行動観察における観察頻度の決め方（第2報）―～低出現率を示す観察対象事象は，どの程度，観察頻度を少なくすることができるのか～

Determination of observation frequency in behavior observation (Second report): The extent to which observation frequency can be lowered for events with low occurrence rates

要旨：本研究は，5％未満の低い出現率を示す観察対象事象に対して，どこまで観察頻度を粗くすることができるのか調べることを目的とした．母集団とした10時間連続観察における観察対象事象の出現率と，15秒～60分間隔の11通りでシミュレーションサンプリングした標本の出現率の間に差があるか比較した．その結果，605標本中599標本に有意差は認められず，有意差を認めた6標本は1.59～0.03％の非常に低い出現率に対する10～60分間隔でシミュレーションした標本であった．5％未満の低出現率を示す観察対象事象は，30～60分間隔といった少ない観察頻度で連続観察と同程度の予測が可能であることを実証した．

　This study investigated the extent to which the observation frequency of events with a low occurrence rate of ＜5% could be lowered even further. The occurrence rates of events observed continuously for 10 hours were compared with those of 11 types of simulation samples performed at intervals of 15 sec to 60 min. The results showed no significant differences for 599 of the 605 samples, and the six samples for which significant differences were observed were from simulations performed at intervals of 10 to 60 min with very low occurrence rates of 0.03% to 1.59%. These results indicate that for events with a low occurrence rate of ＜5%, a similar level of prediction using continuous observation can be achieved with a low observation frequency of intervals of 30 to 60 min.