The technical term for a coincidence is a correlation. Correlations, observed patterns in the data, are the only type of data produced by observational research. Correlations make it possible to use the value of one variable to predict the value of another.
For example, one could use Daniel Stern's finding from the previous page, that mothers and newborns with a good relationship tend to synchronize their movements. From this one might predict that babies how are "out of sync" with their mothers might fuss and cry significantly more than other babies.
If a correlation is a strong one, predictive power can be great. Consider this figure, from data produced by a study at the University of Illinois. Researchers asked 56, students about their drinking habits and grades, to see how drinking might correlate with performance in school.
A negative correlation. The results are clear. The more a student drank, the worse was that student's grade-point average. B students those with a 3. Using the correlation shown in this graph, you could predict that a person who drank a six pack of beer every day would be likely to flunk out of school. This is a negative correlation, which means that one variable goes up as the other goes down. As the amount of alcohol consumed goes up on the graph, the corresponding GPA goes down.
A positive correlation is one in which variables go up or down together, producing an uphill slope. To be more accurate, we should change the label on the X-axis of the graph to "Self-report of drinks consumed per week. This is a good time to apply some critical thinking skills to an operational definition. We do not know if self- reports of drinking accurately reflect real levels of drinking.
Perhaps people who get good grades are more likely to lie about drinking. The data do not rule out this explanation. All we know from this study is what people said about how much they drank. This type of modeling usually involves mixed or multilevel statistical models, which allow for individuals to be nested into aggregates. Although that statistic appears to indicate a strong linear relationship, such a conclusion would only be appropriate for the top left graph.
The other three violate assumptions of the statistical analysis, emphasizing the importance of plotting data first to choose a suitable analysis. To avoid assuming two variables are independent because their correlation equals zero, the data must be plotted to make sure it is monotonic. If not, one or both variables can be transformed to make them so. In a transformation, all values of a variable are recalculated using the same equation, so that the relationship between the variables is maintained but their distribution is changed.
Different types of transformations are used for different distributions; for example, the logarithmic transformation compresses the spacing between large values and stretches out the spacing between small values, which is appropriate when groups of values with larger means also have larger variance. Without access to the original data, it is impossible to know whether this error has been committed. Correlation errors are as old as statistics itself, but as the number of published papers and new journals continues to increase, errors multiply as well.
Although it is not realistic to expect all researchers to have an in-depth knowledge of statistical methods, they must continuously monitor and extend basic methodological knowledge. Ignorance or uncritical assessment of the adequacy and limitations of statistical methods used often are the source of errors in academic papers. Involvement of biostatisticians and mathematicians in a research team is no longer an advantage but a necessity.
Some universities offer the option for researchers to check their analysis with their statistics department before sending the article to review with a publication. Although this solution could work for some researchers, it provides little incentive for the researcher to take this extra time. The process of scientific research requires adequate knowledge of biostatistics, a constantly changing field. To that end, biostatisticians should be involved in the research from the very beginning, not after the measurement, observations, or experiments are completed.
On the other hand, basic knowledge of biostatistics is essential in the critical appraisal of published scientific papers. A critical approach must exist regardless of the journal in which the paper is published. A more careful use of statistics in biology can also help set more rigorous standards for other fields. To avoid these problems, scientists must clearly show that they understand the assumptions behind a statistical analysis and explain in their methods what they have done to make sure their data set meets those assumptions.
A paper should not make it through review if these best practices are not followed. To make it possible for reviewers to test and replicate analyses, the following three principles must become mandatory for all authors intending to publish results: publishing data sets as supplementary information alongside articles, giving reviewers full access to the software code used for the analysis, and registering the study in a publicly available database online with clearly stated study objectives before the beginning of research, with mandatory submission of summary results to avoid publication bias toward positive results.
These steps could speed up the process of detecting errors even when reviewers miss them, provide increased transparency to bolster confidence in science, and, most important, avoid damage to public health caused by unintentional errors.
Skip to main content. Login Register. Page 26 DOI: Illustration by Tom Dunne. Facebook Twitter. Bibliography Aldrich, J. Correlations genuine and spurious in Pearson and Yule. Statistical Science — Andrade, A. Grande, C. Talsness, K. Grote, and I. A dose-response study following in utero and lactational exposure to di- 2-ethylhexyl -phthalate DEHP : Non-monotonic dose—response and low dose effects on rat brain aromatase activity.
Toxicology — Anscombe, F. Graphs in statistical analysis. American Statistician — David, H. A historical note on zero correlation and independence. Hill, A. The environment and disease: Association or causation? Although this seems like a minor change to the research design, it is extremely important. Now if the exercisers end up in more positive moods than those who did not exercise, it cannot be because their moods affected how much they exercised because it was the researcher who determined how much they exercised.
Likewise, it cannot be because some third variable e. Thus experiments eliminate the directionality and third-variable problems and allow researchers to draw firm conclusions about causal relationships. Practice: For each of the following statistical relationships, decide whether the directionality problem is present and think of at least one plausible third variable.
Chiang, Dana C. Skip to content 6. Explain why a researcher might choose to conduct correlational research rather than experimental research or another type of non-experimental research. Interpret the strength and direction of different correlation coefficients. Explain why correlation does not imply causation. Key Takeaways Correlational research involves measuring two variables and assessing the relationship between them, with no manipulation of an independent variable. Correlation does not imply causation.
A statistical relationship between two variables, X and Y , does not necessarily mean that X causes Y. The sign indicates the direction of the relationship between the variables and the numerical value indicates the strength of the relationship. Exercises Discussion: For each of the following, decide whether it is most likely that the study described is experimental or correlational and explain why.
An automotive engineer installs different stick shifts in a new car prototype, each time asking several people to rate how comfortable the stick shift feels. A social psychologist tells some research participants that they need to hurry over to the next building to complete a study.
She tells others that they can take their time. Then she observes whether they stop to help a research assistant who is pretending to be hurt. People who eat more lobster tend to live longer. People who exercise more tend to weigh less. College students who drink more alcohol tend to have poorer grades.
Bushman, B. Effects of televised violence on aggression. Singer Eds. Thousand Oaks, CA: Sage. Chocolate consumption, cognitive function, and Nobel laureates.
New England Journal of Medicine, , Previous Section. Next Section. License 6.
0コメント