If you want to do some research on any topic in the countries where there is more terrorism, decrease in the arrivals is mainly due to terrorism, due to which other variables results are biased. How to control it?
I think you should define terrorism first. That definition will give you the variables wich you need to measure.
TERRORISM: The UN General Assembly Resolution 51/210 defines terrorism as: "Criminal acts intended or calculated to provoke a state of terror in the general public, a group of persons or particular persons for political purposes are in any circumstance unjustifiable, whatever the considerations of a political, philosophical, ideological, racial, ethnic, religious or any other nature that may be invoked to justify them." Let's take this definition as an acceptable one.
METHODOLOGICAL DESIGN: The question asks for a design to measure the effect of terrorism on tourism. The followings are needed: (i) tourist arrival data from terror and non-terror condition in one country (if not possible then one needs a proxy country with non-terror condition); (ii) time period t1 to t2 for event measurement; and (iii) control groups must consists of both terror and non-terror condition countries outside of the country of study.
TIME SERIES APPROACH: If a country had time series that did not have terror condition and that during such a time there are data on tourism arrival available, that data should be collected. Then at a particular time, terror condition exists and (i) continue or (ii) dissipate and return to normality. This is an easy case, there is no need for a control group. If a condition comes and go like this, one can compare their two counts Poisson distribution.
A second scenario is no terror condition before, but at a particular time a terror condition was introduced and that condition continues. Here one would test for integration. Define terror condition as "shock" and the object of the test is to verify whether the effect of the shock is integrated to the series. If the shock is integrated into the series, the current mean would not revert to its long run equilibrium observed prior to the introduction of the terror condition. One can verify by introducing error correction mechanism (ECM); if after ECM is introduced, the data still does not revert to its long-rum equilibrium, it means that the integration of the effect of shock is permanent. the series has lost its original memory. The terror condition has changed the regime.
CONTROL: We need a country in the same region with similar market condition, except the terror condition. If we can find two countries outside of the subject country: (i) one country with similar infrastructure, but no terrorism; and (ii) another country with similar infrastructure, but has terror condition. Do paired comparison analysis to verify if terror is a significant factors for tourism arrival.
A = country of study with terror condition
B = country for comparison; has terror condition
C = country for comparison; has no terror condition
Country B is used to prevent Type 1 error. Country C is used to prevent Type 2 error. The pair design consists of: AB, AC and BC.
TEST STATISTICS: See attached link for statistical tests appropriate for the design described above. Cheers.
http://fcc-statistics.wikispaces.com/file/view/100+Statistical+Tests.pdf
I agree with Paul:
1. Use time-series data preferably over a sufficiently long period of time (prior to the terrorism threat/incidences). Preferably built around the tourism cycle most relevant to your research problem (i.e. 1 year)
2.Note the volatility (e.g.in arrival) due to seasonality effects, and "discount"/adjust it for the time-series data. This should provide the baseline series for comparison with the time-series after the terrorism threat.
3.Consider time-series over the same duration as in point 1 that is after the terrorism threat). Note the seasonality volatility. And assuming that your targeted tourism cycle is 1 year it would also be good to consider that particular year when the terrorism occur and how it affected the arrival/other variables
4. Compare and contrast the seasonality effects from the time-series discussed in point 2. with that of point 3.
5.You should be able to isolate/control for the "terrorism effect"
6. This is still a very rough and dirty approach, to fine-tune it perhaps consider Granger causality and co-integration tests. This is partly to control for "percolating factors". Do a Google search with the following query search string: "granger causality and cointegration tourism terrorism filetype:pdf"
7. Other considerations: example when post-terrorism policies design to mitigate the threat that actually augmented the arrivals, and other similar positive/negative feedback loops in the system. You may need to consider a systems approach to this.
I would say look at the research into this question at countries as a guide but of course you will find that most actually dont want to look into this seriously and actually cover up the issues as much as possible. In Thailand and the Philippines where i have done most of my research and where tourism is particularly important (especially in Thailand RE the Deep South being close tourist enclaves) the relationship between tourism and terrorism has been regularly 'ignored'. Your question is a little muddled but there are those of us looking at these two issues in certain countries.
- Badges
- Science topic
- Similar topics
- Geoscience
- Asia
- India