conducted to see if SRC frequency, SRC severity and gender significantly predicted each of the cognitive scores, and whether the predictive values of SRC frequency and SRC severity were moderated by gender. Bonferroni’s correction was employed to control for multiple testing, with an adjusted alpha level of α = .05/10 = .005. Visual inspection of scatter matrixes did not identify any correlations between predictor variables. Possible collinearity was observed through interpretation of variation inflation factor values in the second hierarchical stage when the interaction term was entered in each model. When using multiplicative interaction terms within the same model as their component variables, the values near zero stay near zero, and the high numbers get much higher, so a degree of multicollinearity is expected. This does not affect the overall fit of the model. Centring of the predictor variables was used to minimise collinearity to acceptable limits (Aiken & West, 1991; Field, 2013). For example, the variable “SRC Frequency” was initially coded as “0 = none”, “1 = low”, “2 = medium”, and “3 = high”, and unacceptable levels of collinearity were observed. These variables were subsequently recoded as “-1.5 = none”, “-0.5 = low”, “0.5 = medium”, and “1.5 = high”. Regression summary tables are available in Appendix
K.
Regression 1 (Table K1). Detection was the dependent variable. SRC
frequency and gender were entered at stage one. The interaction between SRC frequency and gender was entered at stage two. At stage one, the regression equation explained <0.1% of the variation in detection scores. This was not significant (F(2, 77) = 0.33, p = .723), R2 = .01. Introducing the interaction variable
in stage two explained an additional 2.2% of the variation in detection scores. This was not a significant change in R2, (F(2, 76) = 0.80, p = .495), R2 = .03.
Regression 2 (Table K2). Identification was the dependent variable. SRC
frequency and gender were entered at stage one. The interaction between SRC frequency and gender was entered at stage two. At stage one, the regression equation explained 1.0% of the variation in identification scores. This was not significant (F(2, 77) = 0.41, p = .668), R2 = .01. Introducing the interaction variable
in stage two explained less than 0.1% additional variation in identification scores. This was not a significant change in R2, (F(2, 76) = 0.34, p = .797), R2 = .01.
Regression 3 (Table K3). One card learning was the dependent variable.
SRC frequency and gender were entered at stage one. The interaction between SRC frequency and gender was entered at stage two. At stage one, the regression equation explained 0.1% of the variation in one card learning scores. This was not significant (F(2, 77) = 0.05, p = .954), R2 < .01. Introducing the
interaction variable in stage two explained less than 0.1% additional variation in one card learning scores. This was not a significant change in R2, (F(2, 76) =
0.06, p = .617), R2 = .02.
Regression 4 (Table K4). One back test was the dependent variable.
SRC frequency and gender were entered at stage one. The interaction between SRC frequency and gender was entered at stage two. At stage one, the
regression equation explained 0.5% of the variation in one back test scores. This was not significant (F(2, 77) = 0.20, p = .819), R2 < .01. Introducing the interaction
variable in stage two explained 2.8% additional variation in one back test scores. This was not a significant change in R2, (F(2, 76) = 0.86, p = .467), R2 = .03.
Regression 5 (Table K5). SSRT was the dependent variable. SRC
frequency and gender were entered at stage one. The interaction between SRC frequency and gender was entered at stage two. At stage one, the regression equation explained 5.4% of the variation in SSRT. This was not significant (F(2, 77) = 2.21, p = .117), R2 = .05. Introducing the interaction variable in stage two
explained 0.1% additional variation in SSRT. This was not a significant change in R2, (F(2, 76) = 1.48, p = .227), R2 = .06.
Regression 6 (Table K6). Detection was the dependent variable. SRC
severity and gender were entered at stage one. The interaction between SRC severity and gender was entered at stage two. At stage one, the regression equation explained <1.8% of the variation in detection scores. This was not significant (F(2, 77) = 0.73, p = .487), R2 = .02. Introducing the interaction variable
in stage two explained an additional 2.0% of the variation in detection scores. This was not a significant change in R2, (F(2, 76) = 1.01, p = .392), R2 = .04.
Regression 7 (Table K7). Identification was the dependent variable. SRC
severity and gender were entered at stage one. The interaction between SRC severity and gender was entered at stage two. At stage one, the regression equation explained 1.0% of the variation in identification scores. This was not significant (F(2, 77) = 0.40, p = .674), R2 = .01. Introducing the interaction variable
in stage two explained an additional 0.3% of the variation in identification scores. This was not a significant change in R2, (F(2, 76) = 0.35, p = .791), R2 = .01.
SRC severity and gender were entered at stage one. The interaction between SRC severity and gender was entered at stage two. At stage one, the regression equation explained 2.0% of the variation in one card learning scores. This was not significant (F(2, 77) = 0.78, p = .463), R2 < .02. Introducing the interaction
variable in stage two explained an additional 3.3% of the variation in one card learning scores. This was not a significant change in R2, (F(2, 76) = 1.41, p =
.247), R2 = .05.
Regression 9 (Table K9). One back test was the dependent variable.
SRC severity and gender were entered at stage one. The interaction between SRC severity and gender was entered at stage two. At stage one, the regression equation explained 0.4% of the variation in one back test scores. This was not significant (F(2, 77) = 0.16, p = .852), R2 < .01. Introducing the interaction variable
in stage two explained 2.6% additional variation in one back test scores. This was not a significant change in R2, (F(2, 76) = 0.79, p = .501), R2 = .03.
Regression 10 (Table K10). SSRT was the dependent variable. SRC
severity and gender were entered at stage one. The interaction between SRC severity and gender was entered at stage two. At stage one, the regression equation explained 4.3% of the variation in SSRT. This was not significant (F(2, 77) = 1.74, p = .1.82), R2 = .04. Introducing the interaction variable in stage two
explained 0.1% additional variation in SSRT. This was not a significant change in R2, (F(2, 76) = 1.17, p = .327, R2 = .04.
The hierarchical regressions indicated that SRC frequency and SRC severity were not significantly predictive of outcomes on CogState tests or SSRT. The influence of these predictors was not moderated by gender.
Hypothesis 2: SRC dosage predicting SRC symptoms, moderated by