Effects of WSWM
Table
2 shows an overview of the mean scores (SD) on the dependent variables at pre-test (T0) and post-test (T1) for both intervention and comparison groups, and the F-test statistic for the time x group interaction effect. Below we discuss the significant interaction effects on the different outcome variables.
Table 2
Mean items scores or percentages for experimental and control groups at pre- test and post-test
Knowledge
| | | | | |
Knowledge about non-causes of STI (petting, fondling, deep kissing) | .37 (.56) | .31 (.37) | .32 (.37) | .28 (.34) | .38 |
Knowledge about causes of STI (anal sexual intercourse; vaginal sexual intercourse; mutual masturbation) | .47 (.46) | .64 (.43) | .44 (.37) | .58 (.38) | 2.20 |
Knowledge about non-causes of HIV (petting, fondling, deep kissing) | .38 (.55) | .34 (.38) | .27 (.34) | .30 (.34) | 6.02* |
Knowledge about non-causes of pregnancy (petting, fondling, deep kissing) | .93 (.45) | .89 (.34) | .90 (.20) | .90 (.21) | .16 |
Knowledge causes of pregnancy (vaginal sexual intercourse; not anal intercourse) | .26 (.44) | .23 (.42) | .22 (.32) | .20 (.31) | .28 |
Beliefs
| | | | | |
Beliefs about causes of pregnancy | 3.76 (.99) | 4.18 (.89) | 3.78 (.98) | 4.03 (.89) | 11.16** |
Beliefs about causes of HIV | 3.60 (.96) | 3.74 (.98) | 3.53 (.97) | 3.72 (.93) | .27 |
Beliefs about causes of STI | 3.27 (.96) | 3.58 (.89) | 3.25 (.96) | 3.51 (.87) | .67 |
Risk perception (perceived risk and perceived severity)
| | | | | |
Risk perception towards HIV | 3.92 (.94) | 3.90 (.93) | 3.89 (.94) | 3.95 (.90) | 1.07 |
Risk perception towards STI | 3.45 (1.00) | 3.55 (.95) | 3.49 (.97) | 3.55 (.94) | .73 |
Risk perception towards pregnancy | 4.00 (.90) | 4.04 (.88) | 3.94 (.89) | 4.02 (.87) | .21 |
Delay
| | | | | |
Attitudes | 4.10 (.97) | 4.22 (.94) | 4.11 (.95) | 4.16 (.97) | 1.51 |
Perceived social norm (single item) | 3.59 (1.37) | 3.89 (1.24) | 3.69 (1.33) | 3.78 (1.29) | 7.15** |
Self-efficacy | 4.02 (1.03) | 4.17 (.97) | 3.94 (1.04) | 4.12 (.98) | .03 |
Intention-1 | 4.01 (1.27) | 4.16 (1.15) | 3.93 (1.29) | 3.94 (1.26) | 4.11* |
Intention-2 | .91 (.28) | .92 (.28) | .92 (.28) | .87 (.34) | 6.98** |
Condom use
| | | | | |
Attitude (wise to use a condom) | 3.78 (1.09) | 3.82 (1.04) | 3.81 (1.07) | 3.74 (1.12) | 4.56* |
Attitude (pleasant to use a condom) | 3.52 (1.29) | 3.27 (1.16) | 3.54 (1.29) | 3.37 (1.22) | 1.02 |
Perceived social norm towards condom use | 3.79 (1.21) | 3.68 (1.24) | 3.86 (1.23) | 3.80 (1.20) | .26 |
Self-efficacy towards condom use | 3.31 (1.05) | 3.37 (1.01) | 3.34 (1.04) | 3.27 (1.01) | 5.09* |
Performance behaviour (buy, carry a condom etc.) | 1.94 (1.14) | 2.14 (1.11) | 2.23 (1.25) | 2.53 (1.22) | .46 |
Intention to use a condom | 3.70 (1.35) | 3.79 (1.27) | 3.82 (1.26) | 3.75 (1.27) | 3.96* |
Non-coercive sex
| | | | | |
Performance behaviour to escape and avoid situations where unwanted sex could happen | 2.24 (1.33) | 2.21 (1.27) | 2.28 (1.35) | 2.25 (1.30) | .000 |
Attitudes towards using force for getting sex | 3.65 (1.04) | 3.78 (.97) | 3.64 (1.06) | 3.69 (1.04) | 1.82 |
Self-efficacy in dealing with coercive sex | 3.74 (.82) | 3.96 (.79) | 3.76 (.77) | 3.88 (.76) | 5.51* |
Intention to deal with coercive sex | 3.93 (1.01) | 4.02 (.91) | 3.95 (.96) | 3.93 (.99) | 2.69 |
Knowledge about risky sexual behaviour leading to STI, HIV or pregnancy: Multilevel analysis revealed a reversed effect of intervention on knowledge scores relating to non-causes of HIV (petting, fondling and deep kissing). At pre-test, a significant difference between groups was found, F (1, 1182) = 18.27, p < .000), with the intervention group scoring significantly higher on knowledge of non-causes of HIV than the comparison group. At post-test, students who followed WSWM had, on average, lower knowledge scores as compared to pre-test, whereas students from the comparison group improved their knowledge at post-test as compared to pre-test. However, the intervention group still scored significantly higher at post-test than the comparison group, F (1, 1257) = 4.22, p < .05). For the other knowledge scales, no significant effects of intervention were found.
