Do parents perceive student anxiety to reduce following secondary school students’ participation in an indicated CBT intervention programme?
Experimental hypothesis
Null hypothesis Dependent Variable
Parents of participants in the CBT-based
intervention condition will report significant
reductions in student anxiety between pre-test and post-test measures. These changes will not be observed in parents of participants in the wait-list comparison group. There will be no statistically significant difference between parental perceptions of student anxiety reported by parents of participants in the experimental condition and parents of participants in the wait- list comparison condition, between pre-test and post-test measures.
Parental responses on the Spence Children’s Anxiety Scale for Parents.
Table 4.4 – A table to show the Experimental and Null hypotheses for Research Question Two
4.2.5.1 Descriptive Statistics
Table 4.4 outlines the descriptive statistics for research question two. This data is derived from pre-test and post-test data provided by parents’ completion of the SCAS-P measure. The means and standard deviations are provided for the intervention and wait-list conditions and the total sample.
148 Observations:
The mean anxiety score reported by parents of participants in the intervention group is lower than that reported by parents of participants in the wait-list group at time 1.
The time 1 scores reported by intervention group parents are clustered more closely around the mean, with lower standard deviation than those scores recorded by parents in the wait-list condition.
The intervention group demonstrated reductions in parent-reported scores of student anxiety between time 1 and time 2, suggesting that these parents perceived their children’s anxiety to have reduced over the course of the intervention (Figure 4.5).
The wait-list group demonstrated an increase in parent-reported scores of student anxiety between time 1 and time 2, suggesting that these parents perceived their children’s anxiety to have increased over the course of their wait-list phase (Figure 4.5).
The time 2 scores reported by intervention group parents are clustered more closely around the mean, with lower standard deviation than scores recorded by parents in the wait-list condition.
149 Time Condition N SCAS-P Results Mean Standard Deviation Time 1 (Pre-test) Intervention condition 8 23.25 9.20 Wait-list comparison condition 8 34.38 23.31 Total 16 28.81 18.05 Time 2 (Post-test) Intervention condition 8 22.37 10.25 Wait-list comparison condition 8 41.13 25.27 Total 16 31.75 20.99
Table 4.5 – A table to show the descriptive statistics for Research Question Two
150
Figure 4.5 – A graph to show the mean parent-report anxiety scores for both conditions at both pre-intervention and post-intervention
151
4.2.5.2 Test of Normality
The Shapiro-Wilk analysis was undertaken to explore the normality of distribution of the data across both conditions. The results are illustrated in Table 4.6. Shapiro-Wilk data Intervention condition Wait-list
condition Total sample
Statistic .278 .147 .936
Degrees
freedom 8 8 16
Significance .068 .200 .304
Table 4.6 - A table to show the results of the Shapiro-Wilk test data at Time 1 for the intervention condition, wait-list condition and total
sample
These results show that the Shapiro-Wilk analysis was not statistically significant (p>0.05) for either the intervention, wait-list or total samples. The data can therefore be assumed to be normally distributed. Figure 4.6 represents a Q-Q plot of the distribution of the total data set across both conditions at time 1. The closer the data lies to the line, the more confidently we can say that the data is normally distributed. Normally distributed data enables the implementation of parametric tests.
152
Figure 4.6 – A Q-Q plot illustrating the distribution of parent data at Time 1 within the total sample
4.2.5.3 Test of Homogeneity of Variance
Levene’s test of Equality of Variances was used to assess the spread of scores on the dependent variable (SCAS-P anxiety scores) between the two conditions at time 1. The results of this analysis were statistically significant (F(1,14)=8.354. p=0.012), indicating that equality of variances was not apparent across both conditions.
A smaller sample size, such as that within the current study, may impact upon the power of the Levene’s test to detect differences between the variances (Field, 2013; 10). There was therefore a need to inspect the individual variances for each condition at time 1, with intervention variance (9.192=84.45) being substantially smaller than the wait-list variance (23.312=543.41), suggesting that homogeneity of variances was not apparent.
153 There was therefore a need to proceed with caution, as equality of variances between the conditions cannot be assumed (Cohen, Manion and Morrison, 2011; 650). Several authors state that parametric testing may still be considered when this assumption is violated, but equal sample sizes exist across conditions (Dancey and Reidy, 2011; 156; Field 2013).
4.2.5.4 Test of Equality of Means
Independent t-tests were used to compare the mean anxiety scores for the independent groups at time 1. The results of this analysis were not statistically significant (t(14)=-1.256, p=.240, equal variances not assumed) indicating that the two conditions were comparable in terms of mean anxiety score at time 1.
4.2.5.5 Two-way Mixed ANOVA
A two-way mixed ANOVA was use to analyse whether there was a significant interaction between the between-groups independent variable (condition) and the within-subjects factor (time) upon the dependent variable (parent- reported scores of student anxiety).
The results of the two-way mixed ANOVA indicate that the main effect of the within-subjects factor, time, was not statistically significant (F(1,14) = 2.597, p = .129, partial 2 = .156), nor was the main effect of the between subjects
factor, condition (F(1,14) = 2.708, p = .122, partial 2 = .162). The condition
by time interaction was also not significant (F(1,14) = 4.374, p = .055, partial
2 = .238), albeit this interaction did approach statistical significance (p>0.05).
The results of the ANOVA are presented in appendices 42 and 43.
4.2.5.6 Summary of Research Question Two
Those trends in parent-reported student anxiety scores illustrated in Figure 4.5 and the descriptive statistics reported in Table 4.5 demonstrate a slight reduction in parent-reported anxiety scores for the intervention condition between time 1 and time 2, compared to an increase in parent-reported anxiety scores for the wait-list condition between time 1 and time 2.
154 These trends tentatively indicate that parents within the wait-list condition perceived their children’s anxiety to increase during the wait-list phase, compared to the slight reduction in student anxieties reported by parents in the intervention condition. This is an unexpected trend and was not predicted within the initial experimental hypotheses for research question two. However, these findings must be interpreted with caution as results of the two-way mixed ANOVA suggest that this condition-by-time interaction is only approaching statistical significance and does not achieve full statistical significance (F(1,16) = 4.374, p = .055, partial 2 = .238). Explanation of this
effect will be considered in section 5.2.2.
Therefore, when the experimental hypothesis is considered; ‘Parents of
participants in the CBT-based intervention condition will report significant reductions in student anxiety between pre-test and post-test measures. These changes will not be observed in parents of participants in the wait-list comparison group.’ it would appear that the condition that participants were
in (intervention or wait-list comparison) did not have a significant impact upon parent-reported levels of student anxiety.
Subsequently, the following null hypothesis must be accepted: ‘ There will be
no statistically significant difference between the student anxieties reported by parents of participants in the experimental condition and parents of participants in the wait-list comparison condition between pre-test and post- test measures’.
Further consideration of the findings from research questions one and two will take place in the discussion chapter.