Table 1
Participants and Information Included in Add Health Waves I, III, and IV.
| Students | Parents | Information Used in Current Study | ||
|---|---|---|---|---|
| Wave | Core Sample | Oversamples | ||
| Wave I In-Home Interview 1994–95 | Random selection from stratum of US students in grades 7–12 (N = 12,105) | Purposeful selection for oversamples plus saturation samples of 16 schools (N = 8,640) | Parents of participating students (N = 17,670) | * Demographic variables * Health report * Items for measurement of perceived belonging * Grades * Report of symptoms |
| Total in-home interview sample of grade 7–12 students at Wave I (N = 20,745) | ||||
| Wave III In-Home Interview 2001–02 | All Wave I respondents (now aged 18–26 years) who could be located and re-interviewed (N = 15,197) | * Information about on-time high school completion | ||
| Wave IV In-Home Interview 2007–08 | All Wave I respondents (now aged 24–32 years) who could be located and re-interviewed (N = 15,701) | * Information related to diagnosis of chronic medical conditions in childhood * Student level weights | ||
Table 2
Characteristics of Participants in the Study.
| Variable | Range | WAVE I Full Data Set (N = 20,180) | Subsamples | Wave IV Full Data Set (N = 10,813) | ||
|---|---|---|---|---|---|---|
| Design-Based M (se) | Unweighted M (SD) | Healthy (N = 897) M (SD) | Chronic Ill (N = 629) M (SD) | Design-Based M (se) | ||
| Age | 12–19 | 15.85 (.12) | 16.11 (1.70) | 16.07 (1.73) | 16.0 (1.69) | 15.82 (.12) |
| Gender | %female | 50.5% | 50.5% | 47% | 63% | 52% |
| Race | %minority | 27.9% | 42.3% | 42% | 32% | 27% |
| Parent Educ | 1–5 | 2.67 (.05) | 2.68 (1.14) | 2.71 (1.14) | 2.59 (1.07) | 2.76 (.05) |
| Grade Avg | 1–4 | 2.85 (.02) | 2.75 (.77) | 2.79 (.77) | 2.72 (.76) | 2.90 (.02) |
| Belong Scale | 1–5 | 3.71(.02) | 3.68 (.75) | 3.74 (.71) | 3.59 (.82) | 3.74 (.02) |
| Health Report | 1–5 | 4.01 (.01) | 3.97 (.76) | 4.06 (.73) | 3.74 (.82) | 4.03 (.02) |
| Symptoms Score | 0–31 | 0.78 (.01) | 0.76 (.45) | 7.70 (4.50) | 10.60 (5.78) | 0.77 (.01) |
| Complete HS | %Yes | 92.7% | 92.3% | 94.4% | 90.1% | 93.7% |
[i] Note: Gender: 0 = male, 1 = female; race: 0 = white, 1 = minority; HS completion: 0 = No, 1 = Yes.
Table 3
Final Regression Model for Perceived School Belonging Using Design-Based Wave I Data.
| Stand. Coefficient | Linearized se | p | |
|---|---|---|---|
| Age | –.029 | .008 | <.001 |
| Gender | .001 | .020 | .947 |
| Race | –.026 | .025 | .307 |
| Parent Education | –.009 | .009 | .344 |
| Grade Average | .197 | .014 | <.001 |
| Initial Model R2 | .067 | ||
| Wave I Health Report | .120 | .015 | <.001 |
| Symptoms Mean Score | –.377 | .027 | <.001 |
| Model R2△ | .072 | ||
| Total Model R2 | .139 |
[i] Note: Gender coded male = 0, female = 1; race coded white = 0, minority =1; bold = variance explained by control variables, additional variance explained by health variables and total variance explained.
Table 4
Model Comparisons for Healthy and Chronically Ill Student Groups.
| Model | X2 | df | p | X2 difference** | df difference | Significance (difference) | RMSEA | CFI | SRMR |
|---|---|---|---|---|---|---|---|---|---|
| 1. Same Form | 301.71 | 98 | <.001 | – | – | – | .052 | .974 | .028 |
| 2. Same Loadings | 302.07 | 106 | <.001 | .36(1v2) | 8 | ns | .049 | .975 | .032 |
| 3. Same Loadings and Err Variances | 407.17 | 133 | <.001 | 105.10(2v3) | 27 | Sig. | .052 | .965 | .051 |
| 4. Same Loadings and Intercepts | 341.29 | 119 | <.001 | 39.22(2v4) | 13 | Sig. | .049 | .972 | .039 |
[i] Note: Target parameters for good model fit are RMSEA < .05, CFI > .95, and SRMR < .05; RMSEA = root mean square error of approximation, CFI = comparative fit index, SRMR = standardized root mean-squared residual. The goal for X2 is non-significance; however, large sample size may lead to a significant X2 result.
** Model elements are constrained to compare between the groups, 1 v. 2 = comparison between fully unconstrained model and model with constrained factor loadings, 2 v. 3 = comparison between constrained loadings only and constrained loadings plus error variances, 2. v. 4 = comparison between constrained loadings only and constrained loadings and intercepts.
Table 5
Group Goodness-of-Fit Parameters (SRMR) for Comparison Models.
| Observations | 1. Same Form | 2. Same Loadings | 3. Same Loadings and Error Variances | 4. Same Loadings and Intercepts | |
|---|---|---|---|---|---|
| Healthy | 924 | .024 | .027 | .046 | .034 |
| Chronically Ill | 627 | .032 | .037 | .055 | .044 |
[i] Note: SRMR = standardized root mean-squared residual, target < .05.
Table 6
Design-Based Coefficients and Odds Ratios for Final Logistic Regression Model to Predict High School Completion by the Age of 19 Years (N = 10,253).
| Variable | b (log odds) (Linearized se) | p | OR (Linearized se) | 95% CI for OR |
|---|---|---|---|---|
| Age | –.3252 (.034) | <.001 | .7223 (.024) | .6759, .7719 |
| Gender | .3185 (.123) | .011 | 1.3751 (.169) | 1.0786, 1.7530 |
| Race | –.4326 (.146) | .004 | .6488 (.095) | .4861, .8659 |
| Parent Education | .5008 (.071) | <.001 | 1.6500 (.117) | 1.4342, 1.8983 |
| Grade Average | .9250 (.076) | <.001 | 2.5219 (.192) | 2.1697, 2.9313 |
| W1 Health Report | .1703 (.078) | .031 | 1.1857 (.093) | 1.0158, 1.3840 |
| Diagnosis Status | –.5055 (.751) | .502 | .6032 (.453) | 1.1365, 2.6651 |
| School Belonging | –.3373 (.130) | .010 | .7137 (.092) | .5523, .9222 |
| Diag X Belonging | .4882 (.158) | .003 | 1.6293 (.258) | 1.1911, 2.2287 |
[i] Note: High school completion coded no = 0, yes = 1; Gender coded 0 = male, 1 = female; Race coded 0 = white, 1 = minority; diagnosis coded 0 = no, 1 = yes