Content area
The purpose of this study was to examine the relationship between gender composition and group cohesion in U.S. Army combat support and combat service support units. Five studies were compared in this analysis, including survey data obtained in the Persian Gulf during Operation Desert Storm; in Somalia during Operation Restore Hope; in Haiti during Operation Uphold Democracy; and in two garrison samples.
Full text
The gender integration of the United States Armed Services has become a controversial issue in the wake of recent sex scandals that began with the revelation that several female recruits at Aberdeen Proving Ground, Maryland, had been coerced into sexual activity with their drill instructors over a two year period.l This was followed by other scandals involving adultery, fraternization, and consensual sex.2 Some conservative analysts have expressed the view that the emerging scandals are an indication of the negative impact of women's increased presence in the military. It has been argued that sexual aggression is an important component of being an effective warrior, and that gender integration will thus inevitably lead either to sexual assault or to the weakening of the warrior spirit.3
The effects of women's presence on unit performance and readiness was not found to be negative, according to two studies conducted in the mid-1970s. In the first study, soldiers from forty combat support and combat service support companies were evaluated over an eighteen-month period during their Army Training and Evaluation Program (ARTEPS).4 The presence of women was found to have no significant effect on the operational capabilities of the unit. The second study found a negligible impact of women's presence on unit performance during ten-day field exercises.5 A more recent study similarly found that gender integrated basic training had no negative effects on the performance of either gender, and actually showed some positive effects for women.6 However, certain military social scientists have argued that technical proficiency is not the most important factor to be considered in the expansion of women's military roles. They allege that the presence of women in the unit has a negative impact on male bonding, and therefore on unit cohesion and military effectiveness?
Cohesion refers to the comradeship of the small unit, which is widely believed to be essential both to mission accomplishment and individual survival.8 Cohesion involves a kind of bonding among soldiers that sustains their commitment to one another, and to their unit, and most importantly, their will to continue fighting despite the stresses of combat or the mission.9 Cohesion refers to interpersonal associations at the level of the primary group, the smallest of which is the two-person buddy system,10 while the upper bounds are at the company level.11 Demographic homogeneity has been found to be helpful in the development of cohesionfor example, similarities in social background, ethnicity, age, values, and upbringing.l2 However, this kind of homogeneity is not attainable in a large heterogeneous society, such as the United States, where inequality and segregation are socially unacceptable. In this regard, commonality of experience through extended periods of close contact has become the glue that holds the unit together.l3 The question is whether gender is a special case of demographic heterogeneity that remains an impediment to cohesion in the present-day United States Army.
The relationship of gender composition to unit cohesion was addressed in two recent studies conducted by the Walter Reed Army Institute of Research. The first of these studiesl4 based on data collected in 1988 found a significant negative correlation between percentage of women in the work group and horizontal cohesion among male junior enlisted soldiers. This finding did not support Kanter's tokenism hypothesis,15 which posits that the increased presence of women would have positive organizational outcomes for women. However, it supported a competing hypothesis developed by Blalockl6-the minority-proportion-discrimination hypothesis, which posits that the increased presence of a minority will lead to increased discrimination because of the perceived threat of competition.
In the second study conducted in 1995,17 we found no significant relationship between the percentage of women in the work group and unit cohesion. The two studies were conducted seven years apart. Intervening events, which involved the increased participation of women in combat arenas-particularly the Persian Gulf War-were believed to have an impact on gender relations in the service, and the increased perception of women as soldiers without a gender tag.Is However, the spate of current crises indicates that in the minds of many, the gender tag is still very prominent. The question thus arises as to what extent these two studies represent real before and after changes, rather than two chance findings that have little potential for generalizability.
The purpose of the present study was to conduct a meta-analysis of the relationship between gender composition and group cohesion across five studies, including the two that have already been described. The study will use the group level method of data analysis, and will also include an examination of attitudinal and demographic characteristics of the samples that may account for variations in the findings.
Method
Subjects
This analysis incorporates data from five separate studies of soldiers from combat support and combat service support companies conducted between 1988 and 1995.
Garrison 1. A study of cohesion, psychological well-being, and readiness was conducted among soldiers at an infantry post in the United States in May 1988.19 Participants included 1203 male soldiers and 284 female soldiers across 19 companies.
