PENNSYLVANIA RURAL COMMUNITIES AND INDIVIDUAL MENTAL HEALTH

 

Steve Jacob , A.E. Luloff & Jeffrey C. Bridger

 

 

ABSTRACT

 

Mental health has usually been described as dependent upon individual circumstances and events, and stress is often a key factor.  Stress has been studied extensively as an individual phenomenon.   Little research has examined the role of the community in the stress process.  This represents a significant gap in the literature as the community is an important context in the individual's life.  Wilkinson (1979; 1991) offers strong theoretical support that links the rural community to individual mental health, while most assume that community contributes to an individual's mental health, this relationship has not been empirically established.   In this research, analyses revealed that few community-level factors directly influenced mental health.  However, the influence of community factors were seen in an interaction model.  The theoretical and ameliorative implications of these findings are discussed.

 

 

INTRODUCTION

 

 

Human beings lead complex lives which expose them to varying levels of stress.  Few have recognized that rural life can be as stress-inducing as city life, and that rural communities do not always provide idyllic living conditions (Willits & Luloff, 1995).  The farm crisis of the 1980s helped to focus popular and academic attention on rural issues, including the existence of non-farm rural stress and associated outcomes such as depression (Srole, 1978; Mazer, 1983; Mazer, 1976; Wagenfeld, Goldsmith, Stiles, & Manderscheid, 1988).  Despite this work, the dimensions and contexts of rural stress are still not fully understood.  Chief among the gaps in knowledge is the lack of research that examines how the characteristics of rural communities may affect mental health.

           

Stress is a generic term one associated with mental tension or strain and is viewed as a generalized response of the body to a stimulus (Geller, Bultena, & Lasley, 1988; Krannich, Riley, & Leffler, 1988).  The subjective feeling of stress is derived from a stimulus (stressor) and from environmental demands (; Krannich, Riley, & Leffler, 1988).  Because stress is a generalized and subjective experience, individuals have unique responses to stress.  Outcomes of stress can be physical and/or psychological.   Depression, anxiety, alcohol and drug abuse, and domestic violence are cited as frequent outcomes of stress (Pearlin, 1989; Belyea & Lobao, 1990).  Often individuals display multiple outcomes (Thoits, 1995).

           

Aneshensel (1992) pointed out that groups differ in their vulnerability to stress.  Factors that help establish the social status of the individual have been shown to be important in explaining stress outcomes (Aneshensel, 1992; Pearlin, 1989; Thoits, 1987).  Additionally, stress research often is conducted together with life events research, which focuses on undesired changes such as the death of a spouse or legal troubles (Pearlin, 1989).  Such events are called stressors and they produce stress and associated outcomes including depression (Cohen, Karmarck, & Mermelstein, 1983; Mirowsky & Ross, 1979).  Coping and mediation can lessen the impacts of stressors.  Coping, or the ability to handle stress without negative consequences, differs from mediation, which implies the lessening of the effects of stress.  Coping is an individual-oriented action, but is learned from one's reference groups and membership in other groups (Pearlin, 1989).  Mediators are social supports, including sympathetic friends, who help to alleviate or lessen stress (Pearlin, 1989; Maton, 1989).  Both coping and meditation, as operationalized in much of the literature, are recorded as individual or interpersonal traits and characteristics (Maton, 1989).  Concepts such as sense of belonging to a community, associations, and other forms of community-oriented social support are left as abstractions (Maton, 1989).  This is notable, as Thoits (1995) has hypothesized that it is possible that whole groups, such as churches, neighborhoods, and communities, may function as sources of perceived social support. 

           

Frequently, indicators of depression are employed as the dependent variable in stress research (Pearlin, 1989; Aneshensel, 1992).  Depression is a complicated retardation of positive emotions and motor functions and includes symptoms such as loss of appetite, insomnia or hypersomnia, and inability to concentrate (Eaton & Kessler, 1981).  Depression varies in severity from mild mood swings to extreme psychoses (Eaton & Kessler, 1981).  In this research individual-level indicators of status, stressors, and social supports are used to predict depression index scores.  These factors represent a conventional modeling approach to depression.  Additionally, community-level factors will be included in some analyses.  Individual-level factors are expected to proximally affect depression index levels.  The community-level factors are expected to influence depression index levels through main effects and through interactions with individual-level variables.  Interactions are anticipated because the community-level factors provide a context for individual experience.  The community-level variables may be more influential through modifying the relationships of individual-level variables to the dependent variable than what is seen through the main effects.      

           

This is not to say that the community is the only context of a person’s life.  The stress process is generally conceived as almost exclusively the function of individual responses and circumstances (Pearlin, 1989). Usually these circumstances are empirically measured by indicators of status, such as personal characteristics.  However, there are other important circumstances that have received an increasing amount of attention in the literature.  For example, Eckenrode and Gore (1990) have challenged this static approach and argued that the context of the individual’s life must be considered.  They felt that “...life events could no longer be viewed as purely external environmental assaults on individuals, unconnected to the structure of people’s daily lives” (Eckenrode & Gore, 1990:3).  Similarly, they noted that social supports cannot be thought of as unchanging resources that neutralize the impact of stressors (Eckenrode & Gore, 1990).  They developed a conceptual model that sets the immediate contexts of work and family.  These contexts offer connections between the key junctures in the stress process (Eckenrode & Gore, 1990).  Considerable research has since examined the effect of the contexts of family and work on stress (see Thoits, 1995, and  Eckenrode & Gore, 1990 for recent reviews).  However, relatively little has been done with the context of community.  This study explores the community as an important context in the individual’s life.  While some family and work factors are included, only the community-level factors will set the context in these analyses. 

     

These variables have been used frequently in well-being and stress research.  In general, conditions associated with higher socioeconomic status reduce stress outcomes (Thoits, 1995).   This stems from the idea that coping strategies and the occurrence of many negative life events are distributed most favorably to those higher up in the social stratification system, and in general this seems to hold true (Thoits, 1995).  However, there is quite a bit of variation in these responses by social demographic category.  Social psychologists and others attribute much of this variation to role identity, meaning that personal values towards a role, beliefs, and other circumstances can make those with relatively high status more vulnerable to the effects of stress than a strict stratification interpretation would suggest (Thoits, 1995).  Increased social supports should reduce stress outcomes, though there may be significant variation across characteristics and social roles (Thoits, 1995).  In this research, the emphasis is upon social supports that are community or group-oriented, as they have been given less attention in the literature.  A major theme in the literature has been the “buffering” effects that certain coping and mediation strategies, such as social supports, represent for people faced with stressful events.  Thoits (1995:64) described social supports as a social fund that people can draw upon when handling stressors.  This social fund can work to buffer the individual from damaging physical and mental impacts of life events and chronic strains (Thoits, 1995:64). 

     

These categories were later used in the analyses as a variable called community type and reflect sustenance organization and also potential barriers to social interaction.   Economic change reflects the sustenance organization of the community.   Communities with newer-styled economies may experience higher levels of stress outcomes, which could possibly be attributed to the stress-inducing shift in economic structure that many rural communities in Pennsylvania could have made.  The second component in this measure impacts opportunities for social interactions.  Population change can occur through inmigration, outmigration, natural increase, or natural decreases.  In these Pennsylvanian communities, migration is the leading contributor of population change.   Population loss is expected to be correlated with individual-level depression, because as people migrate, they disrupt their social networks. 

           

Population decline has a great social and economic impact on rural communities.  A significant share of such losses is often in the outmigration of younger residents who are forced to move to pursue economic opportunity.  When this occurs, these people often leave relatives and friends who make up the majority of their primary ties.  Maintaining these ties is possible but difficult, because the social networks between migrants and those who remain within the locality are disrupted. 

