The Role of Validity, Relevance and Generalizability - Coursework Example

 THE ROLE OF VALIDITY, RELEVANCE AND GENERALISABILITY Introduction Validity, reliability and generalizability of both quantitative and qualitative data play a crucial role in all the steps involved in the research process from formulation of the research question all through findings and recommendations[Dav02]. The validity, reliability and generalizability of data play an explicit part and an integral role in the research process[Cre00]. Reliability is basically encompassed on the precision of the measurement. Validity is based on how truthful a measurement is while generalizability is defined as the extent to which the findings may be applicable to different settings from the original. Validity, reliability and generalizability are important because they give any measure credibility. Quantitative data basically refers to data expressed in numerical form. It ranges from age to height of individuals and it is measured by the use of ratio, ordinal and interval scales. However, qualitative data cannot be expressed in numerical form and is measured by the use of a nominal scale. A good example of qualitative data is gender and complexion. The aim of this paper is to critically investigate the role that reliability, validity and generalizability plays in both the quantitative and qualitative data. Validity of Quantitative and Qualitative Data Validity in both quantitative and qualitative is encompassed on how truthful the data collected is. However, there is a great difference in terms of validity in quantitative and qualitative data. To begin with, in quantitative research validity is based on the objectivity and neutrality of the data collected. It is also encompassed on predictability of the data and how random the samples collected are[Lec00]. The forms of validity of data include construct, face, predictive, concurrent and theoretical among many others. Validity tells us more on how the instrument used for research is used to enable researchers achieve the research’s objective. Validity in quantitative data determines whether the true intention of a research is achieved and the level of the truthfulness of the results. In most cases researchers usually estimate the validity of quantitative data by asking a range of questions and in also search for the solutions in the research carried out by others. Validity in quantitative data is based on whether the instrument strikes the “bull’s eye” of the research object[Sma02]. Some research describes the role of validity in qualitative research as being what kind of data is to be collected and how it is to be collected. They also state that most quantitative researchers affect the relationship between the data and construct in order to make their investigation valid. Therefore, their involvement during their research process greatly affects the validity of the data collected. Validity in quantitative data can be established by triangulation, a persistent and prolonged observation, samples conducted theoretically and by making comparisons[Mos00]. However, the validity of quantitative data can be threatened by failure or the inability to fully describe the independent variables and the unreliability of the instruments used for measuring. However in qualitative research validity of data is encompassed in the uniqueness, holism and uniqueness of the data. Validity in qualitative data is mainly concerned with the richness, scope, the feeling strength, honest and the subjectivity of the data[Min03]. Validity in qualitative data is described broadly. To begin with, validity in qualitative data is a “contingent construct”. There have been concerns that in qualitative data there are no concerns of validity and therefore it lacks applicability. However, some qualitative researchers agree to some form of qualification for their research. This is basically the need for a measure in regards to their research. According to [Cre00], the validity of qualitative data is determined by how the researchers perceive it and the choice of the assumption in the model or paradigm[Ahr06]. Due to this reason, the researchers have created their understanding and concepts of validity as generated by their own terms of validity. According to them validity in qualitative data plays the role of trustworthiness of the data, its rigor and finally the quality of the data[Rol06]. Therefore, most researchers perceive the validity of qualitative data in terms of quality and this determine whether the research is proper. Validity in qualitative data is determined by an assessment of explanations and searching for all the possible invalidity sources among other ways. Reliability of Quantitative and Qualitative Data Reliability is encompassed on the accuracy of the data to the extent that the data is reliable in a way that it is useful. Reliability in quantitative research is viewed in the following types. To begin with, there is reliability viewed as being stable. In this form there is consistency of the data over the samples and duration which in our case is represented by time. To continue, reliability is viewed in terms of equivalence and in this form there is inter-rater reliability which basically means the number of estimates that are consistent among the diverse researchers or the observers[Coz01]. There is also reliability being viewed in terms of its consistency internally. In this category we have the split half reliability where the correlation between two sets determines the reliability of the quantitative data. The test retest reliability occurs when the same test is conducted to individuals twice. This method brings out stability. This is because if the measure is stable then the results from the quantitative data should be similar. As stability increases then the reliability increases too and this means that the results can be repeated. According to [Jop00] there exists a problem in the method of test-retest because the instrument is made unreliable. This is due to the fact that method impels the responses which are given. The differing responses are mainly brought about by the changes in attitudes. It is thus the role of the researcher to ensure high levels of consistency in tests as they greatly affect reliability. Due to the vast role that reliability plays on ensuring there is reliability in the quantitative data, there are ways of improving reliability[Jea03]. To begin with, it is important that all the external sources that might affect the variation of data be greatly minimized. The conditions under which the quantitative data measurement takes place should be greatly standardized and the outliers should