A
Quantitative Review of Associative Patterns in the Recall of Persons
Devon D. Brewer
Interdisciplinary Scientific Research
e-mail: http://www.interscientific.net/contact.html
Giovanni Rinaldi
Istituto di Analisi dei Sistemi ed Informatica del
Consiglio Nazionale delle Ricerche
Andrei Mogoutov
Ecole de Hautes Etudes en Sciences Sociales
Thomas W. Valente
This
paper was presented in the Lin Freeman Festschrift session at the 20th
International Sunbelt Social Network Conference,
ABSTRACT:
The order in which people freely recall a set of words, persons' names, or
other items indicates how they organize those items in memory. An individual's
cognitive structure of the persons with whom he or she has some particular
relation can be described by noticing how he or she associates from one person
to the next during recall. Using improved statistical measurement, we review in
detail five studies that systematically examined associative patterns in the
recall of persons and evaluate competing hypotheses about the nature of these
patterns. Individuals in these studies recalled their acquaintances, coworkers,
and friends. Across studies, the results consistently show that persons
recalled adjacently or successively are perceived to interact more with each
other than those not recalled adjacently. No other factor describes associative
patterns as well as this notion of perceived social proximity. These results,
along with related research, imply the influence of social networks on memory
for persons and suggest a universal feature of human social cognition.
The order in which people freely recall a set of words, persons' names, or other items can indicate how they organize those items in memory (Puff, 1979). Associative patterns are one important aspect of the way people recall. Associative patterns refer to the connections or relationships between adjacently recalled items. The way an individual associates from one item to the next in free recall can reveal his or her cognitive structure of those items. Associative patterns may be identified by measuring clustering in recall. Clustering of items by a particular factor occurs when successively recalled items are more likely to share some characteristic or have some relationship than items not recalled successively.
Social psychologists have proposed and tested a number of different hypotheses about associative patterns in the recall of persons. Bond and Brockett (1987) postulated that memory for acquaintances is organized on two levels. At the broader level, individuals cluster acquaintances by the social contexts in which they encounter them (e.g., school, work, family, church, etc.). That is, people should tend to list acquaintances from the same social context adjacently in recall. Within social contexts, they asserted, acquaintances are organized in memory according to personality types. In their study, Bond and Brockett (1987) observed that undergraduates clustered acquaintances moderately to strongly by social context. Within social context clusters of adjacently recalled acquaintances, however, they found that subjects clustered by personality traits only weakly.
Fiske (1995) posited that acquaintances belong to one of four relationship mode categories (communal sharing, authority ranking, equality matching, and market pricing) and that individuals cluster acquaintances in recall according to these modes. He found that adults clustered acquaintances in recall moderately by relationship mode. Both Bond and Brockett (1987) and Fiske's (1995) hypotheses are categorical. Acquaintances are assumed to belong to only one social context, personality type, or relationship mode category, and associative patterns are hypothesized to correspond to these categorical structures.
Brewer (1995b) presented a somewhat different hypothesis about the organization of persons in memory. His hypothesis is that individuals recall and think about the people whom they know primarily in social network terms. Brewer (1995b) reviewed several studies that reported results consistent with this hypothesis. Specifically, these studies suggested that perceived social proximity (or social interaction) is the principal associative factor when people recall persons whom they know. In other words, persons recalled adjacently or successively tend to interact more with each other than those not recalled adjacently. These studies showed that subjects almost uniformly display highly statistically significant clustering by perceived social proximity. However, the most relevant studies reported no information on the magnitude of clustering by social proximity at the level of individual subjects because no methods were available for computing the standard measure of clustering with respect to non-categorical structures (e.g., perceived interaction patterns). This shortcoming prevented quantitative comparisons of the degree of clustering by different variables or with the results of other studies.
The calculation of clustering by such non-categorical variables as social proximity involves a formidable computational problem. In this paper, we describe this problem and our solution to it. Then we report on reanalyses of data from two previously published studies and present results for three other studies. The last of these studies tests the social proximity, personality, and relationship mode hypotheses directly. Furthermore, we compare quantitatively the results of these studies with other studies on associative patterns in the recall of acquaintances. Finally, we evaluate the different hypotheses about associative patterns in the recall of persons based on this research and note the implications of these results on techniques for eliciting personal and social networks.
