BEYOND AGENCY TO STRUCTURE:
The Research Problem
Sense-Making studies have from the beginning emphasized a focus on how actors see their movements in situations to be blocked or impeded. Fundamental to the Methodology is the assumption that whatever impacts structures have on individuals, these are mediated by actor sense-making. Thus, a series of Sense-Making studies of information seeking all hypothesized and found that actor assessments of blocks in their movements explained far more variance in information seeking and use than demographic descriptors of actors. 
While it can be argued that demographic descriptors of information seekers are a way of incorporating a concern for structural conditions of power arrangements into information seeking research, Sense-Making has traditionally assumed that these descriptors imposed as social markers are too far removed from the sense-making work that information seekers do as they move through their lives. Sense-Making assumes this for two primary reasons: One is that by being imposed from the outside, demographic descriptors do not capture directly enough the material conditions of actor lives that provide the context within which actors do their sense-making work. The second is that demographic descriptors capture too large chunks of time-space.
In a number of Sense-Making studies, demographic predictor variables have been labeled as "across time-space" measures and contrasted with "time-space bound" measures. Sense-Making's long-time use of actor assessments of how their movement is stopped in specific situations is an example of a time-space bound measure which has been shown to be a powerful predictor in past studies. In one early study (Dervin, Nilan & Jacobson 1981), for example, demography accounted for an average of 1.6% variance in information use criteria while time-space bound descriptors of the information seeking situation accounted for an average of 17.4%. Later studies have focused on determining the conditions under which demography should be a stronger predictor, hypothesizing that the more a sense-making behavior, external or internal, is constrained by societal conditions, the more likely demography will be a strong predictor. One very recent study (Dervin and Shields, 1999) of phone user sense-making found that demographic measures accounted for more variance in whether phone users took action to protect their privacy while situational measures predicted better how users evaluated the results of their actions. In that study, the authors' general conclusion emphasized that restricting understanding of users to demographic predictions alone yielded both a too shallow and a too chaotic picture of users' worlds.
The present study intervened in this argument in a different way. One difficulty with using demography to tap structural conditions of power is that demography is ascribed as attributes of people but, by definition, structural conditions of power are attributes of society or societal context. What the present study proposed, then, was that it would be useful to add assessments of structural conditions to Sense-Making's roster of predictors. The emphasis would be on doing so in such a way that predictors were not imposed on people as attributes carried from place to place but rather on situations and societal contexts within which people do their sense-making work.
Further, this study assumed that using actor assessments of structural conditions as predictors of information seeking would itself be a useful intervention in the "structure and agency" debate which pervades social sciences today. Traditionally, structure has been defined as being imposed upon human action but in the most recent versions of social theory, owing in large part to the generative work of Giddens , structure is seen as being maintained, mediated, resisted, changed, and even transcended by human agency. 
Briefly, one version of the structure-agency issue is this: How can we study the interplay of the constraining impacts of power structures and the not-totally-constrainable actions of humans as agents at the same time. One other quantitative Sense-Making study has addressed this same meta-question (Dworkin 1987), but in media studies rather than the information seeking and use research tradition.
Thus, the purpose of the study reported here was to assess the potential utility of adding explicit attention to structural power arrangements to quantitative Sense-Making studies of information seeking and use. Underlying this intention was an interest in assessing structural power arrangements within a Sense-Making framework while still conceptualizing them as attributes of societal conditions.
This contingent conceptualization of actor assessments of structural conditions as being at the same time both situated and pertinent across situations is what is seen as explicitly addressing the recent attention in social theory to the mediation of structure and agency. Further, this conceptualization of actor assessments of structural conditions rests on work in social theory that shows the persistent ways by which power arrangements are replicated across situations. At the same time, social theory also notes the often uncanny awareness of these arrangements by actors in situations and how this awareness arises from what might be considered lay sociological theorists' understandings of societal conditions. It is this uncanny awareness that forms the backdrop for actors' struggles and movements. 
