need today
Nokcool
DQStatsweek3.docx1. Yu Chen
Sep 28, 2020Sep 28 at 6:57pm
Having a curiosity towards a certain relationship or scientific phenomena, and asking a scientific question is the very beginning of conducting scientific research. But before a researcher just dives into the study, the researcher must first form an opinion, and that is the hypothesis. Hypothesis in my opinion is the guide to performing the test and validating that assumption. However, one should most definitely try to avoid confirmation bias in which the study is modified and curbed to confirm the hypothesis.
One way to avoid bias in a study is to set a level of significance reasonable to the field of study. For example, for medical studies, the level of significance should be set to 0.01 or 1% chance to have a false positive. For regular studies such as find the relationship between eating cotton candy and level of happiness, the level of significance can be much higher such as 0.1 or 10% chance of having a false positive.
Another way to avoid bias is to select a sample that represents the population well. Much like the quote, "A small sample, we repeat, is rarely the big scientific problem. Interpretation is," correct sampling in many ways will affect the outcome of the study. For example, if a researcher likes to find out how supportive the public are of the government, sampling should consist of people from different political views such as conservative, liberal, neutral, etc. If the sample consists only of conservative and republican is in office, we can very well assume that the result is going to be biased towards a favorable view.
Yu
2. Cameron Izzi
Sep 29, 2020Sep 29 at 7:31am
Hypothesis testing is a must for conducting important and valid research. The hypothesis is the back bone of why someone is performing a research study, and gives the researcher a foundation on which to base their testing on. Without a hypothesis a researcher would be blindly testing and trying experiments with no real goal or scientific driving factor behind it. Another important factor of hypothesis testing it not allowing it to affect the results of the experiment. This being if the results of the experiment do not agree with the original hypothesis, do not allow this to influence a change in the experiment to try to prove the original hypothesis right. This is key because if the hypothesis testing proves the original hypothesis wrong, then that is just as good as it proving it correctly. This is the whole point of hypothesis testing; to begin to understand what is thought to be a cause of something. If it does prove the original hypothesis wrong, it is important to go through the data and fully understand why the hypothesis was wrong in the first place. Ensure there were no wrong assumptions made, and the data is being analyzed correctly. As stated in the quotes above, "“Be careful of averages and how they’re applied. One way that they can fool you is if the average combines samples from disparate populations. This can lead to absurd observations such as: ‘On average, humans have one testicle.’”—Daniel J. Levitin, A Field Guide to Lies: Critical Thinking in the Information Age (2016). This is extremely important to do when concluding hypothesis testing and analyzing the data to be sure that the experiment was fully understand and the results are meaning full.
3. Cameron Izzi
TuesdaySep 29 at 8:33am
I researched how hypothesis testing was using to during a study of neuroimaging. In this study the statically tested hypothesis each element of a field for which they hypothesis. Dalmau-Cedeño et al. stated right off in the paper that, "In areas of scientific research where imaging is involved (e.g., neuroimaging, remote sensing, etc.), it is often necessary to test statistical hypotheses at each element of a 2 or 3-dimensional field of sites (pixels or voxels)" (p. 24). The main reason was images have pixels that contain 2 to 3 fields of information, and this lead to reason to needing to do hypothesis on each field to validate their assumptions. They used a large amount of different techniques including hyper-parameter tuning, false positive testing, bayesian formulation, and statistical studies looking at null distribution. The results of the study are they were able to increased computational complexity for 2-dimensional images and feel this study could easily be expanded to 3-dimensional data. They also stated that these techniques can be expanded to any local hypothesis testing than just neuroimaging.
I found this research really cool because they are applying satatiscal techniques to hypothesis testing that in the end can really help people in the medical field. This research can lead to quickly applying techniques to image processing of medical images that can lead to faster and earlier discovery of illness within a person.
References:
Dalmau-Cedeño, O. S., Alvarado-Carrillo, D. E., & Luis Marroquín, J. (2020). Regularized Hypothesis Testing in Random Fields with Applications to Neuroimaging. Revista Mexicana de Ingeniería Biomédica, 41(2), 22–39. https://doi-org.proxy-library.ashford.edu/10.17488/RMIB.41.2.2
WednesdaySep 30 at 9:45am
The study I found focused on tests designed to detect and improve employee predictability turnover. Every test used did not use hypothesis testing, so there’s specific questions I will point out from within the study. In the study, a research question was stated that sought to determine if employees, classified as “reluctant leavers within their profession”, if they have the same temperament as individuals classified as enthusiastic stayers. The hypothesis and null were that reluctant leavers and enthusiastic stayers have similar attributes and job attitudes or temperament. The researchers used post hoc comparisons and power analysis to show that findings between the two were not due to low statistical power. The researchers were also aided by the Bayes factor, which references the ratios generated from the division of the odds of having reviewed the standardized effect size below an alternative hypothesis by the odds of having the observed effect size under the null hypothesis (Li, Lee, Mitchell, Hom & Griffeth, 2016).
The areas (sample size) analyzed were job satisfaction, job embeddedness, and affective commitment. The answers reported from the purported four states sampling were graphed to ease for interpretation. What the researchers were able to do accomplish was that there wasn’t much of a difference (in the responses) from enthusiastic stayer employees or reluctant leaver employees, which helped the research. It helped add to the research that in turn will contribute to a much wider study of increasing the predictability variables of employee turnover.
References
Li, J., Lee, T. W., Mitchell, T. R., Hom, P. W., & Griffeth, R. W. (2016). The effects of proximal withdrawal states on job attitudes, job searching, intent to leave, and employee turnover. Journal of Applied Psychology, 101(10), 1436–1456. Retrieved from https://doi-org.proxy-library.ashford.edu/10.1037/apl0000147