           

The loss of population base can impact those who remain in other ways.  As the younger and better educated residents leave, a significant loss in human capital occurs, which makes the recruitment of high technology and service sector businesses less likely (Jensen & McLaughlin, 1995; Killian & Beaulieu, 1995).  Similarly, fewer people remain to pay local taxes, making average fixed costs higher for those who remain.  Losses in population also are likely to be reflected in corresponding reductions in social services (Hassinger & Pinkerton, 1986).  Finally, the dependency ratio increases.  This means there are fewer working-aged residents to help support those who are retired or very young.  O’Brien, Hassinger, and Dersham (1994) showed that a “less viable” community, which among other things had shown decreased enrollment in high school and population loss, had residents with higher depression index scores than residents of a “more viable” community.

           

Other rural areas face a different set of problems.  "Boomtowns" are areas of rapid growth and a topic of continuing debate (Krannich & Greider, 1984).  Some argue that rapid growth areas create great amounts of strain and stress on their populations (Cortese, 1982; Cortese & Jones, 1977).  Gilmore (1976) describes a situation in a boomtown where goods and services cannot be supplied to residents, and alienation, divorce, abuse, alcoholism, school dropout rates, and crime all radically increase.  Wilkinson (1991) questioned the validity of Gilmore's assertions and the research of others that relied on Gilmore as a primary source, as they believed Gilmore relied upon anecdotal evidence to support his claims.  In other cases, extreme increases in various rates are easily explained by low or no occurrence of phenomenon, followed by the report of a small number of incidents.  The number of incidents is reported as a huge percentage increase, but when considered with the population size and the increase of policing, these few occurrences seem less important (Wilkinson, 1991).  Nonetheless, concern about rapid growth and quality of life remain well-founded (Wilkinson, 1991).

           

The percent of population aged 18-64, reflected not only the age distribution of the population, but also the percentage of population that was of prime wage earning years.  A larger percentage here reflected a population base that was more likely to be in the labor force (Hassinger & Pinkerton, 1986).  In this situation, a larger percentage also indicated that there was a smaller dependent population.  The dependent population is generally not in the labor force and includes children and retirees.  A larger percentage of residents in the dependent population in a community can strain services through increased demand, while there are fewer wage earners to pay supporting taxes (Hassinger & Pinkerton, 1986).  Though there is no research that indicates at what point a community is “strained,” the information does allow for comparisons across the communities.  The mean for this variable was .58, the standard deviation was .03, and the range was .48 to .65.

           

The percentage of population with college degrees reflected education levels within the ecological community.  Increased levels of human capital and retention of better educated residents suggests that there are good economic opportunities within the community (Jensen & McLaughlin, 1995).  Further, communities with better educated residents tend to be more active in social and economic development (Martin & Wilkinson, 1985).  Education levels are important stratification variables, and those communities with residents with higher levels of education are likely to have better life circumstances.  These conditions within the community are expected to increase well-being and reduce reporting of depression symptoms.  The mean for this variable was .12, the standard deviation was .04, and the range was .04 to .32.

           

Average household income reflected the general affluence of the community.  Communities that are more affluent generally provide more comprehensive and better public and private services and have healthier economies (Wagenfeld, 1990; Hamilton, Broman, Hoffman, & Renner, 1990).  Therefore, it was likely that communities with higher average household incomes had residents that reported lower occurrences of depression.  The mean for this variable was $29,336, the standard deviation was $5,546, and the range was $15,150 to $45,506.

      

Last, female-headed households reflected the gender distribution of family heads within a community.  Most female-headed households face economic and social problems that put them at special risk for stress and distress outcomes (Jacob & Willits, 1994).  Further, Kassab and Luloff (1993) found that types of households are a key factor in economic strain experienced by the family.  These family strains may be reflected in increased use of social welfare supports and services within the community.  Divorced and single-parent families may reflect undesirable social conditions within a community (Ross, Bluestone, & Hines, 1979).  As such, it was expected that higher percentages of female headed households in a community would increase reports of depression.

 

     

Theoretical Perspective: The Community and Individual Well-Being

     

Individual well-being is, according to Wilkinson (1991:68), affected proximally by private experiences and intimate relations, including the contexts of work and family.  Nonetheless, the community has a role in fostering individual well-being.  The link between community and individual well-being can be generalized to three main domains.  First, community is the setting where individuals meet their basic needs for survival (i.e. sustenance).  Second, the community provides opportunities for social interaction, which influences development of the self, personality, self-actualization, and participation in local social fields. Third, the community provides an arena for collective involvement and actions.

                       

Sustenance

 

According to Wilkinson's (1991) perspective, community emerges as a dynamic process from people interacting with each other on place-relevant issues that comprise a common life.  An important component of this interaction includes the allocation and management of resources.  Though in complex societies pecuniary interests predominate, sustenance issues remain important.  Localities provide the place or setting where sustenance is secured and daily needs are met.  In the attempt of individuals to meet their primary needs, local society can develop.  Local societies have been described as cohesive and complete groupings, often with distinct cultures and institutions.  These local societies are self-perpetuating, constrained within a bounded area, and comprised of residents who often share a sense of place (Wilkinson 1991). 

 

Wilkinson (1991:2) described local society as "a comprehensive network of associations for meeting common needs and expressing common interests."  Localities and local societies provide what Wilkinson (1979) called sustenance organization, which meets needs for food, safety, shelter, and gratification.  This organization provides solutions to the basic problems of survival and self-serving desires.  To the degree that such an organization is efficient a greater amount of human energy and attention can be placed on self-actualization.  Self-actualization is described by humanist psychologists as a striving for independent competency in a socially responsible fashion or realization of one's potential, often stated as higher order needs (Wilkinson, 1979; Allport, 1955; Maslow, 1954).  Self-actualization increases self-respect and worth, which in turn enables an individual to respect and trust others.

                       

Social Interaction

 

The self, as related by Mead (1934) and Blumer (1969), is a social experience that rises out of phases or processes of social interaction and connects the individual to others.  If community conditions affect social interactions negatively, the self may only be experienced in partial or distorted form.  Social conditions that minimize interference with social interactions allow individual well-being, a natural state, to occur (Wilkinson, 1991).  As such, "social conditions only set the stage" for individual well-being; they are not a cause (Wilkinson, 1991:70).  Social conditions do not drive well-being, but rather impede its emergence.  Outside of one's immediate family, the interactions that occur within the bounds of community most directly impact the development of self (Bridger, 1994).  Thus, the composition of the community, such as the age distribution and education levels, can impact the nature of social interactions.

     

Both Bridger (1994) and Wilkinson (1991) state unequivocally that anything that inhibits interaction suppresses the emergence of community.  Luloff and Swanson (1995) developed this idea further, specifically noting that bigotry, anti-Semitism, racism, sexism, ageism, and other causes of discrimination are just some of the barriers that cause disaffection in community interaction and retard the emergence of community.  Obviously, communities characterized by such "isms" undermine their residents ability to realize their potential individual well-being, particularly in terms of social support, but also in their ability to meet higher order needs of collective involvement and action (Wilkinson, 1991). 

      

Collective Involvement and Actions

      

In his field theory of community, Wilkinson (1991) described the community field.  Like any social field, the community field consists of actors, associations, and activities.  However, the community field has a structural aspect that links parties and coordinates actions (Wilkinson, 1991). The emergence of the community field can enhance the capacities of the collective to take actions that can directly improve the lives of community residents (Wilkinson, 1991; 1993).  Though community is thought to be ubiquitous, the emergence of the community field is a more specialized case that brings about a social state conducive to individual well-being and self-actualization.  The community field does not pursue any single interest, as most social fields.  Rather, it pursues the general community interest (Wilkinson, 1991).  The actions in the community field serve to coordinate other social fields (Wilkinson, 1991).

           

When community actions occur a generalized linking structure among the social fields develops that can be used again for action.  Purposive community actions can improve the lives of individuals by strengthening local sustenance organization and improving social conditions.  Participation in collective involvement and actions is an important component of individual well-being.  Wilkinson (1991:79) described the link of collective involvement and actions to individual well-being as the validation of the self as an active contributor to the process of improving a shared life.  This affirms social responsibility and esteem in the individual.