Persian Gulf. A study of deployment stress and coping among soldiers deployed to the Persian Gulf during Operation Desert Storm was conducted in May 1991. The sample comprised 856 male and 169 female soldiers across 48 companies.20
Somalia. A survey of deployment stress and coping was conducted among soldiers deployed to Somalia in 1993 as part of Operation Restore Hope. Participants included 1887 male and 265 female soldiers from 30 companies.21
Haiti. A study of deployment stress and health was conducted among soldiers deployed to Haiti in 1995 as part of Operation Restore Democracy. Study participants included in the present analysis were 848 male and 128 female soldiers from 27 combat service support companies. Soldiers from combat arms companies were excluded from this analysis.22
Garrison 2. A study of gender integration and sexual harassment was conducted among soldiers at three Army posts in the United States between May and July 1995. The sample comprised 939 male soldiers and 251 female soldiers across 34 companies.23
Data Collection and Response Rates
In the two garrison studies, questionnaires were administered to soldiers in auditoriums or classrooms at the posts where they were stationed. All soldiers available on the days of the survey administration were given time off work to complete the survey. In the Garrison 1 study, an average of about two-thirds of the soldiers in each unit participated in the study. In the Garrison 2 study, about one third of the soldiers in each unit participated. Response rates among those who were asked to participate was above 95 percent in both studies.
In the deployment studies, surveys were administered to soldiers on site. Because large auditoriums were frequently unavailable, surveys were typically administered and collected through the chain of command. Attempts were made to distribute surveys to as many units as possible during each deployment. Precise estimates of response rates are nearly impossible to calculate. However, every effort was made to ensure that representative samples of units were included in the studies, and that all available soldiers in participating units were given an opportunity to complete the survey questionnaire. The researchers' impressions were that most soldiers who were actually given the questionnaires filled them out. Nonresponders seem to have been those whose duties precluded their receiving the survey. There was no evidence to suggest that the samples were not representative of the units surveyed or that the data was biased by non-participation.
Measures
Horizontal cohesion. All five studies included a measure of horizontal cohesion based on scales developed at the Walter Reed Army Institute of Research (WRAIR) across a variety of studies since the mid 1980s.24 The construct measures the degree of bonding among peers at the small unit level, and focuses on off-duty relationships. Each study used a slight variation of the scale, although certain core items were common to all five studies. Because of differences across studies, we used two measures of horizontal cohesion in the correlational analyses-the first was the full scale used in each specific study, while the second was the core group of items common to all studies. Alpha coefficients for the horizontal cohesion scales in the five studies ranged from .75 to .91. Those for the core items ranged from .72 to .83. A list of items used in the various studies in provided in Appendix B.
Support for the mission. Questions regarding soldiers' support for the mission were asked in the three studies conducted during deployments. In the Haiti and Somalia studies, soldiers were asked to rate a statement regarding their belief in the mission on a scale of strongly agree to strongly disagree. In the Persian Gulf study, soldiers were asked whether belief in the mission was used as a coping mechanism in dealing with stress. Those who responded positively were asked how helpful belief in the mission was as a coping mechanism. Response choices ranged from very helpful to very unhelpful.
Gender composition of the work group. In four of the studies (all except Garrison 2), gender composition of the work group was defined as the percentage of women in the company calculated on the basis of those who participated in the surveys. Two percentages were calculated-one being the percentage of female junior enlisted soldiers in relation to male junior enlisted in the company, and the other being the percentage of all women in relation to all soldiers in the company. In the Garrison 2 study, soldiers were asked the number of men and women they worked with on a daily basis. The percentage of female coworkers was calculated for each soldier in relation to the total number of coworkers.
Sociodemographic characteristics. Five sociodemographic variables were examined in this analysis-gender, age, ethnicity, rank, and marital status. Two ethnic categories were recognized in the study-white and minority. Two categories of marital status were recognized-married and unmarried. Three rank categories were recognized: junior enlisted (those in the ranks of private through corporal), noncommissioned officers (NCOs), and officers. Number of children living at home or away from home was not included in the demographic comparisons since this variable was not assessed in a consistent manner across studies.
