• No results found

3.2.1 Choosing on the depended variables

The case selection strategy is based on selecting cases based on the dependent variables, which is the existence of an innovation system in the selected cases. All chosen cases were investigated based on the occurrence of the phenomena’s outcomes (innovation clusters on different industrials focus in each country). The main reason was that we could not infer by selecting cases only on the independent variables, such as universities, financial institutions, government research institutions, and industries. Investigating the independent variables were the phenomena do not occur (innovation clusters) is a waste of time and effort because the reason for the investigated phenomena may not exist.

Another important reason for the difference between the three cases is to increase the validity of the results and to support the findings. Based on the mill method of a difference the chosen countries are India, South Korea, and Saudi Arabia (Mill, 1847).

All three countries’ innovation systems are different in their outcomes (industrial specialization).

- Outcomes difference:

In the 1990s, India decided to be the world hub of information technology. To achieve this goal, India concentrated its activities and effort to develop its institutions to cope with this objective (India, New Delhi: Planning Commission, 2013). South Korea chose to take a different path and decided to be the world high technological hub in the early 1990s. To achieve this goal, South Korea’s efforts were concentrated to develop the institutional and industrial infrastructure that supported this objective (OECD, 2016).

Saudi Arabia also followed a different path and decided to increase its effort and be a pioneering country in oil and petrochemical production (Lempinen, 2011). The country’s effort was concentrated in developing this industry and built a robust infrastructure and institutions that helped to achieve this goal.

- Administrative difference:

India has been considered as a federal country where each state has its unique power over its territory (India, New Delhi: Planning Commission, 2013). In South Korea, the case was different in that it was ruled by a dictatorship until the late 1970s when it was transformed to be a republic country (Mahlich, 2007). Saudi Arabia is different in that it is ruled as a complete monarchy country were central decision-making is important to support the economy. This difference shows that even if there were a variation in the administrative level the policies to support the innovation systems and to build a Regional Innovation System would have commonalities.

- Population difference:

India is different than the other countries that it has more than 1billion inhabitant (India, New Delhi: Planning Commission, 2013). South Korea population is way below India, South Korea’s population is around 51 million, which will give the dissertation richness in dealing with such conditions (OECD, 2016). Finally, Saudi Arabia population is around 30 million, which varies significantly from the other two cases.

3.2.2 Addressing selection bias

Selection bias is one of the major problems of inference. Selection bias is defined as

“systematic error that arises either when cases are selected according to an

unrepresentative sampling rule, or when some (often unknown) nonrandom process assigns cases to cases” (Brady, Henry E., Collier, David, 2004).

In their book Designing Social Inquiry, King et al. (1994), describes two major types bias involved in selecting cases. The first bias occurs when selecting cases on the dependent variable and the second bias rise upon selection on the explanatory variable. The major problem with selecting cases on the dependent variable is that variation is trimmed where observations do not include the full range of variation possible in a particular case. Thus,

“any selection rule correlated with the dependent variable attenuates estimates of the causal effect on average” (Achen, 1986). In qualitative research, this means that the true causal effect may be larger than that estimated by the scholar. However, King et al. (1994), claim that bias resulting from the selection of the dependent variables can be mitigated by:

o Avoiding cases where there is no natural variation in the dependent variable. These circumstances do not support causal inference because the result (the dependent variable) will be the same even if the cases and the independent variable vary. These have been succeeded by choosing three different cases in different countries and

different industrial focus. India focuses on IT industry; South Korea focuses on Heavy industry and High technology; and the Saudi industry focuses on Oil.

o Preparing a list of alternate cases that encompasses similar circumstances as the selected cases to use when insufficient data may threat the validity of findings.

Other cases that can be investigated that have a list of successful are China such as Hong Kong Innovation cluster and Taiwan Hsinchu Science Park (Chen, Wu, &

Lin, 2006).

o Preparing lists of alternative data sources to use in the possibility that restricted access to some data may affect the dependent variable. Alternative data sources are available from the OECD and the UN data and research investigation.

The most different case selection strategy can be biased when the full range of variables is not representative. However, the process of causal analysis begins by identifying a universe of possible observations to theses variables. In our case, the observations are policies that affect our independent variables. These policies range from economic, organizations, education, innovation, taxes and incentives anything that can lead to the successful creation of RIC. Some common yardsticks then measure the effectiveness of these policies such as codes (Abbott, 2004). Thus, unsuccessful coding is one of the most threatening factors in the most different case selection strategy.