           

The emergence of the community field may help to integrate individuals into a community.  When individuals participate in collective involvement and actions, there are more opportunities for social interaction.  Importantly, these actions are almost always voluntary.  Often, the parties that interact may ordinarily have nothing in common.  However, their shared participation in a community-oriented collective action and/or displays of solidarity can function as a mechanism for social integration.   It may be that individuals who are well-integrated into their communities are able to buffer the stresses that are associated with contemporary life (Maton, 1989).  In this sense, community-oriented participation and displays of solidarity may directly enable community residents to cope with and mediate life stress. 

           

This community participation may have increased the social ties of an active individual.  Community provides an arena for social interactions which can be primary, secondary, or tertiary in nature.  Generally, relationships with immediate family members, close relatives, and friends are thought to be primary.  These relationships play an important role in personality and self-development due to the intensity and frequency of interactions.  Further, primary relationships are conducted under specific norms in face-to-face contact that often occurs with many or most of the primary group members present.  Social network theorists refer to these relationships as "strong ties" (Granovetter, 1973).  Interactions outside of the family often are less intense, yet still can be quite influential in personality and self-development.  These relationships are secondary and tertiary in that they are less intense, occur less frequently, are not typically conducted under special "group norms," and rarely are all group members present.  Secondary relationships are defined as those that involve direct interactions, while tertiary relationships are associational or very casual.  Both are considered to be characteristic of the "weak ties" designation of social network theorists (Granovetter, 1973).  Community participation and the accompanying relationships that are produced would be considered to be secondary and tertiary.

 

                       

METHODS AND RESULTS

     

This research explores the basic questions of: how the rural community’s sustenance, social interaction, and collective involvement and action characteristics influence an individual's mental health and whether these community factors interact with individual-level factors.  These potential relationships with mental health are explored through the following research goals:

 

        1) To develop a multi-level framework that incorporates both individual and community-level

                 factors in an exploration of a measure of mental health (a depression index).

         2) To estimate multivariate models that examine:

                    a) individual-level correlates of a depression indicator.

                    b) the independent effects of community factors on a depression indicator. 

                    c) interactions between individual and community factors on a depression indicator.

             

This research effort involved both primary and secondary data collection.  Contextual data were aggregated to the community-level, while other data were collected and remained at the individual-level.  Primary (telephone and mail) and secondary data (from the U.S. Census) were used in the analyses.  Secondary data were drawn from the 1970, 1980, and 1990 Bureau of the Census Summary Tape File 3.  The relevant tables for Pennsylvania's Minor Civil Divisions (MCDs) were used for the development of ecological communities and sampling frameworks for two primary data collection surveys.  As part of a larger study, each Pennsylvania MCD was ranked according to population size (Claude & Luloff, 1995).  A computer algorithm based on central place theory was then used to identify and assign MCDs to an ecological community consisting of a central place (Nord, et. al.,, 1994:5).  In Pennsylvania, 212 central places were defined and then assigned rural or urban status in accordance with the Census Bureau-defined status of their central place  (Nord, et. al.,, 1994:5).  Of the 212 central places, 173 were rural (containing 61 percent of all the Commonwealth's MCDs) and are the focus of this study.  The central place and surrounding hinterland make a good definition of community because it ensures that most needs of daily life can be met locally, a key condition of community. 

           

As part of a broader effort to identify community characteristics, actions, and projects over the last ten years, a mail survey was conducted.  Local knowledgables or key informants1 were identified by calling the chief municipal officer in each of the 173 rural central places and asking for the names of specific individuals who were perceived as being very knowledgeable about the local area (Claude & Luloff, 1995).  A total of 1038 key informants were identified and contacted, 6 for each of the 173 communities.  Key informant responses from each community ranged from zero (3 communities) to six (3 communities).  The median survey response was three key informants.  The response rate for these initial key informants was 49 percent and covered 170 of the 173 rural ecological communities (98.3 percent) (Claude & Luloff, 1995). 

           

To generate a single community response, the method of modal “yes” was used.  All unanimous “yes” responses and multiple “yes” responses that outnumber “'no” responses were treated as “yes.”  When there were only two responses and one was “yes,” the response was treated as “yes.”   If there was one “yes” and two “no,” the final response was “no.”  Where the key informants failed to respond to a question or answered “don't know,” the aggregated response became a “no” (Claude & Luloff, 1995:4).  These community response data were then matched with individual-level data via a lookup table. 

           

A telephone survey sample was randomly drawn from the 173 rural communities.  Phone listings for the appropriate communities were identified through a MCD-to-telephone exchange lookup table.  A total of 804 persons in 149 of the 173 rural ecological communities completed the telephone survey.  The response rate from the phone survey (completed divided by eligibles) was 64 percent. 

 

Measuring the Dependent Variable

 

Measuring aspects of mental health in a population is made easier by the existence of a number of questions that have demonstrated high levels of reliability.  Questions were drawn from two batteries to form a single depression index used in this study.   The Center for Epidemiological Studies depression index (CES-D) and the Health Opinion Survey (HOS) measure aspects of health, depression, and anxiety.  Both the CES-D (Weissman, Sholomskas, Pottenger, Prusoff, & Locke,. 1977; Robins, Helzer, Croughan, & Ratcliff,. 1977; Husaini, Neff, Harrington, Hughes, & Stone, 1980; O'Brien, Hassinger, & Dersham, 1994) and the HOS (MacMillian, 1957; Leighton, 1963; Murphy & Leighton, 1965; Husaini, Neff, & Stone, 1979) have been used extensively.  Both batteries are fairly long, with a combined 47 questions.  Five questions from each battery were chosen to include in the instrument that was used for the local residents telephone survey.  These items were selected to cover the more global aspects of depression.   For this study, distinguishing subtleties was less important than measuring a general form of distress, such as depression.

 

Respondents were asked  “How Frequently Have You Felt The Following Ways?”: 1) I felt sad; 2) I had crying spells; 3) my sleep was restless; 4)I talked less than usual; 5) I was bothered by things that usually don't bother me;  6) My appetite was poor; 7) I felt I couldn't shake the blues; 8) I had trouble keeping my mind on tasks; 9) I felt depressed; and 10) I felt everything was an effort.  Response categories were: 1) never; 2) seldom; 3) frequently.  When a factor analysis was conducted to ensure uni-dimensionality, only one factor was extracted that explained 45 percent of the variation.  The index score was derived by summing the responses for the ten items and taking the mean for each respondent.  Reliability for this index was established via Chronbach's alpha.  The depression index had an alpha of .80, a mean of 1.63, and a standard deviation of .39, and the index responses ranged from one to three (see Appendix 1 for the descriptive statistics for all variables in the analyses). 

      

The Independent Variables

 

The independent variables in this study were divided into six sets.  Three sets of variables were at the individual-level: 1) status; 2) stressors; and 3) social supports.  In later analyses, three additional sets at the community-level were added: 1) community type, 2) community context, and, 3) community measures.  The community-level sets of variables reflect sustenance, social interactions, and collective involvement in varying degrees.

      

Status

 

Status was measured by the socio-demographic variables of age, education, income, gender, marital status, and employment status.  Average age within this sample was 47.6.  Average education was slightly more than high school (where: 1) was less than high school; 2) was high school or equivalent; 3) some college; 4) college; and 5) graduate or professional).  Median household income fell within the $15,000-$29,999 range.  The following response categories were used: 1) less than $15,000; 2)$15,000 to $29,999; 3) $30,000 to $44,999; and 4) $45,000 and above.  Half the sample was female (coded 0 = female, 1 = male).  Nearly two-thirds of the respondents were married (64 percent) and marital status was coded 0 = unmarried, 1 = married.  The overwhelming majority of respondents were employed full-time (90 percent), coded 0 = non-fulltime employment, 1 = fulltime employment.

