CREDIBILITY OF THE RESEARCH

that guided the completion of the questionnaire and the design and layout were revised several times until an acceptable quality was attained in this study.

5.5.4.3 Interpretability

Cooper and Schindler (2011:286) further explain that interpretability should be given attention if someone other than the designer must interpret the results and or purpose of the study. In order to overcome this practical problem, the researcher submitted a cover letter detailing the problem being investigated and the objectives of the study with every questionnaire.

2011:280; Leedy & Ormrod, 2014:103). Whereas external validity concerns itself more with generalizability of research findings across persons, settings, times, et cetera, the internal validity of a research study is the extent to which its design and the data it yields allow the researcher to draw accurate conclusions about cause-and-effect and other relationships within the data (Cooper & Schindler, 2011; Leedy & Ormrod, 2014).

While both forms of validity were relevant to the current study, internal validity was more relevant and also being mindful that the current study was descriptive and exploratory in nature. The three broad forms of internal validity (content validity, criterion validity and construct validity) as offered by various authors (Cooper &

Schindler, 2011:280; Zikmund et al., 2013:304; Leedy & Ormrod, 2014:91) are discussed next.

5.6.1.1 Ensuring content validity

Content validity of an instrument refers to the extent to which the measuring instrument adequately covers the domain of interest (Cooper & Schindler, 2011:281; Zikmund et al., 2013:304). An evaluation of content validity (face validity) begins with identifying the constituents of the concept being measured or the instrument being representative of the sample of the content area (domain) being measured (Leedy & Ormrod, 2010:92). For the current study, it was necessary to identify the elements from literature that constitute social networking, innovation, competitiveness and firm performance. The specific element of every construct (for social networking see section 2.9, innovation- section 4.9.1, competitiveness- section 4.9.2 and firm performance- section 4.9.3) under investigation was used to compile investigative questions in the questionnaire such that the questions sufficiently covered the whole spectrum of the concepts and constructs to comply with Siddiek’s (2010:137) criteria for content validity.

5.6.1.2 Ensuring criterion validity

Criterion validity demonstrates the success of an instrument in either predicting or estimating outcomes including practicality of the measure (Bryman & Bell, 2011:165;

Cooper & Schindler, 2011:281; Zikmund et al., 2013:304) or whether it is meaningfully related to some other measure of behaviour (Goodwin, 2010:132). Thus the measure should be able to predict with some degree of confidence the existence of the relationship(s) between two or more score on a future criterion. Since the current study is descriptive and claims no causality of variables, predictive validity of the questionnaire was therefore not an issue and was not pursued by the current study.

5.6.1.3 Ensuring construct validity

Construct validity refers to the degree to which the researcher measures what he/she intends to measure and to which proper identification of the independent and dependent variables were included in the investigation (Hair et al., 2008: 338).

Goodwin (2010:132) contends that construct validity considers whether the test adequately measures some construct, and whether it connects directly with what is now a familiar concept.

Dzansi (2006:86) also contends that construct validity is the degree to which a test measures an intended hypothetical construct. To enable proper measurement of the various constructs of the study a detailed literature review was conducted and the constructs, social capital, entrepreneurship, innovation, competitiveness and firm performance, were operationally defined in Chapters 2;3 and 4 respectively. Also, as there are several existing instruments for measuring the various constructs under study, the measurement instrument used in the current study was largely developed from literature, with a few adjustments to the existing ones.

5.6.2 RELIABILITY

The reliability of a measurement instrument is determined by the extent to which it produces consistent results if applied at different times (Bryman & Bell, 2011:164;

Cooper & Schindler, 2011:283; Kumar, 2011:181; Zikmund et al., 2013:305). Internal consistency of an instrument demonstrates the extent to which different indicators of a concept converge on a common meaning. It shows the extent of homogeneity among the different items of a multi-item measurement instrument. Values above 0.80 are deemed to possess excellent consistency, between 0.70 and 0.80 to be good; and between 0.60 and 0.70 are considered fair. Values below 0.60 are considered poor (Bryman & Bell, 2011:158; Cooper & Schindler, 2011:283; Zikmund et al., 2013:305).

After testing the questionnaire amongst 60 ECFs, the researcher slightly adapted the phrasing of some questionnaire items in the instrument in order to improve the understanding of the items.

The current study used Cronbach’s Alpha value to examine the internal consistency of the instrument. The questionnaire’s reliability statistics for the constructs built in this study are summarised in Table 5.2 below. A Chronbach’s Alpha value of at least 0.700 would indicate a reliable research instrument or a construct within the instrument.

Table 5.2 shows that some constructs are not very reliable (Chronbach’s Alpha statistic <0.700) while others are very reliable (Chronbach’s Alpha statistic =0.700).

The questionnaire as a whole was found to be highly reliable (Chronbach’s Alpha statistic=0.904). The reliabilities indices of the constructs are shown and comments, indicating if such a construct is reliable, are also included in the table.

5.7 POPULATION, SAMPLING FRAME AND SAMPLING