Method of Analysis
This study examined group-level correlations between the two main study variables-horizontal cohesion and percentage of women in the group. The unit of analysis in this study was the company, which comprised anywhere from 4 to over 300 soldiers (see Table 5). Companylevel analyses were selected over individual-level analyses because both cohesion and percentage of women in the unit can be considered company-level attributes. Horizontal cohesion at the company level was measured by averaging cohesion scores from members of the company. This approach is thought to provide a more objective assessment of a shared social phenomenon because it is based on the aggregation of scores of multiple respondents, thus eliminating individual variability. Scores provide measures of the group' s assessment of a particular climate variable, and is therefore thought to reflect social processes operating beyond an individual level phenomenon.25
Before conducting analyses involving aggregate scores, it is often instructive to determine whether the study groups vary significantly on the construct of interest. One way to test for significant differences is to conduct a one-way random effects analysis of variance (ANOVA), where the construct of interest (in this case cohesion) is the dependent variables.26 The results of this analysis indicated that there were significant differences among companies in terms of cohesion in all samples except for the Garrison 2 study. This sample was nevertheless included in the group-level analysis for comparison's sake. The following is a summary of the main methodological issues that were taken into account in developing an approach to these correlations:
Measures of Cohesion. As we have already noted, each study used a slightly different measure of horizontal cohesion. Differences in measures could account for potential differences in findings across studies. Therefore, two measures of cohesion were examined in each study. The first was the full cohesion measure used in the study, while the second comprised core items that were common to all or most studies (see Appendix B). Five core items were common to the Garrison studies, the Somalia study, and the Persian Gulf study. For the Haiti study, only three items common to the other studies were available. However, a scale of four items was used for the core variable, by including an item that was similar to core items in the other studies.
Weighted versus Unweighted Correlations. In order to be included in the group-level analysis, a company was required to have a minimum of four soldiers participating in the study and to have at least three valid horizontal cohesion scores. These scores were summed and then divided by the total number of soldiers in the company to derive an average score for the company. In an earlier article, which we published on the Garrison 1 study, we used unweighted group level correlations to examine the relationship between cohesion and percentage of women in the work group. This method did not take group size into account in the calculation of the correlation. Both small and large groups were given equal weight. Correlations were computed between group mean scores regardless of the sample size. Some researchers have suggested that large variations in group size may result in inaccurate estimates of group correlations, although this methodological issue remains unresolved.27 The solution is to weight the mean score of the group by the number of individuals in the group. Weighting is accomplished by assigning group-level scores (mean cohesion and percentage of women) back to the members of each company and then calculating the individual level correlation. In this study, both weighted and unweighted correlations are presented for the purposes of comparison and to provide a more complete evaluation of our previous published findings.
Junior Enlisted Soldiers versus All Soldiers. In our Garrison 1 study, a significant correlation between percentage of women and unit cohesion was found for junior enlisted men only. In the Garrison 2 study, we were not able to examine junior enlisted soldiers separately because of insufficient numbers for group level analysis. In the three samples of deployed soldiers, it was therefore necessary to examine correlations for both junior enlisted soldiers and all soldiers in order to determine whether these effects typically apply only to junior enlisted soldiers, or whether this finding was specific to the Garrison I study.28
Total Sample versus Men Only. The decision of whether to include or exclude women in conducting these analyses is related to both theoretical and methodological considerations. The literature specifically addresses the effects of women's presence on male bonding, suggesting that it is the men's assessment of unit cohesion that would be negatively affected by the increased presence of women. This would suggest that men should be examined separately. On the other hand, since cohesion is a climate variable that pertains to the whole group, not just a portion of it, the inclusion of women would seem to offer a more accurate assessment of the group climate.
The methodological problem is that women tend to score lower on horizontal cohesion measures than men, which could itself account for lower levels of horizontal cohesion in units with more women. This problem can be overcome in two ways: the first is by excluding women from the analyses, or by analyzing men and women separately, as we did in the Garrison 1 study.29 The second solution is to conduct a statistical test known as a z test that can determine whether the correlation based on weighted group scores is significantly larger than the correlation based on individual scores. If the z test is not statistically significant even though the group correlation itself is significant, this could indicate that the accumulated effect of women's lower horizontal cohesion scores is responsible for the significant group level correlation between cohesion and percentage of women. On the other hand, if the z test is statistically significant, this would suggest that the relationship between cohesion and percentage of women is a true group phenomenon. The z test employed in this study is derived from a method known as Within and Between Analysis (WABA),30 and is only applicable to the samples that included both male and female soldiers. It is not applicable to the men only samples because there was no variation in gender in those samples. However, group correlations were calculated for the men only samples by assigning to each male soldier a score based on the percentage of women in the unit.