Conceptualization of Predictor Variables
It is fair to say that while Sense-Making studies have often implicitly incorporated attention to issues of actor empowerment, most of the quantitative studies of information seeking and use have not addressed structural power arrangements in systematic and explicit ways. The strongest predictor in Sense-Making studies of information seeking and use has been a measure called situation movement state which locates actors in terms of how they see their movement impeded at the moment of information seeking and use. This idea is rooted in Sense-Making's conceptual foundation drawn from the work of Carter (1991) focusing on discontinuity. Sense-Making, thus, assumes that bridging gaps (internal and external, between time one and time two, between space one and space two, between self now and self yesterday, and so on) is axiomatic of the human condition. This does not imply that actors are problem-solving because Sense-Making assumes problem-solving is only one kind of gap-facing. Rather it assumes that gap-bridging is constant. In this framework, Sense-Making mandates that we study the phenomenon of sense-making (in this study, operationalized as information seeking) as gap-bridging.
With the application of the situation movement state construct, actors are conceptualized at specific moments of information seeking as, for example, on a road with two or more roads ahead of them (a decision situation); as being dragged down a road not of their own choosing (problematic situation); or, as following someone down the road who has traveled the road before (being led situation). This conceptualization of the actor's stop in movement at the instance of information seeking thus potentially addresses issues of power but indirectly and in a way that does not focus explicitly on structural power arrangements. Situation movement state as a predictor of information seeking, then, represents the perceptual description of sense-making movement constraints but does not directly attend to the socio-political context within which this movement is framed.
In order to explicitly add attention to structural power arrangements, this study operationalized two measures not used in previous Sense-Making studies. Both, as is consistent in Sense-Making studies, asked the information seeker, as the actor in the situation, to make a situational assessment. One of these measures was perceived relative status, focusing on the actor's assessment of his/her status relative to others in the information seeking situation. The second of these measures was perceived openness of communication, focusing on the actor's assessment of the communication environment which impelled the information seeking. These measures were drawn primarily from the attentions of critical scholars such as Habermas.  The specific assessments of power arrangements were done by actors in terms of how they saw things in specific situations. Actors were asked, for example: "In this situation, did you see yourself of higher status than other(s)? Did you feel welcome to communicate?"
It is important to re-emphasize here that the treatment of power arrangements in the present study was operationally anchored in specific situations and in this sense was both situated and contextualized, a basic mandate of the Sense-Making Methodology. Yet, at the same time the treatment of power arrangements in this study was assumed to pertain to across situation power arrangements (i.e. societal conditions).
The three predictor variables for this study were then: (1) perceived relative status (2) perceived openness of communication; and (3) situation movement state. In traditional analysis of variance terms, this study asked how much variance would be accounted for in information seeking by these three predictors operating individually and/or conjointly. The first two hypotheses were:
These two hypotheses rested on the assumption that structural assessments would be a useful addition to Sense-Making studies of information seeking and use. However, H2 was tempered by the emphasis in Sense-Making on the moment of gap, which is assumed to impel information seeking, and how this moment of gap is never fully constrained by structure. The assumption is that there is an inherent discontinuity between structure and action, which can potentially be bridged by creative agency.
The present study introduced an additional complexity with Hypothesis #3 which positions what is known in analysis of variance terms as "subject" variance conceptually as a possible predictor of information seeking. Hypothesis #3 focuses on the possibility of consistency in information seeker behaviors across situations and expects that while subject variance may be significant it will not account for as much variance in information seeking as situation movement state, perceived relative status and perceived openness of communication.
In introducing subject variance in this way, this study conceptualizes the individual information seeker as a potential carrier of constraints beyond those measured by the present study. The logic here is as follows: Given that information seeking and use is impelled by the unique conditions of a given moment or instance of sense-making, a moment of gap; and given that gap facing and bridging is internally driven by a unique moment of thinking, feeling, and sensing; then if we find consistency of individuals across situations, we can assume that this may be evidence of additional structural variables which were not tapped by this study.
What these structural variables might be is, of course, open to a diversity of arguments in current social science literatures.  At one extreme there is the nature versus nurture debate which argues about whether human consistencies across time-space (generally tapped by measures of demography, personality, cognitive styles, habits, and so on) arise from inherent transcendental structures of the person or from structures inculcated by societal conditions and material experiences. This debate is sometimes alternatively labeled the trait versus state debate. At the other extreme are various post-modern theorists who challenge the entire concept of transcendental human identity as being a manufactured idea, and argue that it is more appropriate to think of identity as de-centered. It is beyond the purpose of this article to enter this debate. Suffice it to say that our assumption was if we found consistency in information seeking across situations that this in some way pointed to untapped structures.
