Does enlistment propensity predict accession? High school seniors' plans and subsequent behavior

Armed Forces and Society

Volumn 25, Issue 1, 1998, Pages 59-80

  Bachman, Jerald G ^a   Segal, David R ^a   Freedman Doan, Peter ^a   O'Malley, Patrick M ^a  

^a NONE

Author keywords

[No Author keywords available]

Indexed keywords

EID: 0032148294     PISSN: 0095327X     EISSN: None     Source Type: Journal    
DOI: 10.1177/0095327X9802500104     Document Type: Article

References (29)

1. 1. Charles C. Moskos, "The Marketplace All-Volunteer Force," in The All Volunteer Force after a Décade, eds. William Bowman, Roger Little, and G. Thomas Sicilla (Washington, D.C.: Pergamon-Brassey's, 1986), 15-22.
2. 2. Gary R. Nelson, "The Supply and Quality of First-Term Enlistees under the All-Volunteer Force," in The All Volunteer Force after a Decade, 23-51.
3. 3. See, for example, Mady Wechsler Segal, "Women's Military Roles Cross-Nationally," Gender and Society 9 (December 1995): 757-775.
4. 4. Brace R. Orvis, Martin T. Gahart, and Alvin K. Ludwig, Validity and Usefulness of Enlistment Intention Information (Santa Monica, CA: RAND Corporation, 1992).
5. 5. Beth Asch and Bruce Orvis, Recent Recruiting Trends and Their Implications: Preliminary Analysis and Recommendations (Santa Monica, CA: RAND Corporation, 1994).
6. 6. Jerald G. Bachman, Lloyd D. Johnston, and Patrick M. O'Malley, The Monitoring the Future Project after Twenty-Two Years: Design and Procedures, MtF Occasional Paper 38. (Ann Arbor, MI: The Institute for Social Research, 1996).
7. 7. A base year (hereafter, BY) sample is drawn each year using a three-stage probability sampling design to select approximately 130 public and private high schools representative of those in the 48 contiguous states. Professional interviewers from the Institute for Social Research supervise survey activities at the school site, usually during regular classroom periods in March, April, or May. All respondents are asked to Fill out one of six forms of a 45-minute, paper and pencil, self-administered questionnaire. Student response rates vary from school to school, between 75 percent and 100 percent, producing sample sizes of roughly 17,000 seniors each year. Because of changes in the questionnaire design, we report data from classes beginning in 1976.
8. 8. From each senior class, 2,400 seniors are selected for follow-up, and randomly divided into two groups, each numbering about 1,200. Members of one group are mailed questionnaires one year after graduation, and every two years thereafter; those in the other groups are mailed questionnaires two years after graduation, and every two years thereafter. Thus, individual participants are surveyed on a two-year cycle, beginning either one or two years after graduation, for a total of up to seven follow-ups. Respondents are paid $5 ($1 beginning with the class of 1991) for each follow-up participation. The follow-up samples are drawn so as to be largely self-weighting; however, because the primary focus of the study is on drug use, users of illicit drugs are oversampled for follow-ups (by a factor of three to one). Weights are used in all analyses to adjust for the differential selection probabilities. In this report, we use base-year data from the classes of 1976 through 1991, with follow-ups through 1995. All respondents from the classes of 1976 through 1989, and one-half of the class of 1990, have had the opportunity to respond to three follow-ups; the class of 1991 has had the opportunity to respond to only two follow-ups.
9. 9. Since Monitoring the Future base-year data are collected in schools on a single day, absent students do not complete a questionnaire; consequently, we adjusted for the effects of excluding absentees in our estimates. Students are asked how many days of school they have missed in the previous weeks. Using this variable, individuals are assigned to different strata as a function of how often they are absent. Actual base-year participants in each stratum are weighted to represent all students in their stratum, including absentees on the particular date of administration. Our procedures produce weighted numbers of cases that are somewhat smaller than the actual numbers of follow-up cases (39,761 actual follow-up respondents and of those, 35,693 who answered the base-year propensity question).