Results
Comparison of Demographic Variables Across Samples
Table 1 shows differences in types of units across the five studies. Because of the large variety of units involved, patterns are difficult to detect. There appears to be a concentration of transportation, supply, and maintenance units in the Persian Gulf study. The Haiti and Somalia studies have a greater variety of units than the Persian Gulf study. This probably reflects the kinds of units that were available to participate in the survey and differences in the missions.
Table 2 shows age differences by gender across the five studies. In the Persian Gulf and two Garrison studies, male soldiers had an older mean age than female soldiers, but this was not evident for the Somalia and Haiti studies. Table 3 shows marital status and ethnic differences by gender across the five studies. Male soldiers were significantly more likely to be married than female soldiers. However, this did not apply among the junior enlisted, except in the case of Somalia. Across all studies, female soldiers were more likely to belong to minority ethnic groups than males. Ethnic differences across gender were most pronounced in the Garrison 2 study, which had the largest percentage of minority women of all studies, and was the least pronounced in the Garrison 1 and Haiti studies. Demographic comparisons for the junior enlisted in the Garrison 2 study are not presented because this sub-sample was not included in the group level analyses owing to insufficient numbers of junior enlisted in some of the units.
Table 4 shows differences across the five studies with regard to gender composition and group size. The Garrison 2 study had the largest average percentage of women, while Haiti had the smallest average percentage, followed by Somalia. For junior enlisted soldiers, the Garrison I study had the largest average percentage of women. Average company size was largest in the Somalia and Garrison I studies, and smallest in the Persian Gulf study.
Comparison of Horizontal Cohesion and Support: for the Mission by Gender across Samples. In all studies except for the Garrison 2 study, horizontal cohesion scores were significantly higher for men than for women (see Table 5). Because of the different numbers of items in the cohesion scales used in each study, scores were divided by the total number of items in the scale, thus facilitating comparisons across studies. Men reported significantly more support for the mission than women in the Somalia study, and marginally more support than women in the Persian Gulf study. In the Haiti study, on the other hand, female junior enlisted soldiers reported more support for the mission than male junior enlisted, though there was no gender difference for the study sample as a whole (see Table 6).
Group Correlations of Gender Composition by Cohesion across Samples. Table 7 presents group correlations of percentage of women by horizontal cohesion across the five studies. For each of the three studies of deployed soldiers, sixteen versions of this correlation are presented, that is, weighted and unweighted correlations for two versions of the horizontal cohesion scale across four versions of the sample: the total sample (men and women); (2) the total sample for men only; (3) the junior enlisted sample (men and women); and (4) the junior enlisted men only sample. For the Garrison 1 study, information was not available on the core items for the total sample because certain key cohesion questions were designated as inapplicable to higher-ranking respondents in that study. For the Garrison 2 study, information was not available on the junior enlisted sample because of insufficient numbers for a separate group level analysis of junior enlisted soldiers only.
Negative correlations between percentage of women in the unit and horizontal cohesion were evident in four of the five studies-Somalia, Persian Gulf, Garrison 1, and Garrison 2. In the Garrison 2 study, the unweighted correlations for the total sample were only marginally significant. However, the corresponding unweighted correlations for the men only sample were not significant. In the Haiti study, correlations that were not close to zero were actually in the positive direction, and one correlation (the unweighted correlation for the junior enlisted sample) was significant with p < .10. Though several weighted correlations reached statistical significance in the Persian Gulf study, only one z test-that for the core items for the total sample-indicated a significant true group effect. In addition, two of the unweighted correlations reached statistical significance at the p => .01 level, and three were significant at the p < .10 level. The strongest findings were for the Somalia study, in which fourteen out of the sixteen correlations reached statistical significance at or above the p < .OS level, (the highest correlation being -.80). The remaining two correlations reached significance at the p < .06 level. In the Garrison 1 study, all but one of the correlations for the junior enlisted sample were significant at the p < .OS level, and one was significant with p = .06 Two reached significance as group correlations based on the z test. Correlations for the total sample were not significant. In the Garrison 2 study there was insufficient information to conduct correlations for junior enlisted soldiers Correlations were only conducted for the total sample.
Discussion
While different methods of calculating group correlations between percentage of women and unit cohesion produced slightly different results, overall patterns indicate a consistent negative relationship between these two variables. The only study to display no evidence of this negative relationship was the Haiti study. The strongest relationship was found in the Somalia study, where all of the correlations were statistically significant.