The next sections of this article provide operational definitions for the three predictor variables and the criterion variables, descriptions of the study design, the questionnaire and its fielding, and statistical analyses of the resulting data.
The predictor variables were conceptualized in advance with respondents being asked to recall actual situations that they saw as fitting the researchers' definitions. A description of this process is included in the section below on the questionnaire and fielding. The three predictor variables were operationalized as follows:
Situation Movement State
Based upon Sense-Making's movement metaphor, this variable tapped different ways that people see their movement as stopped. Sense-Making assumes that it is at the moment of stop that actors are impelled to make or remake sense and that it is at such moments that actors are more likely to engage in information seeking. The three situation movement states used in the present study were drawn from prior work: 
Perceived Relative Status
Perceived Openness of Communication
The criterion variables for this study were gleaned from an in-depth re-examination of information seeking items derived inductively in past Sense-Making studies. For purposes of the present study, information seeking is conceptualized as the asking of questions. A set of 18 different kinds of questions asked in sense-making situations were extracted from prior studies. Table 1 presents a list of the 18 questions. Each respondent was asked to indicate the extent to which each question was a question he/she had in each specific situation. These measures became, then, the set of original 18 information seeking criterion variables.
These original 18 information seeking criterion variables were factor analyzed to reduce the complexity of the results reported here. Six factor analytic information seeking variables constitute the criterion measures for this study. The selection of the specific factor analysis procedures used to create the criterion measures involved an intersection of a conceptual analysis of the 18 original measures with examination of a variety of different rotations (varimax and quartimax) and different sized factor solutions. The factor analysis selected for creating the criterion measures met three criteria: (1) Eigen values greater than 1.0; (2) The pattern of factors was consistent with the conceptual analysis; and (3) The pattern of significances and the plots of means reflected the pattern of findings obtained by examining predictive results for the original 18 items.
According to these criteria, a six-factor, varimax solution was judged to be optimal. The eigen values of the factors ranged from a high of 4.12 to a low of 1.02 and together accounted for 62% of the variance in the 18 original criterion variables. The naming of factors was driven by a conceptual analysis in conjunction with the highest factor loading, the communality, and factor purity. The resulting criterion measures consisted, then, of factor scores indicating the relative emphasis each respondent in each different situation placed on each kind of question asking. Factor scores ranged from -1.0 to +1.0 with a mean of 0.0. Table 1 shows which of the original 18 information seeking measures loaded highest on each of the final six factor score measures. The final six criterion measures tap the extent to which each respondent asked six different kinds of questions.
Table 1. Original 18 information seeking measures presented in descending order of their factor loadings according to their clusterings into the six final factor score measures of information seeking.
The Study Design
This study, in keeping with most quantitative uses of Sense-Making, focused on the capacity of different kinds of predictors to account for variance in information seeking criterion measures.
The chosen study design was a repeated measures design with equal n in each cell in a completely crossed 3x2x2 factorial design involving all combinations of the three predictor variables. Each of the factor analytic criterion measures were tapped once for each of the 12 cells in the matrix. Figure 1 shows the study design graphically and Figure 2 shows the database design.
The Questionnaire and Fielding
The questionnaire began with an introduction that described the Sense-Making notion of sense-making movement by employing a "path" metaphor and gave extensive definitions of the predictor variables, which were called "situational conditions." For each cell of the 3x2x2 design described above, each respondent was asked to select a specific situation which met the "situational conditions" (e.g. a situation "...where you had to make a decision, where the individuals in the situation were of higher status than you, and where two-way communication between you and the others did not seem easy or even possible"). In this way, each respondent described his/her information seeking in each of 12 situations. These situations were called "communication situations in your life, situations that were in some way troublesome to you or which required your attention."
After describing each of their 12 situations, respondents were asked to specify in words how each met the three situational conditions, which defined it. Then in the context of this anchoring in an actual situation from the respondent's life, each respondent rated each of the 12 situations in terms of the extent to which each of 18 different sense-making questions was "a question of mine." Ratings were on a seven-point Likert scale (1 = not at all a question of mine; 7 = very much a question of mine). Three different versions of the questionnaire were prepared to control for order effects, each with a different, random ordering of the "situational conditions."
