The test-retest reliability of free association norms

Abstract

The test-retest reliability of free association norms J 146 Ss gave free associations to 158 stimulus words on two occasions separated by 10 weeks. The probability of responding with an identical associate (LA) after 10 weeks was .32. A correlation of .84 was found between number of Primary (P) responses and number of lAs for the 146 Ss. A high positive correlation was found between dominance of the Primary (DP) and number of lAs for the 158 words, while a negative correlation was found between frequency in the language of the stimulus word (LF) and number of lAs. The coefficient of multiple correlation yielded R1.23= .92, indicating that a high degree of reliability can be predicted for words with low LF and high DP.

Normative and idiosyncratic sources of recall and misrecall.

“…On the other hand, if 6" produced a normatively lowprobability response, that word is not likely to remain high in his hierarchy, and it is also not likely to be recalled. However, the data for Group N suggest that high-probability normative responses can be well recalled even when they have not 3 It is of interest to note that these reliability values agree reasonably well with those of other investigators (Braun, Constantini, Link, & Ehmer, 1967;Brotsky & Linton, 1967;Fox, 196S;Hall, 1966). The average probability of repeating an Ii was ,49 and of repeating an I 8 was .33. been produced by S. Figure 3 presents recall of Rgs as a function of their position in 5"s idiosyncratic hierarchy, and it can be seen that although recall varies with idiosyncratic rank, the recall of an R 2 which is not in -S"s hierarchy is still high relative to recall of a nonnormative word (see Fig.…”

“…If a longer interval is inserted between FA and study-recall, it becomes necessary to obtain some measure of the reliability of idiosyncratic hierarchies. The present study extends the work on reliability (Brotsky & Linton, 1967;Hall, 1966;Jung, 1967) to consideration of 5"s first three associates (Ij-s). This is a necessary extension since the data presently available only show the probability with which a word occurs as an I ± on two successive occasions.…”

Association time, commonality, and the test-retest reliability of free association responses

“…Agreement between Phase 1 and Phase 2 was 60% for male Ss and 64% for female Ss. These values were substantially higher than those found by the authors in an earlier study (Brotsky & Linton, 1967) but more in line with the results of Hall (1966), who found about 50% lAs on test and retest. The high percentage of test-retest agreement in the present study may be due to some combination of long list, short test-retest period, and individual-oral administration procedure.…”

“…When any frequency based index such as C or frequency of occurrence of the most common response (dOminance of the primary) is used to predict individual behavior, the assumption is made that there is a lawful relationship between the response hierarchy derived from a normative group and the hierarchy of the individual S. If this assumption is warranted, it would be predicted that high C, short AT responses would tend to be repeated by a given S on two or more separate occasions. Following the work of Hall (1966), BrotskY & Linton (1967 investigated the probability of Sa' giving identical associations (IA) to 148 words on two separate occasions. Although the test-retest reliability was, in general, much lower than that reported by Hall (.32 vs .50), a correlation of .88 was found between dominance of the primary and the number of lAs.…”

Association tests and outdoor ecology education

“…However, despite differences in personal backgrounds, such associative frameworks are considered to be rather stable. In other words, while individual associations may change because of everyday experiences, the respective frameworks are considered to be rather constant (cf., Brotsky & Linton, 1967;Schaeffer, 1991). The tendency, therefore, to associate in certain directions has to be regarded as a valuable feature of the concepts, and it has been taken into consideration that associations are highly influenced from a subliminal association level.…”