• No results found

Administration of the technique for non-clinical and clinical groups

The planned administration of the CEAT was the same for the non-clinical and the clinical groups. It would be conducted individually with students in school hours, in a private room provided by the school.

All children would be asked to confirm willingness to participate prior to commencing the administration. Any now declining would be excused from the study, and another participant found.

All stories would be recorded verbatim in writing by the researcher as the child spoke.

4.7 Planned data analyses

Data collected in the Pilot Phase would be transcribed using a word processing program, and then coded using the Coding Scheme developed in the Design Phase. With coding completed, statistical analyses of the results would be conducted. Following this, some qualitative analyses would be possible.

4.7.1 Matching of non-clinical and clinical samples and statistical analyses of Pilot Phase

Of the non-clinical sample of 20 boys and 20 girls 10 of each were to be selected to be match as closely as possible with the 10 boys in the clinical sample. The variables decided as critical for matching were, grade level, birth order, number of siblings, intact or broken family, and family income. The resulting selected non- clinical group of 10 boys and 10 girls would be included in the data analysis.

Statistical analyses of the Pilot Phase would comprise an investigation of the internal consistency of the CEAT, and analyses related to the hypotheses of the study.

4.7.1.1. Internal consistency of CEAT

It was anticipated that the internal consistency of the CEAT would not be able to be assessed by means of a parametric statistical technique. This was because the pilot use of the CEAT had generated data resulting from the coding of participants’ story responses that were obviously not normally distributed as required by parametric

tests, and the clinical sample to be recruited was unlikely to be large enough to compensate for this.

Therefore, it was decided to use the non-parametric method known as the Sign Test to evaluate internal consistency, first preparing the data in the way described below.

(a) First, it would be determined whether the four attachment style scores coded for a participant’s response to a picture were higher or lower than that participant’s scores coded for the other pictures. To achieve this, each individual’s responses were to be normed across their scores for each of the four attachment styles for each picture.

(b) Then, the normed scores were to be compared with the individual’s other coded scores, for each of the remaining pictures. When the comparison was found to be higher than the norm, a plus was to be awarded, and when the comparison was found to be lower than the norm, a minus was to be awarded. When the comparison generated identity, a zero was to be awarded.

The Sign Test would then be used to determine whether the total number of pluses and minuses were significantly different than what could be expected by chance.

Finally, it was planned to assess how the four attachment style scores of the 30 participants for each picture compared, in order to determine if any of the cards were biased toward producing scores associated with particular styles of attachment. Here again, the normed scores were to be compared, and the pluses or minuses awarded these comparisons were to be totaled by adding all the pluses and minuses for each of the four attachment styles across the 30 sets of responses for each picture. Then, the Sign Test was to be applied to determine if these totals were significantly higher or

lower than could be expected by chance. Data related to cards found to be biased could then be removed from the data sets prior to hypothesis testing.

4.7.1.2 Statistical analysis relating to hypotheses of the Pilot Phase

It was planned to test the two hypotheses in the Pilot Phase of the study that: (1) there would be no significant difference between the scores of the male (n=10) and

female (n=10) participants in the non-clinical group on security of attachment; and (2) the clinical group (n=10) would have a significantly lower mean score on security

of attachment than the non-clinical group (n=20).

Analysis of variance would be used to detect differences in the mean security of attachment scores of the three groups, namely the clinical boys, the non-clinical boys and the non-clinical girls. A contrast test Helmert would be used to determine the nature of differences emerging.

4.7.1.3 Exploratory analysis of data in Pilot Phase

It was planned to conduct a general exploratory analysis of the qualitative data collected in the Pilot Phase of the study that would involve:

(1) discernment of commonalities in stories belonging to the most frequent patterns of attachment quality revealed in the Pilot Phase;

(2) examination of differences in the responses of girls and boys and the three grade levels to the CEAT activity;

(3) evaluation of how suitable the activity is for the different age groups participating in the study; and

(4) evaluation of the technique’s clinical usefulness.

It was hoped that the process of the analysis would draw attention to issues related to the suitability of the technique for boys and girls in this age group, and perhaps bring to light indicators of attachment in middle childhood that had not been previously noted.