The questionnaire was self-administered by students in an introductory communications class over a two-week period after a one-hour presentation of the instrument and respondent instructions. A total of 162 students returned questionnaires that were completely filled out.
There were three different analyses of the study data. The first was a content analytic "check" of the respondents' descriptions of the situational conditions to ensure that they were as intended. Employing standard content analytic procedures, a reliability coefficient was generated between a graduate student coder's judgment of predictor variable categories and respondent descriptions of the situational conditions. Lack of parallelism between the two perspectives was coded as a disagreement using a simple percentage agreement index. The index was computed by dividing the number of times the graduate student coder's judgments of situations agreed with the students' judgments by the total number of assessments made. The resulting reliability coefficients for the three predictors were all above 94%.
The second analysis was the factor analysis of the criterion measures as described above.
The third was a series of six analyses of variance for repeated measures -- one per information seeking criterion variable -- which were conducted in order to interpret the study's results in terms of the 12-cell factorial design described above. One adjustment was made to procedures because the typically used common error term (the mean square of the interaction between subjects and the predictor variable in question) was judged to be too liberal. A more conservative error term (the mean square of the sum of squares for all interaction terms involving subject variance) was chosen after extensive consultation with statistical experts.  This was seen as a more robust test of the variance accounted for because it compensated for the lack of independence inherent in the repeated measures design.
The analyses of variance for repeated measures indicated whether or not there was a significant relationship for the main effects or interactions or subject variance. In addition, the relative strengths of the individual predictor and subject sources of variance was estimated following procedures suggested by Vaughn and Corbalis (1969, p. 208). This provided a conservative formula for estimating percentage of variance explained.
Plan for Presentation of Results
Results are presented below in two sections. First, in order to anchor the results of the predictive analysis based on significance and variance accounted for, we focus on how the three predictors worked separately and conjointly for each information seeking criterion. The purpose is to provide a qualitative context grounded in actual respondent reports before examining the statistics pertaining to hypotheses which otherwise might seem highly removed from these reports. Following this qualitative presentation, we focused on the results across criterion measures because it was these comparisons which addressed the study's specific hypotheses.
In presenting our results, we drew from three tables. Table 2 presents the means for each criterion variable for each cell in the design, following the study design presented in Figure 1. Table 2 also includes the results of t-tests for correlated means for each of the six factor analysis-derived criterion variables showing which sub-means were significantly different from each other within a given application of the 3x2x2 analysis of variance test. Table 2 is presented primarily as necessary background information, for the discussions below are actually drawn from Tables 3 and 4. Table 3 extracts only the main effect means between the sub-groups of each of the three predictor variables showing which sub-groups were significantly higher or lower on a given information seeking criterion. Table 4 presents a summary of estimated variance accounted for by each of the predictor variables as well as the results of the analysis of variance for repeated measures across the six factor analytic criterion measures. Table 4 also reports on significant main effects for the three predictor variables and for subjects as well as for interactions among the three predictor variables.
Table 2. Within cell mean factor scores on the six criterion information seeking measures arranged according to the study design, including results of significance tests between all mean pairs in the 3x2x2 design.
Table 3. The main effect factor score means for each predictor variable sub-group with results of significance tests between predictor variable sub-group mean pairs.
Table 4. Summary of estimated variance and significance for the 3x2x2 repeated measures factorial analyses of variance across all six factor analytic criterion variables.
Results for Each Information Seeking Criterion
Because our purpose here is to provide a qualitative context for the discussion of statistical results pertaining to hypotheses, in the interest of brevity we focus in this section on findings that illustrate the significant main effects relationship reported in Tables 3 and 4.  Each of the six information seeking criteria is presented in turn below.
This information seeking criterion tapped actor emphasis on defining assessments -- "How are elements related to each other?" "What are different ways of looking at this?" Table 4 results showed that both subjects and situation movement state were significant sources of estimated variance accounted for with this criterion. Table 3 showed that respondents asked defining questions significantly more often in decision situations than in problematic situations (average factor scores of +.05 versus -.06 respectively) with their asking of defining questions in being led situations falling in between (-.01).
