Research in Developmental Disabilities 43–44 (2015) 51–60
Contents lists available at ScienceDirect
Research in Developmental Disabilities
Validity and reliability of the Medical Outcomes Study Short-Form Health Survey version 2 (SF-12v2) among adults with autism Rahul Khanna a,*, Krutika Jariwala b, Donna West-Strum c a Faser Hall 236, Department of Pharmacy Administration, School of Pharmacy, The University of Mississippi, University, MS 38677, United States b Faser Hall 211, Department of Pharmacy Administration, School of Pharmacy, The University of Mississippi, University, MS 38677, United States c Faser Hall 223, PO Box 1848, Department of Pharmacy Administration, School of Pharmacy, The University of Mississippi, University, MS 38677, United States
A R T I C L E I N F O
A B S T R A C T
Article history: Received 31 October 2014 Received in revised form 8 June 2015 Accepted 15 June 2015 Available online 4 July 2015
Background: The purpose of the study was to assess the validity and reliability of the Medical Outcomes Study Short Form-12 version 2 (SF-12v2) instrument among adults with autism. Methods: Study data was collected using a cross-sectional online survey of adults with autism enrolled with the Interactive Autism Network (N = 291). Factorial validity was assessed using confirmatory factor analysis technique. Item–scale correlations were examined for convergent validity. Known-groups validity was assessed by examining the variation in Physical Component Summary (PCS) and Mental Component Summary (MCS) scores by autism severity. Cronbach’s alpha was determined for internal consistency reliability. Floor and ceiling effects were also assessed. Results: A two-factor model with correlated error terms was found to have a good fit. The PCS scale strongly correlated with the underlying items representing the scale. The MCS scale had strong to moderate correlation with its underlying items. For known-groups validity, the MCS score varied as expected with lower score observed among adults with high severity as compared to low severity; however, PCS score varied inversely. Internal consistent reliability of the SF-12v2 was good, and there were no floor and ceiling effects. Conclusions: Except for known-groups validity, all other psychometric indicators performed well for the SF-12v2. ß 2015 Elsevier Ltd. All rights reserved.
Keywords: HRQOL SF-12v2 Autism Adults
* Corresponding author. Tel.: +1 6629151651; fax: +1 6629155102. E-mail address: [email protected] (R. Khanna). http://dx.doi.org/10.1016/j.ridd.2015.06.006 0891-4222/ß 2015 Elsevier Ltd. All rights reserved.
52
R. Khanna et al. / Research in Developmental Disabilities 43–44 (2015) 51–60
What this paper adds?: Recent studies have indicated that adults with autism have lower health-related quality of life (HRQOL), which is one of the most commonly used health outcomes metric, as compared to their peers in the general population. HRQOL can be assessed using either disease-specific or generic instruments. Given the lack of autism-specific HRQOL instruments, researchers typically rely on generic instruments such as the Medical Outcomes Study Short-Form-12 version 2 (SF-12v2) for HRQOL assessment in this population. Though the SF-12v2 is a well-established scale that has been used across several disease areas, it is not known as to how this instrument performs among adults with autism. To date, no study has assessed the psychometrics (validity and reliability) of this instrument among adults with autism. Collecting data from an online survey of adults with autism, this study assessed the factorial, convergent, and known-groups validity of the SF-12v2. Further, the internal consistency reliability and floor and ceiling effects were also studied. Study results showed the SF-12v2 to have good factorial and convergent validity; however, results for known-groups validity were inconsistent. The SF-12v2 had good reliability and no floor and ceiling effects. This study showed the SF-12v2 to perform generally well among adults with autism. Given these results, researchers can gain confidence in their use of the SF-12v2 for HRQOL assessment among adults with autism.
