurate to approach the problem that way, it is much more complicated to understand, and I sincerely doubt that in the present instance it gives a substantially different result than I have determined.

Dick Wagner • Van Nuys, CA (12/16/00) NG_688m

back.gif (2777 bytes)