Measurement of Clustering by
Non-categorical Variables
Roenker, Thompson, and Brown (1971) introduced the Adjusted Ratio of Clustering (ARC), now the standard measure of clustering in recall by a single variable for individual subjects. The ARC equals (o - e)/(m - e), where o is the observed clustering score, e is the expected (by chance) clustering score, and m is the maximum possible clustering score for a given subject. The ARC takes a value of 1 when clustering is maximal, a value of 0 when clustering is at the level expected by chance, and negative values when clustering is less than expected. The distribution of possible clustering scores is usually not symmetric (rather, typically skewed to the right), and therefore the ARC has no universally defined lower bound.
The ARC is easily computed when the variable is binary and categorical. In this case, o is the number of adjacently recalled pairs of items that belong to the same category. The expected clustering score, e, is computed as (pw/pt) * (n - 1), where pw is the number of pairs of recalled items that are in the same category, pt is the number of all recalled pairs of items, and n is the number of items recalled. The maximum clustering score, m, is the number of items recalled minus the number of categories represented in the recalled items.
Somewhat different procedures are required to compute the ARC when measuring clustering by a variable that is not categorical. First, for a non-categorical variable, a square hypothesized associative structure matrix must be constructed that contains the hypothesized associative strengths (such as social proximities) between each pair of recalled items. These strengths may be binary or valued. (The hypothesized associative strengths for categorical variables, or category memberships, may also be represented in this matrix form, e.g., with cells for same category pairs containing "1" values and cells for all other pairs containing "0" values). In the case of a non-categorical variable, o is the sum of the hypothesized associative strengths for the adjacently recalled pairs of items. The expected clustering score, e, is the mean off-diagonal cell value (hypothesized associative strength) multiplied by the number of items recalled minus one.
The maximum clustering score, m, is much more difficult to compute for a non-categorical variable than a categorical variable. The task of computing m actually is a case of the classic Traveling Salesman Problem (TSP) (Lawler, Lenstra, Rinnooy Kan, & Schmoys, 1985). In the TSP, the goal is to find the shortest route among a set of cities (like that for a traveling salesman) such that each city is visited once and only once. The TSP typically is discussed in graph theoretic terms, with cities referred to as nodes and the distances between pairs of cities as edges. In the clustering context, the hypothesized associative structure matrix may also be conceived of as a graph, with items as nodes and hypothesized associative strengths as edges. The goal of finding the shortest Hamiltonian path (a path in which each node is visited once and only once) in the TSP corresponds to the goal of finding the maximum clustering score, once proximity data have been appropriately converted into distance data.
When the number of items recalled is few, all possible permutations of nodes (or items in a subject's recall sequence) can be enumerated and the corresponding path lengths (or clustering scores) calculated. The maximum clustering score found in this enumeration is m. When the number of nodes/items recalled is greater than 10, the enumeration approach becomes infeasible. Mathematicians and computer scientists have struggled for decades to develop algorithms to solve the TSP to optimality (Lawler et al., 1985). We used Padberg and Rinaldi's (1991) algorithm to obtain provably optimal solutions to the TSP and thus m. This algorithm produces optimal solutions even in cases with large numbers (thousands) of nodes and with non-Euclidean data on the distances between nodes.
In this paper, we compare ARCs for different variables (including categorical and non-categorical variables) and different studies, focusing on ARCs for social proximity. We define social proximity broadly to include measures of interaction- and/or sentiment-based social ties, such as social interaction, friendship, and knowing. In the studies we describe, the hypothesized associative strengths for the social proximity variables refer to the social ties between the recalled persons.