1. Introduction Autism is a neurological and developmental disorder characterized by socio-communicative limitations and restricted and repetitive behaviors (American Psychiatric Association, 2000). Recent research has indicated the prevalence of autism to be 1.5% among children aged eight years across eleven community sites in the United States (US), with prevalence among males to be 4.5 times as those in females (Wingate et al., 2014). Though epidemiological data on the prevalence of autism among adults in the US is limited, research conducted in Britain has found it to be 1% (Brugha et al., 2011). Over the past few years, there has been an increasing interest in studying health outcomes among adults with autism, especially their quality of life (QOL)/health-related quality of life (HRQOL) (Jennes-Coussens, Magill-Evans, & Koning, 2006; Kamio, Naoko, & Tomonori Koyama, 2013; Kamp-Becker, Schro¨der, Remschmidt, & Bachmann, 2010; Khanna, Jariwala, West-Strum, & Mahabaleshwarkar, 2014; Renty & Roeyers, 2006). In the only such study among a US sample of adults with autism, Khanna et al. (2014) found lower physical and psychological HRQOL among adults with autism as compared to their peers in the general US adult population. Similar results have been reported elsewhere, with the general trend toward lower HRQOL among adults with autism (Jennes-Coussens et al., 2006; Kamio et al., 2013; Kamp-Becker et al., 2010). These results are not surprising considering that motor impairments and comorbid mental illnesses are common in this population, besides the adverse influence placed by the disorder itself (Kamio et al., 2013; Khanna et al., 2014; Matson, Matson, & Beighley, 2011). HRQOL is a multi-dimensional construct that pertains to psychosocial and physical health of an individual (Calvert & Freemantle, 2003). While QOL is a broad construct that encompasses both health and non-health (environmental, political, geographical) related dimensions, HRQOL focuses exclusively on health status assessment. Research across different clinical settings has indicated HRQOL to be a predictor of mortality and healthcare use (Centers for Disease Control and Prevention [CDC], 2014; Dorr et al., 2006; Mapes et al., 2003; Singh, Nelson, Fink, & Nichol, 2005; Tsai, Chi, Lee, & Chou, 2007). Information concerning HRQOL can be measured using both generic and disease-specific instruments. Generic HRQOL instruments (e.g., Medical Outcomes Study (MOS) 36-Item Short Form (SF-36) (Ware & Sherbourne, 1992), MOS 12-Item Short-Form version 2 (SF-12v2) (Ware, Kosinski, & Keller, 1996), EuroQol Five-Dimension Questionnaire (EQ-5D) (The EuroQol Group, 1990)), as the name suggests, can be used across different diseases, populations, and clinical settings and provide an overall profile of an individual’s physical and psychosocial health. However, the generic measurement approach often makes these instruments unsuitable for assessing disease-specific symptoms and treatment responsiveness. Diseasespecific instruments (e.g., Quality of Life in Epilepsy (QOLIE-89) (Devinsky et al., 1995), Pediatric Asthma Quality of Life Questionnaire (PAQLQ) (Juniper, Guyatt, Feeny, Ferrie, Griffith, & Townsend, 1996)) overcome these limitations by being specific to a particular disease, population, or function. But their strength also leads to a weakness that these instruments cannot be used to make comparisons across different populations. The use of generic or disease-specific instrument depends on several criteria including research objectives, patient population, and resource availability. Considering the lack of a disease-specific instrument, studies in autism have typically used a generic measurement approach to assess QOL/HRQOL among individuals with autism. Generic instruments including the World Health Organization Quality of Life-BREF (WHOQOL-BREF) (Jennes-Coussens et al., 2006; Kamio et al., 2013; Kamp-Becker et al., 2010) and SF-12 (Khanna et al., 2014) have been used to that extent. The WHOQOL-BREF is a 26-item QOL instrument that includes assessment of physical, psychological, social, and environmental domains (Skevington, Lotfy, & O’Connell, 2004). The SF-12v2 is an abbreviated version of the parent instrument SF-36, which is the most commonly used generic HRQOL instrument (Johnson & Coons, 1998). Unlike QOL, HRQOL is amenable to healthcare interventions, thereby making it a valuable outcome among policy makers and healthcare providers for patient health assessment in autism.
