Internal and External Validity in Clinical - Research Paper Example

Before a study can be conducted, its methods should be planned and organized. One of the considerations in making research proposals is the sampling plan. The sampling frame to be used should be both appropriate and feasible. Similarly, the sampling method should be a balance of usability and validity. Sampling Frame Sampling frame is defined as the list from which the sample of the study is selected. Broadly, it can either be population registers (PR), electoral registers (ER) or healthcare registers (HR). Direct effect on internal and external validity It should be able to account for all members of the population for the study to be accurately descriptive of the population. Aside from determining external validity, sampling frame also affects internal validity. Selection bias can occur due to absence of certain specific groups of the population being studied from the sampling frame (Wolf, Kuulasmaa, Tolonen, Sans, Molarius, & Eastwood, 2005). For example, if a study claims to measure the prevalence of stroke among hypertensive individuals, then recruiting only those patients admitted to a hospital would probably result to a higher value. It is important that both the inclusion and exclusion criteria are appropriate to the population addressed (Dans, Dans & Silvestre, 2008). Effect on response rate The response rate, which is also vital in the validity of the study, may also be determined by the chosen sampling frame. The most updated PR can provide more accurate information about the current contact details and addresses of the sampling frame. In fact, according to the study of Wolf, Kuulasmaa, Tolonen, Sans, Molarius, & Eastwood (2005), which investigated the relationship between sampling frame type and response rates to assess the impact of non-response bias, differences in contact rate of the samples are mainly due to the quality of the sampling frame, such that the more updated PR provides the highest proportion of samples that can be contacted. Low contact rates are caused by the persons having died or moved away from the address indicated in the sampling frame. Sampling frame: the most important part of sampling plan Although other components, such as sampling method, are important as well, the sampling frame is the most important part of the sampling plan, because even if the best sampling method, randomization, (Kowakczyk, n. d.) will be used, it may still fail to sample for certain subgroups of the population, depending on whether the sampling frame to be used is complete or not. Sampling Methods Broadly, random sampling, non-random sampling and repeat sampling are the types of sampling methods. This section discusses each of them on the basis of their functions, methods and disadvantages. Random sampling This can either be simple, systematic, stratified, cluster, random walk, or staged sampling. It is the random selection of the samples from the sampling frame, either by drawing out pieces of paper, written in which are numbers corresponding to items on the frame (simple), through sampling every nth item of the list (systematic), by randomizing from a subgroup of the frame (stratified), through randomizing the subgroup, all members of which will be obtained data from (subgroup), using specific random instructions where to go to sample people (random walk), or by sampling within samples (staged). Which to use depends on the availability of a sampling frame (if unavailable, use random walk), and number of items in the sampling frame (large numbers favor systematic, stratified, and cluster). Despite being regarded as best way of sampling, randomization may not be feasible in a population that is geographically dispersed. For those using subgroup or cluster, care must be taken not to exclude other subgroups that are important (Mayoux, 2009). In addition, one cannot randomize people to a certain exposure, especially if the outcome to be studied is negative or harmful (Dans, Dans & Slivestre, 2008). Non-random sampling Examples of non-random sampling are quota, purposive, chain sampling or snowballing, and genealogy-based sample. This can be done by preferably sampling more in particular subgroups, depending on issues investigated (quota or purposive, if diversity is the objective), or through sampling from the contacts (chain sampling) or families (genealogy-based) of other samples. Since the sampling is made easier by being limited to certain subgroups or network, they are sometimes referred to as convenience sampling. Because subgroups of the population are consciously missed by using these sampling methods, the results of these studies may be biased (Mayoux, 2009). Indeed, validity is greatly compromised if a large percentage of those sampled are not part of the population in question, or a significant proportion only represents a particular subgroup. In such case, the findings will not accurately describe the population supposedly studied. However, it should be remembered that validity is not an absolute quality, but a continuum. Even if non-random sampling is used, it does not mean that the study is not internally valid anymore (Taylor & Asmundson, 2008). If bias is minimized by sampling for as much subjects as possible to increase the chances of obtaining a set of sample that represents the population being studied, then the validity of the study is greatly improved. In fact, sampling error is equal to the inverse of the square root of sample size. Thus, a greater sample size can greatly minimize any inaccuracy of the results from the true value that should be caused by sampling. Repeat sampling methods Finally, repeat sampling method is the use of the same sample of people used in a previous study. This is used to follow through processes over time (Mayoux, 2009). Summary: The Effect of Bad Sampling Plan In summary, for a study to be externally valid, the sample should represent the population. To achieve this, the sampling plan should allow all members of the population to have a positive and quantifiable probability of being sampled. Furthermore, data should be obtainable from these sampled individuals (Wolf, Kuulasmaa, Tolonen, Sans, Molarius, & Eastwood, 2005). If the sample does not represent the population in question, both internal and external validity will be compromised, making its results insignificant and inapplicable. References Dans, A. L., Dans, L. F., & Silvestre, M. A. A. (2008). Painless Evidence-Based Medicine. England: John Wiley and Sons, Ltd. Taylor, S. & Asmundson, G. J. G. (2008). Internal and external validity in clinical research. In D. McKay (Ed.), Handbook of Research Methods in Abnormal and Clinical Psychology. California: SAGE Publications, Inc. Wolf, H. K., Kuulasmaa, K., Tolonen, H., Sans, S., Molarius, A. & Eastwood, B. J. (2005). Effect of sampling frames on response rates in the WHO MONICA risk factor surveys. European Journal of Epidemiology, 20, 293-299. Mayoux, L. (2009, Jan. 14). Whom Do We Talk To? Issues in Sampling. Retrieved from: www.sed.manchester.ac.uk/research/iarc/ediais/word-files/Sampling.doc Read More