The strongest test of the hypothesis that high percentages of women are related to low unit cohesion is provided in the men only sample. By eliminating women from the analysis, one can control for the fact that women's cohesion scores tend to be lower than men's cohesion scores. In the men only analysis, all of the correlations between percentages of women in the unit and unit cohesion were in the anticipated direction with the exception of the Haiti study. The correlations between the two variables for the Garrison 2 study were not statistically significant, but they were significant for the Persian Gulf, Somalia, and Garrison 1 studies.
An even more restrictive test of the hypothesis was presented in the junior enlisted men only analysis. In this analysis, as with the men only analysis, all of the correlations (except those from Haiti) were in the anticipated direction. The correlations for the Persian Gulf study were nonsignificant, but those for Somalia and Garrison 1 were statistically significant.
In addition to the men only and junior enlisted men only analyses, the z test provided yet another method of testing the same hypothesis using the total sample and total junior enlisted sample. The z test provided a method of controlling for the effects of the individual's gender on self-reported cohesion, while simultaneously including the cohesion scores for both genders in the calculation of the group level correlations. Using this method, it was possible to determine whether significant correlations represented true group effects and were not simply a result of individual women reporting significantly less cohesion than individual men. In cases where the z test reached significance (as indicated by the asterisks beside the weighted correlations in Table 7), this may indicate an effect related to women's increased presence in the group, and not simply an increase of the number of individuals with lower cohesion scores. In the Somalia study, all weighted correlations indicated significant group effects using the z test. In the Garrison 1 study, only weighted effects for the junior enlisted sample were significant at the group level, and in the Persian Gulf study, only the weighted effect for core items for the total sample was significant at the group level. There were no significant group effects based on the z test in the Garrison 2 or Haiti studies.
The Haiti study presents a contrast to the other four, in that the correlations between percentage of women and cohesion, though nonsignificant, were actually moving in the positive direction. This was particularly evident in the junior enlisted sample, where one unweighted correlation was marginally significant. This could not be accounted for by individual level scores, because the correlation was in the positive direction despite the fact that women's cohesion scores were lower than those of men. While the effects are too weak to draw any conclusions about the positive effects of women on cohesion, they certainly indicate that negative effects appear common but are by no means universal. Several factors were considered in explaining differences across studies:
Demographic Differences other than Gender
Although there are some demographic differences across studies, none of them appears to account convincingly for differences in the major findings. For example in three of the five studies, the women were significantly younger than the men. However, both the Somalia and Haiti studies show no significant age differences between male and female soldiers and were also most similar in average percentage of women in the unit. However, these two studies were most dissimilar with regard to the major findings. All studies showed a higher proportion of women with minority group status than men, but the differences were least pronounced in Haiti and most pronounced in the Garrison 2 study. However, the Haiti and Garrison 2 studies differed the least with regard to the major findings.
Characteristics of Units: Size and Mission
Average group size could be a factor in accounting for some of the differences in findings. The Somalia and Garrison 1 studies both had the largest average group size. However, in the Garrison 1 study, significant correlations between the major study variables was found for the junior enlisted sample only, while the largest average group size was found for the total sample as well as for the junior enlisted sample. The large variety of units both within and across studies makes it difficult to draw conclusions about the effects of type of unit.
Gender Differences in Support for the Mission
Support for the mission may be a factor in explaining the absence of negative effects in the Haiti study, in which junior enlisted female soldiers reported greater support for the mission than junior enlisted male soldiers. Furthermore, the one marginally significant positive correlation in the Haiti study occurred only in the junior enlisted sample, not in the total sample. The Persian Gulf study showed the opposite effect: support for the mission was higher among men in the total sample, but not among the junior enlisted. Again, it was in the total sample, not among the junior enlisted, that significant negative correlations were found for the major study variables. In the Somalia study, however, significant negative correlations between the major study variables were found both in the junior enlisted sample and the total sample. Greater male support for the mission was only found in the total sample. Gender differences in support for the mission were rather weak, and are therefore unlikely to be the sole explanation for the major study findings.
Timing of Studies
Another factor to consider is the timing of the studies. The two most recent, conducted in 1995, both showed the least negative effects. However, the study with the most negative effects was conducted in 1993, while the Persian Gulf study, with moderate or weak effects, was conducted in 1991. It is possible but unlikely that major changes in the climate for women occurred between 1993 and 1995.
