External, Internal, and Construct Validity

Introduction

When crafting the research method and design, it is essential for the researcher to evaluate issues that are likely to have an impact on the validity of the study results. Various types of validity help in ascertaining whether the conclusions, inferences or propositions made are consistent with the truth (Gratton & Jones, 2010). A number of factors have to be taken into consideration when developing a rigorous research methodology. These factors include construct, internal and external validity. This paper compares construct, internal and external validity and identifies the likely threats to construct and external validity. In addition, this paper draws upon the envisioned research to outline the impact of validity.

Comparison of External, Internal and Construct Validity

External validity is primarily concerned with the sample used in the study and the degree to which the conclusions of the study can be related or applied to the broader population. In this regard, if the conclusions made from a particular research have been established to be valid for other people located in different places at different time horizons, then it can be asserted that the inferred conclusions are likely to be reliable (Trochim & Donnelly, 2008). The external validity of a scientific study denotes the degree to which the findings of the study can be generalized. As a result, external validity is primarily concerned with the ability to generalize conclusions, inferences and propositions using the findings of a study. To this end, sampling models are only considered externally valid if the samples are representative of the population that is being generalized (Tracy, 2012). There are two main methods that researchers can adopt to make sure that their results can be generalized to the wider population. The first method involves the researcher identifying the population of interest from which generalized conclusions, inferences and propositions will be made. This is followed by developing an equitable sample from the one is that the researcher is faced with the challenge of determining the population from which the findings are going to be generalized, something that is not always possible for the case of large populations. The other issue relates to developing an equitable sample from the population that is being studied (Cozby, 2012). The second method that can be used to achieve external validity is using of proximal similarity model, whereby generalization of the findings is permitted on grounds that the study sample is similar to the wider population in terms of the times, places and people being described (Tracy, 2012). With a significantly similar study sample with regard to the population, the researcher can make generalizations with more confidence, particularly when the study sample is closely similar to the population. On the other hand, the confidence with which the findings are generalized reduces when there is a low level of similarity between the study sample and population that is being studied (Trochim & Donnelly, 2008).

Internal validity refers to the degree to which changes observed in one variable leads to changes observed in another variable. Therefore, causality is a significant factor with regard to internal validity since it denotes the relationship between a variable of interest and another one, how it changes, and how this change is likely to affect another variable. Nevertheless, it is imperative to note that causality does not imply that the same variable changes usually result in a specific outcome. Instead, causality means that a specific outcome increases the likelihood that a change in a variable will result in a specific outcome. In addition, the latter is less probable when a similar change in a variable does not have an impact on another variable’s probability distribution (Trochim & Donnelly, 2008). Thus, is impossible for causation to be observed without association; nevertheless, an observed association does not necessarily translate to a causation. A crucial factor to be taken into account when ascertaining causality is the likelihood that the observed changes in the variable may be attributed to some reason that has been unidentified by the researcher or that the observations can be attributed to the intervention of the phenomenon that was being investigated. Internal validity seeks to make sure that empirical evidence resulting in the change in a variable is indeed a contributor to a specific outcome observed in another variable. In order to establish the cause and effect, three conditions have to be met; these include covariation of the cause and effect, temporal precedence, and plausible alternative. With regard to temporal precedence, it must be established by the researcher that the cause occurred before the effect (Gratton & Jones, 2010). As for the covariation of the cause and effect, they must have some form of a relationship, which simply implies that a change in a variable will result in a particular outcome for another variable. Similarly, lack of change in a variable will imply that a specific outcome of another variable will not be observed. Plausible alternative explanations help the researcher to recognize whether the covariation between two variables can be considered causal. Therefore, the causal relationship between two variables can be ascertained when measures have been taken to eliminate all the plausible alternative causes to a specific observed outcome (Trochim & Donnelly, 2008). Probable alternative explanations can be eliminated through the use of a control group, which refers to a sample obtained from the population that is not subjected to the changes in the variables of interest. If observations in the control group show no specific outcome, likely alternative explanations can be dismissed to be the cause.

At first glance, it is apparent that internal and external validity are contradictory in the sense that achieving internal validity through the exclusion of interfering variables is likely to reduce external validity owing to the fact that the research will be performed in a manner similar to the laboratory setting. On the contrary, in observational studies, it is impossible to control variables that may interfere; however, the natural environment can be measured including the setting and time, but this reduces internal validity (Trochim & Donnelly, 2008).