This information seeking criterion tapped actor emphasis on assessing action-related possibilities -- "How can I decide among my options/alternatives?" "What are my options/alternatives?" Table 4 showed that subjects, situation movement state, and perceived openness of communication were significant as main effect sources of estimated variance accounted for with this criterion. Table 3 showed that respondent asking of doing questions differed significantly between all three situation movement states. When in decision situations respondents showed the highest means (+.32), followed by problematic situations (-.08), and then being led situations (-.25). Further, when respondents were in open communication situations they were significantly more likely to report asking doing questions (mean factor score of +.06) than when in closed communication situations (-.06).
This information seeking criterion tapped actor emphasis on seeking other people's orientations -- "Does anyone agree with me?" "How do other people see this situation, what are their motives/reasons/plans?" Table 4 showed that subjects and situation movement state were significant main effect sources of estimated variance accounted for with this criterion. Table 3 indicated that respondents in problematic situations were significantly more likely to ask connecting questions (mean factor score of +.09) than when in being led situations (-.09) with decision situations falling in the middle of these two extremes (-.00).
This criterion focused on information seeking oriented towards helping the respondent escape the situation or at least understand the antecedent conditions -- "How can I make this situation go away?" "What caused this situation?" Table 4 showed that subjects, situation movement state and perceived openness of communication were significant main effect sources of estimated variance accounted for. Table 3 showed the likelihood of respondents asking removing questions differed significantly across the three situation movement states. When in problematic situations respondents asked the most removing questions (mean factor score of +.52), followed by decision situations (-.16), and then being led situations (-.36). Further, respondents were more likely to ask removing questions if they perceived the communication potential to be closed (+.18) than if they perceived it to be open (-.18).
This criterion focused on information seeking oriented towards future conditions -- "What will result from this situation?" "How can I avoid bad consequences?" Table 4 showed that subjects, situation movement state and perceived relative status were significant main effect sources of estimated variance accounted for with this criterion. Table 3 showed respondents were significantly more likely to ask projecting questions when in decision situations (mean factor score of +.09) than when in either problematic or being led situations (-.01 and -.08 respectively). Further, respondents were significantly more likely to engage in this type of information seeking when they saw themselves of lower status than when they saw themselves with higher status (+.04 and -.04, respectively).
This information seeking criterion tapped respondent attempts to motivate themselves -- "How can I get motivated?" Table 4 showed that subjects, situation movement state and perceived relative status were significant main effect sources of estimated variance accounted for with this criterion. Table 3 showed that respondents were significantly more likely to seek motivation when they were in being led situations (average factor score of +.11) than when they were in either problematic or decision situations (-.03 and -.08, respectively). Further, Table 3 showed that when respondents saw themselves as low status they were more likely to ask motivating questions (+.08) than when they saw themselves as high status (-.08).
Results for Each Hypothesis
In this section, we focus on the results across criterion measures because this comparison addresses this study's specific hypotheses.
This hypothesis addressed the utility of adding actor assessments of structural conditions as predictors in Sense-Making studies of information seeking.
Across the six criterion measures of information seeking, Table 4 showed perceived relative status related significantly to respondent likelihood of asking projecting and motivating questions (p<.01 and p<.001 respectively). Further, perceived relative status showed three significant interactions: with situation movement state for the asking of doing questions (p<.05): with situation movement state and with perceived openness of communication for the same criterion measure (p<.05) and for the asking of connecting questions (p<.05). In all, perceived relative status showed significance on 2 of 6 main effect tests; and 3 of 32 interaction tests.
Across the six criterion measures of information seeking, Table 4 showed perceived openness of communication related significantly to respondent likelihood of asking doing and removing questions (p<.001 on both). Perceived openness of communication also showed four significant interactions: with situation movement state for the asking of removing and projecting questions (both p<.001); and with both situation movement state and perceived relative status for the asking of doing and connecting questions (both p<.05). In all, perceived openness of communication showed significance on 2 of 6 main effect tests; and 4 of 32 interaction tests.
This hypothesis focused on the predictive power of situation movement state in comparison with that of the two structural variables.