R. Khanna et al. / Research in Developmental Disabilities 43–44 (2015) 51–60
53
The purpose of this study was to determine the validity and reliability of the SF-12v2 among adults with autism. Specifically, factorial validity, convergent validity, known-groups validity, internal consistency reliability, and floor and ceiling effects of the SF-12v2 were tested among a US sample of adults with autism. With the absence of autism-specific HRQOL instrument, researchers have generally relied on the use of generic instruments to assess HRQOL among adults with autism (Jennes-Coussens et al., 2006; Kamio et al., 2013; Kamp-Becker et al., 2010; Khanna et al., 2014; Renty & Roeyers, 2006). If determined to be psychometrically sound, together with its ease of administration and low respondent burden, the SF-12v2 may present a good alternative to measuring HRQOL among adults with autism until disease-specific instruments are developed. 2. Methods 2.1. Study design and sample Using a cross-sectional online survey design, data was collected from adults with autism who were registered with the Interactive Autism Network (IAN). The IAN is an online registry run by the Kennedy Krieger Institute, and includes both individuals with autism and family members as participants. The IAN database has been clinically validated (Lee et al., 2010), and also verified by a review of medical records provided by parents and professionals (Daniels et al., 2012). Among adults (18 years of age and above) with autism registered with the IAN, those with the capacity to self-report with minimal help were approached for participation through email. To protect the identity of registry participants, the IAN performed sample identification based on study criteria and emailed the cover letter with survey link. Voluntary participation was emphasized in the cover letter email. Qualtrics online software system (Qualtrics Inc., Provo, UT) was used to administer the survey. A total of 297 survey responses were received. We deleted responses with 15% or more of missing data. Of the 297 responses, 6 fulfilled the criteria and were therefore deleted. The final sample constituted 291 responses. A $15 Amazon gift-card was provided to survey participants. This study was conducted as part of a larger grant funded by the Organization for Autism Research (OAR). The University of Mississippi Institutional Review Board approved the study under exempt status. 2.2. Measures 2.2.1. SF-12v2 The SF-12v2 is a 12-item instrument, which provides information on eight HRQOL subdomains including general health (GH, one item), physical functioning (PF, two items), role physical (RP, two items), bodily pain (BP, one item), vitality (VT, one item), social functioning (SF, one item), role emotional (RE, two items), and mental health (MH, two items). These subdomains can be collapsed into summary physical and mental health scales (Ware et al., 1996). General US population norm-based scores (mean 50, standard deviation [SD] 10) for summary physical component scale (PCS) and mental component scale (MCS) were calculated using the QualityMetric SF Health Outcomes Scoring Software. Higher scores reflect better HRQOL. The SF-12v2 has been found to be valid and reliable across different patient populations (Jakobsson, Westergren, Lindskov, & Hagell, 2012; Maurischat, Ehlebracht-Ko¨nig, Ku¨hn, & Bullinger, 2006; Maurischat, Herschbach, Peters, & Bullinger, 2008; Okonkwo, Roth, Pulley, & Howard, 2010; Resnick & Nahm, 2001) including severe mental illness (Salyers, Bosworth, Swanson, Lamb-Pagone, & Osher, 2000). 2.2.2. Adult Short Autism-Spectrum Quotient-10 Items The Adult Short Autism-Spectrum Quotient-10 Items (AQ-10) was used to assess autism severity among adults with autism. The AQ-10 assesses information concerning attention to detail, communication, difficulty in switching attention, imagination limitations, and social skill limitations (Allison, Auyeung, & Baron-Cohen, 2012). The 10 items are measured on a four-point Likert-type scale with endpoints ‘‘definitely agree’’ and ‘‘definitely disagree’’. A total score is calculated by summing the score on the individual items. 2.3. Statistical analysis Item-level descriptive statistics including total responses, missing data, mean (standard deviation (SD)), and normality (skewness and kurtosis) were determined. A range between 1.25 and 2.0 for skew values and between 1.0 and 8.0 for kurtosis are indicative of univariate normality (Harlow, 1985). Case mean substitution method was used for missing data replacement. Confirmatory factor analysis (CFA) technique was used to determine the factorial validity of the SF-12v2. Two different iterations of the SF-12v2 two-factor structure were tested in the study. These iterations were based on the original factor structure of the SF-12v2 and prior studies examining the factorial validity of the SF-12v2 (Maurischat et al., 2006, 2008; McBride, Adamson, Bunting, & McCann, 2009; Ware et al., 1996; Wilson, Tucker & Chittleborough, 2002). In both iterations, the PCS and MCS latent factors were intercorrelated. In model one, as with the original proposed model of the SF-12v2 (Ware et al., 1996), items representing the BP, GH, PF, and RP subdomains were loaded on the latent factor PCS, while the MH, RE, SF, and VT subdomains items were loaded on the latent factor MCS. Model two was a modified version of model one with
