背景與目的：針對回塑性議題的大型抽樣調查資料結構特性，指出傳統分析方法的缺失，進一步提出更恰當的模型-多層次離散時間存活模型，希望能更精確釐清影響青少年發生早齡性交與初次性交不使用避孕措施的相關因素。
研究設計：採用多階段抽樣的橫斷性調查。屬於二手資料分析。
研究對象：採用國民健康局於2000年所進行的「台灣地區在校學生性知識態度與行為研究調查」之資料，篩選15-19歲完訪者，共計7336在校生。
主要測量變項與統計方法：依變項為初次性交是否使用避孕措施。以隨機效用多項邏輯特模型進行分析。
結果：僅9.8%的男生與8.2%的女生曾發生過性行為，而有發生過性經驗的男生中，初次性交有避孕者占58.8%；女生則有55.4%。其中認為自己可以從事婚前性行為、目前有抽菸、認為同齡同儕有人發生性交是導致男女早齡性交發生的共同因子。而影響男生初次性交避孕使用的危險因子包括：認為自己可以從事婚前性行為（Z統計量＝-2.54，P-value＝0.01）、認為家人愛且關心自己（Z統計量＝-1.96，P-value＝0.03）；保護因子則為：曾吸食毒品（Z統計量＝1.91，P-value＝0.03）、及母親認為可以發生婚前性行為（Z統計量＝1.82，P-value＝0.04）；影響女生初次性交避孕使用的危險因子包括：曾看色情書刊（Z統計量＝-1.66，P-value＝0.05）、認為家人愛且關心自己（Z統計量＝-1.80，P-value＝0.04）、父親對婚前性行為持沒有意見的態度（Z統計量＝-2.00，P-value＝0.02）、就讀高職（Z統計量＝-1.61，P-value＝0.05）；保護因子則為：曾看色情電影或錄影帶（Z統計量＝3.17，P-value<.01）、與父親關係良好（Z統計量＝2.77，P-value<.01）、不希望繼續升學（Z統計量＝2.03，P-value＝0.02）。
結論與建議：同儕關係和親子互動是影響青少年早齡性交的主因。但同儕對性交時是否採取避孕措施的影響卻不顯著；而親子兩性溝通和父母親的性態度則對青少年初次性交是否採取安全性行為扮演重要角色，突顯家庭教育的重要。此外，男生早齡性交的發生和避孕使用在不同學校間差異不大，但女生卻會受極大的影響，因此期待學校教育能針對不同特質的學生加強關懷與宣導。

Objectives：In regard to the data structure of large-scale sample survey of retrospective issue, we pointed out the lack of traditional method and proposed more appropriate model, multi-level discrete time survival model, to analysis. Through the model, we hoped to get the more accurate results about the risk and protective factors for the age at the first sexual intercourse and contraceptive use.
Design：It is a multi-stage cross-sectional survey and a secondary data analysis.
Participants：The data were obtained from a high school survey of sexual knowledge, attitude and behavior conducted by the Bureau of Health Promotion in 2000. The total sample of 7336 students aged 15-19 years.
Main Outcome Measures and Statistical Method：The dependent variable is dose teenagers use contraception measures or not at first sexual intercourse. The relationship of person, family, and school factors to age at first sexual and contraceptive use were estimated by multinomial logit model with random effects.
Results：9.8% of boys and 8.2% of girls were sexually experienced. Among the boys and girls who had sexual behavior, there were 58.8% of boys and 55.4% of girls had used of contraception at the first sexual intercourse. Both boys and girls, the common significant risk factors for the age at debut of sexual intercourse were believing that they can engage in premarital sex, Current smokers, and having friends who were sexually experienced. The significant risk factors for contraceptive use at the initiation of sexual intercourse for boys were believing that they can engage in premarital sex (Z-statistics＝-2.54，P-value＝0.01), and thinking that family love and care about their (Z-statistics＝-1.96，P-value＝0.03). On the other hand, having abused drugs (Z-statistics＝1.91，P-value＝0.03), and mother felt premarital sex can occur (Z-statistics＝1.82，P-value＝0.04) were protective factors. Among girls, risk factors were looking at pornography (Z-statistics＝-1.66，P-value＝0.05), thinking that family love and care about their (Z-statistics＝-1.80，P-value＝0.04), fathers’ neutral attitude towards premarital sex (Z-statistics＝-2.00，P-value＝0.02), and studying in alternative high schools (Z-statistics＝-1.61，P-value＝0.05). Protective factors were watching pornographic films or videos (Z-statistics＝3.17，P-value<.01), good relationship with their fathers (Z-statistics＝2.77，P-value<.01), and did not wish to continue their studies (Z-statistics＝2.03，P-value＝0.02).
Conclusions：Peer relations and parent-child interaction were the two main reasons that affected teenager have sex at early age. The factors of peer were not significant for contraceptive use, and the family factors like parent-child communication and parental attitudes about sex were played an important role in contraceptive use at the debut of sexual intercourse. It is highlight the importance of family education. In addition, the occurrence of sexual intercourse at early age and contraceptive use were not different for boys in the different schools, but greatly different for girls. Hence, we looked forward that school education could enhance care and advocacy for different characteristics of students.