Table 4 showed the estimated variances accounted for by situation movement state across all six criterion variables in comparison with that accounted for by the two structural variables. Situation movement state accounted for a higher percentage of variance explained with a total across the six information seeking criteria of 34.17% and a mean of 5.70%. This is compared with totals of 2.26% for perceived relative status and 5.01% for perceived openness of communication. The comparable averages across the six criteria were .38 and .84, respectively for perceived relative status and perceived openness of communication. With the exception of the criterion measure tapping the asking of motivating questions where both situation movement state and perceived relative status each explained 1.72% of the variance, situation movement state explained more variance than either of the structural assessment variables. Further, situation movement state showed a significant main effect relationship with all six information seeking criteria compared to only 2 of 6 for each of the structural assessment measures.
This hypothesis focused on whether we found consistency across situations in respondent information seeking. The expectation was that if such consistency was found it would account for less variance than that accounted for by the three situationally-based predictors. In Table 4, the test of this hypothesis involved comparing the non-subject variance accounted for with the subject variance accounted for under main effects. Non-subject variance was computed by summing variance accounted for from all main effects and all interactions, while excluding subject main effects.
Results in Table 4 showed that overall, across situation subject consistency accounted for less variance than situational consistencies. Thus, subject main effects totaled 13.16% across the six information seeking criteria, an average of 2.19%; compared to a 43.51% total and 7.25% average for the situational measures.
Despite this marked difference, it is important to note that subject variance was significant at p<.001 for all six information seeking criteria. For two of the six information seeking criteria (asking removing and doing questions), subject variance accounted for did not reach even 10% of that accounted for by the situational factors. For two others (asking projecting and motivating questions), subject variance accounted for was roughly 66% of that accounted for by the situational factors. And, for the remaining two information seeking criteria (asking defining and connecting questions) subject variance accounted for was greater than situational variance accounted for. For the latter measure, the two were almost equal. But for the asking defining questions criterion, subject variance accounted for was about ten times greater than situational variance (5.41% accounted for versus .56%).
Summary and Conclusions
In this study, structural assessment measures were not as strongly predictive of information seeking in terms of variance accounted for as our hypotheses expected. Only four out of 12 possible main effects and 5 of 24 interactions were significant. Given, however, that very conservative approaches to statistical presentations were used and that the resulting significances were beyond what would be expected by chance, the results suggest that attention to actor assessments of structural conditions is a promising avenue of attention for Sense-Making studies of information seeking.
Hypothesis #1, thus, was moderately supported by the results of this study. One or the other of two structural variables was a significant predictor of four of the criterion variables. In general, the structural predictors were least active for those aspects of information seeking and use which on the surface involved the actor implementing a sense of freedom to move (i.e. asking doing and connecting questions). In contrast, they were most active where information seeking focused on high constraints either external or internal (e.g. wanting the situation to go away as indicated by asking removing questions, or being concerned about getting movement started as indicated by asking motivating questions).
A comparison between the significances for the two structural predictors was also interesting. perceived relative status was the stronger of the two in predicting the asking of projecting questions (e.g. What will result? How can I avoid bad consequences?) and motivating questions (e.g. How can I get motivated?), while perceived openness of communication was the stronger of the two in predicting the asking of doing questions (e.g. How can I decide among my options/alternatives?) and removing questions (e.g. How can I make this situation go away?). It is telling that the status predictor related more strongly to information seeking criteria that indicated a struggle with motivating self and the future while the openness of communication predictor related more strongly to the here-and-now situation. While it is beyond the purpose of this report to delve deeply into the possible explanations for these results, the results have an underlying coherence to them suggesting that there is potential value for future study.
It is possible to conjecture why the structural assessment measures did not play a stronger role. One explanation coming from critical theory involves the ways in which power structures do their work out of consciousness. It may be that while actors can identify the forces at play they are not as ready to implement a sense of agency vis-a-vis these forces. It may be that a more in-depth interview is required as Dervin (1999, in press) has suggested. This possibility is discussed below.
One important aspect of this study to note here is that respondents were asked to rate each of the original information seeking criterion items, regardless of whether or not they actually asked the question out loud. It may be that in situations that are very highly constrained by external structural conditions, respondents do not need to define the situation or examine their connecting options because the possibilities are clearly inherent in the situation itself.
Hypothesis #2 was strongly supported by this study. Thus, situation movement state was confirmed in this study as it has been in past studies as the strongest Sense-Making predictor of information seeking. Situation movement state clearly taps constraints and stops in movements in ways that respondents resonate with. There may be, as Dervin suggests, a kind of universal interface in human focus on movement. Dervin terms this a "verbing' conceptualization (Dervin, 1999) and suggests that it allows actors to name the substantive aspects of their own worlds while at the same time allowing them to focus on their own struggles and achievements.