54
R. Khanna et al. / Research in Developmental Disabilities 43–44 (2015) 51–60
correlated error terms for items representing subdomains PF, RP, MH, and RE, respectively. Fit indices including chi-square statistic, Comparative Fit Index (CFI), Goodness of Fit Index (GFI), Adjusted Goodness of Fit Index (AGFI), Root Mean Square Error of Approximation (RMSEA), and Akaike Information Criterion (AIC) were used to determine fit of the two models. The lower the value of chi-square statistic, the better the model fit. CFI and GFI score >0.90 and AGFI score >0.80 indicate a good fit (Clara, Cox, & Enns, 2001; Hu & Peter, 1999). RMSEA value
Contents lists available at ScienceDirect
Research in Developmental Disabilities
Validity and reliability of the Medical Outcomes Study Short-Form Health Survey version 2 (SF-12v2) among adults with autism Rahul Khanna a,*, Krutika Jariwala b, Donna West-Strum c a Faser Hall 236, Department of Pharmacy Administration, School of Pharmacy, The University of Mississippi, University, MS 38677, United States b Faser Hall 211, Department of Pharmacy Administration, School of Pharmacy, The University of Mississippi, University, MS 38677, United States c Faser Hall 223, PO Box 1848, Department of Pharmacy Administration, School of Pharmacy, The University of Mississippi, University, MS 38677, United States
A R T I C L E I N F O
A B S T R A C T
Article history: Received 31 October 2014 Received in revised form 8 June 2015 Accepted 15 June 2015 Available online 4 July 2015
Background: The purpose of the study was to assess the validity and reliability of the Medical Outcomes Study Short Form-12 version 2 (SF-12v2) instrument among adults with autism. Methods: Study data was collected using a cross-sectional online survey of adults with autism enrolled with the Interactive Autism Network (N = 291). Factorial validity was assessed using confirmatory factor analysis technique. Item–scale correlations were examined for convergent validity. Known-groups validity was assessed by examining the variation in Physical Component Summary (PCS) and Mental Component Summary (MCS) scores by autism severity. Cronbach’s alpha was determined for internal consistency reliability. Floor and ceiling effects were also assessed. Results: A two-factor model with correlated error terms was found to have a good fit. The PCS scale strongly correlated with the underlying items representing the scale. The MCS scale had strong to moderate correlation with its underlying items. For known-groups validity, the MCS score varied as expected with lower score observed among adults with high severity as compared to low severity; however, PCS score varied inversely. Internal consistent reliability of the SF-12v2 was good, and there were no floor and ceiling effects. Conclusions: Except for known-groups validity, all other psychometric indicators performed well for the SF-12v2. ß 2015 Elsevier Ltd. All rights reserved.
Keywords: HRQOL SF-12v2 Autism Adults
* Corresponding author. Tel.: +1 6629151651; fax: +1 6629155102. E-mail address: [email protected] (R. Khanna). http://dx.doi.org/10.1016/j.ridd.2015.06.006 0891-4222/ß 2015 Elsevier Ltd. All rights reserved.
52
R. Khanna et al. / Research in Developmental Disabilities 43–44 (2015) 51–60
What this paper adds?: Recent studies have indicated that adults with autism have lower health-related quality of life (HRQOL), which is one of the most commonly used health outcomes metric, as compared to their peers in the general population. HRQOL can be assessed using either disease-specific or generic instruments. Given the lack of autism-specific HRQOL instruments, researchers typically rely on generic instruments such as the Medical Outcomes Study Short-Form-12 version 2 (SF-12v2) for HRQOL assessment in this population. Though the SF-12v2 is a well-established scale that has been used across several disease areas, it is not known as to how this instrument performs among adults with autism. To date, no study has assessed the psychometrics (validity and reliability) of this instrument among adults with autism. Collecting data from an online survey of adults with autism, this study assessed the factorial, convergent, and known-groups validity of the SF-12v2. Further, the internal consistency reliability and floor and ceiling effects were also studied. Study results showed the SF-12v2 to have good factorial and convergent validity; however, results for known-groups validity were inconsistent. The SF-12v2 had good reliability and no floor and ceiling effects. This study showed the SF-12v2 to perform generally well among adults with autism. Given these results, researchers can gain confidence in their use of the SF-12v2 for HRQOL assessment among adults with autism.