    

Stressors 

 

The group of variables, stressors, had four indices and were derived from the telephone survey of residents.  Stressors are expected to increase the occurrence of stress outcomes.  The first is the well known stressful life events inventory (SLE) in a modified and parsimonious form where 0 = non-occurrence and 1 = occurrence 2 (Holmes & Rahe, 1967; Dohrenwend & Dohrenwend, 1978; Geller, Bultena, & Lasley, 1988).   These events, occurring in the previous calendar year, included: a member of the household or family (other than the respondent) experience injury or illness; marriage; marital separation; divorce; partner’s death; loved one’s death; legal trouble; job loss; and, retirement.  The most conservative scheme of simply summing the events was employed2.  The mean score was 1.90, the standard deviation was 1.22, and the observed range was zero to six.

           

The second index was known as the community evaluation index, and consisted of seven issues that were rated by the respondents as being: 1) not serious; 2) hardly serious; 3) somewhat serious; and 4) very serious (Bourke & Luloff, 1994).  Items in this index included: 1) access to health care; 2) poverty; 3) crime; 4) quality of education; 5) high tax rates; 6) environmental quality; and, 7) disagreements among local residents.  The mean score was 2.03, and the standard deviation was .60.  The alpha reliability for this index was .77.

           

The next variable in this group was a global health question that asked the respondent: "Do you have any physical/health problems at present?" (Cohen, Karmarck, & Mermelstein, 1983).  Answer categories were a simple yes (coded as 1) and no (coded as 0).  The mean percent with physical/health problems was .26 and the standard deviation was .44.

           

Global stress was addressed by a single question which asked: "How much of a problems is stress for you?" (Cohen, Karmarck, & Mermelstein, 1983; Krannich, Riley, & Leffler, 1988).  Answer categories and responses for this question were: 1) not a problem (20.5 percent); 2) hardly a problem (37.0 percent); 3) somewhat of a problem (35.6 percent); 4) very much a problem (6.9 percent).  The variable range was one to four, with the higher score reflecting a higher level of stress.  The mean for this variable was 2.29 and the standard deviation was .87.

      

Social Supports

 

The social supports group consisted of four variables and was derived from the telephone survey of area residents.   The first index in the group was the Kasarda and Janowitz (1974) community attachment index.  The first question asked: "How much at home do you feel in your community?"  Responses included: 1) not at all; 2) slightly at home; 3) neither at home or not at home; 4) somewhat at home; and 5) very much at home.  The second question asked: "How interested are you in knowing what is going on in your community?"  Answer categories were: 1) not at all; 2) slightly interested; 3) neither interested nor disinterested; 4) somewhat interested; and 5) very interested.  The last question asked: "Suppose you had to leave your community, how sorry or pleased would you be to leave?"  Answer categories included: 1) very pleased to leave; 2) somewhat pleased to leave; 3) it would not make any difference; 4) somewhat sorry to leave; 5) very sorry to leave.  The index had an alpha reliability of .65.  The indexed was summed and a mean was derived for each respondent.  This mean was 4.25 and the standard deviation was .69. 

          

The second variable was a single item dealing with neighbors and asks:  "Of the ten houses nearest to yours, how many have you been in personally?" (Luloff, Bourke, Jacob, & Seshan, 1995).  Responses ranged from 0 to 10, the mean was 5.17 and the standard deviation was 3.17. 

           

The third variable was the social networks index which catalogs how frequently the respondent interacts with: 1) friends; 2) acquaintances; and 3) neighbors.  The index had an alpha reliability of .56.  Scores ranged from: 1) never; 2) a few times a year; 3) once a month; 4) once a week; and 5) more than once a week.  The mean was 4.71 the standard deviation was .97.

           

The last variable was a single item dealing with religiosity.  The question simply asks: "How frequently do you attend religious services?"  Responses included: 1) never; 2) several times a year; 3) once a month; 4) once a week; and 5) several times weekly.  The mean for this question was 3.08 and the standard deviation was 1.32.

      

Community Type

      

A typology was constructed to classify the 173 rural ecological communities according to  population and economy.  A typology focuses on specific phenomena whereby data are ordered and classified to create ideal types.  While typologies are conceptions, they lead to further understanding of the empirical world (Luloff & Nord, 1992).  Population change and employment change were cross tabulated to form four categories: 1) slow population growth or decline with a more traditional rural economy (21.4 percent); 2) slow population growth or decline with a more service-based rural economy (30.1 percent); 3) population growth with a more traditional rural economy (25.4 percent);  and 4) population growth with a more service-based rural economy (23.1 percent).  In the multivariate analysis, the last category (old economy/population growing) serves as the reference category.  The means reflect the percentage of respondents surveyed from each community type.  As such, the mean for each category was .25 and the standard deviation was .43 (for each category).

           

Community Context

 

The group of items labeled community context, consisted of four variables: 1) percent of the population aged 18-64; 2) percent of the population with college diplomas; 3) average household income; and 4) percent of families that are female headed households.  All variables in this group were derived from the 1990 decennial Census and were aggregated to the ecological community-level.  These variables reflect aspects of sustenance organization.

 

The mean for this variable was .11, the standard deviation was .03, and the range was .05 to .18.

      

Community Participation

     

This first index, labeled Community Actions, asked "Has your community initiated or completed the following?": 1) programs to retain or expand firms; 2) committee or program to recruit new business or industry; 3) information program of business, industry, or investors; 4) built an industrial park; 5) created or changed industrial zoning; 6) encouraged cooperation to build local highways; 7) encouraged cooperation for development of telecommunications; 8) encouraged cooperation for a local airport; 9) sought funds for economic development; 10) developed a formal job training program; 11) made efforts to improve downtown shopping; 12) made other efforts to increase jobs; and 13) sponsored a local business program.  The 13 items produced an alpha reliability of .87.  The questions were summed and the mean for this index was 8.09 and the standard deviation 3.63.

 

The second index, developed by Young and Young (1973), later modified and used by Lloyd and Wilkinson (1985) and Zekeri, Wilkinson, and Humphrey (1994) measured local identity through the presence of community-oriented activities and displays.  Called Community Solidarity, this index used the same coding and summing schemes as the community activity index.  The battery asked, "Does your community have the following?": 1) an organization that tries to improve community life; 2) a community band, team, or performing group; 3) an organization that encourages growth and development; 4) any community wide project to improve the community; 5) a periodic celebration or festival; 6) a war monument; 7) a building for community wide meetings; 8) a town square, central park, or commons; and, 9) a special welcome sign. The nine items produced an alpha reliability of .76.  The mean for this index was 8.29 and the standard deviation was 1.96. 

 

The Multivariate Analysis of Depression

 

Table 1 contains the regression analyses of depression3.  The regression coefficients (b) are reported along with the standardized regression coefficients (B).  Model I presented the individual-level variables.  Model I will remain the comparison point for the Full Model on Table 1.  Model improvement F scores were based upon the improvement (if any) over the standard error in Model I and can be thought of as a goodness of fit test (Lunneborg, 1994).  Model I includes only the individual-level variables (Status, Stressors, and Social supports).  The Full Model adds the community-level variables (Community Type, Community Context, and Community Participation) to Model I.  The Interaction Model adds all the statistically significant (p < .05) two-way interactions to between the individual and community-level variables to the Full Model.  The analyses were conducted with listwise deletion of missing data, with 687 cases having complete information and included in these analyses.

     

For Model I, within the status variables the main effects of education and gender were statistically significant.  Those with higher levels of education and males had lower depression index scores than did their opposites.  Age, income, marital status, and employment status had weak and statistically insignificant relationships in the model.   

           

A greater number of stressful life events, perceived community problems, and global stress all were correlated positively with depression index scores.  Global stress and stressful life events were particularly strong predictors, as indicated by their relatively large standardized regression coefficients.  All variables in this block reached statistical significance except the health variable.

