Internal Validity and Causal Relationships in Research
However, as with any research, there exist threats to its integrity. For example, single group threats whereas the research only studies a single group, and multiple group threat when the research studies several groups in the study. Another threat is a social threat. A social threat would occur when research is conducted in “real-world” human contexts where people react to what affects them and what is happening to people around them. A strategy to eliminate or reduce these threats is to establish a methodology which is very broad on one hand, and thorough on another level.
INTERNAL VALIDITY:
STANDARDIZED TESTING ANXIETY
CAUSES
CAUSES:
Ã?· Teaching “to the test”
�· Assessment policies
�· Educator expectations
�· Graduation requirement
When you refer to external validity, you are actually referring to generalization. This validity is how the research would represent for other people in other places and at different times. The hardest aspect about external validity is you might be wrong in your generalization. Not all people, places, and times are the same. How can you possibly compare education in the United States to that of education in Iraq? The difference in people, places, and time may cause threats to your study.
The way to decrease threats to external validity is to use random selections within your population. Another way would be to provide a lot of data showing the similarity among groups of people, places, and times.
EXTERNAL VALIDITY:
SAMPLE
Sample generalizes the population in respect to people, places, and time.
I might have left the conclusion validity last, due to it’s overwhelmingly lack of understanding. Although it might represent the most important of the types of validity, it is also the most misunderstood. Conclusion validity is the level to which conclusions we determine about relationships in our statistics are reasonable. Conclusion validity is similar to internal validity because it also deals with causal relationships.
There are many issues that can attribute to the threat to conclusion validity. For example, low reliability, a weak relationship, and/or lack of enough information. The way to improve these factors is by gaining control and confidence over the ability to recognize relationships and to accumulate the necessary amounts of data to support such a relationship.
CONCLUSION VALIDITY:
DATA COLLECTED AFTER STUDENTS TAKE STANDARDIZED TESTS:
*Questionnaire about their anxiety level prior/during/after test.
Is the data a reasonable conclusion that shows a relationship between standardized tests and anxiety? Was there another factor that led to the anxiety? Given the data, if it establishes this relationship then conclusion validity exists. If it does not, then we do not have internal validity. The data must show a relationship.
It might not be good to place labels on data, however, in construct validity, labeling is the issue. Construct validity is similar to external validity. It is similar because it is related to generalization. However, instead of referring to people, places, and time, construct validity refers to generalization of your research and data to the concept of your research and data. In construct validity you are attempting to establish that your research and data are valid. Construct validity may represent the most important of the validities. The measures made in data and research need to demonstrate both convergent and discriminatory validity. The biggest threat to this validity is the attempt to establish a pattern. This pattern must represent your findings and must also represent the label you have placed on this study. In order to eliminate these threats, you must utilize process pattern matching. In order to access construct validity you must understand pattern matching.
CONSTRUCT VALIDITY:
Theoretical Realm: Observational Realm:
�· Sweating
�· Sense of worry
�· Stomach pain
�· Headache
�· Lack of concentration
�· Erratic heartbeat
Along with validity is the concept of reliability. Measuring reliability is not an exact science. In measuring reliability, you must establish estimates. The four types of reliability mentioned in the module are: inter-observer, test-retest, parallel-forms, and internal consistency. Inter observer refers to the consistent estimates made by different observers. Test-Retest refers to the consistency over a period of time. Parallel-Forms refers to the way two tests measuring the same content can be consistent. Lastly, the Internal Consistency measures the results within a test.
