The Method of Constructing an Attitude Scale - Essay Example

Internal Validity Internal validity with regard to research design describes how effectively a study has been carried out and how much confidence canbe placed upon the conclusions derived solely because of the independent variables without having the effect of extraneous variables (Campbell & Stanley, 1966). Threats to Internal Validity Following are the threats to internal validity such that because of these threats, the confidence upon the results of the study hampers. 1. History History becomes a threat to internal validity, as there is likelihood of previous events posing impact in the existing experiments. 2. Maturation Maturation means that the changes in the dependent variable occur because of continuous lapse of time. 3. Statistical Regression Because of statistical regression, independent variables are regressed but their results tend to be closer to their mean posing a threat to validation of the results. 4. Testing Testing becomes a threat when the posttest results are affected by the pretest conditions and circumstances. 5. Instrumentation Instrument of a research design can pose significant threats to the internal validity in case of unreliability and non-objectivity of the data set. 6. Selection Selection becomes a threat when non-equivalent groups are included in the study such that their earlier or later absence in the study compromises the internal validity of the study. 7. Experimental Mortality Because of attrition, internal validity hampers because of the inclusion or dropping of some of the variables in the study. 8. Selection Interactions Selection becomes the threat to internal validity when it interacts with other threats to internal validity e.g. history, maturation, instrumentation, etc. External Validity External validity refers to the phenomenon in which generalization of the results of an existing study is applied to other contexts and people (Issac & Michael, 1971). Threats to External Validity Mainly there are four threats to external validity, which creates severe problems upon the generalization of the results, are given as follows: 1. Interaction This becomes the threat to the external validity because of selection and treatment of the subject. For instance, changes in the number of observations or changes in the number of variables. 2. Pretesting Pretesting becomes a threat when it can harm the results such that the results would have been different had there been no pretesting occurred. 3. Experimental Settings This can be a threat to external validity in case when the participants of the experiment are aware of the experiment as compared to the situation where participants are not aware of the same. The results might have been different. 4. Multiple Treatment Interference In case of using different treatments on the same subjects, it is quite difficult to generalize the results of both the treatments thus posing a threat to external validity. Difference between the Probability and Non-Probability Sampling Design The major difference between the probability and non-probability sampling design arises because of the use of random sampling under probability-based sampling (Gravetter & Wallanau, 2008). Probability sampling works on the principle random sampling such that every unit in a population has equal chance of being selected as a sampling unit. However, under non-probability sampling, sampling units are selected on any other basis but not based on selected randomly. Types of Probability Sampling Following are the types of probability sampling design described as under: 1. Simple Random Sampling In this type of random sampling, every unit has the equal chance of being selected and normally through computerized functions; samples are selected on purely random basis. 2. Stratified Random Sampling Under stratified random sampling, the whole population is split into homogenous subpopulation named as strata and then the random samples are picked out from each strata. 3. System Random Sampling Systematic random sampling involves the design of sample in which samples are drawn in some systematic manner. For instance, every fifth item from a number wise population should be selected as a sampling unit, etc. 4. Cluster Random Sampling Cluster random sampling divides the population based on geographical area and then the random samples are drawn from each cluster. Geographical area is the main distinguishing factor between cluster and stratified random sampling design. Non-Probability Sampling Any other method of drawing samples from a given population other than random sampling is referred to as non-probability sampling. Generally, there are two broader categories of non-probability sampling such as convenience sampling and purposive sampling. Purposive sampling itself has further five types and discussed in the subsequent paragraphs. Convenience Sampling Under convenience sampling, the researcher draws the sample as per his/her own convenience. For instance, researchers conducting social research use college or university students as their target samples because it is easy to collect their responses. In clinical researches, patients are the major respondents of the researcher because it is easy to obtain their responses. However, the biggest problem that lies with this method is that the samples drawn are not actually the true reflection of population and thus creates the threat to external validity of the research design. Purposive Sampling Purposive sampling mainly involves the selection of sampling units based on some purpose behind their selection. This method is the best method of sampling in case where the target population is of a specific type. So in order to select the samples from that population other non-relevant respondents are ignored and only relevant are considered. 1. Modal Instance Sampling Modal instance sampling is based upon the statistical concept of mode such that those sampling units are considered in a sample, which occur frequently, thus creating a “typical” sample of the most frequent respondents. 2. Expert Sampling Under this sampling, people with specific skills, expertise and experience are allowed to develop the sample so that the sampling units drawn should be based upon the most correct and rational logic. 3. Quota Sampling Quota sampling is that kind of purposive sampling in which a particular proportion or quota of sampling units is fixed to be drawn. For instance, in order to create a sample in which 20% of Chinese ethnicity are to be drawn from a population of mixed ethnic groups, then samplings units will be kept drawing until 20% of Chinese ethnic people are selected in the sample. 4. Heterogeneity Sampling This kind of sampling works on the principle of obtaining all the ideas, opinions, or views which are included in the given population. Unlike most frequent units, this sampling includes those sampling units, which are different or diverse from other sampling units in some form. 5. Snowball Sampling Snowball sampling is the one in which the respondents are included in the sample upon meeting a specific criteria. When a respondent is included in the sample, that respondent is asked by the researcher to refer another respondent who also meet the same criteria. This method of sampling is especially quite useful when it becomes quite hard to access the respondents in a population. Likert Scale Construction Likert Scaling is the method in which opinion or feeling of the respondent is reflected through some numbers (Likert, 1932). For instance, in response to the question “how important is the role of sports in your daily life?” the respondent can be given different degrees of such importance, which might be 1) Strongly Agree, 2) Agree, 3) Neither agree nor disagree, 4) Disagree, and 5) Strongly Disagree. Rating from 1 to 5 is given to each state and then the respondents are asked to select their state based upon their own opinions. Purposes of Research Following are the three main types of research: 1. Exploration The purpose of research holds the phenomenon of exploring new ideas, new facts, new dynamics, and originality of a new phenomenon. Based upon this research, further structured research areas are explored for future researches. 2. Description This purpose of research mainly describes the highlights of an existing phenomena, trends, patterns, performance in the past periods and similar information. 3. Explanation Explanation is constituted as one of the major purposes of research such that it provides the explanations and details of an existing phenomenon, situation, circumstance, event, etc. This purpose resolves the question of “why” and provides the rationale and justification of a particular phenomenon. Importance of Time in Research Time plays an important role in the research such that research studies can be bifurcated into two broader distinctions namely cross-sectional and longitudinal studies. Under cross-sectional studies, research being conducted covers a specific point in time and the whole research is subject to that particular point in time. On the other hand, longitudinal studies work on the basis of studying a particular phenomenon over a time period. In this way, the impact of time can be observed in the longitudinal studies such that different sorts of trends and patterns can be reflected as a behavior of subject under research. References Campbell, D. T., & Stanley, J. C. (1966). Experimental and quasi-experimental designs for research. Chicago: Rand McNally. Gravetter, F., and Wallnau, L. (2008).Statistics for the Behavioral Sciences. Belmont, CA: Cengage Learning. Isaac, S., & Michael, W. B. (1971). Handbook in research and evaluation. San Diego: EdITS. Likert, R. (1932). The method of constructing an attitude scale. Archives of Psychology, 140, 44-53. Read More