Cross-lagged dynamic panel models examining intensity of IPV perpetration (n=2479)
| Model 1 | Model 2* | |||||
| Coef | SE | P value | Coef | SE | P value | |
| Time variant variables | ||||||
| Food insecurity | 0.07 | 0.04 | 0.045 | 0.09 | 0.04 | 0.031 |
| Housing status | – | – | – | 0.12 | 0.05 | 0.009 |
| Relationship status | – | – | – | 0.12 | 0.04 | 0.153 |
| Time invariant variables | ||||||
| Age at baseline | – | – | – | −0.10 | 0.03 | <0.001 |
| Childhood abuse | – | – | – | 0.21 | 0.03 | <0.001 |
| Alpha | 0.77 | 0.05 | <0.001 | 0.59 | 0.06 | <0.001 |
| Fit indices | ||||||
| Chi2 | 6.56 | 7.64 | ||||
| Chi2 p value | 0.010 | 0.106 | ||||
| Df | 1 | Lower bound | Upper bound | 4 | Lower bound | Upper bound |
| RMSEA | 0.044 | 0.015 | 0.082 | 0.016 | 0.000 | 0.035 |
| CFI | 0.984 | 0.994 | ||||
Models account for past use of violence and bidirectional nature of association (ie, IPV perpetration leading to later food insecurity).
CFI, comparative fit index; Coef, standardised coefficient; IPV, intimate partner violence; RMSEA, root mean squared error of approximation.