2. A Few Keywords and Definitions for Understanding Statistics – V (contd.)

-

- by Dr. Prafulla Dikshit          (3 minutes read)

2.9    Reliability and Validity 



All the measures discussed in the previous posts vary in the extent to which they are at a risk or have a possibility of being biased or yielding faulty results. However, there are ways to determine and to some extent ensure that the bias or errors in the measures may be minimized. This helps ensure the effectiveness of the measures. One such means to determine the effectiveness of the measures you are using is to assess their Reliability, which is the consistency or the stability of the measuring tool or instrument. It is also important to point out the distinction between the property or phenomenon, being measured, its measure, and the measuring tool at this juncture. The distinction between the property or quantity being measured is largely self-explanatory as that between the measured and the measurement. However, the distinction between the measures and the measuring tools is inconsistent and varies between the different measures. At a conceptual level, a measure is a system developed to measure the property or quantity, whereas the tool or the instrument is the physical or tangible aspect of it that interacts and records the property in terms of the defined measures. The tool has a measuring interface on one hand and an observer or user interface on the other. Mostly, the measures and the measuring instruments as explained in the previous posts are distinct from each other. However, in many cases such as the psychological scales, there is a high degree of overlap between the measure and the tool. In others, where the tool or the instrument is a physical and tangible object, this distinction is much clearer, for example in the case of a thermometer, wherein the temperature is a physical measure, measured through the thermometer, which is a clear and distinct measuring tool. The error in the measuring tool, however, reflects not just the error in the tool but may represent a bias in the entire system or measure in the way it is defined or deployed.

The errors discussed so far make the measurement less reliable and lead to inaccuracy and wrong judgment about entities, properties, or phenomena. However, what if the tools or measures are applied erroneously and tend to measure something other than what is intended to be measured or what it claims to measure? This raises the question of whether the measure is valid or truthful and or Validity of the results. As such reliability and validity are distinct concepts but may occur in an overlapping manner in nature. Thus, data and results may be valid but not reliable and vice versa; they may be both reliable and valid, or they be neither reliable, nor valid (See Figure 1).

Figure 1

Reliability versus Validity in Statistics

 Both reliability and validity are critical statistical concepts from a research point of view, and there are several different types of reliability and validity. Thus, we will discuss the same in detail in future posts. However, it suffices to introduce them here for us to get familiar with the basic concepts. 

                                                                         ****  

 


Comments

Post a Comment

Popular posts from this blog

2. A Few Keywords and Definitions for Understanding Statistics – II (contd.)

-
Image
 by Dr. Prafulla Dikshit     (6-8 minutes read)      2.6   Types of Data Now that we know something about data and statistics, let’s talk a little about the types of data. There are many different categorizations of data, which we will eventually study. However, at the most basic level, we can see data in terms of the characteristics we are trying to measure through the data. Thus, a broad categorization of the attributes of the population or the sample would give us different types of data. In statistics, there are specific terms for these characteristics. The characteristics of a population denote parameters , whereas the corresponding characteristics of a sample drawn from the given population are called statistics . Ideally, they may be measuring the same - basic attributes or properties, but their values will differ by the source (population or sample) from which the data for the same are drawn. Parameters are numbers that summa...
Read more

2. A Few Keywords and Definitions for Understanding Statistics – VI (contd.)

-
Image
                2.10         Research Methods     In the last post, we touched upon the concepts of reliability and validity and how they help determine the effectiveness of the measures we use in statistics. We are now inching closer to finishing basic know-how of terms and keywords to get acquainted with the field of statistics. We have now arrived at a point where we are transitioning from the 'what' and 'why' to the 'how' of statistics while exploring the basic ideas and principles to get a feel of statistics.    In this context, there is an important distinction we need to make between how we implement all these basic ideas and principles toward our research goals. This has to do with the methods of research that tie the ideas discussed thus far into applicable knowledge. These make the most of statistical learning as related to research applications, however, c...
Read more

2. A Few Keywords and Definitions for Understanding Statistics – III (contd.)

-
Image
(6 minutes read)     - By Dr. Prafulla Dikshit 2.7   Four Levels of Measurement So far, we have seen how the continuity versus disjointedness of data points, can be the greatest differentiator between different types of data. We can also look at types of data from a measurement level perspective to get a very relevant view of data types. There are broadly four levels of measurement:             2.7.1  Nominal – A nominal level of data measurement is at the most a categorization. Importantly, the categories are not ordered . For example, if 'place of birth' is a variable we are interested in, the data we collect is nominal, since we may categorize the participants by place of birth, but we can't order it to say that this one, say London, or that, like Tokyo, is the best 'place of birth' and then another one, and then another and so on. Similarly, we can categorize people by race or color of skin but c...
Read more