Beliefs about STI, HIV and pregnancy: Students from the intervention schools scored significantly better than the comparison students at post-test as compared to pre-test when it came to wrong beliefs concerning pregnancy, like "a girl cannot get pregnant the first time she has sexual intercourse". At pre-test, there was no significant difference between intervention and comparison groups when it came to wrong beliefs concerning pregnancy, F (1, 1810) = 1.12, p = .289). However, at post-test, the intervention group scored significantly higher than the comparison group, F (1, 1501) = 21.98, p < .001. No significant effects were found for beliefs about STIs, or for beliefs about HIV.
Risk perception: No significant effects were found on risk perception.
Delaying sexual intercourse: We found an interaction effect for intervention over time for perceived social norm. At pre-test, there was no significant difference (F (1, 1745) = .13, p = 1.30) between intervention and comparison groups. At post-test, both intervention and comparison groups agreed significantly more with the statement, "My friends believe that people my age should postpone sexual intercourse until they are older", as compared to pre-test, but this change was only significant for the intervention group. Furthermore, students who had followed WSWM were more convinced at post-test as compared to pre-test that they would wait with sexual intercourse until they were older (condition x time interaction effect), (F (1, 1385) = 11.18, p = .001), whereas there was no significant increase in agreement with this statement for the comparison group at post-test as compared to pre-test. At pre-test, there was no difference between intervention and comparison groups (F (1, 1655) = 1.01, p = 3.15. The students from the intervention group also held a stronger intention to delay sexual intercourse at post-test, F (1, 1257) = 7.22, p = .007, again with no significant difference between the two groups at pre-test, (F (1, 1695) = .01, p = .937). No significant effects were found for attitudes towards delaying sexual intercourse, nor on self-efficacy towards delaying sexual intercourse.
Condom use: The analysis revealed a positive effect on attitude towards using condoms ("It is wise to use a condom" and "everybody should use one"). At pre-test, there was no significant difference between intervention and comparison groups, F (1, 1623) = .34, p = .853. The difference at post-test was marginally significant, F (1, 1319) = 3.27, p = .07, with an increase in positive attitude towards condom use among the intervention group students and a decrease in positive attitude among the comparison group students. After having followed WSWM, the intervention group students also scored significantly higher on self-efficacy using a condom than the comparison group, but this effect was primarily due to a marginally higher mean score on self-efficacy among the intervention group students at post-test as compared to pre-test, and a decrease in the mean score on self-efficacy among the comparison group students. At both pre-test (F (1, 1747) = 2.12, p = .146) as well as post-test (F (1, 1415) = 1.16, p = .282), there were no significant differences between intervention and comparison groups. Finally, the analysis showed an effect of intervention on the intention to use condoms. At pre-test, a significant difference was found between intervention and comparison groups, F (1, 1555) = 4.59, p = .032, with the intervention group being less inclined to use a condom than the comparison group. At post-test, students who followed WSWM showed an increase in intention to use a condom, whereas students from the comparison group had on average lower scores on intention to use a condom as compared to pre-test. Although the difference at post-test between the mean scores for intention to use a condom between the intervention and the comparison groups was not significant (F (1, 1275 = .749, p = .387), the time x group interaction was. This means that the intervention was relatively effective in increasing the intention to use a condom. No significant effects were found for past performance behaviour, or for perceived social norm towards condom use.
