• No results found

CHAPTER 4: RESEARCH DESIGN AND METHODOLOGY

4.10 VALIDITY AND RELIABILITY OF RESULTS

Validity and reliability are measures that help researchers to determine whether their studies are worthy and and should be believed and used to solve an idenitifed problem. The ultimate aim of research is to provide solutions to a problem. Validity is measure of the extent to which the research instruments have helped the researcher to collect the intended information for a study. Reliability is a measure confidence in the study, determining whether the results can be relied on to solve an identified problem or not. In reliable studies, results are replicable or transferable. The following section explains the concepts of validity and reliability and how they were taken care of in this study.

4.10.1 Validity

According to Vanderstoep and Johnson (2008), Lodico et al. (2006) and Mukherji and Albon (2015), validity is a measure of the trustworthiness of a study. A study whose tools measure what it is intended to measure is valid. Though this study was not entirely quantitative, truthfulness of the measure (internal validity) was ascertained through the nature of responses from different respondents. Vanderstoep and Johnson (2008) explain that findings can meet external validity when the findings from an investigation can be generalised to other samples, populations or settings. The same results may still be found if the study were conducted in other parts of the country. The nature of similarity in responses collected from the different places provided an indication that the questions were clear. There was adequate representation of responses from all provinces. This was achieved through cross tabulations, and correlations that were run to confirm the trends of responses in different provinces against the understanding of certain concepts such as curriculum adaptation. Thus, construct validity was achieved. This is line with Creswell (2014a: n.p.) who said; “validity using the convergent approach should be based on establishing both quantitative validity (e.g. construct) and qualitative validity (e.g. triangulation) for each data base”. The sample sizes in

quantitative data were equal across the sampled sites, i.e. 40 respondents per province. The use of various methods equally ensured the validity of the qualitative component of this study. The responses that were provided met answered the research questions which focused on curriculum development.

4.10.2 Reliability

Reliability is an attribute of research which measures the consistency of data or research findings. This means such findings would be the same had the study been conducted somewhere else where there are similar respondent characteristics. Vanderstoep and Johnson (2008) define reliability as the extent to which a measure yields the same scores across different times, groups of people, or versions of the instrument. Ary, Jacobs, Sorensen & Razavieh (2010:236) “the reliability of a measuring instrument is the degree of consistency with which it measures whatever it is measuring. This quality is essential in any kind of measurement.” Instruments are tested for internal reliability in order to eliminate sources of error and ensure confidence in results. Since this study employed a mixed-methods design, both measures to test for reliability were considered. This study attained reliability through various tests.

 The pilot test of the instruments provided a litmus test of the nature of responses the study was going to bring forth. It was observed through the administered questionnaire that this study is reliable because it produced same responses as was seen in the pilot.

 The use of different instruments to collect data yielded same results. Questionnaires, interviews and teacher observations yielded consistent results. For instance, the challenges faced in the implementation of the curriculum, the understanding of curriculum adaptation and the status of teacher skills in teaching LSENs were established through all the three instruments.

 Even within each instrument, responses from different respondents in different places were similar. Ghosh (2015:244), says in a test for reliability, if “the same questionnaire can be tried on two similar samples and if the percentage of response are similar, the samples are to be regarded as reliable”.

 The data was collected at different times (time reliability) and in different places (geographical) but the results were consistent regardless of when and where the respondents were located.

 Cross tabulations in SPSS helped to cross-check certain variable relationships which were established by comparing provinces.

The questionnaires were tested for internal reliability using Crounbach’s alpha measurement. Johnson and Christensen (2012:142) explain the Cronbach alpha as a coefficient alpha providing, “a reliability estimate that can be thought of as the average of all possible split half correlations, corrected by the Spearman-Brown formula.” Since this study adopted a mixed methods design, the questionnaire had mixed questions. Only quantitative data scored on SPSS was subjected to the reliability test. Out of the 39 selected items tested, the stress inventory was (39 items; α = .773) which is higher than the accepted (0.7). This tells us that the instrument was reliable. “A value of Cronbach alpha size above 0.70 can be used as a reasonable test of scale reliability.” (Gaur & Gaur (2009: 134). The questionnaire was divided into four major categories. The table 4.4 shows the distribution of Cronbach internal consistence scores.

Table 4.4

Reliability scores of questionnaires

Category items # Cronbach alpha

Knowledge and involvement in CDP 8 0.585

Curriculum adaptation 5 0.590

Materials and strategies 13 0.798

Curriculum adaptation strategies 13 0.790

Inventory stress value 39 0.773

As can be observe in table 4.4, the inventory stress value gives the confidence of reliability in the research instruments used.

Basically, when mixed methods are used, validity is ascertained through triangulation or the use of multiple methods (David & Saeipoor 2016). As earlier alluded to, this study used methodological triangulation using the principles of complementarity, completeness, expansion, corroboration, compensation and diversity. These principles, each in its own way helped to ascertain the validity of data.