Table 6. RPE assigned level rows versus SRTB total score Cols
RPE | SRTA level |score
9–11 12–14 15–17 18–20 21–23 24– 26
27–29 30–32 33–35 36–38 39–41 42–44
0+ 0 0 0 0 1
1 0
1 1
1+ 0 0 0 1 1
2 0 0 0 0 1
2 1 1
0 0 2+ 0
1 2
2
0 0 0 0 0 3
0 0 0 0 2 1 1
1 2 3+ 0
2 1 4 1
4 0 1 5 3 1
44+ 0
1
0 0 4+
1 1
2 3
Here again we see the spread. Anyone with a score of 30–35 on final form SRT A or B could be RPE level 2 to 4.5.
In general terms, we could derive some “cut-off” scores for RPE as we did for SLOPE. However, because of the variable nature of the RPE assigned levels in this study, it is difficult to estimate what the error would be in
trying to generalise to any new population being assessed using the SRTs.
4.4 Relationship Between Preliminary and Final Form SRTs
Thirty-one subjects who took the Preliminary SRT also took the two Final Form SRTs SRT A and SRT B. That is, these subjects repeated the same two sets of sentences. The first time these sentences were embedded in the
Preliminary SRT, yielding the extracted scores for those sentences. The second time these sentences comprised the final form SRTs A and B, yielding the final form scores. In this section we investigate whether or not the
extracted scores taken from the Preliminary SRT are significantly different from the scores on the same sentences in final forms A and B. To compare the extracted scores with the scores on the final forms we
performed t-tests for paired samples. The results are as follows: FORM
A FORM B
Mean Extracted scores = 28.8 Mean Extracted scores = 27.8
Mean Form A SRT = 30.6
Mean Form B SRT = 30.9
Difference =
1.8 Difference
= 3.1
St error
= .55
St error
= .72
t = -3.31
t = -4.22
p = .0024
p = .0002
n = 31
n = 31
For both forms SRT A and SRT B the subjects performed significantly better p.01 on the final forms than on the same sentences in the Preliminary SRT. This could be due to one of several factors:
1. a learning effect - the subjects do better on the Final Forms because they have learned or become familiar with the sentences;
2. increased familiarity with the test methodology; or 3. there is less fatigue, and therefore concentration is better, with fifteen sentences than with sixty-three
sentences. If the effect is due not to experience having already carried out the first round test, but due to reduced strain
because of a shorter set of sentences, the suggestion is that calibrations based on extracted scores may not be entirely valid when the SRT tool is used on the field. A subject’s L2 proficiency level may be overestimated.
4.5 Further Analyses
Further analyses are presented and discussed in appendices. Appendix 7 discusses the use of correlation and regression analyses in this study. Appendix 8 comments on alternative approaches to selecting final form SRT
sentence sets.
5 Conclusions and
Recommendations
5.1 Conclusions From This Study