Overview
It is theorised that the beliefs that the MSOCS measure are relatively stable over time. It is partly the hypothesised temporal stability of these beliefs that enables them to have an influence on health over time. It was therefore necessary to demonstrate the stability of the MSOCS over time.
This study was conducted to investigate the temporal stability of the MSOCS and used a test-retest design. While it was hypothesised that the subscales would be relatively stable over time, it was also expected that there would be some variation between the subscales because the beliefs concern very different 'means' and the nature of these 'means' will lead to different degrees of change. For example, beliefs regarding the availability of money are likely to be less stable than beliefs regarding ability. As the MSOCS follows a similar approach to control beliefs as the SPOCQ (Wellborn et al., 1989) and the CAM! (Skinner et al., 1988) the temporal stability of the subscales in these measures is worth considering. The test-retest reliability of the SPOCQ over an 8-week interval ranged from .39 to .64. It is understandable that some of these values are low given that the participants were children. It is expected that higher temporal reliability will be found for the subscales of the MSOCS.
Method
Participants
The participants in this study were an opportunity sample of undergraduate students. These students were from Massey University (Palmerston North) and were completing a second year psychology course in research methods. Participation in the exercise was voluntary. Forty-two students completed the questionnaire on both occasions. The participants had an age range of 19 -44 and a mean age of 23.58 (median = 20.(0).
Thirty-one of the participants were female while 1 1 were male. Details of the ethnicity and highest educational qualification are contained in Table 6.6. The majority of the
participants identified themselves as NZ EuropeanlPakeha (59.5%) while 1 6.7 percent identified themselves as NZ Maori. Most participants had a higher school qualification
such as University Entrance (83.2%) while only 7 . 1 % had a university degree, certificate
or diploma and none of the participants had a trade certificate or professional certificate or diploma.
Table 6.6
Ethnicity and Highest Educational Qualification
Number of Percentage respondents of Ethnicity NZ Maori Asian NZ EuropeanlPakeha Pacific Islander Other 7 3 25 2 5 Highest educational qualification
No school qualification 3
School certificate passes 4
Higher school qualifications 32
Trade certificate or Professional certificate or 0
diploma
University degree, diploma, or certificate 3
Post graduate qualification 0
Questionnaire respondents 1 6.7 7 . 1 59.5 4.8 1 1 .9 7. 1 9.5 76.2 o 7. 1 o
The questionnaire included the items from the MSOCS which were chosen for further testing at the end of the EFA (see Chapter Five) and questions concerning age, gender, ethnic group, and highest educational qualification. A copy of the questionnaire is attached in Appendix Ten.
Procedure
The purpose of the study and what the study involved was explained to the students during class. Students who wished to be involved with the study completed the questionnaire immediately during class time. The questionnaire was completed twice with a six-week interval between administrations.
Results
Missing Data
82.3% of the items had no missing data. The items that had missing data had only small amounts with the most data missing for any one variable only being 5.7%. Most people
(69%) missed no items. On average each person missed only .66% of the items with a range of 0-6% items missed. EM imputation was used to replace the missing data.
Subscale Scores
Mean subscale scores were calculated using the items included in the CFA analysis and the sample means and standard deviations are shown in Table 6.7.
Assumptions
The assumptions of normality and linearity were evaluated through SPSS 10.0
DISCRIPTIVES. Normality was assessed through the examination of skewness and kurtosis statistics and histograms. Many of the variables were non-normal and
transformations were performed to normalise the distribution of responses. Analyses were performed with the transformed and untransformed data and there was very little
difference between the two sets of results. Consequently, the results reported are those for the untransformed data because these are easier to interpret.
Table 6.7
Means and Standard Deviations of MSOCS Subscales in the Test-Retest Analysis Mean Standard Deviation
Ability Capacity 1 3 . 1 6 .54 Ability Strategy 1 3 . 1 1 .38 Family Capacity 1 3.70 .39 Family Strategy 1 3.21 .46 Friends Capacity 1 3.34 .46 Friends Strategy 1 2.40 .45 Luck Capacity 1 3.43 .7 1 Luck Strategy 1 1 .65 .54 Money Capacity 1 2.37 .3 1 Money Strategy 1 2.43 .50 Ability Capacity 2 3. 1 1 .64 Ability Strategy 2 3 . 1 2 .43 Family Capacity 2 3.70 .4 1 Family Strategy 2 3 . 1 7 .54 Friends Capacity 2 3.27 .61 Friends Strategy 2 2.35 .54 Luck Capacity 2 3.50 .75 Luck Strategy 2 1 .83 .48 Money Capacity 2 2.46 .35 Money Strategy 2 2.43 .55
The scores from the first data collection can be identified by the number 1 after the subscale name and the scores from the second data collection can be identified by the number 2 after the subscale name.
Table 6.8 shows the test-retest reliability of the MSOCS subscales. The test-retest reliability for the subscales varied markedly from Money Capacity (.49) to Ability Capacity (.87).
Table 6.8
Test-Retest Reliability for MSOCS Subscales
Subscale Ability Capacity .87 Ability Strategy .53 Family Capacity .63 Family Strategy .72 Friends Capacity .78 Friends Strategy .67 Luck Capacity .65 Luck Strategy .67 Money Capacity .49 Money Strategy .59 Control .68
Discussion
The results show that, as expected, most of the subscales have reasonable stability over a six-week interval. The two sub scales which had the worst temporal reliability were Ability Strategy and Money Capacity. The relatively low reliability of Money Capacity is
understandable given that the participants were students. Money Capacity essentially measures whether the person believes that they have access to the money they need. It
would be expected that, with the nature of student finances, there would be change over a period of six weeks. The low temporal stability of Ability Strategy is less understandable and does cause some concern. It may be that the participation in academic activities changes students' belief in the role of ability to achieve their goals. As the sample was quite young, it would be expected that the stability coefficients would be higher with a more representative sample which included older people. Students are at a time of change
and their attitudes and beliefs may be relatively unstable. However, the responses were
more stable than those reported for the SPOCQ (Skinner et al., 1988).
It is recognised that the sample used in this study is small and is not representative of the general population. Another limitation is that the time interval between administrations is only six weeks. The mechanisms outlined in Chapter Three through which control beliefs are associated with health are likely to operate over extended time periods. Consequently, it would be expected that control beliefs would need to be fairly stable over extended periods of time. In the future it would be ideal to test the temporal stability of the subscales again with a larger more representative sample with a more extended interval between administrations.
However, although there are these limitations, the study does give some indication of the temporal stability of the measure and does suggest that with this sample, over a six-week period, the beliefs measured by the MSOCS are fairly stable.
Summary
This chapter has presented analyses used in the final development and assessment of the properties of the MSOCS. The fust analysis was a confirmatory factor analysis which compared the hypothesised model to alternative theoretical models. This analysis showed that the hypothesised model fits the data better than any of the alternative models which supported the hypothesised structure of the measure. However, the model fit was only marginal and as a result of efforts to improve the model two items were removed. Analysis of the internal reliability of the subscales showed that the resulting subscales had
acceptable internal reliability. The third analysis looked at the relationship of the MSOCS subscales to themselves and to a number of other similar and dissimilar psychological measures. It was found that the relationships were of the size and direction that would be expected given the theoretical understanding of the measures. Lastly, the temporal stability of the subscales over six weeks was explored. This showed that the subscales generally had good temporal stability over this period of time. From the analyses
performed in this chapter it was concluded that while improvements could be made to the MSOCS in the future, it would be suitable for exploratory analyses.