The ARCs for two variables cannot be meaningfully compared if either is categorical. An ARC for a categorical variable can be high even when clusters defined by that variable account for few adjacently recalled pairs of persons. In such a circumstance and to the extent that subjects' recalls are patterned, some other variable(s) must underlie the associations among the other pairs of adjacently recalled persons. Moreover, even though a subject may display clustering by a categorical variable, he or she may also cluster by a second variable within clusters of the first variable. To address this latter issue, when data were sufficient and available, we measured clustering by social proximity within clusters defined by a categorical variable. Such analyses control for clustering by the categorical variable. If the categorical variable is the fundamental associative factor in a subject's recall, associations within clusters defined by that category should be essentially random with respect to other variables. If subjects cluster by social proximity within clusters of a categorical variable, then it indicates that social proximity provides a fuller, more detailed description of subjects' associative patterns and suggests that clustering by the categorical variable may be a coincidental byproduct of clustering by social proximity. For these control clustering analyses, we modified Bond and Brockett's (1987) control ARC measure. For each cluster defined by a categorical variable in a subject's recall, we computed the o, e, and m clustering scores. Then we summed the observed scores together, the expected scores together, and the maximum scores together across clusters to arrive at grand o, e, and m scores for computing a control ARC. We calculated a control ARC only when the subject had at least one cluster defined by the categorical variable with four or more persons or at least two clusters defined by the categorical variable with 3 or more persons each. Appendix A shows examples of how the ARC and control ARC are calculated.
Study 1
Method
Subjects. Subjects
were 25 college-aged members of a church-affiliated Christian fellowship of
Taiwanese and Taiwanese-American young people in southern
Procedure. Subjects performed two tasks: a recall task and a pile sort of persons by social proximity. All 25 subjects did the recall task, while only 11 (5 females and 6 males) did the pile sort task. For both tasks, subjects were interviewed individually by Yang (who was a member of the fellowship), usually after fellowship meetings or church services in a private room or a quiet setting out of sight and earshot of other fellowship members. For most subjects who performed both tasks, there was a 2-3 week interval between the recall and pile sort tasks; a few subjects performed both tasks during the same interview, with the pile sort task always following the recall task.
Yang gave the following instructions orally and bilingually to subjects for the recall task:
Who are all the people involved with the [name of the fellowship]? In giving your answers, please try to give first and last names, or as much of the person's name as possible. You do not need to mention your name or my name. List aloud the names of as many people involved in the [name of the fellowship] that you can think of.
No instructions were given regarding the order in which subjects were to list names. Subjects were given 10 minutes to mention persons (all subjects finished within 9 minutes, and the mean amount of time used by subjects was 2 minutes, 25 seconds).
After 20 subjects had done the recall task, the full name (or as much as was known) of each different person mentioned in the recall interviews was written on a separate 3" x 5" note card in both English letters and Chinese characters. Cards with the names of the few persons first mentioned in the last five recall interviews were added to the set as they became available. The instructions for the unconstrained single pile sort by social proximity followed in large part those used by Freeman, Freeman, & Michaelson (1988). Subjects were first asked to separate out from the set of randomly shuffled cards those persons whom they did not recognize, i.e., could not match the name with a face. Then, subjects were instructed to sort the cards into piles of persons who tended to like, interact with, and hang around each other, both at fellowship meetings and elsewhere (see Weller & Romney, 1988, for other details on the single pile sort). Subjects' responses to this task constituted their perceptions of the social proximities among persons in the fellowship--i.e., perceptions of the fellowship's social network. Individuals' reports of social proximity in pile sort tasks have been shown to be highly accurate with respect to observed interaction patterns (Freeman, et al., 1988; Webster, 1993/1994). For brevity, perceived social proximity will be referred to here simply as "social proximity."
For the measurement of social proximity ARCs, Brewer and Yang (1994) created a social proximity matrix including all the persons recalled by subjects. Following Freeman et al. (1988), the cells in this matrix contained proportions referring to the number of subjects who placed a pair of persons in the same social proximity pile in the pile sort task divided by the number of subjects who recognized both persons. Brewer and Yang constructed a social proximity associative strength matrix for the persons recalled by each subject based on this larger social proximity matrix. Social proximity data were available for 99 of the 105 persons recalled by subjects because a few persons first mentioned in the last five recall interviews were inadvertently not included in the last few pile sort interviews. These six persons were subsequently omitted from subjects' recall sequences when computing social proximity ARCs. In addition, persons who did not actually belong to the fellowship (intrusions) and self-mentions were also omitted from subjects' recall sequences for all analyses. Observed clustering scores (o) for recall sequences with repetitions were reduced by the number of repetitions multiplied by the ((uncorrected o for sequence including repetition[s]/(number of responses [including repetitions] - 1)) (see Brewer & Yang, 1994, p. 354). We made similar corrections, as necessary, when analyzing the data from the other studies as well.