Events in Theater not Specifically Examined in the Studies
Final factors to be considered are events that occurred in theaters that are not measured in these studies. For example, U.S. troops in Somalia were at one time under a Marine Corps Commandant who instituted a policy of treating men and women differently, thus maximizing separation of the sexes. Although the study was conducted after his command had ended, it is still possible that his influence lingered. This does not explain the strong finding among junior enlisted in the Garrison 1 study; however, other unknown leadership factors may have been in effect at that post during the time of the survey.
Perhaps a more compelling argument could be made for a possible effect of the relative level of threat of death or injury to troops in the different environments. In the deployment studies, the strongest negative effect of percentage of women occurred in Somalia, where troops had the greatest likelihood of being shot at despite the fact that this was officially a peacekeeping deployment. In the Persian Gulf study, where troops were less likely to be shot at, the effect on cohesion of the percentage of women was moderately negative. In Haiti, where there was little or no likelihood of troops encountering shooting violence, the effect of the percentage of women on cohesion was neutral or positive. This does not explain differences in the two garrison studies. However, in the Garrison 2 study, which occurred in the same time frame as the Haiti study, some of the participating units were preparing for deployment to Haiti, and training conditions may have simulated those in the Haiti environment.
In conclusion, this meta-analysis of the relationship between gender composition and unit cohesion found that while the negative effects of increased female presence on group cohesion have occurred in a variety of settings, both deployed and non-deployed, the findings are by no means universal or even consistently strong. No specific factor has hitherto been found that could account for all of the differences, but some that should be examined in future research include size of the unit, soldiers' support for the mission, level of violence in theater, and the effects of leadership policies regarding the treatment of the genders.
Notes
AUTHORS' NOTE: The views of the authors do not purport to reflect the position of the Department of Defense or the Department of the Army (para 4-3, AR 360-5).
1.Jackie Spinner, "Aberdeen Sergeant Convicted of Rape," Washington Post, Wednesday, 30 April 1997; P. T. Kilborn, "Sex Abuse Cases Stun Pentagon, but the Problem Has Deep Roots," New York Times, Monday, 10 February 1997, B8. 2. Tamara Jones, "U.S. Military Takes Aim at Adultery," Washington Post, Monday, 28 April 1997; Jackie Spinner and Dana Priest, "Consensual Sex was Widespread at Army Base," Washington Post, Sunday, 30 March 1997, A7; F. Shen, "At Army Post, Fraternization is the Enemy," Washington Post, Sunday, 24 November, 1996, A24.
3. Stephanie Gutmann, "Sex and the Soldier," The New Republic, 24 February 1997, 18-22.
4. U.S. Army Research Institute for the Behavioral and Social Sciences, Women Content in Units Force Development Test (MAXWAC), (Alexandria, VA: U.S. Army Research Institute, 1977).
5. Cecil F. Johnson, Bertha H. Cory, Roberta W. Day, and Laurel W. Oliver, Women Content in the Army Reforger 77 [REFWAC77] (Alexandria, VA: U.S. Army Research Institute, 1978).
6. J. A. Mottern and Z. M. Simutis, "Gender Integration of U.S. Army Basic Combat Training," in Proceedings of the 36th Annual Conference of the International Military Testing Association, (Rotterdam, The Netherlands, 25-27 October, 1994), 24 29.
7. Paul L. Savage and Richard A. Gabriel, "Cohesion and Disintegration in the American Army: An Alternative Perspective," Armed Forces & Society 3 (1976):340-376, 349.
8. Frederick J. Manning, "Morale and Cohesion in Military Psychiatry," in Military Psychiatry: Preparing in Peace for War, ed. Franklin D. Jones, Linette R. Sparacino, Victoria L. Wilcox, Joseph M. Rothberg, (Falls Church, VA: Office of the Surgeon General, U.S. Department of the Army, 1994), 1-18. 9. Edward C. Meyer, "The Unit," Defense 82, (February 1982): 1-9. 10. R. Little, "Buddy Relations and Combat Performance," in The New Military, ed. M. Janowitz (New York: Russell Sage Foundation, 1964), 1995-224. 11. Larry H. Ingraham, The Boys in the Barracks, (Philadelphia, PA: Institute for the Study of Human Issues, 1983).
12. A. L. George, "Primary Groups, Organization, and Military Performance," in Handbook of Military Institutions, ed. R.W. Little (Beverly Hills, CA: Sage, 1971), 293318.
