Psychometric Properties of the Parent and School Survey

Background:

The federal government and various studies have ideologically promoted and verified the effectiveness of parental involvement in children education. Despite of the importance of parental involvement, defining it remains a challenge. If parental involvement is a meaningful predictor, or even cause, of better academic achievement, it is critical to find an agreement as to what parental involvement means and how it should be measured.

The Parent and School Survey (PASS) is an instrument designed to quickly, easily, and accurately measure parental involvement in their children’s education (Ringenberg, et al., 2005). The 24-item PASS with five Likert-scale responses were developed to reflect Epstein’s six-construct parent involvement framework. The PASS items have been tested for test-retest reliability, sufficient variance, and accurate qualitative interpretation by subjects. Since the refined version in 2005, the PASS has not been tested with diverse sample and its validation studies have not been found yet.

Purpose:

The study aimed to examine the psychometric properties of the PASS and to test the construct validity with six-construct model.

Methods:

Using the self-administered survey, the study consisted of 317 parents who had a child who attending an elementary school in the state of Mississippi. The majority of the parents were mothers (88.6%), African Americans (55.8%), with high school graduates (53%), in relatively low income status (<$19,999, 48.6%) as being single not married (48%).  Descriptive statistics and item statistics were used to screen the norm score, internal consistency for subscales, and deviant (irrelevant) items. To test the Epstein’s six-construct model, a series of confirmatory factor analysis (CFA) was used in comparison with one global factor model (parental involvement with 24 items).  

Results:

This study revealed that while the overall scale has acceptable reliability (α= .842), two sub scales such as decision-making & collaborating failed to reach the acceptable reliability. By deleting five deviant items, the Cronbach Alphas for each subscale improved somewhat: parenting (.716), communicating (.708), volunteering (.571), learning at home (.697), decision-making (.441), and collaborating with community (.589).  Both one global factor (PASS) and intercorrelated six factors models failed to reach the acceptable fit statistics. The second-order model with six measured variables and two-factors (Parenting at Home and Involvement with School and Community) was emerged as the best fitted (CFI = .98, SRMR =.052) with significantly loaded variance from .72 to .35. Criterion validity was confirmed with grouping (e.g., race, gender) outcome variables (e.g., academic achievements)

Discussion and Implications:

Despite of acceptable reliabilities, several items failed to reflect the concepts, including parental involvement in decision-making and collaboration with community, which dimensions may not be relevant to this sample. The PASS was highly significantly correlated with children’s behavioral problems and academic achievements.

The findings suggest that a common definition and measurement tool can help to compare across studies and clearer dialogues about parental involvement. Based on the established system framework (parental involvement at home, school, and community), school social workers can use this revised instrument to assess and promote parental role when dealing with troubled children in school and community.