70 opposite sexed dyads were observed and videotaped in 3 laboratory settings:

A. Session 1. The dyad comes into an unstructured situation with a male experimenter. The room contains a couch and video camera. The E gives the following instructions and starts the video. “The first thing I would like you to do is to come in here and get acquainted with each other. I would like to tape your conversation Though. So just have a seat on the couch. Say whatever you like, and I’ll be back in about 5 minutes.”

B. Session 2. Same as Session 1, but 4 weeks later and with a different opposite-sexed partner.

C. Session 3. Unexpected debate, immediately following Session 2. E comes in and hands both Ss a clipboard and says, “The next thing I do is hand each of you a pad of paper because some people like to be able to take some notes during the next part of the experiment. That is because the next part calls for the two of you to have a little debate. Specifically, the topic we have people debate is the use of capital punishment, because most people can come up with at least some arguments of both sides of that issue. I’ll just flip a coin and have [ ] call it. If it comes up what you call, you will be in favor of capital punishment and if it doesn’t then you will be against it. [Flips coin]. Okay, the debate will last about 5 minutes. I’ll just give you a short minute to collect ;your thoughts and then we’ll start. After a brief pause, the E says ,”Begin”, starts the video, and leaves the room.

Each S in the study got 1 or 2 people that knew the subject “best” to rate the S on the California Q Sort, which consists of 100 items that are sorted into a 9 category, forced-choice distribution, ranging from 1-not at all characteristic to 9-highly characteristic. A sample item might be “Is cheerful”. In this forced choice distribution, the final distribution of ratings must resemble a normal curve. Thus, only 5 items each can get ratings of the extremes 1 and 9. Less extreme ratings of 2 and 8 must have 8 items each, 3 and 7 must have 12 items each, 4 and 6 must have 16 items each, and the Neutral 5 must have 18 items. A similar distribution was used in the sorting of the modifications of the 100 item Cal. Q sort. that had fewer items. The Calif. Q sort will be shown in class.

In addition, trained raters viewed the videos (as often as necessary) and rated both Ss on a modified California Q sort that contained 62 items, 41 from the original California Q sort and 21 items created because they were relevant to the laboratory settings. This was called the behavioral Q sort (BQ). An average of 6 coders saw each session at least twice. No coder viewed more than a single behavioral episode for each subject. The mean Spearman-Brown reliability of the BQ was r = .64 (moderate).

RESULTS

Cross-situational correlations for the BQ:

Sessions 1 and 2: 45 of 62 sig: M = .37

Sessions 2 and debate: 41 of 62 sig: M = .27

Session 1 and debate (D): 30 of 62 sig: M = .20

Some items had large cross-situational correlations:

“Speaks in a loud voice,” .70 for 1&2, .65 for 2&D

“Exhibits an awkward interpersonal style”, .66 for 1&2, .31 for 2&D

“Laughs frequently”, .63 for 1&2, .43 for 2&D

“Behaves in a cheerful manner”, .60 for 1&2, .41 for 2&D

“Behaves in a fearful and timid manner”, .60 for 2&D, .65 for 1&2.

“Has high enthusiasm and high energy level,” .53 for 2&D, .59 for 1&2.

“Is expressive in face, voice & gesture”, .52 for 1&D.

“Is reserved and unexpressive”, .41 for 1&D.

There were also situational effects upon behavior:

Behaves in a fearful manner, and Expresses insecurity or sensitivity, were greater in 1 than 2.

Exhibits social skills and Behaves in a cheerful manner, were greater in 2 than 1.

Expresses warmth, seems to like partner, and initiates humor, were greater in 2 than D.

Discusses philosophical issues with interest, Expresses skepticism or cynicism, and Acts in an irritable fashion, were greater in D than 2.

Session 1 was obviously more psychologically similar to session 2 than the debate. The situation effects (the magnitude of change in ratings across situations) was unrelated to the magnitude of the cross-situation correlation of any given BQ item, average r = -.02.

The consistent items on the BQ were consistent across all 3 settings; the inconsistent BQ items were inconsistent across all 3 settings, indicating that some behaviors are more trait like than others.

Personality judgments by peers were related at a low level with BQ ratings of lab behavior by raters, ave. R = .35, max R = .41.

F&C infer that there is probably more consistency across lab settings than between lab settings and real life; although procedural differences may contribute to this. For example, the peer raters and lab raters were differentially trained.

F&C ask the reasonable question if the differences in consistency among different BQ items were because of differential reliability and variance of the BQ items. F&C answer, “Yes, somewhat”. These psychometric issues do contribute to the differences, but are not complete explanations. It is also very likely that items with a greater breadth of situational relevance (meaning that they were likely to occur in a wide range of situations; e.g., “Speaks in a loud voice, Behaves in a fearful or timid manner”) were more cross-situationally consistent than items with a narrow range of situational relevance (e.g., “Expresses interest in fantasy and daydreams, Demonstrates interest in topics related to power.”)

Also, items that reflected more “Operant” behaviors were more consistent than items reflecting more “Respondent” behaviors.

CONCLUSIONS: Data impressively broke the r = .3 to .4 barrier. Behavior is much more predictable from traits when behavior is viewed in terms of its psychological meaning, not its concrete manifestations. Situational specificity and behavioral consistency are not necessarily in opposition. “EVEN THOUGH SITUATIONS PROFOUNDLY AFFECT WHAT PEOPLE DO, PEOPLE CAN STILL MANAGE TO PRESERVE THEIR DISTINCTIVE BEHAVIORAL STYLE ACROSS SITUATIONS”