13. Manning, "Morale and Cohesion in Military Psychiatry," 1994. 14. Leora N. Rosen, Doris B. Durand, Paul D. Bliese, Ronald R. Halverson, Joe M. Rothberg, and Nancy L. Harrison. "Cohesion and Readiness in Gender Integrated Combat Service Support Units: Impact of Acceptance of Women and Gender Ratio," Armed Forces & Society 22 (1996): 537-553.
15. Rosabeth M. Kanter, "Some Effects of Proportion on Group Life: Skewed Sex Ratios and Responses to Token Women," American Journal of Sociology 82 (1977): 965990.
16. Hubert M. Blalock, Towards a Theory of Minority Group Relations, (New York, Capricorn, 1970).
17. Leora N. Rosen and Lee Martin, "Sexual Harassment, Cohesion, and Readiness in U.S. Army Support Units," Armed Forces & Society 24 (1997): 221-244. 18. D. H. Marlowe, "Personnel and Manpower: Change and Evolution in the Human Dimensions of Military Service" in American Defense Annual, ed. C. F. Herman (New York: Lexington, and Mershon Center: Ohio State University, 1994), 158-159. 19. Rosen et al., "Cohesion and Readiness," 1996 20. Previously unpublished data.
21. Previously unpublished data.
22. Ronald R. Halverson, Paul D. Bliese, Robert M. Moore, and Carl A. Castro, Psychological Well-being and Physical Health Symptoms of Soldiers Deployed for Operation Uphold Democracy: A Summary of Human Dimensions Research in Haiti (Washington, DC: Walter Reed Army Institute of Research, 1995).
23. Rosen and Martin, "Sexual Harassment, Cohesion and Readiness." 24. J. Griffith, "Measurement of Group Cohesion in U.S. Army Units," Basic and Applied Social Psychology 9 (1988): 149-171; D. H. Marlowe, "New Manning System Field Evaluation (Washington, DC: Technical Report No. 1, Walter Reed Army Institute of Research, 1986).
25. Paul D. Bliese and Ronald R. Halverson, "Individual and Nomothetic Models of Job Stress: An Examination of Work Hours, Cohesion, and Well-being," Journal of Applied Social Psychology 26 (1996): 1171-1189; K. J. Klein, F. Dansereau, R. J. Hall, "Levels Issues in Theory Development, Data Collection, and Analysis," Academy of Management Journal 19 (1994): 195-229; S. E. Markham, "Pay-For-Performance Dilemma Revisited: Empirical Example of the Importance of Group Effects," Journal of Applied Psychology 73 (1988): 172-180; S. E. Markham and G. H. McKee, "Group Absence Behavior and Standards: A Multilevel Analysis," Academy of Management Journal 38 (1995): 1174-1190. 26. Markham, "Pay-For-Performance Dilemma Revisited," 1988. 27. Paul D. Bliese and Ronald R. Halverson, "Group Consensus and Psychological WellBeing: A Large Field Study," Journal of Applied Social Psychology 28 (1998):563580.
28. Rosen et al., "Cohesion and Readiness," 1996. 29. Markham, "Group Absence Behavior," 1995.
30. In WABA. the z score is obtained by subtracting the z transformation of the weighted between-group correlation (the group correlation) from the z transformation of the weighted within-group correlation. The within-group correlation is based on the individual's position above or below the group mean. In other words, the group score is subtracted from the individual's score and the resulting score is then used to compute the within-group correlation. In the case of gender, the average percentage of women in the group is subtracted from the individual's gender. Gender is thus treated as a quantifiable variable, and is dummy coded as O=female and 1=male.
LEORA ROSEN is a social science analyst at the National Institute of Justice, where she is responsible for assessing and implementing national level research dealing with violence against women and family violence. She previously worked in the Division of Neuropsychiatry, Walter Reed Army Institute of Research (WRAIR). Her research focused on the military family and women in the military. Address for correspondence: Leora Rosen, Ph.D., 11110 Stephalee, Rockville, MD 20852,
PAUL BLIESE, Ph.D., is an Army captain and research psychologist in the Division of Neuropsychiatry, WRAIR. His research focuses on soldier stress and adaptation to military operations.
KATHLEEN WRIGHT, Ph.D., is the Deputy Director of the Division of Neuropsychiatry, WRAIR. She has conducted research in the areas of military operational and traumatic stress.
ROBERT K. GIFFORD, is an Army colonel assigned as Executive Officer of WRAIR. He also serves as research psychology consultant to the Surgeon General of the Army.
Copyright Transaction Inc. Spring 1999