TIME EFFECTS ON BEHAVIOR 



Roger Bakeman and Robert Helmreich 

 The University of Texas at Austin 



Throughout this report we have used observational data primarily in highly 

 reduced, summary form. For example, we have referred to the total percentage 

 of time aquanauts slept during missions but have, in general, ignored the ways 

 in which time spent sleeping changed during the course of a mission. It seems 

 clear that collapsing time series data during initial analyses is justified, 

 yet it is equally clear that much richness remains to be tapped. Such analyses 

 will undoubtedly reveal details and indicate mechanisms only vaguely suggested 

 by first order generalizations based on summary data. 



The problems presented by time series analyses are difficult but not insur- 

 mountable. There is a risk of being simply inundated by mounds of detail. 

 Here, as elsewhere, it makes sense first to determine at a molar level if a 

 phenomenon holds interest, then, if it appears promising, to examine it in more 

 detail. Accordingly, our first approach to time series analyses has been 

 simply to compare time spent in various activities during the first half of a 

 mission with time spent on the same activities during the second half. This 

 "split mission" comparison indicates differences not only with missions but 

 suggests different patterns of change between short (14-day) and long (20-day) 

 missions . 



Specifically, trials by subjects analyses of variance were computed for two 

 groups: those aquanauts participating in the short missions and those partici- 

 pating in the long missions. In addition, these analyses were run by team both 

 for scientists only and for all five aquanauts. Since patterns of significance 

 are almost identical with and without the engineers, the results reported below 

 refer to the analyses including all aquanauts. Thus, using our standard activ- 

 ity variables, differences between the first-half and last-half of both the 

 short and the long missions were examined (see Table 9) . 



The most striking finding from this analysis is that 8 of the 9 variables used 

 showed a significant change from the first-half to the last-half of the short 

 missions, but that only 4 variables were significant in the case of the longer 

 missions. This suggests that the long missions were long enough to gain some 

 stability over time, but that the short missions were subject to a "Night- 

 Before-Christmas" excitement effect, i.e., people entered the short missions 

 excited, worked very hard initially, burned out somewhat, and then relaxed more 

 during the second-half of the mission. The data bear out the tenability of 

 this notion. 



Of the significant variables for the short mission, 4 were highly significant 

 (p <.001); these were: total work," total leisure," total marine science, and 

 co-recreation. Direct marine science, habitat maintenance, and gregariousness 

 differed less significantly (p <.01) than the four mentioned above but were 



" Total work is a composite variable consisting of: direct marine science, 

 marine science support, habitat maintenance, and maintenance of others. 

 Total leisure consists of: co-recreation, solitary recreation, and relaxing, 

 resting and idling. 



VIII-46 



