The Statistical Test Used
The study aims to use a multivariate analysis of variance (MANOVA) in comparing the effects of cognitive behavior therapy (CBT), behavior therapy (BT), and no treatment (NT) on actions and thoughts related to obsessive-compulsive disorder (OCD). MANOVA is an appropriate statistical test because the data have one independent variable with three categories and two dependent variables on a continuous scale. According to Denis (2016), MANOVA applies in the analysis of data to determine the effect of two or more independent groups on two or more continuous variables. Jackson (2015) holds that the nature of data in independent and dependent variables determines the type of statistical analysis. The independent variable is the treatment method comprising CBT, BT, and NT while the dependent variables are the occurrences of actions (obsession-related behaviors) and the occurrences of thoughts (obsession-related cognitions).
The Basis of the Problem
The basis of the problem is that different methods of psychotherapy have different impacts on the treatment of OCD. As CBT and BT are two common methods employed in the treatment of OCD, NT was added as a control treatment method for comparison purposes to reveal the effect of both CBT and BT. Patients with OCD (N = 30) were sampled, and equal numbers (n = 10) were randomly assigned to CBT, BT, and NT groups. After treatment, the outcomes of these therapies were assessed by measuring the occurrences of actions (obsession-related behaviors) and the occurrences of thoughts (obsession-related cognitions). In this view, a comparative analysis of the outcomes of different therapies using MANOVA is essential to determine the most effective therapy.
Research Questions
- What are the effects of CBT, BT, and NT on the occurrences of thoughts and actions among patients with obsessive-compulsive disorder?
- Which is the best therapy in the treatment of obsessive-compulsive disorder among patients?
Hypothesis
The effects of CBT, BT, and NT on the occurrences of thoughts and actions among patients with obsessive-compulsive disorder are not statistically significant.
In answering the questions mentioned above and testing the hypothesis, the study utilized MANOVA. Warner (2012) explains that MANOVA uses the generalized linear model in establishing the relationships between dependent and independent variables. Descriptive statistics form the basis of MANOVA for they summarize data by highlighting inherent patterns and trends (Karter, 2016). The descriptive statistics show that there are apparent differences in the occurrences of actions and thoughts among patients with OCD in CBT, BT, and NT groups. Essentially, the descriptive statistics reveal that the occurrences of actions among CBT, BT, and NT groups (M = 4.53, SD = 1.456) are lower than the occurrences of thoughts among CBT, BT, and NT groups (M = 14.53, SD = 2.209).
Table 1.
Table 2.
Multivariate analyses indicate that the effects of therapies on actions and thoughts of patients with OCD are statistically significant. According to Pillai’s trace, the effects of CBT, BT, and NT on the occurrences of obsessive actions and thoughts are statistically significant, V = 0.318, F(4,54) = 2.557, p = 0.049. Likewise, Wilkis’ Lambda indicates that CBT, BT, and NT have statistically significant effects on the occurrences of obsessive actions and thoughts, L = 0.699, F(4,54) = 2.555, p = 0.050. However, Hotelling’s trace reveals that CBT, BT, and NT do not have a statistically significant effect on the occurrences of obsessive actions and thoughts among patients with OCD, T = 0.407, F(4, 54) = 2.546, p = 0.051. Similar to Pillai’s trace and Wilkis’ Lambda, Roy’s largest root confirms that CBT, BT, and NT have a statistically significant effect on the occurrences of obsessive actions and thoughts among patients with OCD, Θ = 0.335, F(4,54) = 4.520, p = 0.020.
Nevertheless, univariate analysis reveals that CBT, BT, and NT have no statistically significant effects on the occurrences of obsessive actions (F(2,27) = 2.77, p = 0.080) and the occurrences of obsessive thoughts (F(2,27) = 2.154, p = 0.136)
Table 3.
Post hoc analysis (Table 4) shows that there are no statistically significant differences in the occurrences of obsessive actions and the occurrences of obsessive thoughts between treatment groups (p> 0.05).
Table 4.
According to Pallant (2016) mean plot enhances exploratory analysis of data for it depicts the variation in means among different groups. The comparison of obsessive actions among treatment groups shows that BT has the lowest mean (Figure 1) while the comparison of obsessive thoughts reveals that CBT has the lowest mean (Figure 2).
Discriminant Analysis
The analysis of covariance matrices depicts that obsessive thoughts and actions have no significant relationship in the CBT group. Field (2013) asserts that discriminant analysis is necessary for it provides details about relationships established in MANOVA analysis. Furthermore, covariance matrices show that obsessive thoughts and actions have a positive relationship in the BT group. In contrast, covariance matrices reveal that obsessive thoughts and actions have a negative relationship in the NT group.
Table 5.
The discriminant analysis elucidates how two discriminant functions explain the discriminating ability of functions. Table 6 indicates that the first discriminating function accounts for 82.2% of the variance (R2 = 0.251) whereas the second discriminating function accounts for 17.8% of the variance (R2 = 0.068).
Table 6.
The combination of the two discriminating functions gives a significant differentiation of the treatment groups, L = 0.699, (4) = 9.508, p = 0.05. However, the removal of the first discriminating function shows that the second discriminating function does not have a statistically significant differentiation, L = 0.932, (4) =1.856, p = 0.173.
Table 7.
The structure matrix reveals the correlation between discriminating functions and how treatment outcomes load differently onto both functions. Evidently, obsessive actions load highly onto both the first discriminating function (r = 0.711) and the second discriminating function (r = 0.703). In contrast, obsessive thoughts load very highly onto the second discriminating function (r = 0.817) than the first discriminating function (r = -0.576).
Table 8.
he assessment of the discriminant function plot indicates that the first discriminating function differentiates the CBT group from the BT group. Comparatively, the second discriminating function differentiates the NT group from both BT and CBT groups.
Interpretation
Although descriptive statistics indicate that there are apparent differences in the means of obsessive actions and thoughts, the univariate analysis reveals that the differences are not statistically significant between the treatment groups. Coefficients of MANOVA, namely, Pillai’s trace, Wilks’ Lambda, and Roy’s largest root indicate that CBT, BT, and NT have a statistically significant effect on the occurrences of obsessive actions and thoughts. Post hoc analysis further reveals that there are no statistically significant differences in the occurrence of obsessive actions and thoughts based on treatment groups. Further analysis using the discriminant method suggests that differentiation of discriminating functions reveals underlying dimensions. Although the treatment methods appear to influence the occurrences of obsessive actions and thoughts, their distributions are inherent to OCD. In answering the first question, the data analysis established that the effects of CBT, BT, and NT have statistically significant effects on the occurrences of thoughts and actions among patients with OCD. CBT is the best treatment for obsessive thoughts whereas BT is the best treatment for obsessive actions among patients.
References
Denis, D. (2016). Applied univariate, bivariate, and multivariate statistics. Hoboken, NJ: Wiley.
Field, A. (2013). Discovering statistics using IBM SPSS statistics (4th ed.). Los Angeles, CA: SAGE Publications.
Jackson, S. J. (2015). Research methods and statistics: A critical thinking approach (5th ed.). Belmont, CA: Cengage Learning.
Karter, J. (2016). Descriptive statistics and exploratory analysis of data with matlab. New York, NY: CreateSpace.
Pallant, J. (2016). SPSS survival manual: A step by step guide to data analysis using IBM SPSS. Sydney, Australia: Allen & Unwin.
Warner, R. M. (2012). Applied statistics: From bivariate through multivariate techniques: From bivariate through multivariate techniques (2nd ed.). New Delhi, India: Sage Publishers.