2015 Vol. 31 No. 3 350-359 Psychological Development and Education DOI 10. 16187 /j. cnki. issn1001-4918. 2015. 03. 13 * 1 2 2 3 2 2 1. 510631 2. / 510631 3. 518060 latent profile analysis LPA 12718 644 1 9. 86% 19. 15% 70. 99% 2 Z 2. 6SD 61. 21% 38. 28% 8. 36% 3 8. 93% ~ 35. 26% LPA UPI B844 1 1. 1 2007 2005 UPI 0-1 UPI University Personality Inventory UPI 2012 UPI 1966 Yoshitake 1996 2007 2005 1991 1995 1999 2003 UPI UPI * 14ZDB159 2013JK045. E-mail zhangwei@ scnu. edu. cn 350
τ 1. 2 Vermunt latent profile model LPM 1968 latent class model LCM 1. 3 50 Marsh Lüdtke Trautwein & 2008 latent class Morin 2009 analysis LCA I jk k j L I k k C T 1 P I k = T Latent P C = t P I k C = t 1 t = 1 class Latent classes Finite mixture Finite 20081 mixtures T P C = t = 1 2 1704 64 t = 1 LCA Maarten Naaktgeboren Reitsma Moons & Groot 2 2013 LCA P I k C = t = J P I jk C = t 2 38 6 j = 1 4 3 P C = t t 2 P I jk = l C = t t 2002 j l 1 2007 LCM L P I jk = l C = t = 1 5 DSM l = 1 Latent Profile Analysis LPA - IV SPD SPD LCA 1 f I k = T P C = t f I k μ t t = 1 t 3 SPD Fossati Krueger τ t Markon Borroni & Maffei 2013 Szatmari 1995 component μ t 10 ICD - 10 351
2015 3 LCM 6 1 - π i0 i 0 2013 2008 Bertrand Bénichou Grenier & Chastang 2005 π i1 = P X i = 1 Y = 1 5 Antonius Van & Tim 2010 1 - π i0 = P X i = 0 Y = 0 6 2009 2013 UPI 2013 2014 UPI Golden Standard LPA person - centered 2 2. 1 2012 2013 N 2009 1. 4 = 12718 19. 1 ± 0. 95 64% 2. 2 2. 2. 1 UPI UPI 1995 UPI 60 4 56 25 UPI sensitivity 1 UPI 25 25 2 specificity 25 3 4 5 π i1 i 1 P X i = 1 Y = 1 Y = 1 1 UPI 20 ~ 25 X i = 1 20 25 2 8 16 26 3 352
UPI 0. 914 2. 2. 2 90 SCL - 90 0 ~ 5 12 10 13 10 6 6 10 9 7 SCL90 0. 83 0. 82 0. 87 AIC BIC 0. 83 0. 74 0. 71 0. 78 0. 82 0. 69 2. 3 2013 644 AIC BIC LMRT 2009 UPI 20 ~ 30 Mplus7. 0 SPSS18. 0 3. 1 UPI 55 Kappa - 4 0. 79-0. 78-0. 92 2. 4 2. 4. 2 UPI 2012 6129 2013 6589 LMRT Entropy Entropy 3 25 52% RMSEA = 0. 02 SRMR = 0. 04 CFI = 0. 98 TLI = 2. 4. 1 0. 97 χ 2 /df = 1. 88 UPI 2012 6129 27 = 64% RMSEA = 0. 02 SRMR = 0. 04 CFI = 0. 99 TLI = 0. 99 χ 2 /df = 1. 87 52% 64% Muthén 1989 0. 5 0. 23 0. 53 UPI y* y* A B C < 0. 4 353
2015 3 25 25 1 D 3 E 3 ~ 6 27 3 1 χ 2 = 1073. 07 DF = 2 2 314 RMSEA = 0. 03 CFI =. 97 TLI =. 97 2012 2013 α = 0. 61 0. 59 0. 61 4 ~ 6 α = 0. 70 3 3 4 5 0. 81 3. 2 3 UPI 10% UPI 4 25 3 T1 n = 6129 T2 n = 6752 1 1 2 3 4 5 6 BIC 97376. 26 52384. 56 27587. 48 17185. 18 12463. 93 11340. 68 AIC 97309. 09 52243. 49 27372. 51 16896. 32 12101. 18 10904. 03 Entropy 1. 00 0. 96 0. 99 0. 99 0. 99 0. 99 LMRT 0. 00 0. 01 0. 00 0. 00 0. 00 0. 01 BIC 110492. 80 101462. 4 0 97606. 46 96423. 07 88142. 79 87581. 14 AIC 110424. 60 101353. 30 97456. 47 96232. 17 87910. 99 87308. 43 Entropy 1. 00 0. 88 o. 88 0. 83 0. 93 0. 87 LMRT 0. 00 0. 01 0. 00 0. 00 0. 00 0. 01 2 T1 2012 n = 6129 UPI Z T2 2013 n = 6752 UPI Z n % n % 1 4476 73. 03 -. 24 -. 37 -. 34 -. 49 -. 23 -. 45 4793 70. 99 -. 28 -. 45 -. 37 -. 47 -. 27 -. 52 2 1061 17. 31. 56. 86 -. 00 1. 36. 61 1. 06 1293 19. 15. 57 1. 05. 07 1. 14. 69 1. 07 3 573 9. 35. 86 1. 30 2. 62 1. 27. 67 1. 57 666 9. 86. 88 1. 24 2. 59 1. 23. 66 1. 69 2 9. 35% ~ 9. 86% UPI UPI 1. 5 2. 6 1 T2 2013 UPI 1 70. 99% 354