1. Introduction Autism is a neurological and developmental disorder characterized by socio-communicative limitations and restricted and repetitive behaviors (American Psychiatric Association, 2000). Recent research has indicated the prevalence of autism to be 1.5% among children aged eight years across eleven community sites in the United States (US), with prevalence among males to be 4.5 times as those in females (Wingate et al., 2014). Though epidemiological data on the prevalence of autism among adults in the US is limited, research conducted in Britain has found it to be 1% (Brugha et al., 2011). Over the past few years, there has been an increasing interest in studying health outcomes among adults with autism, especially their quality of life (QOL)/health-related quality of life (HRQOL) (Jennes-Coussens, Magill-Evans, & Koning, 2006; Kamio, Naoko, & Tomonori Koyama, 2013; Kamp-Becker, Schro¨der, Remschmidt, & Bachmann, 2010; Khanna, Jariwala, West-Strum, & Mahabaleshwarkar, 2014; Renty & Roeyers, 2006). In the only such study among a US sample of adults with autism, Khanna et al. (2014) found lower physical and psychological HRQOL among adults with autism as compared to their peers in the general US adult population. Similar results have been reported elsewhere, with the general trend toward lower HRQOL among adults with autism (Jennes-Coussens et al., 2006; Kamio et al., 2013; Kamp-Becker et al., 2010). These results are not surprising considering that motor impairments and comorbid mental illnesses are common in this population, besides the adverse influence placed by the disorder itself (Kamio et al., 2013; Khanna et al., 2014; Matson, Matson, & Beighley, 2011). HRQOL is a multi-dimensional construct that pertains to psychosocial and physical health of an individual (Calvert & Freemantle, 2003). While QOL is a broad construct that encompasses both health and non-health (environmental, political, geographical) related dimensions, HRQOL focuses exclusively on health status assessment. Research across different clinical settings has indicated HRQOL to be a predictor of mortality and healthcare use (Centers for Disease Control and Prevention [CDC], 2014; Dorr et al., 2006; Mapes et al., 2003; Singh, Nelson, Fink, & Nichol, 2005; Tsai, Chi, Lee, & Chou, 2007). Information concerning HRQOL can be measured using both generic and disease-specific instruments. Generic HRQOL instruments (e.g., Medical Outcomes Study (MOS) 36-Item Short Form (SF-36) (Ware & Sherbourne, 1992), MOS 12-Item Short-Form version 2 (SF-12v2) (Ware, Kosinski, & Keller, 1996), EuroQol Five-Dimension Questionnaire (EQ-5D) (The EuroQol Group, 1990)), as the name suggests, can be used across different diseases, populations, and clinical settings and provide an overall profile of an individual’s physical and psychosocial health. However, the generic measurement approach often makes these instruments unsuitable for assessing disease-specific symptoms and treatment responsiveness. Diseasespecific instruments (e.g., Quality of Life in Epilepsy (QOLIE-89) (Devinsky et al., 1995), Pediatric Asthma Quality of Life Questionnaire (PAQLQ) (Juniper, Guyatt, Feeny, Ferrie, Griffith, & Townsend, 1996)) overcome these limitations by being specific to a particular disease, population, or function. But their strength also leads to a weakness that these instruments cannot be used to make comparisons across different populations. The use of generic or disease-specific instrument depends on several criteria including research objectives, patient population, and resource availability. Considering the lack of a disease-specific instrument, studies in autism have typically used a generic measurement approach to assess QOL/HRQOL among individuals with autism. Generic instruments including the World Health Organization Quality of Life-BREF (WHOQOL-BREF) (Jennes-Coussens et al., 2006; Kamio et al., 2013; Kamp-Becker et al., 2010) and SF-12 (Khanna et al., 2014) have been used to that extent. The WHOQOL-BREF is a 26-item QOL instrument that includes assessment of physical, psychological, social, and environmental domains (Skevington, Lotfy, & O’Connell, 2004). The SF-12v2 is an abbreviated version of the parent instrument SF-36, which is the most commonly used generic HRQOL instrument (Johnson & Coons, 1998). Unlike QOL, HRQOL is amenable to healthcare interventions, thereby making it a valuable outcome among policy makers and healthcare providers for patient health assessment in autism.