Results and Discussion
The mean number of persons recalled by subjects was 30.4 (s.d. = 10.6), out of approximately 100 persons who had attended the fellowship in the year prior to data collection. Table 1 shows the ARCs for social proximity as well as several other variables; some of the categorical ARC results were also reported by Brewer and Yang (1994). On average, ARCs for social proximity are moderate as are those for kinship relations, first name, and first letter of first name. Subjects' recalls exhibit modest clustering by sex and fellowship section (high school-aged vs. college-aged) membership.
Table
1. Summary of ARCs for Study 1
|
Variable |
n |
Mean |
Median |
S.D. |
Range |
% positive |
|
Social proximity |
25 |
.40 |
.39 |
.12 |
.05/.57 |
100 |
|
within sex clusters |
23 |
.39 |
.47 |
.27 |
-.26/.83 |
91 |
|
Kinship |
24 |
.45 |
.44 |
.29 |
-.11/1.0 |
96 |
|
Fellowship section |
17 |
.27 |
.25 |
.35 |
-.20/1.0 |
77 |
|
Sex |
25 |
.25 |
.28 |
.16 |
-.13/.58 |
96 |
|
First name |
23 |
.76 |
.84 |
.24 |
.35/1.0 |
100 |
|
First letter of first name |
25 |
.32 |
.25 |
.26 |
-.01/1.0 |
96 |
|
controlling for first name pairs |
25 |
.18 |
.06 |
.30 |
-.12/1.0 |
68 |
Subjects' clustering by social proximity remains moderate even within sex clusters. Clusters of kin related, same first name, and same first letter of first name persons are very small (typically involving only 2 or 3 persons); therefore, no control ARC analyses are possible for these categorical variables. Indeed, clusters defined by kinship and same first letter of first name account for very few adjacently recalled pairs of persons (mean = 3.3 for each variable). We performed no control analyses for social proximity clustering within fellowship section clusters either, because subjects (who were all members of the college-aged section) recalled very few high school section fellowship members. In fact, 8 subjects recalled no high school section members at all. However, Brewer and Yang (1994) reported other evidence, based on data aggregated across subjects, that social proximity clustering is still moderate within clusters of persons defined by these variables. We performed a different kind of control analysis to examine the degree of clustering by first letter of first name after controlling for clustering by first name. For this analysis, we restricted our computation of the observed and maximum scores, o and m, for clustering by first letter to those possible paths or permutations of the recall order that included the observed adjacent same first name pairs (see Brewer & Yang, 1994, pp. 358-362). The results show that the clustering by first letter is almost entirely due to clustering by first name (i.e., persons with the same first name tending to recalled adjacently), as the mean and median of the control ARCs are quite small (see Table 1).
These results indicate that social proximity is the main associative factor in subjects' recalls of fellowship members. Subjects clustered persons in recall moderately by social proximity, even within clusters of persons defined by categorical variables.
Study 2
Method
Subjects. Subjects
were 13 employees (including 11 females and 2 males) of a department in the
public affairs division of a research university in the southwestern
Procedure. Thirteen subjects participated in the study, with 10 participating in two interviews, and 3 participating in only one interview. All interviews were conducted individually and privately. The first interview (for the 10 subjects who were interviewed twice) consisted of a free recall task. Brewer (1995a) gave the following instructions orally to subjects for the free recall task:
Who are all the people who work in the [department's name] Department? Please list aloud the names of all the people who work in the [department's name] Department. You do not need to mention your name.
No instructions were given regarding the order in which subjects were to list names and subjects were allowed as much time as needed to mention all the persons they could. When subjects appeared to be done or said they had listed everyone, Brewer prompted them once by asking if there were any other persons in the department.