61. 21% 25. 49% 4 19. 15% 38. 28% 8. 36% 2 8 9. 86% 3. 3 21. 09% 3. 3. 1 52. 28% SCL90 3 SCL90 T2 2013 SCL90 0 1 2 3 4 254 101 64 69 167 655 % 38. 79 15. 42 9. 77 10. 53 25. 49 777 177 99 83 123 1259 % 61. 72 14. 06 7. 86 6. 59 9. 77 4284 244 66 30 51 4675 % 91. 64 5. 22 1. 41 0. 64 1. 09 566 151 105 116 248 1186 % 47. 72 12. 73 8. 85 9. 78 20. 91 1433 192 83 49 59 1816 % 78. 91 10. 57 4. 57 2. 70 3. 25 3316 179 41 17 34 3587 % 92. 44 4. 99 1. 14 0. 47 0. 95 5315 522 229 182 341 6589 0 ~ 4 SCL90 0 ~ 4 3. 3. 2 4 SCL - 90 6589 655 401 61. 21% 91. 64% 644 39 33 84. 61% 92. 63% 644 39 29 74. 36% 97. 21% 644 39 25 64. 10% 96. 81% SCL - 90 6589 1186 620 52. 28% 92. 44% 644 77 38 49. 35% 95. 01% 644 77 32 41. 56% 93. 27% 644 77 31 40. 26% 93. 27% 90% 4 61% ~ 84% 10% 4. 1 UPI 355
2015 3 UPI UPI Yoshitake 1995 UPI 2007 UPI 2004 UPI 4. 2. 2 4 2005 UPI 12 2012 0-1 2003 2007 Muthén 1989 8. 93% ~ 35. 26% Muthén UPI 1978 1989 Parry & McArdle 1991 UPI 2010 SCL90 4. 2 4. 2. 1 LPA 61% ~ 84% 72. 9% 9. 4% 91. 64% ~ 97. 21% 2014 61. 21% SCID-P 8. 36% 10 ICD-10 39% CCMD-3 25. 49% 4 13 Z 88. 0 ~ 93. 8% 95. 3% ~ 2. 6SD 97. 0% Tao 2010 72. 41% ~ 99. 73% 17. 7% 60. 64% ~ 99. 73% LPA 39% 356
UPI 5. 3 1 LPA 2 3 5 5. 1 1 UPI 52% 64% 2 9. 86% 19. 15% 70. 99% 61. 21% 38. 28% 8. 36% Antonius J. Van R. J. & Tim M. S. 2010. Online video game addiction identification of addicted adolescent gamers. Addiction 106 1 205-212. Bertrand P. Bénichou J. Grenier P. & Chastang C. 2005. Hui and Walter s latent-class reference-free approach may be more useful in assessing agreement than diagnostic performance. Journal of 3 8. 93% ~ 35. 26% 5. 2 Clinical Epidemiology 58 1 21-26. 1 Benjamin D. K. DeLong E. & Steinbach 2 Psychopathy in Community-Dwelling Italian Adults. 6 689-708. Kay N. Li K. Nokkaew N. & Park 3 UPI UPI Education 12 1 16-32. UPI W. J. 2004 Latent Class Analysis An Illustrative Application for Education in the Assessment of Resident Otoscopic Skills. Ambulatory Pediatrics 4 1 13-17. Fossati A. Krueger R. F Markon K. E. Borroni S. & Maffei C. 2013. Reliability and Validity of the Personality Inventory for DSM-5 PID-5 Predicting DSM-IV Personality Disorders and Assessment 20 B. 2009. Hopelessness and suicidal behavior among Chinese Thai and Korean college students and predictive effects of the World Health Organization s WHOQOL-BREF. International Electronic Journal of Health Maarten V. S. M. Naaktgeboren A. Reitsma J. B. Moons K. 357
2015 3 G. M. & Groot J. A. H. 2013. Latent Class Models in Diagnostic Studies When There is No Reference Standard A Systematic Review. American Journal of Epidemiology 179 4 423-431. Muthén B. 1978 Contributions to factor analysis of dichotomous variables. Psychometrika 43 551-560 Muthén B. 1989 Dichotomous factor analysis of symptom data. Sociological Methods & Research 18 19-65 Muthén B. 1998. Dichotomous factor analysis of symptom data. Sociological Methods & Research 18 1 19-65. Muthén L. K. Muthén BO. 2012. Mplus Statistical analysis with latent variables User s guide 1 1 159-257. McDonald R. P. 2002. Principles and practice in reporting structural equation analysis. Psychological Methods 7 2 64-82. Parry C. D. H. & McArdle J. J. 1991 An applied comparison of methods for least- squares factor analysis of dichotomous variables.. 