10. 10. Of the 33,241 weighted cases used in the analyses, 16,619 (49.9 percent) had observations in all three follow-ups. An additional 4,535 (13.6 percent) weighted cases - with some missing data - were assigned a value based on their prior or subsequent responses; if a respondent at the first or second follow-up indicated having entered military service, later missing values were so coded; and if a respondent indicated in a later follow-up that he or she had never entered military service, then prior missing values were also appropriately coded. One or more values were imputed to the remaining 12,077 (36.3 percent) via a variation of the "hot-deck" imputation procedure (see Roderick Little and Donald Rubin, Statistical Analysis with Missing Data [New York: John Wiley and Sons, 1987], esp. pp. 65-71). Each respondent with missing military accession data in any of the follow-ups was matched on senior year propensity, sex, race, and class year with other respondents who provided valid data on military accession at all three follow-ups. Cases with missing values for military accession were "shuffled" in a random order into the set of matched "donor" cases, and their value for military accession was imputed from their nearest neighbor in the stack. We cross-tabulated senior year propensity with subsequent accession in two ways: one was restricted to just those cases with three valid observations, and the other used our total adjusted sample. The two cross-tabulations yielded accession rates from each propensity group that were remarkably similar. In other words, our data imputation did not have significant effects on our central claim that base-year propensity to enlist powerfully related to subsequent accession. However, our adjustments did retain the full proportion of cases with high propensity to enlist. Thus, the effects of our data adjustments were to improve the accuracy of our estimates of overall rates of accession. The appropriate tables and discussion can be found in Jerald G. Bachman, Peter Freedman-Doan, David Segal, and Patrick M. O'Malley, Military Propensity and the Propensity Enlistment Relationship, MtF Occasional Paper 39 (Ann Arbor, MI: The Institute for Social Research, 1997).
11. 10. Of the 33,241 weighted cases used in the analyses, 16,619 (49.9 percent) had observations in all three follow-ups. An additional 4,535 (13.6 percent) weighted cases - with some missing data - were assigned a value based on their prior or subsequent responses; if a respondent at the first or second follow-up indicated having entered military service, later missing values were so coded; and if a respondent indicated in a later follow-up that he or she had never entered military service, then prior missing values were also appropriately coded. One or more values were imputed to the remaining 12,077 (36.3 percent) via a variation of the "hot-deck" imputation procedure (see Roderick Little and Donald Rubin, Statistical Analysis with Missing Data [New York: John Wiley and Sons, 1987], esp. pp. 65-71). Each respondent with missing military accession data in any of the follow-ups was matched on senior year propensity, sex, race, and class year with other respondents who provided valid data on military accession at all three follow-ups. Cases with missing values for military accession were "shuffled" in a random order into the set of matched "donor" cases, and their value for military accession was imputed from their nearest neighbor in the stack. We cross-tabulated senior year propensity with subsequent accession in two ways: one was restricted to just those cases with three valid observations, and the other used our total adjusted sample. The two cross-tabulations yielded accession rates from each propensity group that were remarkably similar. In other words, our data imputation did not have significant effects on our central claim that base-year propensity to enlist powerfully related to subsequent accession. However, our adjustments did retain the full proportion of cases with high propensity to enlist. Thus, the effects of our data adjustments were to improve the accuracy of our estimates of overall rates of accession. The appropriate tables and discussion can be found in Jerald G. Bachman, Peter Freedman-Doan, David Segal, and Patrick M. O'Malley, Military Propensity and the Propensity Enlistment Relationship, MtF Occasional Paper 39 (Ann Arbor, MI: The Institute for Social Research, 1997).