林惠生，台灣地區高中、高職及五專在校學生之性知識、性態度及危害健康行為與網路之使用。行政院衛生署國民健康局人口與健康調查研究中心，2002。
陳科榕，以多層次分析探討學校同儕和家庭因素與青少年初次性交年齡及避孕使用之相關性研究。國立成功大學公共衛生研究所，未發表碩士論文，2006。
溫福星，階層線性模式：原理、方法與應用，初版。臺北，雙葉書廊有限公司，2007。

Agresti A. Categorical Data Analysis, John Wiley and Sons, Inc., Hoboken, New Jersey, 2002.
Allison PD. Discrete-Time Methods for the Analysis of Event Histories. Sociological Methodology 1982, 12: 61-98.
Aspy CB, Vesely SK, Oman RF, Rodine S, Marshall L, McLeroy K. Parental communication and youth sexual behaviour. Journal of Adolescence 2007, 30: 449-466.
Brindis C, Boggess J, Katsaranis F, Mantell M, McCarter V, and Wolfe A. A profile of the adolescent male family planning client. Family Planning Perspectives 1998, 30: 63-6,88.
Brown SS and Eisenberg L. The Best Intentions:Uunintended Pregnancy and the Well-Being of Children and Families, Washington, DC: National Academy Press, 1995.
Clawson CL, and Reese-Weber M. The amount and timing of parent-adolescent sexual communication as predictors of late adolescent sexual risk-taking behaviors. Journal of Sex Research 2003, 40: 256-265.
Corcoran J. Ecological factors associated with adolescent sexual activity. Social Work in Health Care 2002, 30: 93-111.
David HP, Morgall JM, Osler M, Rasmussen NK, and Jensen B. United States and Denmark: different approaches to health care, and family planning. Studies in Family Planning 1990, 21: 1-19.
Edwards S. The role of men in contraceptive decision-making: current knowledge, and future implications. Family Planning Perspectives 1994, 26: 77-82.
Fingerson L. Do mothers' opinions matter in teens' sexual activity? Journal of Family Issuses 2005, 26: 947-974.
Gibbons RD, Hedeker D. Random Effects Probit and Logistic Regression Models for Three-Level Data. Biometrics 1997, 53:1527-1537.
Godeau E, Nic Gabhainn S, Vignes C, Ross J, Boyce W, and Todd J. Contraceptive use by 15-year-old students at their last sexual intercourse results from 24 countries. Archives of Pediatrics and Adolescent Medicine 2008, 162: 66-73.
Goldstein H. Multilevel Statistical Models. 2rd ed. London, United Kingdom: Edward Arnold, 1995.
Hox JJ. Multilevel Analysis: Techniques and Applications, Mahwah, N.J.: Lawrence Erlbaum Associates, 2002.
Hutchinson MK. The influence of sexual risk communication between parents and daughters on sexual risk behaviors. Family Relations 2002, 51: 238-247.
Jones JM, Toffler W, and Reed B. The Declines in Adolescent Pregnancy, Birth, and Abortion Rates in the 1990s: What Factors are Responsible? In: Fanwood, NJ: Consortium of State Physicians Resource Councils, 1999.
Kahn, J.A., Rosenthal, S.L., Succop, P.A., Ho, G.Y., and Burk, R.D. Mediators of the association between age of first sexual intercourse and subsequent human papillomavirus infection. Pediatrics 2002, 109: E5.
Kinsman SB, Romer D, Furstenberg FF, and Schwarz DF. Early sexual initiation: the role of peer norms. Pediatrics 1998, 102: 1185-1192.
Kreft IGG. Are Multilevel Techniques Necessary? An Overview, Including Simulation Studies. California State University, Los Angeles.
Luck D. Multilevel modeling. Newbury Park, CA: Sage, 2004.
Maas CJM, Hox JJ. Robustness issues in multilevel regression analysis. Statistica Neerlandica 2004, 58: 127-137.
Manlove J, Ryan S and Franzetta K. Patternso f contraceptive use within teenagers' first sexual relationships. Perspectives on Sexual and Reproductive Health 2003, 35:246-255.

