Background: Scales are essential research tools for measuring a concept of interest. However, researchers are unsure about how respondents interpret broad constructs measured by scales, such as mental health. When respondents show differing understanding of the construct, usually specific scale items are attributed as bias. Differences in response to scale items can only be considered bias if two assumptions are made: 1) the construct is universal (mental health) and 2) the constructs' items only measure the construct. Differential Item Functioning (DIF) investigates whether this kind of bias occurs across groups of respondents. DIF studies the differing rates of responses to individual items after controlling for the overall level of the construct. The literature consistently shows that mental health is not a universal construct. Differences in the conceptualization of mental illness have been found to vary by cultural norms, socioeconomic status, gender, and other demographic characteristics. One frequent comparison cited in the literature is level of somaticized versus psychologized mental distress between cultural groups. For example, research has suggested that Chinese immigrants are more likely to somaticize their mental health problems as compared to native-born Americans. When mental health is not viewed as a universal concept but rather as culturally conditioned, DIF can be understood as an indicator of these different cultural constructs of mental health rather than a form of item bias. Different cultural conceptualizations of mental health and illness can then be tested using statistical procedures developed to assess DIF. Methods: Using the 2001California Health Interview Survey's (n = 53, 697) implementation of the SF12 physical and mental health scales, DIF for individuals speaking a Chinese language at home or not were compared. A multiple indicator, multiple cause structural equation (MIMIC) model, which allows for the testing of DIF, was fit to the SF12 mental and physical health components. An additional equation was used to test the direct effect of speaking Chinese on the items asking about reduced activity and work as they relate to mental or physical health. Results: After controlling for the effect of demographic variables on overall levels of physical and mental health, speaking Chinese was associated with an increased likelihood of endorsing working less (p<0.01) and participating in fewer activities (p<0.01) due to emotional problems. Speaking Chinese at home was not associated with participating in fewer activities due to physical difficulties. Finally, Chinese speakers were less likely than non-Chinese speakers to endorse working less due to physical complaints (p=0.04). Conclusions: The results suggest that after controlling for demographic characteristics and overall levels of physical and mental health Chinese speaking individuals were more likely to endorse psychological reasons for their problems of living and less likely to endorse physical explanations. Implications: The results of this study provide evidence for training social workers to look beyond the cultural stereotypes such as somatization when assessing mental illness with Chinese speaking clients. The results also point to new tools for mental health researchers to study disparities in mental illness.

Word Count: 478