The second interview (for those 10 subjects who were interviewed twice) occurred 2-3 weeks after the first interview. The second interview began with a recall task. Five subjects were assigned to a free recall task (as in the first interview) and 5 were assigned to an alphabetically directed recall task (see Brewer, 1995a for details on this assignment process). For the alphabetically directed recall task, Brewer gave the same oral instructions as in the first interview, except for the second sentence, which was replaced with: "Please list aloud the names of all the people who work in the [department's name] Department in alphabetical order by their first names as best as you can."
After the recall task in the second interview, subjects performed two quasi-successive pile sort tasks (cf. Boster, 1987; Freeman, et al., 1988). The full name (or as much as was known) of each different person mentioned by subjects in the first interview was written on a separate 3" x 5" note card. (No additional persons were mentioned in the second interview). Subjects sorted persons for two different social relations: how closely persons worked with one another (work proximity) and how much persons socialized with one another (socializing proximity). The order in which subjects performed the pile sort tasks was balanced across subjects. For each pile sort task, subjects were first asked to separate out from the set of randomly shuffled cards those persons whom they did not recognize, i.e., could not match the name with a face. For the work proximity pile sort task, subjects were instructed to:
Sort these
persons into different piles according to how much they work with each other on
job-related activities. Put persons that work with one another into the same
pile.
For the socializing proximity pile sort task, subjects were instructed to:
Sort these
persons into different piles according to how much they socialize with each
other, such as going to lunch together, meeting outside of work after hours,
and/or talking with each other about things unrelated to work or the
[department's name] department. Put persons that socialize with one another
into the same pile.
After the initial sort, a subject was asked to loosen his or her criterion for working (socializing) together and, if possible, join piles of persons into larger groupings on the basis of working (socializing) together. This step was repeated with further loosening of the subject's criterion until the subject did not perceive larger groupings (other than the whole department as one pile). At this point, the cards were rearranged into the piles the subject made in the initial sort. Then the subject was asked to tighten her/his criterion for working (socializing) together and, if possible, split piles of persons into smaller groupings of persons who worked (socialized) more intensely with each other. This step was repeated until the subject did not perceive finer groupings (other than each person as a single pile).
The 3 subjects who participated in only one interview performed the free recall task and the pile sort tasks in the same session. The analysis of these subjects' recalls is presented with the other subjects' first interview recalls.
We created a work proximity associative strength matrix based on data aggregated from individual subjects' pile sort responses. To construct this matrix, we ordered the groupings of persons sorted by a subject into levels from broadest (where the subject could not join any more piles) to narrowest (where the subject could not split any pile further). The work proximity of a pair of persons from the perspective of each subject is indexed by a proportion representing the number of levels the pair was placed in the same pile divided by the total number of levels that subject used in the task. The work proximity values for each pair of persons were averaged across all subjects who recognized both persons in that pair to arrive at the aggregated matrix. We created a socializing proximity associative strength matrix in the same fashion.
Results and Discussion
Subjects recalled a mean of 16.0 persons (s.d. = 2.1) in the first interview. Table 2 presents the summary statistics on the ARCs for several variables based on data from the first interview. Many of the results for the categorical ARCs were first reported by Brewer (1995a). Location distance refers to the shortest walking distances between each of the 19 employees who had offices at the department's main location, as measured on a blueprint of the main department location. The categorical status variable refers to the five organizational status levels in the department that ranged from director to student assistants/interns.
Table
2a. Summary of ARCs for study 2, first interview
|
Variable |
n |
Mean |
Median |
SD |
Range |
% positive |
|
Work proximity |
10 |
.56 |
.58 |
.12 |
.38/.79 |
100 |
|
within status clusters |
8 |
.21 |
.31 |
.59 |
-.82/.94 |
75 |
|
Socializing proximity |
10 |
.39 |
.42 |
.10 |
.23/.52 |
100 |
|
Status |
10 |
.45 |
.48 |
.29 |
-.07/.83 |
90 |
|
Sex |
10 |
.06 |
.15 |
.22 |
-.24/.28 |
60 |
|
First letter of first name |
10 |
.08 |
.07 |
.20 |
-.24/.43 |
70 |
|
Location distance |
|
|
|
|
|
|
|
Non-locationally oriented Ss |
10 |
.25 |
.22 |
.13 |
.11/.54 |
100 |
|
Locationally oriented Ss |
3 |
.63 |
.65 |
.04 |
.58/.67 |
100 |
Brewer (1995a) observed that three subjects' recalls are clearly locationally oriented as revealed by their spontaneous comments and inspection of their recall sequences. All summaries in Table 2 except for the location distance ARCs exclude these subjects. Subjects clustered fellow employees in recall by work proximity moderately strongly. For these subjects, clustering by socializing proximity and status are moderate. Clustering by sex and first letter of first name tend to be very slight to negligible. The non-locationally oriented subjects clustered modestly by location distance.
