• No results found

Chapter 8 Results of the Intervention Study

8.2 Section A: Connectedness

8.2.5 Connectedness analysis of the control and experimental groups over time

A paired t-test analysis comparing the intervention and control group at Time 1, shown in Table 8.4, revealed the overall connectedness mean for the experimental group (206.7) was

142

higher than the overall scale mean for the control group (202.75). The experimental group also had a lower standard deviation compared to the standard deviation of the control group for the same period. Examination of the connectedness subscale means for both external connectedness sources of family and community were higher for the experimental group compared to the control group but the two highest subscales for both participant groups over time. At Time 1 both groups reported the largest variances for both subscales, with the variance for family being higher for the control group (5.10) compared to the experimental group variance of 4.00. In contrast, community variance was higher for the experimental group (5.02) compared to the control (4.32).

In descending order, both groups, at Time 1, reported low means for perceived personal connectedness, followed by other students with the lowest means for connectedness to lecturers. The experimental group reported a personal connectedness mean of 40.70 compared to the control mean of 40.20. In addition, the standard deviation for the experimental group was slightly lower (3.25) than the variance for the control group at 3.44. The subscale mean for connectedness to other students, for the experimental group, was higher and variance were smaller (x̅ = 40.55, s = 3.64) compared to the control group (x̅ = 39.34, s = 4.12). This pattern appeared to be repeated when examining lecturer connectedness mean and variances for the experimental (x̅ = 39.17, s = 3.39) and control groups (x̅ = 38.79, s = 3.66). Overall, at Time 1, the overall connectedness mean and subscale means are higher for the experimental group compared to the control; the overall scale variance and four subscale variances are lower, and more closely grouped, for the experimental group with the exception of community variance.

143 Table 8.4

Connectedness Means, Standard Deviations, F Values and Significance of Change (p Values) Comparing Experimental and Control Group Student Pairs at Time 1 and Time 2

Connectedness Subscales Control Group (n=33) T1 T2 Experimental Group (n=51) T1 T2 x̅ s x̅ s x̅ s x̅ s F p Personal 40.20 3.44 39.45 4.21 40.70 3.25 39.55 3.03 6.006 0.016* Family 43.69 5.10 44.39 5.20 44.69 4.00 44.32 4.77 0.100 0.740 Students 39.34 4.12 38.39 5.57 40.55 3.64 38.73 4.16 7.480 0.008** Lecturer 38.79 3.66 37.88 5.23 39.17 3.39 37.58 4.46 4.974 0.028* Community 40.72 4.32 40.18 5.01 41.62 5.06 41.11 5.14 0.980 0.325 Overall Scale 202.75 20.64 200.31 25.22 206.73 19.34 201.30 21.56 6.165 0.015* * = p <0.05; **; = p <0.01; = increase

144 8.2.6 Connectedness analysis at Time 2

Examination of the connectedness measures at Time 2, as shown in Table 8.4, reveal the overall scale mean decreases for both the control and experimental groups. While the scale mean remained higher for the experimental group (201.30) compared to the control group (200.31), the mean decrease for the experimental group (5.43) was more than twice the decrease reported in the control group (2.44). Over the same period, the overall scale variance for the experimental group increased slightly to 21.56 compared to the much larger control variance which increased to 25.22 at Time 2.

The connectedness influences of family at Time 2, external to the influence of the university support, revealed a slight increase in the control group mean (x̅ = 44.30, s = 5.20) and standard deviation. In comparison, the family connectedness mean for the experimental group decreased slightly (x̅ = 44.39, s = 4.77) while the variance increased at Time 2. An examination of the community connectedness means at Time 2 reported a decreased mean for the control (x̅ = 40.18) and experimental group (x̅ = 41.11). Both reported small increased variances over time. Despite non-significant fluctuations, family and community connectedness subscales means were high and stable for both groups.

The three remaining subscale means retained the same order over time. Personal connectedness means decreased for both the control (39.45) and experimental groups (39.55). Despite the similarity in their means at Time 2 the control group variance increased (s = 4.21) whereas the variance for the experimental group decreased (s = 3.03). Subscale connectedness means for other students and lecturers continued the trend of decreases at Time 2 and both revealed increased standard deviations during the same period. Furthermore, the subscales of other students and lecturer retained the bottom two positions and remained below the top quartile during both collection periods.

Closer examination of the overall scale group means at Time 2 revealed the experimental group dropped by twice the amount (5.43 points) compared to the control group decrease (2.44) over the same period. This pattern was also evident when comparing subscale decreases for other student (1.82) and lecturer means (1.99). Yet again, the experimental group experienced a larger decrease compared to the control group; certainly not an improvement for either group.