R. Khanna et al. / Research in Developmental Disabilities 43–44 (2015) 51–60
53
The purpose of this study was to determine the validity and reliability of the SF-12v2 among adults with autism. Specifically, factorial validity, convergent validity, known-groups validity, internal consistency reliability, and floor and ceiling effects of the SF-12v2 were tested among a US sample of adults with autism. With the absence of autism-specific HRQOL instrument, researchers have generally relied on the use of generic instruments to assess HRQOL among adults with autism (Jennes-Coussens et al., 2006; Kamio et al., 2013; Kamp-Becker et al., 2010; Khanna et al., 2014; Renty & Roeyers, 2006). If determined to be psychometrically sound, together with its ease of administration and low respondent burden, the SF-12v2 may present a good alternative to measuring HRQOL among adults with autism until disease-specific instruments are developed. 2. Methods 2.1. Study design and sample Using a cross-sectional online survey design, data was collected from adults with autism who were registered with the Interactive Autism Network (IAN). The IAN is an online registry run by the Kennedy Krieger Institute, and includes both individuals with autism and family members as participants. The IAN database has been clinically validated (Lee et al., 2010), and also verified by a review of medical records provided by parents and professionals (Daniels et al., 2012). Among adults (18 years of age and above) with autism registered with the IAN, those with the capacity to self-report with minimal help were approached for participation through email. To protect the identity of registry participants, the IAN performed sample identification based on study criteria and emailed the cover letter with survey link. Voluntary participation was emphasized in the cover letter email. Qualtrics online software system (Qualtrics Inc., Provo, UT) was used to administer the survey. A total of 297 survey responses were received. We deleted responses with 15% or more of missing data. Of the 297 responses, 6 fulfilled the criteria and were therefore deleted. The final sample constituted 291 responses. A $15 Amazon gift-card was provided to survey participants. This study was conducted as part of a larger grant funded by the Organization for Autism Research (OAR). The University of Mississippi Institutional Review Board approved the study under exempt status. 2.2. Measures 2.2.1. SF-12v2 The SF-12v2 is a 12-item instrument, which provides information on eight HRQOL subdomains including general health (GH, one item), physical functioning (PF, two items), role physical (RP, two items), bodily pain (BP, one item), vitality (VT, one item), social functioning (SF, one item), role emotional (RE, two items), and mental health (MH, two items). These subdomains can be collapsed into summary physical and mental health scales (Ware et al., 1996). General US population norm-based scores (mean 50, standard deviation [SD] 10) for summary physical component scale (PCS) and mental component scale (MCS) were calculated using the QualityMetric SF Health Outcomes Scoring Software. Higher scores reflect better HRQOL. The SF-12v2 has been found to be valid and reliable across different patient populations (Jakobsson, Westergren, Lindskov, & Hagell, 2012; Maurischat, Ehlebracht-Ko¨nig, Ku¨hn, & Bullinger, 2006; Maurischat, Herschbach, Peters, & Bullinger, 2008; Okonkwo, Roth, Pulley, & Howard, 2010; Resnick & Nahm, 2001) including severe mental illness (Salyers, Bosworth, Swanson, Lamb-Pagone, & Osher, 2000). 2.2.2. Adult Short Autism-Spectrum Quotient-10 Items The Adult Short Autism-Spectrum Quotient-10 Items (AQ-10) was used to assess autism severity among adults with autism. The AQ-10 assesses information concerning attention to detail, communication, difficulty in switching attention, imagination limitations, and social skill limitations (Allison, Auyeung, & Baron-Cohen, 2012). The 10 items are measured on a four-point Likert-type scale with endpoints ‘‘definitely agree’’ and ‘‘definitely disagree’’. A total score is calculated by summing the score on the individual items. 2.3. Statistical analysis Item-level descriptive statistics including total responses, missing data, mean (standard deviation (SD)), and normality (skewness and kurtosis) were determined. A range between 1.25 and 2.0 for skew values and between 1.0 and 8.0 for kurtosis are indicative of univariate normality (Harlow, 1985). Case mean substitution method was used for missing data replacement. Confirmatory factor analysis (CFA) technique was used to determine the factorial validity of the SF-12v2. Two different iterations of the SF-12v2 two-factor structure were tested in the study. These iterations were based on the original factor structure of the SF-12v2 and prior studies examining the factorial validity of the SF-12v2 (Maurischat et al., 2006, 2008; McBride, Adamson, Bunting, & McCann, 2009; Ware et al., 1996; Wilson, Tucker & Chittleborough, 2002). In both iterations, the PCS and MCS latent factors were intercorrelated. In model one, as with the original proposed model of the SF-12v2 (Ware et al., 1996), items representing the BP, GH, PF, and RP subdomains were loaded on the latent factor PCS, while the MH, RE, SF, and VT subdomains items were loaded on the latent factor MCS. Model two was a modified version of model one with
54
R. Khanna et al. / Research in Developmental Disabilities 43–44 (2015) 51–60
correlated error terms for items representing subdomains PF, RP, MH, and RE, respectively. Fit indices including chi-square statistic, Comparative Fit Index (CFI), Goodness of Fit Index (GFI), Adjusted Goodness of Fit Index (AGFI), Root Mean Square Error of Approximation (RMSEA), and Akaike Information Criterion (AIC) were used to determine fit of the two models. The lower the value of chi-square statistic, the better the model fit. CFI and GFI score >0.90 and AGFI score >0.80 indicate a good fit (Clara, Cox, & Enns, 2001; Hu & Peter, 1999). RMSEA value