12. 11. For a fuller discussion of changes over time see Bachman et al., Military Propensity and the Propensity-Enlistment Relationship.
13. 12. Asch and Orvis, Recent Recruiting Trends and Their Implications, 8.
14. 13. The Pearson correlation coefficient assesses the degree of linear association. The eta statistic indicates the strength of the relationship regardless of its linearity; accordingly, the difference between the Pearson correlation and the eta is an indicator of the extent to which the relationship is nonlinear. As can be seen in Figure 1 (and others), the association between propensity and accession is monotonic, but not completely linear (because there is less of a "jump" in the first interval, between the first two categories, compared to the "jumps" associated with the other two intervals). Nevertheless, the association is mostly linear, and the Pearson product-moment correlation coefficient captures the essence of the association. Given its familiarity to most readers, we find it a useful summary statistic.
15. 14. David R. Segal and Jerald G. Bachman, "The Military as an Educational and Training Institution," Youth and Society 10 (1978): 127-134; Bruce R. Orvis, Martin T. Gahart, and Alvin K. Ludwig, Validity and Usefulness of Enlistment Intention Information (Santa Monica, CA: RAND Corporation, 1992); Veronica Nieva, Michael J. Wilson, Dwayne G. Norris, James B. Greenlees, Janice Laurence, and Rod McCloy, Enlistment Intentions and Behaviors: A Youth and Parental Model (Alexandria, VA: U.S. Army Research Institute for the Behavioral and Social Sciences, 1996).
16. 14. David R. Segal and Jerald G. Bachman, "The Military as an Educational and Training Institution," Youth and Society 10 (1978): 127-134; Bruce R. Orvis, Martin T. Gahart, and Alvin K. Ludwig, Validity and Usefulness of Enlistment Intention Information (Santa Monica, CA: RAND Corporation, 1992); Veronica Nieva, Michael J. Wilson, Dwayne G. Norris, James B. Greenlees, Janice Laurence, and Rod McCloy, Enlistment Intentions and Behaviors: A Youth and Parental Model (Alexandria, VA: U.S. Army Research Institute for the Behavioral and Social Sciences, 1996).
17. 14. David R. Segal and Jerald G. Bachman, "The Military as an Educational and Training Institution," Youth and Society 10 (1978): 127-134; Bruce R. Orvis, Martin T. Gahart, and Alvin K. Ludwig, Validity and Usefulness of Enlistment Intention Information (Santa Monica, CA: RAND Corporation, 1992); Veronica Nieva, Michael J. Wilson, Dwayne G. Norris, James B. Greenlees, Janice Laurence, and Rod McCloy, Enlistment Intentions and Behaviors: A Youth and Parental Model (Alexandria, VA: U.S. Army Research Institute for the Behavioral and Social Sciences, 1996).
18. 15. Asch and Orvis, Recent Recruiting Trends and Their Implications.
19. 16. Bruce Orvis, personal communication, September, 1996.
20. 17. Segal and Bachman, "The Military as an Educational and Training Institution."
21. 18. Bachman et al., Military Propensity and the Propensity-Enlistment Relationship.
22. 19. Bachman et al., Military Propensity and the Propensity-Enlistment Relationship.
23. 20. Bachman et al., Military Propensity and the Propensity-Enlistment Relationship, and Segal and Bachman, "The Military as an Educational and Training Institution."
24. 20. Bachman et al., Military Propensity and the Propensity-Enlistment Relationship, and Segal and Bachman, "The Military as an Educational and Training Institution."
25. 21. David R. Segal and Jerald G. Bachman, "Change in the All-Volunteer Force: Reflections in Youth Attitudes," in Marching Towards the 21st Century: Military Manpower and Recruiting, eds. Mark J. Eitelberg and Stephan L. Mehay (Westport, CT: Greenwood Press, 1994), 149-166.
26. 22. Segal and Bachman, "Change in the All-Volunteer Force," in Marching, 153, and Bachman et al., Military Propensity and the Propensity-Enlistment Relationship.
27. 22. Segal and Bachman, "Change in the All-Volunteer Force," in Marching, 153, and Bachman et al., Military Propensity and the Propensity-Enlistment Relationship.
28. 23. Jerald G. Bachman, David R. Segal, Peter Freedman-Doan, Patrick M. O'Malley, "Who Chooses Military Service? Correlates of Propensity and Enlistment in the United States Armed Forces," Military Psychology (in press).
29. 24. Jerald G. Bachman, Lee Sigelman, and Greg Diamond, "Self-Selection, Socialization, and Distinctive Military Values," Armed Forces & Society 13 (1987): 169-187.