Methods and Challenges in Data Collection

1. FOREWARD Authors as Adams, Khan, Hafiz and Raeside (1), suggest some method for data collection, basing on the situation, warning from possible threats to the validity and reliability of data collected. Whatever the method of data collection chosen (observations, experimentation, survey, interviews, diary method, case study, data storage, triangulation), there are several hypothesis that need to be considered since the beginning (1); the challenges born from the nature of the research and level of detail the researcher want to reach, then by time and budget available, so careful consideration and planning of data collection is required.

There are some common principles, for examples try to eliminate as much as possible human errors, analyze all useful data instead of the only one which seems to fit in the theory, run multiple tests to check eventual errors. Collecting data is crucial in many different field of business interest, e. g. from concurrency evaluation to create a model for the estimation of pipe price, before to meet the supplier for the final negotiation.

For example, first strategy adopted from bid and proposal department, for the evaluation of piping price impact, is to evaluate raw material steel price and add a certain percentage which consider total cost of ownership. Second strategy can consider different elements which compose final price, starting from source of data instead of estimate a percentage only. This is one of the key elements: Bebell, O’Dwyer, Russel and Hoffmann (2) studied the importance of technology in the last past years to help researcher to evaluate and confute data availability and validity, for example triangulating the same data.

In any case, quantitative methods doesn’t contextualizes in the situation, considering for example the market situation, the human ability to concretize business relationship, … 2. CHALLENGES 3. 1 Source of data World is full of data and opinion, the advent of technology and internet allow to many users all over the world to get access to the web for those who have access, source of millions of articles, opinion, paper, studies, … According to Bebell, O’Dwyer, Russel and Hoffmann (2) the use of laptop and nternet by learners and scholars, in both cases resulted that about 50% or more use technology to make first research and to deliver instruction. The central IT organization in a statistical agency has a very important role in Web-based data collection, since data collection system has two very broad component – an electronic questionnaire, and everything else associated with moving that electronic questionnaire to and from a respondent, including systems and security considerations (3).

Since the best result is get if the questionnaire, interview, survey, … is focused as much as possible to the argument of research and to participant that well know the argument, source(s) of data, have to be identified since the beginning, possibly during the data collection planning stage. Doing this, the researcher optimizes his / her time, avoiding to source data time per time is need. Researcher has to avoid interpretation and misunderstanding in the question, in order to get invalid responses.

This imply that for example, the questionnaires received, if duly filled, may not be very useful because don’t meet the requirements, otherwise, target of the research cannot be reached. Infact rate of response can results too low so unacceptable, and potentially people can decide to not respond since they don’t know about the question. Initial investment of the time to plan the job, avoid creating questionnaires inefficient to the researcher. When we face to questionnaires which don’t know what’s talking about, the first reaction is to leave it blanks or give confused answers.

For these reasons, random sampling techniques, stratified random sampling techniques integrating with pre-test, are crucial in order to avoid eventual fairness, big enemy of the study, even if the researcher has to consider that a pre-test may sensitize or polarize the person’s behavior and consequently, better performance on the post-test. Some methods for avoiding this issues, will be analyzed in the next chapter “strategies” 3. 2 Characteristics of collected data The target of the researcher is to get the data as objective as possible and the best response rate, not only in terms of numbers but as much representative as possible (2).

It means that collects objective data, makes it stronger and unassailable the research, and open to any new research or alternative solutions. Some examples of objective criteria could be: * Market value * Scientific findings * Efficiency of the model * Professional standards defined * Equal treatment * Tradition * Legal (court) * Reasonableness Collecting the right data, allows the researcher to get representative answers which help to find a solution to the problem that he / she places, otherwise the study can be compromise since the beginning, or can drive the researcher to solution not representative of reality.