A difficulty with the movement metaphor, however, which has been unearthed by qualitative applications of Sense-Making, is the time it takes in the research encounter to encourage and assist respondents to connect their struggles and achievements explicitly to societal structures. This process, which Freire labeled as conscientizing (1970), is required if a large number of respondents are to attend specifically to structural issues. As has been shown in Freire's work on critical pedagogy, conscientizing takes time and involvement in dialogue. In an as yet unpublished study of white racism, Dervin (personal communication, 1999) found that while some white students immediately had a great deal to say about how they were taught to be racist, it took most students five or more weeks of intensive self-examination and dialogue to begin to remember their childhood struggles with the lessons they were being taught by a racist society.
From the perspective of the current study and its mandate to utilize actor assessments of structural conditions in quantitative studies, there seem to be three possible alternative strategies. One is to implement actor assessment measures as suggested in the present study and accept that the results will be weaker than hoped for but a potentially useful cut as if into the top of a mountain. The second is to introduce external judgments of structural conditions with the understanding that they will be emphasizing the structure in structure-agency mediation. The third is to construct one's study in such a way that over a period of weeks respondents are invited to reexamine and talk about their situations in terms of power dynamics. The latter alternative is one, of course, which requires a form of researcher direct involvement in the research situation as in action research.
The conceptual basis of Hypothesis #3 was a major assumption guiding this study, i.e. that subject variance may carry with it unarticulated response to constraining structural conditions not tapped by this study or may be reflective of personal behavioral habits which may or may not be related to structural conditions.
In this study, non-subject estimated variance accounted for was far greater than estimated subject variance by a factor of three to one. However, estimated subject variance was a significant source of variance for each of the six information seeking criteria. Further, subject variance tended to play more of a role when variance accounted for by the strong situation movement state predictor was lowest. All of this is suggestive of the idea that subject variance might be usefully reconceptualized as a carrier of untapped and unarticulated structural constraints. This is, of course, an important departure from traditional information seeking studies that have utilized subject variance primarily in one of two ways. One is by assuming that subject variance taps the essential identity of the individual which moves with that individual across time and space. The second is actually a statistical use -- extracting subject variance in order to remove it from error variance so that there is a higher probability of achieving main effect significances.
The primary weakness of this study is intimately tied up in its exploratory nature, i.e. it was an attempt to expand the coverage of Sense-Making studies of information seeking and use by examining the potential utility of adding explicit actor assessments of societal power arrangements as they are manifested in specific situations. In doing this through the use of two measures of structural conditions (perceived relative status and perceived openness of communication), this study limited the potentially vast range of other structural predictors that could productively be used. However, this is also the primary strength of this study. It attempted to move quantitative applications of Sense-Making from its former implicit attention to issues of societal power to explicit attention. In doing so, it provided an exploratory example of an attempt to attend to both structure and agency in one study.
 The present study is drawn from the senior author's dissertation completed in 1985. As a dissertation using Sense-Making, it was the first to pursue the possible introduction of critical concepts of power and structure in systematic and explicit ways; and the first to create a set of criterion measures of information seeking based on a factor analysis of a roster of sense-making questions drawn from prior Sense-Making studies. The presentation here sets the data in current discourses of social sciences and the most recent articulations of Sense-Making Methodology. While the specific details of current arguments in these discourses have changed since the dissertation was completed, the argument the impelled the study is as pertinent today as then. The data is previously unpublished. Because it focuses on actor selected information seeking situations there is no reason to believe it is dated in any way.
 This article relies heavily on literature reviews and assessments of the state of information seeking studies presented in Dervin (1999, in press), Dervin (1992), and Vakkari, Savolainen, and Dervin (1997).
 For examples of authors explicitly attempting to enter the structure-agency intersection and in doing so melding a variety of traditional oppositional approaches to the social sciences, see Murrow (1994), Murdock (1997).
 The authors are grateful to Edward Fink (University of Maryland) and Jeffrey Katzer (Syracuse University) for their assistance with this issue and their help with the derivation of a conservative formula for estimating variance explained.
