-99-
2.1,5.3. Summary
In the first part of the analyses, I tested a model of reciprocal friendship perception which si¬
multaneously includes both friends' perspectives, and, at the same time, differentiates be-
tween aspects of friendship with differing visibility. The results strongly support the assump-
tion, that, when simultaneously investigating both friends' views on the quality of their
friendship, a conceptual distinction can be made between (a) those aspects of friendship that
are more objectively perceivable and thus part of a shared social reality (i.e., in this case, Play
Encounters and Visits), and (b) other aspects that belong to a more objective, non-shared real-
ity (i.e., in this case, Closeness, Fun, and Conflict). The differences in friends' evaluations of
the subjective aspects of their friendship were also supported by the fact, that the correspond-
ing self-rated and friend-rated factors were only moderately correlated. As expected, though,
no general bias (i.e., no general over- or underestimation of the subjective friendship aspects)
existed on either the self-rated or the friend-rated side in the overall group. In summary, the
proposed model of reciprocal friendship perception was strongly supported and can thus serve
as a basis for the subsequent analyses. * No gender effects on the latent friendship constructs
were found, suggesting that no further gender comparisons were necessary. Grade effects
were found for every aspect of friendship except for Closeness, though.
Developmental trends in reciprocal friendship perception - A comparison
2.2.
among grade groups
Theoretical considerations and specifications of the general hypotheses con¬
2.2.1.
cerning the developmental influences on reciprocal friendship perception
In the previous section, the general distinction between subjectively and objectively perceived
aspects of friendship has shown to be valid, when simultaneously investigating both friends'
As mentioned, the possibility of attenuated relations in the factorial model due to the aggregate indices of
more than one friend was tested. For this purpose, the validation analyses were repeated on the basis of only
the top nominated and reciprocated friendship rating. As depicted in Tables 7a and 7b in the Appendix, even
with the reduced power the results were very similar, both regarding the fit indices and factor loadings, as well
as the other results presented above. This indicates that no attenuation of relations occured in the tested factorial
model.