For example, company can decide to capture data of saving from a certain database characterized by having certain accuracy, i. e. two decimal places; at the end of analysis, the researcher have to know that the result is affected by a certain error value. Infact, even if minimal error is occasionally acceptable, in some cases can lead to unacceptable inaccuracy or even to the failure of the project. For this, determine the level of tolerated error is need during the collection of quantitative data. Techniques and devices for the quantitative collection have to be characterized by a certain tolerable range of error. 3. 3 Data collection

Two main different categories can be considered: primary (data not available by previous research, …) and secondary (data are available elsewhere). In both cases, when we’re collecting quantitative data, it is often tempting to record and use only which results that correspond to priori test, experiments or theory, especially when the expected results are so different from the ones got. However, could happen that especially these unexpected data shown problems with the experimental procedures, so these values should not be ignored. Last but not least, assertiveness of the researcher avoid to influence the questionnaire or data search.

For example, supplier A has quoted 100 and supplier B = 70, C = 72, D = 68 for the same identical package. Technical evaluation has been done for all; it means that, the same package has more or less 40% of difference in price compared than A. It may seems an anomaly, in most of the cases that is since one supplier is trying to getting much money, but a careful analysis can lead to evaluate that B and C quoted very low at the beginning, in order to get the PO, foreseeing to recover later on adding some parts, reaching or going over price of A. 3. 4 Cost and time

Data collection process can requires observation of the research phenomenon, over than time for collection, surveys, … This particularly happen in the longitudinal studies, where data have to be analyzed at different time. Nevertheless, changes can occur in the subjects during the observation period, so they can be influenced. Cost can limit the data acquisition phase, limiting the collection and right type of data need to conduct the research. As the size increases, variability decreases. Moreover instrumentation with right accuracy, basing on the accuracy target level of the research, can be a limit for the research. . STRATEGIES TO OVERCOME 4. 5 Maintain original data Reliability and validity can be proved, without manipulation, and maintain the opportunity eventually to examine again, reinforcing the conclusions. It means that, since the best and quick results are gain through computer, memory disk should be necessary to store the data. Other reason is that longer is a study, higher is the possibility that historical data are necessary since the time tends to change the conditions. Moreover, pre-test need, when done, need to be stored. 4. 6 Pre-test

They can influence the subjects, so post-test different from pre-test can avoid this effect. Multiple independent trials minimize error when collecting quantitative data, asking to distinct group to run the test or experiments aimed at collecting specific quantitative data. These 2 groups can compare the results, which should be the same. 4. 7 Clear and easy data blank document In order to avoid low rate of response, it has to be easy to use and clear, in English language or the language of the subjects, allowing the participants to give informative and accurate.

Over this, the blank is to be simple and quickly to be filled, otherwise participants can be discouraged. 4. 8 Double check source and people for data collection When data collection is delegated to other people or relies to the use of internet, the collection is by other people. For example, company which get information through surveys under payment, it’s a very high quality and quantity way to complete surveys, but need to be analyzed whose responder are really working on the answer or are interested to get the reward only.

Temptation to manipulate data to enhance results is common; when happens, the validity of the research becomes doubt. For sure most of the times mistakes are unwanted, and the response need to be identified. One way to solve this problem should be solved using technology (2). For instance, software can help to create an average, classify and evaluate which are completely out of average and why, since they could be representative of the survey or due to the low knowledge of the responders, collect all the evaluable data finding eventual correlation between the variables.

In conclusion, find the middle way in optimizing the additional cost and reduction of time thanks to technology, is a concrete challenge for the researcher which would share his / her research to others, since research designed to solve problems in medium – long terms, rather than short terms, is increasingly required in today’s business environment. REFERENCES 1) Adams, John; Khan, Hafiz T A; Raeside, Robert (2007) “Research Methods for Graduate Business and Social Science Students.

Sage India” 2) Damian Bebell, Laura M. O’Dwyer, Michael Russell, Tom Hoffmann – 2010 Concerns, Considerations, and New Ideas for Data Collection and Research in Educational Technology Studies 3) Richard W. Swartz and Charles Hancock – 2002 Data collection through web-based Technology 4) Reetta Raitoharju1, Eeva Heiro2, Ranjan Kini3, and Martin D’Cruz – 2009 Challenges of multicultural data collection and analysis: experiences from the health information system research