Essay about Review: Psychometrics and Validity Meaning

Submitted By One2Succeed
Words: 1177
Pages: 5

FACTOR Anal based on mathematical procedures NOT STATISTICAL/instrument refinement/assess construct validity/used 4 scales that assess 2 or more dimensions /factors – factor loadings range from -1.00 - +1.00 w values from 0-+1 positive association, higher the number the higher the association(2) Exploratory FA used in context of health behavior/explores internal structure of scale/used in early development of a scale 2 identify # of factors/correspondence between factors & quality of items/reduces variables(items)to a smaller set of variables(factors).Variables that R strongly correlated R grouped together. EFA can determine if each dimension proposed for the scale exists/determine variables/items within each dimension R correlated w each other(more items in other dimensions)identify weak items on a scale/items that don’t define construct should be deleted from the instrument. Used in EARLY stages/used when underlying structure of data is unknown/determine construct of interest/STATISTICAL/uncovers structure of large sets of variables/examines internal validity/NOT for testing hypothesis or theories.Confirmatory FA Chi squared is an appropriate statistical method to evaluate CFA – CFA used to validate the relationship among factors & between items/used when prior information about the scale structure is known(usually from EFA)if correlations among factors and between items can be reproduced, there is evidence of the validity of scale. Should use at least 5 participants per variable.Used in LATTER stages to confirm hypothesis of theories that have been generated by EFA. Allows researcher to test construct or factor. Uses theory/postulates relationships/determines “goodness of fit” Theory testing NOT theory generating.
CONTENT Validity extent to which a measure represents a given construct.
CONVERGENT VALIDITY shows theoretically based similarities between measurable construct & is subcategory of construct validity/scales should have positive correlation w other measures of the same construct using PEARSON R as measure for correlation coefficients. Pearson correlation range is -1.00to +1.00. in reliability coefficients the greater the coefficient, the higher the reliability. CRONBACH’S ALPHA – MOST POPULAR reliability statistic!!! scores are .00-+1.00-alpha score is closer to one when items are correlated & measuring the same construct/DETERMINES INTERNAL CONSISTENCY & instruments overall reliability homogeneity of items. SHOULD be calculated each time a scale is use to see if remains stable over multiple studies. (a Alpha) range normally 0 and 1. IE: >.9=excellent and >.5 unacceptable. MTMM can enhance further validity/complex method use multiple methods matrix or correlations 2 interpret assessment of construct validity. INTERNAL CONSISTENCY degree of homogeneity of scale items but gets confused w homogeneity of scale/ homo – refers tone dimensional scale & interrelatedness is a term that includes both 1 dimensional & multidimensional scales. Int consistency is concerned w interrelatedness of sample test items/homogeneity refers homogeneity.TEST-RETESTmeasures temporal stability how”constant scores remain from one occasion to another”score is influenced by numerous factors such as real change construct, changes due to subject variability or unreliable measure. Individuals can influence reliability w deliberate effect/ONLY reliable measure when the phenomenon has remained stable. ALPHA COEFFICIENT .70 or greater is a good standard to estimate reliability INTER-ITEM CORRELATION used to determine which items should be removed from the scale between .20 & .80 R ideal/closer alpha is to 1, the greater the INTERNAL CONSISTENCY/if an item does not contribute to the IC of the scale – item should be excluded from the scale. EIGENVALUES- amount of variance that is accounted for by each factor in FA is referred to as the eigenvalue/helps researcher to determine factors to keep and which to discard/values less that .1 are