Methodology
Processing the results of the survey of respondents is conducted in parallel in three stages. The primary handling of questionnaires implies checking that they have been completed correctly. The main criteria of this stage are completeness, accuracy, and uniformity of the answers provided. In the case of detection of faulty completed online questionnaires, these results were not taken into account. Survey secondary processing includes a statistical examination of the data set by building joint histograms of answers distribution, tables, and diagrams for specific categories of the sample: women, men, pensioners, and young people.
The use of mathematical tools for quantitative research allows for the systematization and streamlining of data dispersion. For this reason, tertiary data processing is done through correlation analysis. In particular, the correlation coefficient was determined to test hypotheses, as well as to find the relationship between the parameters sought. The method consists in assigning ordinal numbers to each answer in the questionnaire and searching for relations between the answers of respondents and expected results. The data are processed using statistical programs to implement correlation analysis or in Excel spreadsheets. The correlation coefficient identified is compared with reference values to determine the nature of the relationship, after which the conclusion about the reliability of the hypotheses put forward is made.
Results
The results obtained after statistical processing include charts, histograms, and tables showing different dependencies between input parameters. First of all, histograms of answers distribution averaged for all participants of the questionnaire allow to identification primary trends for quick analysis. Pie charts will show the proportionate distribution of respondents’ physiological and social characteristics: gender, age, social status, and employment. It is worth noting that such primitive means of displaying results can lead to unexpected conclusions, therefore, special attention should be paid to finding hidden relationships. Tables can include a demonstration of the frequency of displaying specific answers or a comparison of the most and least popular elections.
The final results are a calculated correlation coefficient and a graph showing the location of points and the linear trend. First of all, the emphasis should be placed on providing reliable correlation analysis data: correlation coefficient r and sampling error probability, which can be conveniently displayed in the form of a table. The scatter plot between the data should contain the linear trend and all the points in the processing.
Discussion
Interpretation of the results allows the conclusion of the reliability of three hypotheses put forward about the interrelation of parameters. If the statistical analysis shows a rather high correlation coefficient, it can be argued with a certain degree of accuracy that there is a strong correlation, and therefore the policies proposed by the organization are effective (Franzese & Iuliano, 2019). In particular, if the answer to question #6 of the survey shows a high level of confidence in clients’ choice of services, the clinic should introduce this measure. Conversely, if tertiary results management shows a low correlation rate, it may not be cost-effective to introduce new services at a lower cost. In this case, the company should either abandon the idea of scale the survey in case of a sampling error. Unexpectedly, the correlation coefficient may show a negative value. Then the correlation between the parameters is inverse to what the organizers of the research expected.
It is essential to say that several factors limit the research. In particular, the survey was conducted only for the products of the Revenue at DeClerck Family Dental company, so there is no guarantee that all the revealed regularities will be relevant for other organizations. In addition, only correlation analysis was used, although it demonstrates the correlation, but does not have a high degree of statistical significance. Qualitative research methods, such as interviews, are not sufficient to study the individual characteristics of respondents. Future developments of this problem should be aimed at introducing a regression model of analysis that reflects the influence of factors on the shift in response truth. The introduction of new services and changes in dental policy should precede sample scaling if a weak correlation is detected.
Reference
Franzese, M., & Iuliano, A. (2019). Correlation analysis. In S. Ranganathan, M. Gribskov, K. Nakai, & C. Schönbach (Eds.), Encyclopedia of bioinformatics and computational biology (pp. 706-721). Elsevier.