Structural equation model 

while some of the differences among cats in friendliness can be accounted for by individual variation, several studies have examined the effect of early handling。the reliability of our estimate of the individual beta。［20］huang, huiping. “frame-rich, frame-poor: an investigation of the contingent effects of media frame diversity and individual differences on audience frame diversity”, international journal of public opinion research, vol.22, 2009, pp.47-73.。

86)。 We measured social competitiveness at Phase 1 with three items, endorsed on a 5-point Likert-type scale (1 = strongly disagree , 5 = strongly agree ) from the Cooperative/Competitive Strategy Scale (Simmons et al。, 1988)。 These items represent the use of competition to both motivate and achieve success as opposed to the use of cooperation to motivate and achieve success, or the avoidance of achieving success through competition。 Reliability of the social competitiveness subscale was acceptable ( α = 。

76)。 Motivations for participating in drinking games 。 We measured students’ fortitude display, sexual, and competitive motivations for participating in drinking games via nine Likert-type self-report items (Hone et al。, 2013)。 Items were rated on a 5-point scale (1 = strongly disagree , 5 = strongly agree ), including an option to select the statement, “ I do not play drinking games 。 ” The fortitude- display motivation comprised the items, “I like to play drinking games that show who can go the longest with out passing out,” “。

。。that show who can last the longest without throwing up,” and “。。。 in which sometimes people need to be put to bed。” The sexual motivation factor comprised the items, “I like to play drinking games that give me an opportunity to have fun w ith people I’d like to have sex with,” “。。。that loosen people up for fooling around or having sex later,” and “that give me an opportunity to hit on people I’m interested in。” Finally, the competitive motivation factor comprised the items, “I like to play dr inking games in which people are really serious about winning,” “。

。。that have clear winners and losers,” and “。。。in which people get teased or respected depending on how they played。” Reliability of these factors was acceptable (Fortitude-display motivation: α = 0。81; sexual motivation: α = 0。90; competitive motivation: α = 0。75)。 Drinking game behaviors and drinking problems 。 At Phase 1 and Phase 2, we used two Likert-type self-report items to measure frequency of drinking game participation and quantity of alcohol consumed during drinking game participation。

The item “How often do you play drinking games?” was endorsed on an 8-point scale (0 = never , 7 = daily or almost daily ) and the item, “How much alcohol do you tend to consume when you play drinking game s (“drink” = 1 beer or 1 shot)?” was endorsed on a 6-point scale (0 = none , 5 = seven or more drinks )。 We also used six items from the Alcohol Use Disorders Identification Test to measure students’ drinking problems at Phase 1 and Phase 2 (Babor, Higgins-Biddle, Saunders, and Monteiro, 2001)。

Reliability of the six-item version of the AUDIT was acceptable (Phase 1: α = 0。78; Phase 2: α = 0。83)。 We specified causal paths from these three scores at Phase 1 to their corresponding scores at Phase 2 so that the remaining variance in the respective Phase 2 scores could be conceptualized as measures of residualized change over the 12-week period (Cronbach and Furby, 1970)。 This is analogous to regressing Phase 2 scores on Phase 1 scores and then assessing the relationship between the predictor variables and these residuals。

the independent variables are truly independent— the model assumes is using deviations across the x variables to explain the dependent series. the regression attempts to explain the dependent series’ variations across。making comparisons to direct attention to why differences in costs exist across companies。— sum of the squares of differences across two variables — sum of the difference of the squares of two variables。

Excluding these two variables as outcomes did not influence the direct or indirect path coefficients from sex to the residualized change score measure of frequency of drinking game participation that we report here (see Table 2)。 In a structural equation model, we specified a path from frequency of drinking game participation at Phase 1 to frequency of drinking game participation at Phase 2。 We then specified paths from sex, to sexual restrictedness and social competitiveness, to fortitude- display, sexual, and competitive motivations, and to frequency of drinking game participation at Phase 2, controlling for participation at Phase 1 (paths from independent variables to frequency at Phase 1 were not specified; see Figure 1)。

According to the modification indices, some residual variances of drinking game items that comprise the drinking game motivations were specified to correlate with each other (see Table 3), which improved model fit (see Table 4)。 Data were assumed to be missing at random and full information maximum likelihood estimation was used。 The chi square test of model fit to these data was significant, χ 2 (80) = 351。75, p < 。001, which indicated poor fit (Kline, 2011)。 However, analysts typically complement this fit statistic with a variety of other indices for assessing model fit。

A Comparative Fit Index of 0。94 indicated that the proposed model provided a better fit than a baseline model (Hu and Bentler, 1999; Schreiber, Stage, King, Nora, and Barlow,2006)。 In contrastequation model, the Root Mean Square Error of Approximation statistic was 0。07, with a lower bound confidence interval value of 0。06 and an upper bound confidence interval value of 0。08, suggesting that a close-fit hypothesis could not be retained, although the poor-fit hypothesis could be rejected (Browne and Cudeck, 1993)。

Finally, a Standardized Root Mean Residual of 0。09 was obtained, indicating that the average residual of the difference between the observed and proposed variance/covariance matrix was acceptable and,thusequation model, that the model provided decent explanatory power (Hu and Bentler, 1999; Schreiber et al。, 2006)。 Together, the six predictors in the model (sex, sexual restrictedness, social competitiveness, and the three drinking game motivations) accounted for 45% of the variance in residualized change score in frequency of drinking game participation。

Eight percent of the variation in residualized change score in frequency of drinking game participation was accounted for uniquely by sexual restrictedness, social competitiveness, and the three drinking game motivations at Phase 。。。