Introduction
The study problem in this project is to investigate the relationship between asthma and the body mass index. Following the effects and the symptoms of the disease among other chronic lung diseases, there will be an investigation of the global reports about asthma, dividing the affected population into age groups and body weights. The research will then investigate how the spread of the aspects of asthma manifests across the age groups. Among the parameters that will be used in the investigation are the episodes or the attacks of asthma, ranging from mild to chronic conditions. The study is based on the report by the US in 2007, which indicated that approximately 7 % of the American population tested positive during the national asthma diagnosis program (Tersa, 2005). The association or otherwise lack of it will be able to explain the growing number of asthma cases from the time of the diagnosis.
Study design
There are various possible options of study design that can be applied to this study to establish the relationship between the two variables. One of the options is to conduct a qualitative interview among the target population, focusing on the vital parameters. The second option is to conduct a qualitative questionnaire with open ended questions that allow the respondents to provide genuine and flexible descriptions. In the questionnaire and the interview methods, the structure and content must take care of the ethical values that the society upholds to avoid legal challenges arising from gross violation of privacy rights (Troy & Beringer, 2006). The third option is to conduct a systematic observation from the affected population groups. The optimal design of the study is the use of questionnaires, since the nature of the research requires the consent of individual respondents in form of writing. From the questionnaire, the responses will be the guideline for analysis, data interpretation and representation to derive accurate conclusions.
Justification of the design choice
Out of the three research designs, the optimal method is the use of a questionnaire, by virtue of the consent requirement and the documented evidence of data source in writing. The questionnaire is a flexible tool that provides confidentiality of the respondents. It enables the respondent to express responses that could otherwise be difficult to express verbally, as it is in the use of qualitative interviews. Observation does not apply in such a case since it leads to a collection of questions and prejudiced information. Again, observation provides room for assumptions that may mislead the research process. In the case of interviews, it is difficult to collect accurate information since most of the respondents may not be willing to co-operate. Others either exaggerate information or deliberately decide to conceal information or to provide false information on sensitive matters. Using questionnaires, the design can be optimized using appropriate statistical measures such as linear correlation analysis, multiple variable regression analysis, T – test analysis and the analysis of variance (ANOVA). The analysis of the relationship between the two variables will show whether they are in direct proportionality or inverse proportionality. If the relationship is direct, then there is an association between the two variables, otherwise, there is no association.
Statistical measures
In this study, the optimal choice for statistical measure of the association is the bivariate correlation analysis. However, we can use a combination of more than one of the measures, for example, bivariate correlation analysis and T-Test or Bivariate analysis together with ANOVA. To illustrate the use of the statistical measures, we use a sample data out of the previous studies on asthma cases in the US (Miller, 2006). The table below shows the data for people with different weights and the numbers that were diagnosed with asthma.
We then run a combination of linear regression and Spearman’s correlation analysis of the data on the STATA system. Below are the results:
Linear regression analysis results
Spearman’s rank correlation results
- Below is the summary of the results for the Spearman’s rank correlation
- Spearman weight number affected, stats (obs p)
- Number of obs = 13
- Spearman’s rho = -0.7363
- Test of Ho: Weight and the number affected are independent
- Prob > |t| = 0.0041
- Spearman weight number affected, stats (obs p)
- Number of obs = 13
- Spearman’s rho = -0.7363
- Test of Ho: Weight and the number affected are independent
- Prob > |t| = 0.0041
The results show that the linear regression coefficient is – 0.8804. At the same time, the coefficient of Spearman’s correlation analysis is – 0.7633. The two design models show a common feature of negative or inverse relationship between the two variables. In the present study, the same method will be followed to ascertain whether the two variables are related or not.
Subject Selection
The decision on the subject area to address depends on the objective of the study, which in this case is to study the relationship between the two variables; the body mass index and the rate of asthma spread. Other factors may possibly emerge with the power to alter the main subject or increase the number of key issues to address. This is subject to the nature of the findings from the data that the questionnaire will collect.
Measurement Issues
Certain issues may arise, relating to both the measurement, the exposure and the results of the data analysis. Kish (2004) states that such issues may include out of range variances, which may cause large standard errors in the study. If we for instance expect the number of people with weights above 100 Kilograms to be 12 and the data shows the number as 2000, then this is a wide difference that may provide inconsistent results. Secondly, there are biases that may be noticed in the display of the output that may expose the inconsistent entries of parameters at the data capture point. It implies that the study may then have to deal with biases and errors which tend to expose the results to irregularity.
Potential biases that the study might face
This research exercise may have to deal with several biases that it may face. What is apparent is the possibility of missing out a certain possible threshold that may affect the results in tremendous ways. For instance, in the analysis of the statistical measures above, the maximum weight covered was 120 Kilograms with the assumption that there was the largest mass index possible. It is a faulty assumption that may cause irregularity since it is possible to find people among the target populations with weights above 120 Kilograms.
How to handle the biases
To solve the challenge of biases, there is need to set a provision for errors for both positive biases and negative biases. It means that the presentation of the calculations should be indicating whether there is a bias or not, and the measures of the availability of biases and errors.
Possible confounding factors
One of the confounding factors has already been addressed as the mass indexes beyond the set threshold. It certainly has an effect on the results since the outliers will be inevitable to appear. In order to overcome the effects of the confounding factors, there is need to make the ranges and the threshold flexible such that adjustments can be possible whenever the data collected exceeds the threshold. For instance, in the example used in this study, it can be possible to adjust the maximum weight to 140 instead of 120.
Effect modifiers
From the calculation of the biases, standard errors and the various measures, there are chances of observing irregularities. There is a range of errors or biases that are expected to occur in the study. However, it is possible that some errors and biases may be beyond the expected limits. For example, if the correlation coefficient shows a negative correlation and the linear regression analysis shows a positive coefficient, then there is an ambiguity in the interpretation of such associations between the variables. It thus reflects the presence of a modifier or an outlier in the inputs.
Overcoming the effect modifiers
This method is able to overcome the effects of modifiers, by conducting critical checks on the outliers in the data capture in order to identify the causes of irregular data presentation. It is possible that just a single item could have been captured with wrong figures. This according to Casella (2008) may require the process of confirmation of the original data gathered from the respondents.
References
Casella, G. (2008). Statistical Design. Boston: Springer.
Kish, L. (2004). Statistical Design for Research. New York: John Wiley & Sons.
Miller, A. P. (2006). New Developments in Asthma Research. New York: Nova Publishers.
Tersa, M. (2005). Trends In Asthma Research. New York: Nova Publishers.
Troy, B. D., & Beringer, P. (2006). Remington: The Science And Practice Of Pharmacy. New York: Lippincott Williams & Wilkins.