Clustering by work proximity within status clusters is somewhat weaker than work proximity clustering overall, yet still noteworthy. In analyses based on data aggregated across the non-locationally oriented subjects, Brewer (1995a) demonstrated that once clustering by location distance is controlled, clustering by work proximity still is present, while clustering by location distance essentially disappears once clustering by work proximity is controlled. The mean work proximity ARC is much higher than the mean location distance ARC (paired t = 6.07, df = 9, p < .001), and all (10/10) non-locationally oriented subjects had a higher ARC for work proximity than for location distance. Similarly, the mean work proximity ARC is substantially higher than the mean socializing proximity ARC (paired t = 2.73, df = 9, p < .05), and 8 of 10 non-locationally oriented subjects had a higher ARC for work proximity than for socializing proximity. Also, ethnographic evidence suggests that work relationships were more important than socializing relationships in shaping how subjects perceived each other (Brewer, 1995a). Therefore, we conclude that socializing proximity and location distance are not fundamental associative factors apart from their relationships to work proximity. These results indicate that, of the variables examined, work proximity appears to be the most general and primary associative factor.
The three locationally oriented subjects clustered by location distance moderately strongly. Their location distance ARCs provide an interesting comparison to the work proximity ARCs for the other subjects. The locationally oriented subjects consciously recalled their fellow employees in a locationally oriented manner. The mean location distance ARC for the locationally oriented subjects approximates the mean work proximity ARC for the non-locationally oriented subjects. This suggests that it is unlikely there is another variable conceptually and empirically distinct from work proximity that provides a better description of the non-locationally oriented subjects' associative patterns. It also suggests that this level of clustering may be close to the maximum level of clustering by any variable that might be observed in an empirical setting like this department, because the level of clustering observed in a subject's recall is likely to be highest when the associative strategy is deliberate and based on a concrete referent (as it was for the locationally oriented subjects).
The free recall subjects in the second interview had similar ARC results as subjects in the first interview (see Table 2b). The ARC results for the alphabetically directed subjects in the second interview parallel those for subjects in the first interview except that the magnitude of their ARCs for most variables is slightly lower and the degree of first letter of first name clustering is notably higher (see Table 2c). This indicates that subjects' natural associative tendencies are still evident even when they attempt to recall persons using a systematic and familiar strategy for organizing names.
Table
2b. Summary of ARCs for study 2, second interview, free recall subjects
|
Variable |
n |
Mean/Median |
Range |
|
Work proximity |
2 |
.64 |
.55/.73 |
|
within status clusters |
1 |
-.29 |
-.29/-.29 |
|
Socializing proximity |
2 |
.60 |
.57/.63 |
|
Status |
2 |
.24 |
.24/.24 |
|
Sex |
2 |
.12 |
-.24/.48 |
|
First letter of first name |
2 |
-.18 |
-.19/-.17 |
|
Location distance |
|
|
|
|
Non-locationally oriented Ss |
2 |
.42 |
.31/.53 |
|
Locationally oriented Ss |
3 |
.57/.66 |
.32/.72 |
Table 2c. Summary of ARCs
for study 2, second interview, alphabetically directed subjects
|
Variable |
n |
Mean |
Median |
SD |
Range |
% positive |
|
Work proximity |
5 |
.39 |
.40 |
.25 |
-.03/.69 |
80 |
|
within status clusters |
3 |
1.0 |
1.0 |
.00 |
1.0/1.0 |
100 |
|
Socializing proximity |
5 |
.20 |
.24 |
.26 |
-.13/.61 |
60 |
|
Status |
5 |
.19 |
.12 |
.17 |
.02/.42 |
100 |
|
Sex |
5 |
.11 |
.21 |
.23 |
-.19/.35 |
60 |
|
First letter of first name |
5 |
.38 |
.38 |
.35 |
-.16/.70 |
80 |
|
Location distance |
|
|
|
|
|
|
|
Non-locationally oriented Ss |
5 |
.19 |
.21 |
.18 |
-.12/.37 |
80 |
Study 3
Method
Subjects. Eleven
Russians who immigrated to
Procedure. Each subject was first asked to name aloud members of his or her personal network ("circle of relationships"). They were allowed to mention as many persons as they wanted and were given no instructions on the order in which to recall persons. Later in the interview, subjects were asked to indicate which other network members each network member knew by referring to a written list of network members. Subjects were also asked to provide information about the individual characteristics of their personal network members (sex, profession, country of current residence, age, and ethnicity/nationality) and the relationships with their network members (including, among other characteristics, duration of relationship and language of communication).
