Development and Validation of the Survey of Knowledge of Internet Risk and Internet Behavior

Abstract

The development of the Survey of Knowledge of Internet Risk and Internet Behavior is described. A total of 1,366 Grades 7 and 8 male and female students from an urban, suburban, and rural school offered agree—disagree responses to 26 statements defining one Knowledge Scale and five behavior dimensions. Literature-based support is presented for content validity. Construct validity support for the hypothesized dimensions is provided through Rasch model analysis of the Knowledge Scale supporting a unidimensional, hierarchically ordered scale. Latent class analyses provided support for the utility of the five categorical behavioral dimensions. Implications for instrument developers and educators are discussed.

Measuring the Youth Bullying Experience: A Systematic Review of the Psychometric Properties of Available Instruments

“…Three of 31 studies (10%) did not report internal consistency (Table ). When examining total scales and independent subscales, it was noted that few studies reported values for Cronbach's alpha that were lower than the desirable level (α > 0.70) . Lam and Li reported a Cronbach's alpha of 0.55 for the mild subscale of the E‐Bullying Scale in their female study cohort, with alpha values of greater than 0.70 for all other groups and subscales.…”

“…Lam and Li reported a Cronbach's alpha of 0.55 for the mild subscale of the E‐Bullying Scale in their female study cohort, with alpha values of greater than 0.70 for all other groups and subscales. Five study groups conducted 1‐parameter item response theory analyses (Rasch analyses), and thus, reported person separation statistics, rather than Cronbach's alpha values and an additional group reported both Rasch and traditional classical test theory (CTT)‐based estimates of internal consistency …”

“…Content validity was reported in 15 of 31 (48%) studies, yet few studies provided well‐described item development procedures. In 6 of 31 (19%) studies, generated items were based on a review of the literature (Table ). People with subject matter expertise were utilized in 8 of 31 (26%) studies; 6 used subject matter experts for item generation, question formatting, and item reduction, and 2 used experts for item/instrument translation .…”

“…Six of 31 papers used Rasch analysis as a means of evaluating construct validity . Gable et al also used latent class analysis to classify students into 1 of 4 groups: bullies, victims, both, or neither. Latent class analysis is a statistical method that places respondents into mutually exclusive groups with respect to a latent trait that is not directly observable.…”

Cyber threats: a study of what middle and high school student know about threatening behaviours and internet safety

“…While this binary response format can result in less response variability than a traditional five-point agreement scale, pilot administrations of the survey supported the researchers' intent to create a non-threatening version of a 'true/false' format familiar to students' other assessment measures. In the section below, the validity/reliability evidence presented by Gable et al (2011) indicated that the categorical response format was psychometrically successful. Responses were scored '1' or '0' to reflect a high level of the attribute measured by the scale (e.g., knowledge) or higher levels of having experienced the attribute (e.g., bully victim, adult notification, and parental involvement) or exhibited the attribute (e.g., bully behaviour; internet behaviour).…”