英语论文网

to be able to recognize the ways that validity can be tested (Cohen et al. list eighteen different types which is probably not exhaustive), which tests are most pertinent (or can be applied to any particular study), what should be examined and the significance of the degree of validity in any particular test.

The first concern I have is the validity of the sample size in my study. Charles and Mertler (2003, p.162) draw attention to the issue that:

‘If too few respondents are interviewed, the data are not likely to represent the population. Investigators hope to interview at least 30 respondents…’

My study has a small number of respondents but, as I mentioned previously, includes the whole sample population.

The internal trustworthiness of this study is hard to evaluate but I feel that the results will not immediately be generalisable without further studies. Certainly, predictive validity may be used to confirm any relationships drawn from the interviews if it were clear that any of the relationships found were measurable and therefore possible to create and test a predictive model for compatibility of individuals in similar teams.

Construct validity has a place in this study as it can assess the validity of the interview schedule and conduct. Patton (1987) stresses interviews must be done in a consistent and neutral manner while Dillon et al. (1994) believe that successful interviewers must adhere to six rules that include interviewer attitude (avoiding condescension and remaining objective), question structure (avoiding yes/no questions; being direct and informal) as well as the situation (an atmosphere that encourages free responses). However, trying to follow the advice is easy. Realising when these guidelines are broken may not be so obvious to an interviewer and interviewer variability could be considerable. (By having built-in to the schedule some interviewer training I hope to have minimised this.)

I have already mentioned using the co-principals to examine the construct of the interview design but they can also comment on the interview process, transcripts, analysis of the data and also any conclusions arising from the study (a type of respondent validity).

I have tried to remove as much bias from being both investigator and subject of the study although I may still bring bias to the interviews (such as to make the co-principalship seem more successful and therefore reflect upon me more positively) or if the findings show distinct deficits working to distance myself from the situation. Kelly (2004) expresses this clearly when she says intimate knowledge, ‘can at times be confusing, marred by questions of ethics and one’s own selfish interests, threatened by politics and the unknowability of social facts, and compromised by efforts to remain both engaged and neutral.’

When studying within my own community as a senior member of staff subjects should be perceived as being vulnerable. Axinn & Pearce (2006, pp.108-109) describe a number of ways that a subject may ‘want to present him or herself more positively to the researcher,’ and cites a number of studies that indicate that both face-to-face interview situations and ‘mail-in’ survey responses have similar problems with ‘social desirability’. I have recently left the school and the subjects were all at the same hierarchical position but there is still the p