# RESPONDING TO CLASSMATE POST X 2

RESPONDINGTO CLASSMATE POST X 2

UniversityAffiliation

RESPONDINGTO CLASSMATE POST X 2

PostOne Response

Inferentialstatistics helps people to attain conclusions beyond the data athand. It is based on many factors as you have mentioned. However, myconcern is on the t-test, and the process of testing correlation. Thet-test is also known as a two sample t-test as it aims to determineif there exists a statistically substantial variance between themeans of two distinct groups. To run an independent t-test, onerequires an independent variable with two groups, and a continuousdependent variable (Bernstein&amp Bernstein, 2012).Nonetheless, my disturbing view is how well can we eliminate theissue of type I and II errors when handling multiple t-tests? This isbecause, you have pointed out that running multiple tests increasestype 1 error, but correcting it through Bonferroni method enhancesthe chance of type II error. In addition to what you said concerningcorrelation, if the value of the correlation coefficient is about ±1,there is a perfect degree of relationship between the twovariables.&nbspConversely, the relationship gets weaker as the valuetends towards 0.

Reference

Bernstein,S. &amp Bernstein, R. (2012). Inferential statistics. NewYork: McGraw-Hill.

PostTwo Response

Concerningthe key statistical tests and their measurement of significance, Iwould like to elaborate on causal modeling and survival analysis.Causal modeling entails mathematical equations and is based onstatistical and causal assumptions. Some of the statisticalassumptions around it are normality and linearity, non-correlation oferror terms, and non-measurement error. On the other hand, survivalanalysis is time-related and can be in years, weeks, and days.Therefore, ordinary linear regression cannot be used as it involvesnegatives, yet survival times must be positive (Mills,2011).If information concerning survival time is incomplete, then theobservations are referred to as censored. However, my area ofinterest to you is how effective should the process of censoring beto ensure survival analysis is free of bias? I believe measuresshould be in place to ensure censoring is effective to produce truesurvival analysis rates.

Reference

Mills,M. (2011). Introducing survival analysis and event historyanalysis. London: Sage.