McNeely C, Shew ML, Beuhring T, Sieving R, Miller BC, and Blum RW. Mothers' influence on the timing of first sex among 14- and 15-year-olds. The Journal of Adolescent Health 2002, 31: 256-265.
Miller BC, and Moore KA. Adolescent sexual behavior, pregnancy, and parenting: research through the 1980s. Journal of Marriage and the Family 1990, 52: 1025-1044.
Miller BC, Norton MC, Curtis T, Hill EJ, Schvaneveldt P, and Young MH. The timing of sexual intercourse among adolescents: family, peer, and other antecedents. Youth and Society 1997, 29: 54-83.
Miller BC, and Benson B. Family relationships and adolescent pregnancy risk: A research synthesis. Developmental Review 2001, 21: 1-38.
Moore KA, and Peterson JF. Parental attitudes and the occurrence of early sexual activity. Journal of Marriage and the Family 1986, 48: 777-782.
Mott FL, and Haurin RJ. Linkages between sexual activity and alcohol and drug use among American adolescents. Family Planning Perspectives 1988, 20: 129-136.
O'Donnell BL, O'Donnell CR, and Stueve A. Early sexual initiation and subsequent sex-related risks among urban minority youth: the reach for health study. Family Planning Perspectives 2001, 33: 268-275.
Rabe-Hesketh S, Yang S, Pickles A. Multilevel Models for Censored and Latent Responses. Statistical Methods in Medical Research 2001, 10: 409-427.
Reardon SF, Brennan RT, Buka SL. Estimating Multi-Level Discrete-Time Hazard Models Using Cross-Sectional Data: Neighborhood Effects on the Onset of Adolescent Cigarette Use. Multivariate Behavioral Research 2002, 37: 297-330.
Santelli JS, Kaiser J, Hirsch L, Radosh A, Simkin L, and Middlestadt S. Initiation of sexual intercourse among middle school adolescents: the influence of psychosocial factors. The Journal of Adolescent Health 2004, 34: 200-208.
Scott-Jones D, and White AB. Correlates of sexual activity in early adolescence. Journal of Early Adolescence 1990, 10: 221-238.
Seiving RE, McNeely CS, and Blum RW. Maternal expectations, mother-child connectedness,and adolescent sexual debut. Archives of Pediatrics and Adolescent Medicine 2000, 154: 809-816.

Steele F, Diamond, Dualao W. The Determinants of Duration of Contraceptive Use in China: A Multilevel Multinomial Discrete-Hazards Modeling Approach. Demography 1996, 33: 12-23
Sturdivant RX. and Hosmer DW. A smoothed Residual Based Goodness-of-Fit Statistics for Logistic Hierarchical Regression Models. Computational Statistics & Data Analysis 2007, 51: 3898-3912
Svare EI, Kjaer SK, Thomsen BL, and Bock JE. Determinants for non-use of contraception at first intercourse; a study of 10,841 young Danish women from the general population. Contraception 2002, 66: 345-350.