145

An examination of variances across the samples, calculated using Levene’s formula and located in Appendix D, revealed non-significant differences. However, significant mean differences p<0.05, shown in Table 8.6, between groups for personal, other student and lecturer connectedness indicated something was occurring during semester of the intervention but not after that point. A summary of the multivariate analysis26, as shown in Table 8.5, was also undertaken to determine between subject/group connectedness differences observing multiple outcome. In the first instance, this was hoped to reveal patterns and relationships that reflect the reality of university student connectedness.

Based on the intervention variable the differences between the control and intervention subjects reported a small F ratio of 0.421 indicating similar means between groups. Furthermore the non-significant value (0.833) confirmed the intervention had no impact. However, changes within subjects revealed a larger F ratio (F = 2.435) signifying mean differences and a significant p value (0.042), confirming changes in individuals over time. In this instance, the mean decreased over time, despite the population not being identified as disconnected early in their course. Change is to be expected with the passage of time and in this instance, students are less connected as they advance through the first year of their university course.

In an effort to add clarification to this outcome, a univariate analysis27 was undertaken to determine which of the five connectedness sources, when examined individually, contributed to the within group differences over time.

26

Multivariate analysis (MVA) is an ANOVA analysis involving several dependant variables which test for significant differences in two or more means (Vincent, 1999). Such information helps to answer if changes in the independent variable (IV) have a significant effect on the dependent variable (DV). In this research it refers to whether changes in the manipulated independent variable (IV), control and experimental condition have an effect on the dependent variable (DV) of connectedness as a way to inform about the broad interactions and relationships.

27

Univariate analysis describes data for one variable. While helpful to summarise data and reveal patterns, it does not does not inform about cause or relationships (Thomas & Nelson, 1990).

146 Table 8.5

General Linear Model (GLM) of Connectedness Analysis for 84 Student Pairs

Analysis Significance result

Multivariate Between subjects (CG and IG) intervention effect is non-significant (F = 0.421, p = 0.833)

Within subjects time effect is significant (F = 2.435, p = 0.042)

Group by time interaction is non-significant (F = 0.366, p = 0.870)

Univariate Within subject family connectedness is non-significant (F = 0.110, df1, p = 0.740)

Within subject community connectedness is non-significant (F = 0.980, df1, p = 0.325)

Within subject personal connectedness is significant at .016 (F = 6.006, df1, p = 0.016)

Within subject other student connectedness is significant at 0.008 (F = 7.480, df1, p = 0.008)

Within subject lecturers connectedness is significant at 0.028 (F = 4.974, df1, p = 0.028)

Within subject overall connectedness is significant at 0.015 (F = 6.165, df1, p = 0.015)

Univariate Group by time interaction28

Family connectedness is non-significant (F = 1.072, df1, p = 0.304)

Community connectedness is non-significant (F = .0001, df1, p = 0.972)

Personal connectedness is non-significant (F = .0286, df1, p = 0.594)

Other student connectedness is non-significant (F = .737, df1, p = 0.393)

Lecturer connectedness is non-significant (F = 0.368, df1, p = 0.546)

Overall connectedness is non-significant (F = 0.889, df1, p = 0.349)

The univariate results for the family (F = 0.110, p = 0.740) and community means (F = 0.980, p = 0.325) were not significantly different. However, high F values and significance levels for personal (F = 6.006, p = 0.016), other students (F = 7.48, p = 0.008) and lecturers (F = 4.974, p = 0.028) were found to contribute to the overall scale differences (F = 6.165, p = 0.015).

28

Group by time interaction –refers to groups changing over time but in different ways. In this instance, the change related to the difference between the intervention and control groups (IDRE, 2007).

147

The overall connectedness scale mean, and most subscales, decreased over time while most variances increased at Time 2. The two exceptions included a non-significant family mean increase for the control group and a decreased personal standard deviation for the experimental group. Furthermore, variances at Time 2 for the control group indicated responses were exceedingly diverse (25.22) compared to that of the experimental group (21.56) at Time 2. The exception was personal connectedness which decreased for the experimental group. Knowledge that student connectedness changes continue across their first year at university was indeed helpful to sustain efforts of the working party to provide first year student support. However, time by group interactions, shown in Table 8.5, confirmed changes were not significantly different regardless of whether students belonged to the intervention or control groups. This finding indicated more work was yet to be done to explain the lack of expected differences in the semester following the intervention.