be excluded. However, reliability in qualitative data is viewed on different perspectives. It is based on the credibility, applicability, transferability, neutrality, consistency, conformability and the trustworthiness of data. Reliability is used in most cases in all the research types. The most crucial test of reliability of qualitative data is its quality. Great quality enables individuals to understand confusing situations. Quality in quantitative data enables individuals is utilized for explanations while in qualitative data it is used for understanding. This difference between qualitative and quantitative data in terms of the purposes of quality makes reliability to become totally not useful in qualitative data. According to [Mos00], reliability in qualitative data is misleading. Any qualitative research discussed in terms of reliability results to bad consequences. However, some researcher view that reliability plays a great role in qualitative data as it is one of the factors that are always considered while planning a study, doing an analysis of the results and making a judgment on the quality of the data collected. The quality of work in a qualitative data is determined by the criteria of relevance placed in it. Therefore, the criterion for quality is based on credibility, consistency and neutrality among other factors. The dependability of a qualitative research is emphasized by an inquiry audit[Cre00]. An examination of consistency and quality of qualitative data determine its relevance. Trustworthiness is very crucial in examining the relevance of qualitative data. In the determination of the quality of the qualitative data one of the key issues is the trustworthiness of the data. However, due to the fact that reliability of qualitative data is concerned with measurements then it is not relevant in qualitative research. The utilization of quality in qualitative data results to consequences that are not good. Therefore, reliability results from validity of the qualitative data. Reliability in qualitative data is dealt with mainly through triangulation. Triangulation is encompassed in reducing biasness and increasing certainty of information gathered thus making it relevance. Triangulation involves tackling biasness, integrating perspectives, validity and multiple theories. The qualitative research process ranges from the collection of data to the analysis. In the qualitative research process, each step is encompassed by neutrality[Cox05]. The true position of the research is determined by triangulation. Triangulation is based on more than one approach. To begin with, there exists triangulation of methods. Here the main focus is the consistency from the findings that result from the diverse methods of collecting data. This in turn enables the researchers of the qualitative data carry out a meaningful analysis on their findings. Furthermore, there is source triangulation where the consistency of diverse data is determined from the similar data collection methods. Qualitative researchers collect data from different points of time but the approach of data collection is similar. There is diversity as information is collected from different individuals. To continue, there is the analyst triangulation whereby a range of analysts do the analysis of data. This enables the qualitative researchers to analyze biasness of the observers and the analysts in areas of analysis that deal with interpretation[Min03]. The aim of triangulation is to have a holistic picture of the data collected and how it affects the different phenomenon. Lastly, there exists the perspective triangulation; this involves the interpretation of the data by the use of a wide range of theoretical perspectives. Generalizability in Quantitative and Qualitative Data Generalizability is used to refer to how the sample information can be used to make conclusions about the whole population. The information is not limited to the sample alone. The results of their research can be applied to a general study. The main concerns of generalizability mainly deal with if the study can be applied to wide outlook rather than been confined to a specific place[Gar05]. A research that is valid to only one area cannot be used for generalization purposes. The settings and the characteristics of a specific study determine whether or not the information can be generalized to the whole population. In quantitative data, it is always crucial to determine if the findings can be applied to other situations. In carrying out quantitative research it is always important that the sample is representative of the whole population. Random sampling is important as collection of data from the whole population is not only impossible but also very costly[Hen09]. Sampling is always more accurate and precise as compared to drawing information from the whole population. Random sampling is also easier as it is more practical to draw information from a sample as opposed to drawing information from the whole population. Random sampling is also efficient and it forms the basis of inferential statistics that basically deals with making conclusions about the whole population based on the information provided by the sample statistics. Random sampling is also utilized by researchers to avoid any cases of biasness during the process of data collection. For generalizability to be important it is always crucial that the picked sample be a representative of the whole population[Nan01]. Generalizability is however divided into various types even though they are not always applicable in quantitative research. To begin with, there is inductive generalization. This form of generalization is where diverse types of random sampling are utilized with the intention of acting as a representative of a population that is seen to be too large. The basis of generalization is the ability of the sample to act as a representative. A quantitative research is usually combined with a random sample that is representative. A quantitative research is therefore used to aid in the identification of whether the sample findings can be generalized to the whole population. It enables researchers to also investigate the cases that are unique or even extreme. However, there exist challenges in the discovery of a great sample statistic representation. This is due the fact that some information may not be known to the researchers. This thus raises questions on whether the information is representative of the whole population[Ben04]. Generalizations based on statistics are not always the best. For example, a researcher cannot generalize from one case to another especially when the both form part of a sample representative