Question: 3 Discussion Questions; I would like a response to each individual discussion question.
(1) RE: Discussion - Week 4 Initial PostCOLLAPSE
Logistic regression implies the prediction of probabilities using one predictor (Warner, 2013). Reporting values in logistic regression are pivotal to convey data in a meaningful and representative way. Values include, for example, conditional probability, conditional odds, odds ratio, logits, and even slope. However, each value retains similarities, benefits for reporting, and limitations. For example, odds have a tendency of presenting difficulties for the audience to interpret (Osborne, 2015). Additionally, odds may indicate an overestimation or offer the possibility of a more significant effect size. Moreover, the odds ratio may also depict the option of misinterpretation or misrepresentation (Osborne, 2015).
According to Osborne (2015), probabilities are easily perceptive by an audience or lay reader and logit predict difficult values, but often defies ordinary least squares regression. The limitations and benefits of using these values to display data to an audience implicate that the odds ratio is the most effective value to present to readers even though probabilities is perceivable, the odds ratio displays the constant observable influence. The odds ratio is the most informative value in statistics in conjunction with logistic regression (Osborne, 2015). Furthermore, the odds ratio provides the necessary information for comprehension of a lay reader. For example, according to Laureate Education (2017e), the odds ratio is the most effective value to relay when presenting a comparison.
Laureate Education (Producer). (2017e). Creating a contingency table in Microsoft Excel [Video file]. Baltimore, MD: Author.
Osborne, J. W. (2015). Best practices in logistic regression. Thousand Oaks, CA: SAGE Publications.
Warner, R. M. (2013). Applied statistics: From bivariate through multivariate techniques (2nd ed.). Thousand Oaks, CA: SAGE Publications.
(2) Neidi de Carvalho
Conditional probability described statistically as: P(B|A) = P(A and B) / P(A), basically assesses the probability of an event occurring, given that another event has already occurred. The main aspect with conditional probability however is that one event does not directly affect the other and it therefore does not guarantee the other will occur, however it increases the chances as it established a range of likelihood. The relative risk (Osborne, 2015), presupposes that there is a higher risk something occurring when certain predisposing variables are available. An example which I live everyday is the risk of me having a car accident as at the times I have to drive are peak traffic times, thus increasing the chances of an accident as there are more vehicles on the road.
The use of conditional probability may be an easier method for the audience to understand as laypersons are able to understand the concept of the chance of something occurring or not. People deal with chances on a daily basis, thus this concept is one which relates to persons and also one which may be used more often by social researchers.
References: Osborne, J. W. (2015). Best practices in logistic regression. Thousand Oaks, CA: SAGE Publications.
(3) Main Discussion Post: Logistic Regression Values
There are two values that build upon one another, logit and odds in binary logistic regression. Binary logistic regression is defined as the model that takes the scores from one or more variables to estimate the odds of a specific outcome coming to fruition for a dependent variable (Warner, 2013). The outcome variable from a binary logistic regression is identified as a logit. A logit is the log of odds and the odds are obtained by dividing the number an outcome does happen by the number of times the outcome does not happen.
An example that comes to mind to assist with the clarity; the odds that children who have parents that have master degrees will pursue and complete higher education equivalent or exceeds that of their parents or not complete or pursue higher education.
Reference: Warner, R. M. (2013). Applied statistics: From bivariate through multivariate techniques (2nd ed.). Thousand Oaks, CA: SAGE Publications.