Abstract
The process of research entails collecting data and making meaningful conclusions from the data. Significant conclusions are influenced by the process of data analysis, which is in turn affected by the type of data collected. Therefore, it is important to decide the appropriateness of data to be collected with respect to the objectives of the study in the initial stages of research.
Introduction
Variables obtained from a research process provide useful information regarding the topic of interest (Creswell, 2009). Deciding on a method of accurately quantifying variables is a complicated task because it influences how the data obtained are analyzed (Green & Salkind, 2014). The instruments of measurements also require authentication to ascertain that they quantify the variables in question precisely. This paper takes into account the quantification of data and contrasts the conceptions of measurement validity to design validity.
Differentiating between the Levels of Measurement
Levels of measurement define the correlation among the values that are consigned to the features of variables. The four levels of measurement that exist are nominal, ordinal, interval and ratio (Frankfort-Nachmias & Nachmias, 2008). In nominal measurements, numerical figures label the attributes inimitably without considering any order. Gender (male or female) and race (Caucasian or Black) are examples of nominal data. Ordinal measurements may involve the ranking of attributes. However, the intervals between the ranks are meaningless. In interval measurements, the gaps between elements are meaningful. Ratio measurements, on the other hand, must contain an absolute zero that is significant.
An Example of Data That Can Be Transformed From One Level of Measurement to Another
Ratio and interval data can be collapsed down to the interval and nominal levels while data at the ordinal level can be transformed to nominal data. Overall, data at higher levels can be transformed to data at the lower levels (Rubin, 2012).
An Example of Data That Cannot be Transformed to a Different Level
Data at lower levels cannot be transformed to data at advanced levels of measurement. For instance, nominal data cannot be changed to any other data level. Similarly, ordinal data cannot undergo transformation to the interval or ratio levels.
Why Some Data Can Be Transformed and Some cannot be Transformed
Data at advanced levels can be transformed to data at lower levels because advanced data contain the elements that are present in data at the lower levels (Rubin, 2012). However, data at lower levels do not contain the attributes that are present in the advanced levels hence cannot be transformed to data at the advanced levels. For example, ratio data contain an absolute zero as well as information that is present in the nominal, ordinal and interval levels. Interval data also contain information that is present in the ordinal and nominal levels. Conversely, nominal data do not contain the attributes of ratio, ordinal and interval levels.
Comparing the Concept of Validity for Design to the Concept of Validity for Measurement
The validity of a measurement denotes the extent to which a measurement approach or tool thrives in quantifying or describing what it is intended to measure (Frankfort-Nachmias & Nachmias, 2008). Systematic and instrumentation errors determine the validity of measurements. Valid measurements must be acceptable to statistics experts. The items incorporated in the measurements also need to represent any probable questions adequately.
The validity of design, on the other hand, is an indication of the extent to which the chosen research design addresses the concepts to be studied, the appropriateness of the methods and the occurrences or processes to be examined (Wainer & Braun, 2013).
Conclusion
The four levels of measurement are vital to data collection and analyses because they determine the design of the research. Therefore, it is necessary to ensure that data collection is done at the right levels that match the statistical methods that one intends to use in analyzing the data.
References
Creswell, J. W. (2009). Research design: Qualitative, quantitative, and mixed methods approaches (Laureate Education, Inc., custom ed.). Thousand Oaks, CA: Sage Publications.
Frankfort-Nachmias, C. & Nachmias, D. (2008). Research methods in the social sciences (7th ed.). New York: Worth.
Green, S. B., & Salkind, N. J. (2014). Using SPSS for Windows and Macintosh: Analyzing and understanding data (7th ed.). Upper Saddle River, NJ: Pearson.
Rubin, A. (2012). Statistics for evidence-based practice and evaluation (3rd ed.). Belmont, CA: Brooks/Cole.
Wainer, H. & Braun, H. I. (2013). Test validity. Hillsdale, New Jersey: Routledge.