Answer

Measurement Strategies: Data Collection Methods

Good evening colleagues and professors.

My group was assigned the topic ‘Uncontrolled Asthma in Children’ but we narrowed it down to a single objective which was to compare the effectiveness of asthma and self-management education to pregnant women and use of medications in the prevention of hospital admissions of asthmatic children.

Through our literature search, we identified two research articles, both of which used randomized clinical trials as the study design. The first article is Tseng et al. (2017). A article is “Management based on exhaled nitric oxide levels adjusted for atopy reduces asthma exacerbations in children: a dual centre randomized controlled trial” by Petsky et al. (2015). In this discussion of measurement strategies, I have chosen to review the first one; A randomized–controlled trial of a lay-educator inpatient asthma education program’. I have cited this in the reference section. We chose these articles because they are most relevant to the objective of our group. Both were looking at management of asthma among adolescents. Both also used RCT since comparison of groups was involved, similar to the objective of our own group work. 

I will not give you a brief background to the article I will be talking about in this discussion. The article reports a two-arm RCT. In the two-arm randomized controlled study reported in the article by Tseng, Chang and Wu (2017), the authors’ objective was to examine the extent to which a self-management program was effective in controlling asthma among adolescents in Taiwan (Tseng, Chang, & Wu, 2017).

Lets now looks at the methods used, and specifically talk about how data was collected, i.e the measurement strategies. A total of 112 youths aged 12-18 years of age were randomly assigned in to either control of intervention arms of the study following an identified recruitment criteria. A before-and-after survey was used to collect data from the participants. Self-administered questionnaires, which measured demographic variables alongside other study variables, were administered at the beginning of the study (baseline) and after a period of four weeks. The variables measured included gender, whether or not there was smoking in the family, how severe the asthma was and how many years it had been since asthma diagnosis. Additionally, an outcome expectancy index, asthma self-efficacy index, asthma self-management index and asthma control index were used to measure these variables (Tseng, Chang, & Wu, 2017).

Levels of Measurement

Depending on the research question, variables can be measured in three levels namely; nominal, ordinal and metric. Gender, whether or not there was smoking in the family and how severe the asthma was, were measured as a nominal variable. The number of years since asthma diagnosis was a metric variable. The indexes used Likert scales to measure outcome variables such as self efficacy. These data can be analyzed as metric data.

Measurement Levels and Statistical Analysis

The choice of statistical analysis techniques is dependent on the level of measurement of a given variable. Nominal and ordinal variables will often requires statistical techniques which are different from the ones which would apply to metric variables. For instance, in describing metric variables, means, standard deviation, variance and range are used. On the other hand, percentages and frequencies are used to describe categorical variables. Inferential statistics also apply to the level of variable measurement. Chi square is applied for categorical variables while correlation, t tests and ANOVA are used when the dependent variable is metric.

Validity and Reliability of Data Collection Methods

A data collection method is considered valid if it measures what it intend to measure. Reliability refers to repeatability. It measures the extent to which a data collection method/tool can give the same result when it is implemented repeatedly. (Kimberlin & Winetrstein, 2008). Self-administered questionnaires are known to have high reliability but their validity is not as high. Likert scales are considered to have high reliability and reliability scores because they have been tested against Cronbach’s alpha values (Tseng, Chang, & Wu, 2017; Lin, Mayer, & Lee, 2019).