12 1 58-62. Applied Psychological Measurement 15 35-46. DOI. 1995. UPI SCL90 9 10. 1177 /014662169101500105 Ran T. Xiuqin H. Jinan W. Huimin Z. Ying Z. & Mengchen L. 3 117.. 1999. 21 2010. Proposed diagnostic Criteria for internet addiction. - Addiction 105 556-564. Szatmari P. Volkmar F. & Walter S. 1995. Evaluation of 570-580.. 2013. diagnostic criteria forautism using latent class models. Journal of the American Academyof Child and Adolescent Psychiatry 34 2 216-222. Yoshitake M. 1995. UPI A study on mental health of freshmen in. 4 1 650-653.. 2008. SCL90 EPQ UPI 3. 3 1 249-255... 2007.. Toyo Women s College by UPI. 27 33-1 1 158-159. 42. Yoshitake M. 1996. UPI A study on validity of UPI. 28 87-103.. 2005.. 16 1 91-96.. 2012.. 2013... 29 1 427-420. 7 1 625-629.. 2014.. 7 7 1093-1095. 22 3 96-101. 32 1 37-39.. 2007.. 42 6 610-614.. 2013.... 2004.. 12 68-171.. 2003. UPI SCL90 17 1 41-43.. 2008.. 25 3 233-236.. 2009. SNPs... 2008.... 2014.. 2007. 1087 UP1.. 2009.. 2012... 18 12 1991-1998. 358
Identifying Psychological or Behavioral Problems of College Students Based on Latent Profile Analysis SU Binyuan 1 2 ZHANG Jieting 2 3 YU Chengfu 2 ZHANG Wei 2 1. Psychological Counseling & Research Center South China Normal University Guangzhou 510631 2. School of Psychology /Center for Studies of Psychological Application South China Normal University Guangzhou 510631 3. Shenzhen University Shenzhen 518060 Abstract To explore the applicability of latent profile analysis LPA in detecting psychological or behavioral problems a total of 12718 college students were tested for psychological health. The psychological status of the 644 students was evaluated by psychologists counselors and class supervisors. Using evaluation results and the 90 Symptom checklist SCL90 positive detection rate as the golden standard for diagnostic accuracy sensitivity and specificity were compared between LPA and the traditional demarcation method. The results showed that 1 Student s psychological and behavioral problems can be divided into three sub-groups high risk group 9. 86% mental confusion group 19. 15% and healthy group 70. 99%. 2 High risk groups were characterized by prominent mental symptoms Z 2. 6SD. The positive symptom of mental health risk in high risk group is 61. 21% which is far above that of mental confusion group 38. 28% and mental health group 8. 36%. 3 LPA improved sensitivity by 8. 93% - 35. 26% and showed better diagnostic accuracy comparing with the traditional demarcation method. Key words latent profile analysis LPA university personality inventory UPI psychological or behavioral problems college students 359