           

For the social supports variables, visits with the ten closest neighbors and social networks had statistically significant main effects.  These correlations were negative; as visits with neighbors and social networks increased, reported depression index scores decreased.  Overall, the adjusted R2 for this model was .32. 

           

Table 1. The regression models for the depression index and the

various community variables considered with the individual-level variables.

 

 

 

Model I

Full

Model

Interaction Model

 

Variable

 

b

 

B

 

b

 

B

 

b

 

B

THE INDIVIDUAL-LEVEL BLOCKS:

 

 

 

 

 

Status

 

 

 

 

 

Age

.000

.008

.000

.009

.000

.006

Education

-.026*

-.073

-.030*

-.083

-.026*

-.073

Income

.003

.014

.008

.033

.006

.024

Gender

-.063*

-.080

-.055*

-.078

-.054*

-.066

Marital Status

-.052

-.063

-.048

-.058

-.050

-.061

Employment Status

-.009

-.008

-.005

-.005

.008

.007

Stressors

 

 

 

 

 

Stressful Life Events

.071*

.219

.073*

.226

.191*

.592

Community Evaluations

.061*

.091

.064*

.097

.051*

.077

Health

.028

.030

.029

.036

-.211*

-.234

Global Stress

.188*

.411

.190*

.413

.050*

.109

Social supports

 

 

 

 

 

Community Attachment

.031

.053

.029

.051

.031

.054

Visits With Neighbors

-.011*

-.089

-.031*

-.105

-.013*

-.108

Social Networks

-.031*

-.076

-.033*

-.083

-.037*

-.090

Religiosity

.000

-.002

-.002

-.007

-.025

-.082

THE COMMUNITY BLOCKS:

 

 

 

 

 

Community Type

 

 

 

 

 

New Economy/Population Stable or Declining

 

 

.023

.025

-.175

-.191

New Economy/Population Growing

 

 

.046

.051

-.140

-.156

Old Economy/Population Stable or Declining

 

 

-.008

-.009

.064

.070

Community Context

 

 

 

 

 

Percent Population Aged 18-64

 

 

-1.721*

-.112

-1.276*

-.089

Percent Population With College Degrees

 

 

.683

.069

1.876*

.191

Average Household Income

 

 

.000*

-.088

.000*

-.104

Percent Households Female Headed

 

 

.195

.012

.246

.015

Community Participation

 

 

 

 

 

Community Action

 

 

.003

.026

.002

.024

Community Solidarity

 

 

-.004

-.024

-.053*

-.264

INTERACTION TERMS

 

 

 

 

 

 

Stressful Events * Percent College Grads

 

 

 

 

-.975*

-.383

Community Solidarity * Health

 

 

 

 

.028*

.268

Community Solidarity * Stress

 

 

 

 

.054*

.204

New Econ/Pop Stable or Decline * Religiosity

 

 

 

 

.061*

.232

New Econ/Pop Growing * Religiosity

 

 

 

 

.054*

.204

Old Econ/Pop Stable or Decline * Religiosity

 

 

 

 

-.024

-.092

Constant

1.196*

 

2.300*

 

2.437*

 

 

Adjusted R2

 

.32

 

 

 

.33

 

 

 

.37

 

 

Model Improvement F Score

 

 

1.01

 

2.60*

 

N = 687, * = significant at <.05

 

 

 

 

 

 

 

The Full Model adds all the community-level variables to the individual-level variables originally presented in Model I.  Despite the addition of nine new independent variables, the individual-level variables remained remarkably stable from Model I to the Full Model.  Very little change in the standardized regression coefficients (B) was observed from Model I.  Only two community-level variables were statistically significant.  The variables percentage population aged 18-64 and average household income in the communities reached statistical significance at the .05 level. 

 

Those communities with higher percentages of residents aged 18-64 and higher average household incomes had respondents who reported lower scores on the depression index.  Overall, the adjusted R2 for the full model is .33, or roughly 33 percent of the explained variation.  The sum of the effects for the community variables contributed an additional 1.0 percent to the overall explained variation in Model I.  This was not a statistically significant increase.  Despite this ambiguous result, it is possible that the community variables may add understanding to the individual-level model in the form of interaction effects.

      

 

The Interaction Model for Depression

      

The Interaction Model for the depression index is shown on Table 1 and contains all of the two-way interactions between the individual-level and community-level variables found to be statistically significant.  The Full Model served as the basis for the model improvement F score in the Interaction Model.  For the status block, the main effects of education and gender were statistically significant.  As seen in previous research, increased education levels were correlated with reduced depression index scores (Aneshensel, 1992).  Education is an important status variable and was correlated with the depression index in the multivariate framework.  The main effect of gender was also statistically significant.  As evidenced from previous research, females reported higher levels of the depression index (Thoits, 1987).

           

In the stressors group of variables, the main effect of the community evaluation index was statistically significant.  Higher levels of concern displayed in the community evaluations was positively correlated with the depression index.   Perceptions about the local community, as seen in the index, provide evidence that perceptions of community conditions have an impact upon individual mental health.  The stressful life events index, health, and global stress were involved in interaction terms, and are not interpreted. 

           

Within the social supports group, the main effects of visits with neighbors and social networks were statistically significant.  As anticipated, increased visits with neighbors and social networks was correlated with reduced depression index scores (Maton, 1989; Aneshensel, 1992).  Religiosity was involved in the composition of an interaction term.

           

All of the community type variables were involved in interaction terms so these main effects cannot be interpreted.  For the community context variables, the main effect of percent population aged 18-64 was statistically significant.  Residents who lived in communities with higher levels of population aged 18-64 had a negative correlation with depression index scores.  Such communities are likely to have a larger proportion of residents in the workforce and smaller dependent populations.  Further, this measure may offer a crude reflection of economic opportunity in the community.  Average household income in the community also had a statistically significant main effect.  As anticipated, wealthier communities had respondents with lower reported depression index scores.  Average household income levels could be thought of as a measure of the efficiency of a community's sustenance organization.  Lower average income levels in the community may lead to individuals struggling to meet basic needs.  Higher income levels may reflect an efficient economy that enables residents to pursue higher order needs.  The percentage college graduates variable was involved with an interaction.

           

Among the interaction terms, stressful life events by the variable percent college graduates in the community was statistically significant (see Table 2 for estimated scores for the interaction terms).  In all cases increased stressful life events was correlated with higher depression index levels.  However, the community context variable percentage college graduates modified this impact.  For those respondents who experienced only one stressful life event, living in a community with low percentages of college graduates led to a lower reporting of depression index scores than for the other community types with higher percentages of college graduates.  When respondents have experienced two events, percentage college graduates in the community had almost no effect on reported depression index levels.  When respondents have reported three stressful events, communities with higher percentages of college graduates had respondents who reported lower depression index scores than those in communities with low percentages of college graduates.  The effect of living in a community with a lower percentage of college graduates increases the negative impact of life events compared to a respondent living in a community with higher percentage of college graduates.   In this study, respondents that lived in communities with more college graduates were able to handle stressful life events better then residents in communities with fewer college graduates.  This may be due to an increased availability of professional services which can address these stressful life events.  Communities with such amenities and services may be more likely to retain and employ college graduates. Further, communities with higher levels of education may have individuals and groups who become better at problem solving and providing social support.  As such, percent college graduates in the community may reflect a better general community environment as well as higher education levels. 

           

The community solidarity by global stress term was statistically significant.  In all cases the observed correlation was negative; as community solidarity increased, the depression index level decreased.  However, this rate of reduction was greatly influenced by the self-reported stress level of individuals.  For respondents reporting the lowest levels of stress, increased community solidarity was correlated with rapidly reduced reported depression index levels.  For respondents reporting higher levels of stress, this correlation was greatly reduced.  When individuals are very stressed, community solidarity has little influence in reducing depression. However, community solidarity can make a sizeable difference when individual stress levels are lower.  This suggests that events and circumstances in the individual’s life most directly impact depression index levels, but community characteristics, such as solidarity, can make a difference when perceived stress is low.

