Módulo V Validez Psic 3057

The video discusses the concept of validity in psychometric evaluation, focusing on the need for an instrument to measure the construct it claims to assess accurately. The process of establishing validity parallels that of reliability, indicating that a reliable instrument should yield valid inferences. Before examining validity, practitioners calculate reliability as it indicates the instrument's capacity to consistently measure responses over time, providing a foundational check before assessing validity indicators. Content validity is emphasized, which involves ensuring the content of the instrument is representative of the construct's universe. Strategies like specification tables and expert judgment are used to establish content validity.

The video continues to explore content validity and introduces concurrent and predictive validity as methods to evaluate criterion validity, which measures an instrument's capacity to predict short and long-term behaviors. Concurrent validity questions an individual's current success through immediate administration of the test and obtaining the criterion measure right away, showing a positive correlation as evidence of effectiveness. Predictive validity, on the other hand, evaluates future performance by correlating initial test results with future achievements. Examples are provided, demonstrating how intelligence tests predict business decision-making capabilities and performance evaluations.

Construct validity is introduced, evaluating whether an instrument truly measures the intended construct through empirical techniques. Different methods like convergent and divergent validity analyses and multi-method matrices are used to assess this validity. Convergent validity involves correlating two instruments intended to measure the same construct, seeking high positive correlations. Divergent validity, conversely, checks that instruments measuring different constructs show low or negative correlations, ensuring distinct measurements. These methods verify whether instruments differentiate between constructs while being able to measure them accurately.

The multi-method matrix study is a combination of convergent and divergent validity tests done simultaneously to evaluate construct validity. The discussion expands into the dimensionality of measurement instruments, emphasizing the importance of confirming an instrument's capacity to measure one or more dimensions accurately. Dimensionality involves grouping items to measure dimensions which may or may not correlate; theoretical or empirical conceptions of dimensions are discussed. Correct dimensionality ensures that an instrument provides valid inferences, with factor analysis being a key tool used in this process.

The session delves into the significance of understanding the dimensionality in both uni- and multidimensional instruments. Unidimensional instruments focus on a single construct with all items measuring one dimension, summed up to reflect a total scale score. Multidimensional instruments require evaluating correlations between dimensions, critical for diagnosing different constructs without assuming a total score unless theoretically justified. Items should not measure multiple dimensions to avoid validity issues. Examples illustrate varying assumptions on dimensional weight, affected by theories and models guiding the instrument's construction.