in which both cases are involved. The generalizability of a quantitative research is always unknown in cases where the population possesses heterogeneous characteristics and the population elements are unknown. There has to be a provision for allowances due to the heterogeneity of the population[McK04]. This means that generalization from one case to another is based on guessing. Some statistical researcher recommends that sample representation be based on replication reasoning as opposed to sample reasoning. This replication reasoning is mostly utilized in experimental quantitative research. This is because in experimental research there exist seldom measures that ensure the conditions under which the experiment and the control take place are totally representative of the whole population. However, sometimes there exists replication in the experiment and this is always with other subjects. The replication reasoning suggest that the researcher not only makes a selection of cases where they expect the obtained results are a repetition of previous studies made, but also cases in which the researcher expects that the results will contradict the previously obtained results[Gor12]. Therefore, the subsequent studies enable the researchers to affirm or even falsify the hypotheses thereby resulting to the development of a theory. The generated theory then becomes the basis of generalization to the cases that have not yet been studied by quantitative researchers. This now forms analytical generalization. The unstudied case then forms what we call a theory because they are yet to be proven. Therefore, it is indeed true that there exists no generalization from a sample to the whole population. However, there exists generalization from a case to another that is in the scope of the involved theory. This is therefore referred to as theoretical generalization. The theory functions as the driving force to this form of generalization. Furthermore, the variation based generalization is based on variations. It is a good method of generalization. If a study is aimed at investigating the level of unemployment in any given economy then a recruitment of individuals from that economy is made by researchers so that they can represent the population. The condition for this recruitment is that the chosen individuals should be members of the population. The selected individual should be selected randomly based on their age, level of education and gender among other factors. The findings from this experiment should be in consistent with the population. In qualitative data, as opposed to the quantitative view of generalizability being based on the representation of the sample to the population, it refers to the level to which the findings from the study can aid researchers in gaining an understanding of similar situations in order to increase the level in which they represent the targeted population. Qualitative research is always aimed at enabling the researchers to understand similar individuals possessing the same characteristics. As opposed to quantitative research method of generalizability, the qualitative method of generalizability rarely utilized random samples because it is always aimed at increasing the sample representation. Qualitative research always makes way for the justification of the findings for individuals that do not belong to the original initial study. Details on the kind of individuals and the nature of the settings are provided in the initial study and this illustrates how the theory made is applicable on the identification of same individuals and initial settings of the study. According to [Tob04], there is a shortage in qualitative research and it is referred to as anecdotalism and this acts as a threat to generalizability in qualitative research findings. A good example of anecdotalism is a situation where qualitative researchers may find cases that interest them such as events and people that they would like to explore in great depths and details so as to form great stories. In this case, the qualitative researchers are prone to being blinded by such an anecdote of wanting to form great stories. Being prone to an anecdote is rather a rare but not a typical case. Therefore, the results of an anecdote include biasness as the researchers find it really complex and impossible to make a generalization of the findings discovered to other individuals or settings. For this reason, it is crucial to make an argument that qualitative researchers need to make use of clarity on how the cases can be typical so as to avoid problems to do with generalization. The quantitative researchers may utilize the total count number or percentages so as to state the level of how typical a given phenomenon is going to be. Nevertheless, even with the specifications being made, the sample is still not always randomly selected from a given population. For this reason the findings made by the qualitative researchers cannot be generalized to the whole larger and targeted population. This in turn creates inefficiencies. Conclusion In conclusion, reliability, validity and generalizability play a great role in both quantitative and qualitative research. They enable researchers to draw meaningful conclusions and they affect the accuracy of information. They act as control mechanisms against offenses like biasness and lack of objectivity of the researchers. However, the three phenomenons possess different meanings and implications for the different data in our case qualitative and quantitative data. Reliability in both qualitative and quantitative data plays a role in determining the truthfulness of the presented information. Validity on the other hand is centered on the accuracy of the findings while generalization is for the purposes of inferring. This discussion thus offers a great platform for researchers to carry out useful research and make accurate and quality findings that are comparable. Certain specific strategies are to be applied on both qualitative and quantitative data in determination of credible findings and reports. References Dav02: , (Davies & Dodd, 2002, pp. 3-5), Cre00: , (Creswell & Miller, 2000), Lec00: , (Lecompe, 2000), Sma02: , (Smaling, 2002), Mos00: , (Moskal & Leydens, 2000), Min03: , (Miner, 2003), Cre00: , (Creswell & Miller, 2000), Ahr06: , (Ahrens & Chapman, 2006), Rol06: , (Rolfe, 2006), Coz01: , (Cozby, 2001), Jop00: , (Joppe, 2000), Jea03: , (Jeansky, 2003), Mos00: , (Moskal & Leydens, 2000), Cox05: , (Coxon, 2005), Gar05: , (Garher, 2005), Hen09: , (Henry, 2009), Nan01: , (Nancy, 2001), Ben04: , (Benson, 2004), McK04: , (McKee, 2004), Gor12: , (Gordon, 2012), Tob04: , (Tobin & Begley, 2004), Read More