     

The community solidarity by health interaction term was statistically significant.  Where respondents lived in communities with relatively low levels of community solidarity, depression index levels were higher.  This was particularly true for respondents with health problems, compared to those without.  For communities that had average levels of community solidarity, health status made little difference on the reported depression index level.  However, for communities with high levels of community solidarity, respondents with health problems had a negative correlation with depression index levels compared to those with no problems.  This may be due to the social supports that people in communities with high levels of solidarity develop.  When a person in such a community becomes ill, their social networks may become more active and buffer them from the stressful effects of illness and other stressors.

           

The community type interaction terms with religiosity had a statistically significant block F (F = 4.82. p = .002).4  New style economies, regardless of population, showed a correlation where residents who reported higher depression index scores along with increased religious service attendance.  Old style economies had the opposite relationship.  In old style communities, increasing church attendance was correlated with reduced depression index scores.   This suggests religiosity is a more effective social supports strategy in communities with old economies.  New economy community types may have experienced economic and/or population transitions.  Under such circumstances, congregations may have been significantly disrupted, reducing the effectiveness of this social support.

Table 2.  The estimated scores for the depression interaction terms.

 

 

 

 

 

Stressful Life Events

 

Percent College Grads in the Community

 

9

11

13

1

1.59

1.60

1.62

2

1.69

1.69

1.69

3

1.79

1.77

1.75

 

 

 

 

 

 

 

 

Community Solidarity

  Global Stress

 

 

 

 

 

1

2

3

6                                             

1.48

1.62

1.77

8

1.40

1.58

1.76

10

1.33

1.54

1.75

 

 

 

 

 

 

 

 

Community Solidarity

Health Problems

 

 

Yes

No

6

 

1.75

1.70

8

 

1.64

1.65

10

 

1.54

1.60

 

 

 

 

Community Type

    Religiosity

 

Several Times Yearly

Once a Month

Weekly

New Economy/Stable or Decline Pop

1.66

1.74

1.81

New Economy/Growing Pop

1.69

1.75

1.81

Old Economy/Stable or Decline Pop

1.82

1.72

1.62

Old Economy/Growing Pop

1.78

1.73

1.68

 

The interaction model had an adjusted R2 of .37.  The model improvement F score was statistically significant for the appropriate degrees of freedom.  These interaction terms significantly improved the Full Model.  The Interaction Model for the depression index revealed that the community-level factors had more influence through conditional relationships with individual-level factors.

 

      

DISCUSSION

 

Stress and its associated outcomes are problems wherever they exist.  In Pennsylvania, many urban residents do not recognize the stress that is inherent in modern life, regardless of place of residence.  For example, over seven of ten urban residents in a recent survey agreed that "Life in rural communities is less stressful than in other areas" (Willits & Luloff, 1995).  It is a popular belief that stress and stress outcomes, such as depression, are rare occurrences for rural community residents (National Mental Health Association 1988; Srole, 1978). It has been long thought that cities impact the mental health of their residents (Simmel, 1950; Wirth, 1938).  This study was undertaken to explore the influence of the rural community on mental health, which is an important component of individual well-being.  Understanding factors in rural communities that might diminish mental health can be the first step that can lead to ameliorative efforts.

 

           

Implications for Ameliorative Efforts

     

Traditional ameliorative approaches to mental health problems often involve intense individual-oriented therapies (Beck, 1996), and indeed, such treatments have been notably successful in improving the lives of many. However, they are notoriously expensive and time consuming (Hamilton, 1994; Paris, 1990). Further, few private insurance providers supply adequate coverage for the full treatment of mental health problems (Pear, 1996). 

           

Currently, insurance providers are curtailing already meager mental health provisions for millions of Americans specifically to reduce the costs of managed care, especially for Health Maintenance Organizations (Pear, 1996; Goleman,1996).  Congress and the current president have addressed these problematic issues through legislative hearings, but little meaningful action has been taken (Pear, 1996).  In the current political climate, which already has seen the defeat of one model of health care reform, the prospect of any significant mental health insurance reform seems doubtful (Pear, 1996).  The limited access that the general population has had to mental health care continues to be eroded (Goleman, 1996).

           

When stress outcomes such as depression manifest in individuals, they should be treated in the most expedient fashion possible, regardless of cost.  Unfortunately, the reality of the situation is that current business and fiscal restraints do not always allow for the provision of needed care.  One component of a potential solution is to attempt to reduce the occurrence of stress outcomes.  Relatively small improvements in social well-being may appreciably improve individual mental health for many individuals, thus reducing the number of individuals seeking intensive therapies.  In this circumstance, prevention may be cheaper than the cure.

           

Modifying community conditions, or even the relationship of the individual to his/her community could possibly improve mental health.  This is less likely to happen in a direct fashion.  As seen in this study, community circumstances conditioned the efficacy of social supports.  Further, the community also was seen to influence the effect of the individual-stressors.  Though such changes may have only a relatively small impact on aggregate mental health, there could be relatively great reductions in the expensive use of individual therapies.  Improvement in these community-level factors could be preventative mental health measures.  This is undoubtedly more efficient, in terms of costs and human potential, than individual-oriented therapies. 

 

 

Theoretical Implications

 

In this study, community conditions were correlated with the mental health of residents.  This relationship has been assumed frequently in the literature, but rarely shown empirically.  The relationship of community to mental health was most often revealed in these analyses through the interactive effects of the community-level variables with the individual-level variables.  The interaction model revealed that community characteristics often influences well-being in a complex fashion.  At the community-level, education, solidarity, and community type influenced the impact of stressors on an individual's mental health.  

           

The impact of community factors was not overwhelming.  Though at the individual-level, the community-oriented social supports were important predictors of depression index scores in all models.  This was consistent with previous research.  These types of interactions can reveal social integration or the imbeddedness of a person within a community, which gives a sense that one belongs and matters to others.  As in previous research, the individual-level variables were found to be strong predictors of individual mental health.  However, in this contextual modeling strategy, sufficient evidence was found to conclude that variation in certain community factors does impact the individual mental health of community residents.

           

A significant component in the operationalization of this study was the definition and construction of the ecological communities.  If community were defined incorrectly, the context of the analysis would be unrelated to the individual-level data.  However, the community-level data were correlated with the individual-level data.  This relationship itself has theoretical implications.  The definition of community in this analysis was based upon central place theory.  Community, in this study, was operationalized with both an interactional and territorial component.  The ecological community is interactional in that it arises from the social interaction of residents to create, in part, sustenance organization.  Social interaction occurs to meet higher-order needs as well.  It is also interactional because residents from the hinterland travel to the central place to meet some of their needs.  Thus, community and many of the social interactions within it have a territorial component.  The territory may not be the most essential factor in the definition of the community, but it cannot be ignored either. 

           

Though assumed, few studies have empirically examined linked community-level factors to mental health.  Here, broader measures of social and economic structure were chosen over more specific measures to help keep the model parsimonious.  Like any contextual analysis, choosing the variables to represent a complete context was very difficult3.  Because of statistical limitations, relatively few variables could be used in the model.  As such, the contextual variables are broad, and what they represent may be somewhat open to interpretation.  A more careful specification of the context on the basis of these results may yield more concrete interpretations.  The main effects of the percent population aged 18-64 and average household income variables, and the community type variables in interaction terms, lead to some suggestions for further exploration.  The community-level contextual relationship of economy, economic structure, migration, and other forms of sustenance organization to individual mental health should be examined in further detail.

           

This study explored the relationship of community to mental health.  Mental health here was operationalized as a depression index.  Community factors did relate to the depression index with both main and interactive effects.  However, depression is just one of many possible stress outcomes.  Other indicators of stress outcomes and mental health may produce different findings.