Two subjects were interviewed three times over several years. For one of these subjects, only data from the first interview are included in our analysis. For the other subject, only data from the third interview are included in our analysis because the data from the first two interviews are incomplete. Six subjects recalled one or more (range = 1 to 3) couples (e.g., parents, or another husband-wife couple) as a single response. We omitted these couples from analysis because these responses do not refer to individual persons.
The raw knowing data for each subject are nonsymmetric, most likely due to subjects' lapses of attention in responding. Because the knowing relation is by definition symmetric (or virtually symmetric), we symmetrized each subject's knowing associative structure matrix (where 1 = pair knows each other, and 0 = otherwise). Subjects' descriptions of their network members yielded easily classifiable information on seven categorical variables: sex (female/male), country/region of current residence (Russia / France / USA / Germany / Latvia / Israel / Belgium / Africa), age (10 year categories; 0-9 / 10-19 / 20-29, etc.), ethnicity / nationality (Russian / French / Chinese / Latvian / Japanese / Ukranian / Russian-French / Polish-French / Armenian / USA / Italian / Belgian / Polish / Jewish), duration of relationship (0-12 months / 13-60 months / 61-120 months / 121+ months), language of communication (Russian / French / English), and first letter of name (French spelling; each letter of the French alphabet). Network members with missing data on a particular variable were deleted from that subject's recall sequence for analysis on that variable.
Results and Discussion
Subjects recalled a mean of 22.0 personal network members (s.d. = 10, range = 10 to 41). Table 3 presents the ARCs for this study. On average, subjects showed virtually no tendency to cluster network members in recall by sex or first letter of name, and only a slight tendency to cluster by age. However, subjects displayed mild to moderate clustering by duration, language, ethnicity, and country of residence. The control ARC results indicate that clustering by knowing persists undiminished within clusters defined by these variables (we did not conduct control analyses with respect to language becuase it corresponded closely to ethnicity). Therefore, knowing appears to be the fundamental associative factor for these subjects. In addition, the couples who were recalled as "single" network members (and were omitted from our analyses) provide further evidence for association by knowing.
Table
3. Summary of ARCs for study 3
|
Variable |
n |
Mean |
Median |
SD |
Range |
% positive |
|
Knowing |
11 |
.34 |
.29 |
.22 |
.00/.73 |
91 |
|
within duration clusters |
5 |
.43 |
.27 |
.55 |
-.25/1.0 |
80 |
|
within ethnicity clusters |
10 |
.30 |
.38 |
.65 |
-1.0/1.0 |
70 |
|
within country clusters |
11 |
.36 |
.43 |
.34 |
-.25/1.0 |
82 |
|
Duration |
10 |
.25 |
.22 |
.24 |
-.14/.55 |
90 |
|
Language |
8 |
.37 |
.52 |
.41 |
-.26/.77 |
75 |
|
First letter of name |
11 |
-.04 |
.03 |
.16 |
-.25/.23 |
54 |
|
Sex |
11 |
-.03 |
-.17 |
.41 |
-.45/1.0 |
36 |
|
Age |
11 |