Video game U&G 12

ƒ “You can sit there and play it over and over.”

In order to develop the scale, U&G items were generated in Likert form representing each

of the video game play motivations (number of items = 145). A group of seven raters then

eliminated any item that lacked face validity (Emmert, 1989a). The criteria for face validity was

unanimous agreement among the seven raters that an item belonged in one dimension and no

other. This reduced the pool of items to 60, or 10 for each dimension.

*Factor Analysis for Internal Consistency. *The homogenous set of items were pretested on a

sample of 205 young adults and results were factor analyzed for internal consistency (Hunter,

1980). The internal consistency theorem, based on Spearman’s (1904) rank one matrix, states

that if the set of items are alternate indicators of the same construct, then the correlation between

them will be the product of their factor loadings (factor loadings are the correlation between the

item and the underlying construct). The squared factor loading is the reliability for that item in

predicting the underlying construct. Factor loadings are estimated using the Hunter and Gerbing

(1982) method in which factor loadings are estimated by summing the item correlations and

dividing by the square root of the sum of the correlation matrix. The communalities are inserted

in the diagonals and the factor loadings are re-estimated. The process is iterated until a stable

solution emerges.

Once the matrix stabilizes, parameters predicted by the factor model are compared to

obtained correlations. Items with residuals exceeding what would be expected by sampling error

or with low reliabilities are eliminated. Factor loadings for all items are provided in Table 1. A

test of significance, represented as

χ

2

is also available. All items in the six uses and gratifications

scales passed the test for internal consistency (see Table 1).