      

REFERENCES

 

Allport, G.  (1955).  Becoming.  New Haven, CT: Yale University Press.

 

Aneshensel, C. S.  (1992).  Social stress: theory and research.  Annual Review of Sociology, 18, 15-38.

 

Beck,  M.   (1996).   Follow the money: mental health benefits controlled by managed care.  Newsweek. 127(May 20), 23-24.

 

Belyea, M. J., &  Lobao, L. M.  (1990).  Psychosocial consequences of agricultural transformation: The farm crisis and depression.  Rural Sociology, 55(1), 58-75.

 

Blumer, H.  (1969).  Symbolic Interactionism.  Englewood Cliffs, NY: Prentice Hall.

 

Bridger, J. C.  (1994).  Power, Discourse, and Community: The Case of Land Use.  Dissertation in rural sociology. University Park, PA: The Pennsylvania State University.

 

Catalano, R., & Jackson, R..  (1981).  Economic predictors of admissions to mental health facilities in a nonmetropolitan community.  Journal of Social Behavior, 22(September), 284-297.

 

Claude, L. P., & Luloff, A. E.  (1995).  The relationship between community activeness and community success: How do researchers know.  Paper presented at the 3rd annual United States Forest Service Rural Communities Assistance Conference.  Knoxville, Tennessee .

 

Cohen, S., Karmarck, T., &  Mermelstein, R.  (1983). A global measure of perceived stress.  Journal of Social Behavior, 24(December), 385-396.

 

Dohrenwend, B. S., & Dohrenwend, B. P.   (1978).  Some issues in research on stressful life events.  Journal of Nervous and Mental Disease, 166(1), 7-15.

 

Eaton, W. W.,  & Kessler, L. G.  (1981).  Rates of symptoms of depression in a national sample.  American Journal of Epidemiology, 114(4), 528-538.

 

Geller, J. M., Bultena, G., & Lasley, P.  (1988).  Stress on the farm: a test of life- events perspective among Iowa farm operators.  Journal of Rural Health, 4(2), 44-57.

 

Goleman, D.  (1996).  "Critics say managed-care saving are eroding mental care."  New York Times, January 24:C9.

 

Granovetter, M. S.  (1973).  "The strength of weak ties."  American Sociological Review 78(May):1360-1380.

 

Hamilton, J.  (1994).  "A furor over mental health."  Business Week 35(August 8):66-69.

 

Hamilton, V. L., Broman, C. L., Hoffman, W. S.& Renner, D. S.  (1990).  "Hardtimes and vulnerable people: initial effects of plant closings on autoworkers' mental health."  Journal of Health and Social Behavior 31(June):123-140.

 

Hassinger, E., & Pinkerton, J.  (1986).  The Human Community.  New York, NY: Macmillan Publishing Inc.

 

Holmes, T.H., & Rahe, R.H.  (1967).  The social readjustment rating scale.  Journal of Psychometric Research, 11, 213-218.

 

Husaini, B., Neff, J. A., & Stone, R. H.  (1979).  Psychiatric impairment in rural communities.  Journal of Community Psychology, 7(2), 137-146.

 

Husaini, B., Neff, J. A., Harrington, J. B., Hughes, M. D., & Stone, R. H.  (1980).  Depression in rural communities: Validating the CES-D scale.  Journal of Community Psychology, 8(1), 20-27.

 

Jacob, S. G., & Willits, F. K.  (1994).  Objective and Subjective Indicators of Community Well-Being: A Pennsylvania Assessment.  Social Indicators Research, 32, 161-177.

 

Jensen L., & McLaughlin, D. K.  (1995).  Human capital and nonmetropolitan poverty. In Beaulieu, L. J. & Mulkey  D.(eds.).  Investing in People: The Human Capital Needs of Rural America.  Boulder, CO: Westview Press.   Pp. 111-138

 

Kasarda, J. D., & Janowitz, M.  (1974).  Community attachment in mass society.  American Sociological Review, 39(3), 328-339.

 

Kassab, C., & Luloff, A. E.  (1993).  The new buffalo hunt: Chasing the service sector.  Journal of the Community Development Society, 24(2), 175-195.

 

Killian, M. S., & Beaulieu, L. J.  (1995).  Current status of human capital in the rural U.S. In Beaulieu, L. J. & Mulkey  D.(eds.).  Investing in People: The Human Capital Needs of Rural America.  Boulder, CO: Westview Press. Pp. 23-48

 

Krannich, R. S., Riley, P. J. & Leffler, A.  (1988).  Perceived stress among nonmetropolitan Utah residents.  Lifestyles: Family and Economic Issues, 9(4), 281-296.

 

Leighton, D. C., & Stone, I. T.  (1990).  Community development as a therapeutic force: a case study with measurements.  Sociological Practice, 8(1), 72-95.

 

Leighton, D. C., Harding, J. S., Macklin, D. B., MacMillan, A. M. & Leighton, A. H.  (1963).  The Character of Danger.  New York, NY: Basic Books.

 

Lloyd, R. C., & Wilkinson, K. P.  (1985). Community factors in rural manufacturing development.  Rural Sociology, 50 (1), 27-37.

 

Luloff, A.E., & L. E. Swanson.  (1995).  Community agency and disaffection: Enhancing collective resources.  In Beaulieu, L. J. & Mulkey  D.(eds.).  Investing in People: The Human Capital Needs of Rural America.  Boulder, CO: Westview Press. Pp. 351-372

 

Luloff, A.E., Humphrey, C. R., & Smith, S. M.  (1994).  Economic Restructuring and Rural Community Transformation.  Final Report, NRICGP project number 92-37401-8285.  The Pennsylvania State University, University Park, PA.

 

Luloff, A.E., Bourke, L., Jacob, S., & Seshan, S.  (1995).  Final Research Report: Farm and Non-Farm Interdependencies at the Rural-Urban Interface.  Pennsylvania Department of Agriculture, Harrisburg, PA.

 

Lunneborg, C. E.   (1994).  Modeling Experimental and Observational Data.  Belmont, CA: Duxbury Press.

 

MacMillan, A. M.  (1957).  The health opinion survey.  Psychological Reports, 3, 325-339.

 

Martin, K. E. & Wilkinson, K. P.  (1984).  Local participation in federal grant system: Effects of community action.  Rural Sociology, 49(3), 374-88.

 

Maslow, A. H. (1954).  Motivation and Personality.  New York, NY: Harper and Brothers.

 

Maton, K. I.   (1989).  Community settings as buffers of life stress?  Highly supportive churches, mutual help groups, and senior centers.  American Journal of Community Psychology, 17(2), 203-232.

 

Mazer, M.  (1976).  People and Predicaments.  Cambridge, MA: Harvard University Press.

 

Mazer, M.  (1983).  The influence of the rural community on the mental health of its inhabitants.  In McCubbin, H. I. & Figley, C. R. (eds.). Stress and the Family: Volume One.  New York, NY: Brunner/Mazel Milton. Pp. 188-200

 

Mead, G. H. (1934).  Mind, Self, and Society: From the Standpoint of a Social Behaviorist.  Edited by Charles Morris.  Chicago: University of Chicago Press.

 

Mirowsky, J., & Ross, C. E.  (1979).  A comparison of stressful life event weighting schemes.  Journal of Social Behavior, 20(2), 166-177.

 

Murphy, J. M., & Leighton, A. H.  (1965).  Approaches to Cross-Cultural Psychiatry.  Ithaca, NY: Cornell University Press.

 

National Mental Health Association.  (1988).  Report of the National Action Commission on the Mental Health of Rural Americans.  Washington D.C.: Author.

 

Nord, M., Kreahling, K., Christensen, D., Claude, L. P., Fickes, P. D., Humphrey, C. R., Luloff, A. E.,  Schwartz, M. R. & Smith, S. M.  (1994).  Economic restructuring and rural community transformation in Pennsylvania.  AE&RS 246, Department of Agricultural Economics and Rural Sociology, College of Agricultural Sciences, Penn State University, University Park, PA.