Non-coercive sex: The analysis showed a time x group interaction effect for self-efficacy in dealing with sexual coercion. At pre-test, there was no significant difference between intervention and comparison groups, F (1, 1817) = .168, p = .682. At post-test, both intervention group students and comparison students were more confident that they could deal with situations where sexual pressure and force would be used as compared to pre-test, but the increase in the mean score was significantly higher among the intervention group than in the comparison group, F (1, 1467) = 7.73, p =.006). No significant effects were found for past performance behaviour regarding avoiding and escaping risky situations, for attitudes towards sexual coercion, or for intention to deal with unwanted sex and force.
Completeness and Fidelity of implementation: additional analyses
As mentioned before, not all schools had implemented the 14 lessons at post-test (T1). The three intervention schools which had only implemented the first seven lessons (1-7) were deleted from this effect study. In order to determine the role of implementation rate (completeness), we included those schools that were originally dropped, back into the analyses. We then compared the scores of learners from these schools with learners in the intervention schools in the original analyses that implemented at least lesson 1 to 10 (21 intervention schools, 853 students), and those in the comparison schools, again using a multilevel approach to the analyses. This analysis revealed that all of the above mentioned significant effects disappeared for the schools that implemented less than 50% of the lessons. However, unexpectedly, we found two reversed effects for the schools which implemented no more than the first seven lessons. One on the beliefs towards HIV, where both the students in the comparison group as well as the intervention group scored lower at post-test (T1) than on pre-test (T0), but the intervention students significantly lower than the comparison group, F (1, 345)= 4.07, p < .05. Another reversed effect appeared for perceived social norm towards delaying sexual intercourse, F (1, 363) = 5.70, p < .05. Students in the comparison group agreed significantly more with the statement, "My friends believe that people my age should postpone sexual intercourse until they are older" than students from the schools that implemented only lesson 1-7.
We also explored the possible effects of the differences in the fidelity of programme implementation. Based on the results from a process evaluation conducted among 16 (out of the 21) intervention schools, we were able to distinguish between those schools that implemented the programme according to the manual (= full fidelity) and those schools that implemented the programme not totally according to the manual (= partial fidelity). As there were not enough schools which used computers for delivering the programme, we did not distinguish on the basis of computer use. All teachers, whether they would run the programme using computers or not, implemented the programme by using the WSWM manual. Three questions answered by teachers were used: (1) "Did you teach the lessons in the order that is stated in the WSWM manual" (yes = fully = 1; no = partial = 0); (2) "Of all the assignments/exercises that you taught, did you conduct them in the way they were described in the teacher manual (following the exact description)?" (Yes totally = fully = 1; Yes, partially; not sure; not really; not at all = partial = 0) and (3) "I have closely followed the WSWM manual when conducting the lessons" (Totally agree; agree = fully = 1; neither agree nor disagree; disagree and totally disagree = partial = 0).
Full fidelity schools (8 intervention schools; N = 344) were compared with partial fidelity schools (8 intervention schools; N = 325) in a multilevel mixed model design, with gender, age and pre-test measures as covariates. This follow-up analysis revealed that the partial fidelity group scored significantly better on beliefs concerning pregnancy and on knowledge concerning non-causes of pregnancy. At pre-test, the partial fidelity group scored lower than the full fidelity group concerning beliefs towards pregnancy, but not significantly lower. Both partial and full fidelity groups showed an increased mean score at post-test, but the increase among the partial fidelity group was significantly higher than the full fidelity group, F (1, 587) = 8.53, p < .01. Also for knowledge concerning non-causes of pregnancy, the partial fidelity group scored lower on average at pre-test than the full fidelity group. However, students in the partial fidelity group improved their knowledge significantly more than the full fidelity group at post-test, F (1, 514) = 8.51, p < .01.