A friend of yours is absolutely convinced that he has a positive influence on the friendliness of conversations in which his is a participant. He has reached this conclusion on the basis of his everyday observations. You convince him that a systematic study is needed to confirm his hypothesis. Your friend [still smiling] carefully develops an operational definition of the friendliness of a conversation and records a rating for each of the next 50 conversations in which he is a participant. His results show that 75% of these conversations are rated 'very friendly,' 20% are rated 'friendly,' and 5% are rated 'neutral.' Your friend returns to you now convinced beyond a shadow of a doubt that he has a positive effect on the friendliness of a conversation.
Should a different method of observational research be recommended to get a more accurate description of what is going on? Which method? Why?