Pssst… we can write an original essay just for you.
Any subject. Any type of essay.
We’ll even meet a 3-hour deadline.Get your price
121 writers online
Quantitative research allows data to be used to explain, describe, predict, explore, or evaluate phenomena with objectivity and control of related factors (MacInnes, 2016). In order to infer that a causal link exists between an independent, or causal, variable and a dependent variable, three factors of causality must exist. First, there must be an empirical association between the two variables. There must also be an appropriate time order. Finally, there must be a nonspurious relationship between them, or an association that cannot be explained by a third outside variable’s influence.
First, relationship or an association between the two variables must be established. The correlation between them must be such that when one variable changes, the other does as well. Without this association, causation cannot exist. With the establishment of the relationship, the validation of time order, or direction of order, should be validated to demonstrate that a change in causal factor influenced the dependent variable (University of Southern California Libraries, 2016). The cause must precede the effect. Finally, the third condition that must be met in order to postulate a causal link is the need for the relationship to be isolated from all other outside factors—are there other plausible variables that could threaten internal validity?
Internal validity is the extent in which the covariance of two variables is isolated from, and unaffected by, outside influences. It is necessary to demonstrate a causal relationship between the independent and dependent variables. In quantitative research, change in the independent variable is not caused by any other variable in the study; it is the variable that is manipulated during the research project (MacInnes, Internal validity [PowerPoint], 2016). The dependent variable is responsive to the conditions set forth by the causal variable.
Internal validity is a crucial aspect of quantitative research, as it establishes a confidence in the cause and effect relationship.Threats to Internal ValidityTo have high internal validity, results must demonstrate a level of causality that demonstrates little chance of outside factor influence; threats correlate to a compromise in the ability to demonstrate such a relationship and therefore it is imperative that potential internal validity threats are controlled (Michael, 2016). Such threats include time, group, mortality and atypical behavior (MacInnes, Internal validity [PowerPoint], 2016).
Time can be a threat to an experiment when factors outside of the independent variable affect the outcome due to causes other than experimental manipulation, such as history, maturation and instrumentation (MacInnes, Threats to internal validity [PowerPoint slides], 2016). Examples of such threats could come from external or environmental unanticipated events, normal developmental changes over time, test reactivity and instrumental changes between tests. To counter such, a researcher should consider using a control group, spacing tests out to fit the needs of the experiment and having a standardized testing instrument for all tests within the process.Group.Group can be another threat to the internal validity of an experiment.
MacInnes (2016) reported that this can be caused by opposing explanations based on differences in the groups, such as a regression to the mean, due to pretests and posttests with extreme groups, and selection by time interactions, as participants may have differing experiences due to age or environment. Solutions to such threats include using a control group, avoiding the comparison of extreme groups and controlling for selection conflicts by using random assignment and matching.
Mortality is another threat to the internal validity of a study, and therefore the causal relationship of variables. This threat refers to the leaving of participants from a study for different, random reasons, which in turn causes the participant sample to be different between the pretest and posttest. Using a control group does not counter this threat; instead, the researcher should ensure that the project is engaging in order to keep participants interested, shorten the length of the study and examine the pretest scores of those participants who left the study compared to those who did not (MacInnes, Threats to internal validity [PowerPoint], 2016; Ohlund & Yu, 2018).
Finally, atypical behaviors are also threats to the internal validity of a study and refers to the behaviors that can be exerted by groups when they receive less-desirable treatment than the other group(s) . Exmples of such include resentful demoralization among participants, compensatory rivalry (also known as the John Henry effect) and the diffusion of treatment. To control for such, communication between groups must be monitored (MacInnes, Threats to internal validity [PowerPoint slides], 2016).
Statistical validity examines the the extent of a relationship between two variables. Validity holds when, using statistical analysis, the conclusions drawn are adequate and logical (Garcia-Perez, 2012). It describes the appropriateness of concluding a covariance between variables based upon inferences made after interpreting statistical results. The data should have the ability to be generalized to the population represented in the sample, which can be acomplished through the use of an adequate sample size in the research. The appropriate statistical test should also be used to ensure proper analysis of the data. Statistical validity and internal validity differ in the idea that the former is not so much concerned with whether there is a causal relationship between variables like the latter, but instead whether there is a relationship at all (Adams, 2008).
From a statistical conclusion standpoint, the hypothesis—the null hypothses is always assumed to be true at the beginning of the experiment—is either said to be true or is rejected. However, when coupled with what is really true, there are a possible four outcomes. First, there is a Type I Error, when the null hupotheis is rejected, but it is true. This error is the most severe that a researcher can make in terms of statistical validity, as it is drawing conclusions between variables that are unfounded in the population. Another possible conclusion is a Type II Error. This occurs when the null hypothesis is accepted, but it is false. This is a failure to reject when rejection is the proper response.
Data is showing a variable association tht isn’t representative of the population. Accepting a true null is the third outcome, which is when the null hypothesis is accepted and is true. The final, most desired outcome is that of power. This occurs when the null hypothesis is rejected, and it is indeed false. This true alternative demonstrates a desired level of power in the data. There are four influences over power in a quantitative study: sample size, significance level, effect size, and variance. The first three influences have a positive correlation with power—as one increases, so does the other; the last influence, variance, has an inverse relationship—as it increases, power decreases.Threats to PowerThere are two threats to power. The first is an increase in Type I Error Rate, as the desire is to have this rate be low. As this increases wth each hypothesis test completed, a researcher can combat this threat by limiting the number of tests completed (limiting fishing). The second threat to power is a decrease in such, which can be combatted by ensuring the influences discussed above are controlled and accounted for.
Experimental design has the ability to demonstrate strong validity, and therefore covariance, between independent and dependent variables when implemented well. In order to improve internal and external validity, there are methods that can be implemented in the experiment, including random sampling, using a control/comparison group and random assignment. MacInnes (2016) stated that there is no perfect study that eliminates all validity threats.
Simple random sampling is a technique which randomly selects a sample of individuls from a larger population. Because of the random nature, each individual has an equal chance of being chosen for participation and each selection is independent of all others, which removes selection bias, an external threat, and in theory should result in generalizations to the greater population to be made after the study concludes (Cohen, 2008; MacInnes, Designs [PowerPoint slides], 2016).
Control/comparison groups allow for the detection and explanation of possible confounding variables (MacInnes, Designs [PowerPoint slides], 2016). By including this dynamic in a study, a researcher can decrase the internal validity threat of time, as well as the external threat of selection bias.
Random assignment of participants results in equivalent groups, as there is an equal chance to be in the treatment group, which combats several internal validity threats, including group selection and interactions with selection. There is an increase in isolation by removing the relationship between the treatment and other confounding variables.
To export a reference to this article please select a referencing style below:
Sorry, copying is not allowed on our website. If you’d like this or any other sample, we’ll happily email it to you.
Your essay sample has been sent.
Want us to write one just for you? We can custom edit this essay into an original, 100% plagiarism free essay.Order now
Are you interested in getting a customized paper?Check it out!