Benhabib, S. (1992). Situating the self. New York: Routledge.
Bronner, S.E. & Kellner, D.M. (Eds.). (1989). Critical theory and society. New York: Routledge.
Carter, R.F. (1991). Comparative analysis, theory, and cross-cultural communication. Communication Theory, 1 (May), 151-158.
Dervin, B. (1983). An overview of Sense-Making: Concepts, methods, and results. Paper presented at the annual meeting of the International Communication Association, Dallas, TX, May. (available from junior author Dervin).
Dervin, B. (1992). From the mind's eye of the user: The Sense-Making qualitative-quantitative methodology. In J. D. Glazier & R. R. Powell (Eds.), Qualitative research in information management (pp. 61-84). Englewood, CO: Libraries Unlimited.
Dervin, B. (1999, in press). On studying information needs and seeking methodologically: The implications of connecting meta-theory and method. Information Processing and Management.
Dervin, B. & Shields, P. (1999). Adding the missing user to policy discourse: Understanding US user telephone privacy concerns. Telecommunications Policy, 23, 403-435.
Dervin, B., Jacobson, T. L. & Nilan, M. S. (1982). Measuring aspects of information seeking: A test of a quantitative-qualitative methodology. In M. Burgoon (Ed.) Communication Yearbook 6 (pp. 419-444). Beverly Hills, CA: Sage Publications.
Dervin, B., Nilan, M. S. & Jacobson, T. L. (1981). Improving predictions of information use: A comparison of predictor types in a health communication setting. In D. Nimmo (Ed.), Communication Yearbook 5 (pp. 807-830). New Brunswick, NJ: Transaction Books.
Dworkin, M. (1987). Making sense of television news: Situation, context, and psychology of the audience experience. Unpublished PhD dissertation, University of Washington.
Faubion, J. D. (1995). Rethinking the subject. Boulder, CO: Westview Press.
Freire, P. (1970). Pedagogy of the oppressed. New York: The Seabury Press.
Freire, P. (1994). Pedagogy of hope. New York: Continuum.
Fraser, N. (1989). Unruly practices: Power, discourse, and gender in contemporary social theory. Minneapolis, MN: University of Minnesota press.
Giddens, A. (1984). The constitution of society: Outline of a theory of structuration. Cambridge, UK: Polity Press.
Habermas, J. (1979). Communication and the evolution of society. (T. McCarthy, Trans.). Boston: Beacon Press.
Habermas, J. (1981). The Theory of Communicative Action, Volumes 1 & 2. (T. McCarthy, Trans.). Boston: Beacon Press.
Habermas, J. (1987). An alternative way out of the philosophy of the subject: Communication versus subject-centered reason. In J. Habermas The philosophic discourse of modernity: Twelve lectures. Cambridge, MA: MIT Press.
Hollinger, R. (1994). Postmodernism and the social sciences. Thousand Oaks, CA: Sage.
Kramer. E.M. (1997). Modern/postmodern. Westport, CN: Praeger.
Murrow, R.A. (1994). Critical theory and methodology. Thousand Oaks, CA: Sage Publications.
Murdock, H. (1997). Thin descriptions: Questions of method in cultural analysis. In J. McGuigan (Ed.), Cultural methodologies (pp. 178-192). London: Sage.
Nilan, M. S. (1985). Structural constraints and situational information seeking: A test of two predictors in a Sense-Making context. Unpublished PhD dissertation, University of Washington, Seattle.
Outhwaite, W. (1987). New philosophies of social science: Realism, hermeneutics, and critical theory. New York: MacMillan.
Rabinow, P. & Sullivan, W.M. (Eds.) (1987). Interpretive social science: A second look. Berkeley, CA: University of California Press.
Vakkari, P.; Savolainen, R. & Dervin, B. (Eds.) (1997). Information seeking in context. London, UK: Taylor Graham, 13-38.
Vaughan, G. M. & Corballis, M. C. (1969). Beyond tests of significance: Estimating strength of effects in selected ANOVA designs. Psychological Bulletin, 72, 204-213.
Copyright 1999 Communication Institute for Online Scholarship, Inc.
This file may not be publicly distributed or reproduced without written permission of the Communication Institute for Online Scholarship, P.O. Box 57, Rotterdam Jct., NY 12150 USA (phone: 518-887-2443).