 

O'Brien, D. J., Hassinger, E. W., & Dersham, L.  (1994).  Community attachment and depression among residents in two rural midwestern communities. Rural Sociology, 59(2), 255-265.

 

Paris, E.  (1990).  Sigmund Freud, meet Jean-Baptiste: Say managed mental health care.  Forbes, 145(February 19), 148.

 

Pear, R.  (1996).  Wider mental care insurance in now feasible, experts say: Support for Senate position requiring equivalent coverage for mental and physical coverage.  New York Times, May 2, A1.

 

Pearlin, L. I.  (1989).  The sociological study of stress.  Journal of Health and Social Behavior, 30(September), 241-256.

 

Robins, L N., Helzer, J. E., Croughan, J., & Ratcliff, K. S.  (1977).  National Institute of Mental Health diagnostic interview schedule.  Archives of General Psychiatry, 38(April), 381-389.

 

Ross, C. E., & Mirowsky, J.  (1979).  A comparison of life-event weighting schemes: Change, undesirability, and effect-proportional indices.  Journal of Health and Social Behavior, 20(2), 166-77.

 

Ross, P. J., Bluestone, H., & Hines, F. K.  (1979).  Indicators of social well-being for U.S. counties.  Rural Development Research Report no. 10, Economics, Statistics, and Cooperative Service, United States Department of Agriculture.  Washington D.C.:  U.S. Printing Office.

 

Simmel, G.  (1950).  The metropolis and mental life. In Wolff, K. H.  The Sociology of Georg Simmel.  New York, NY: The Free Press. Pp. 409-24

 

Srole, L.  (1978).  The city versus town and country: New evidence on an ancient bias.  In Leo Srole and Anita K. Fischer.  Mental Health in the Metropolis: The Mid-Town Manhattan Study.  New York, NY: New York University Press. Pp. 433-440.

 

Thoits, P. A.  (1987).  Gender and marital status differences in control and distress.  Journal of Health and Social Behavior, 28(March), 7-22.

 

Thoits, P. A.  (1995).  Stress, coping, ans social support process: Where are we? What next?  Journal of Health and Social Behavior, 36(Extra Issue), 53-79.

 

Wagenfeld, M. O., Goldsmith, H. F., Stiles, D., &  Manderscheid, R. W.  (1988).  Inpatient mental health services in metropolitan and nonmetropolitan counties.  Journal of Community Psychology, 9(2), 13-28.

 

Wagenfeld, M. O.  (1990).  Mental health and rural America: A decade review.  The Journal of Rural Health, 6(4), 507-522.

 

Weissman, M., Sholomskas, D., Pottenger, M., Prusoff, B. A., & Locke, B. Z.  (1977).  Assessing depressive symptoms in five psychiatric population: A validation study.  American Journal of Epidemiology, 106(3), 203-214.

 

Wilkinson, K. P.  (1979).  Social well-being and community.  Journal of the Community Development Society, 10(1), 6-16.

 

Wilkinson, K. P.  (1991).  The Community in Rural America.  New York, NY: Greenwood Press.

 

Wilkinson, K. P.  (1993).  Community Distress in Rural Areas: The Exploration of an Interactional Concept.  Paper presented at the Annual Meeting of the Rural Sociological Society. Orlando, FL (August).

 

Willits, F. K., & Luloff, A. E.  (1995).  Urban residents' views of rurality and contacts with rural places.  Rural Sociology, 60(3), 454-466.

 

Wirth, L.  (1938).  Urbanism as a way of life.  American Journal of Sociology, 44(1):3-24.

 

Young, C.l E., Giles, Jr., D. E., &  Plantz, M. C.  (1981).  Natural networks: Help-giving and help-seeking in two rural communities.  American Journal of Community Psychology, 10(4), 457-469.

 

Young, F. W., & Young, R. C.  (1973).  Comparative Studies of Community Growth.  Rural Sociological Society, Monograph Number 2.  Morgantown, WV: West Virginia University Press.

 

Zekeri, A. A., Wilkinson, K. P., &  Humphrey, C. R.  (1994).  Past activeness, solidarity, and local development efforts.  Rural Sociology, 59(2), 216-235.

 

End Notes

           

     1  General information was gained by identifying a set of community-wide leaders and active local residents, called key informants.  These individuals may have held official positions, directed local agencies, or were individuals that several residents described as "knowing a lot about this place."

     2  A weighting scheme for the items tested and reported by Holmes and Rahe (1967) and Ross and Mirowsky (1979) was explored as well.  However, this method produced an index that was correlated highly (r=.97) with the method used.  Additionally, the chosen strategy avoids current criticisms of the validity of the weighting scheme.

     3 Contextual analysis can present severe multicolinearity problems.  Careful use of informal diagnostics such as the zero-order correlations and scatter plots of the standardized residuals revealed little impact of the effects of multicolinearity.  Further, a formal diagnostic statistic, the variance inflation factor (VIF), was utilized.  This statistic indicates how the inclusion of a dependent variable in the model inflates the standard error of the other independent variables.  VIFs greater than 10 are generally thought to have undue influence in the Ordinary Least Squares estimation (Lunneborg 1994).  In no case were the VIFs in the following analyses equal to 10 or greater.

     4  The block consists of the three community types contrasted with the reference category.  The effects of these variables are considered as a block and the block F and its associated probability provide an appropriate test of significance.

 

Appendix 1.  The descriptive statistics for the variables in the analyses.

 

 

Variable

 

 

 

Mean

 

 

Standard Deviation

 

 

 

Minimum

 

 

 

Maximum

 

 

 

Valid N

 

Dependent Variable

 

 

 

 

 

 

 

 

 

 

 

Depression Index

 

1.63

 

.39

 

1

 

3

 

785

 

Status

 

 

 

 

 

 

 

 

 

 

 

Age

 

47.6

 

16.47

 

19

 

89

 

800

 

Education

 

2.71

 

1.10

 

1

 

5

 

787

 

Income

 

2.84

 

1.8

 

1

 

5

 

769

 

Gender

 

.50

 

.50

 

0

 

1

 

804

 

Marital Status

 

.64

 

.48

 

0

 

1

 

790

 

Employment Status

 

.90

 

.36

 

0

 

1

 

770

 

Stressors

 

 

 

 

 

 

 

 

 

 

 

Stressful Life Events

 

1.90

 

1.22

 

0

 

6

 

788

 

Community Evaluations

 

2.03

 

.60

 

1

 

4

 

792

 

Health

 

.26

 

.44

 

0

 

1

 

803

 

Global Stress

 

2.29

 

.87

 

1

 

4

 

803

 

Social Supports

 

 

 

 

 

 

 

 

 

 

 

Community Attachment

 

4.25

 

.69

 

1.33

 

5

 

786

 

Visits With the Ten Closest Neighbors

 

5.17

 

3.17

 

0

 

10

 

791

 

Social Networks

 

4.71

 

.97

 

1

 

6

 

778

 

Religiosity

 

3.08

 

1.32

 

1

 

5

 

774

 

Community Context

 

 

 

 

 

 

 

 

 

 

 

Percent Population Aged 18-64

 

.58

 

.03

 

.48

 

.65

 

804

 

Percent Population With College Degrees

 

.12

 

.04

 

.04

 

.32

 

804

 

Average Household Income

 

$29,336

 

$5,546

 

$15,150

 

$45,506

 

804

 

Percent Households Female-Headed

 

.11

 

.03

 

.05

 

.18

 

804

 

Community Participation4

 

 

 

 

 

 

 

 

 

 

 

Community Action

 

8.09

 

3.63

 

6

 

13

 

804

 

Community Solidarity

 

8.29

 

1.96

 

6

 

10

 

804