What is P-Value and R-Value? and what are these stands for, is there any easy way for medical students to calculate it there-selves?
Because in most of the Universities students are teach bio-statistics but kept them dependent on bio-statistician for data analysis.
p-value is a representation of the statistical significance of a result. It is an estimate for the probability that the results being tested could occur in a naturally-occurring system (random distribution). Statisticians have coalesced around 0.05 or less than 1 in 20 chance of the result occurring naturally as the point that is considered "significant". The p-value has fallen into disfavor and many statisticians prefer to report the confidence interval. The confidence interval is the range of values one would expect the true value to occur in 95% of studies.
R-squared is a representation of the amount of the variance in the dependent variable (the thing being predicted) which is explained by the model specified. This is a measure of the ability of the model to replicate the underlying data.
Both of these can be calculated by hand, however in more complex models and regressions it is preferrable to use a statistical package for accuracy and efficiency.
~~~~~~
As to your comment regarding keeping physicians dependent, statistics is a very active field of study with ongoing research, new techniques, developments, and breakthroughs. Statisticians are exploring new and better ways to model systems and learning how older techniques may have produced biased.
p-value and r-squared are metrics that help to interpret the goodness of fit and the significance of a model assuming that the correct model is applied. Statisticians are interested in bettering science by assuring the best model is used for the situation at hand and that these metrics are interpretted correctly.
In medicine, this would be similar to examining a patient with high potassium levels. The interpretation of this value is very different in a patient with end-stage renal disease requiring dialysis versus an otherwise healthy patient with hyperkalemic emergency. One cannot simply look at the potassium level and act unilaterally. Statisticians often feel similiarly about these metrics.
p-value is a representation of the statistical significance of a result. It is an estimate for the probability that the results being tested could occur in a naturally-occurring system (random distribution). Statisticians have coalesced around 0.05 or less than 1 in 20 chance of the result occurring naturally as the point that is considered "significant". The p-value has fallen into disfavor and many statisticians prefer to report the confidence interval. The confidence interval is the range of values one would expect the true value to occur in 95% of studies.
R-squared is a representation of the amount of the variance in the dependent variable (the thing being predicted) which is explained by the model specified. This is a measure of the ability of the model to replicate the underlying data.
Both of these can be calculated by hand, however in more complex models and regressions it is preferrable to use a statistical package for accuracy and efficiency.
~~~~~~
As to your comment regarding keeping physicians dependent, statistics is a very active field of study with ongoing research, new techniques, developments, and breakthroughs. Statisticians are exploring new and better ways to model systems and learning how older techniques may have produced biased.
p-value and r-squared are metrics that help to interpret the goodness of fit and the significance of a model assuming that the correct model is applied. Statisticians are interested in bettering science by assuring the best model is used for the situation at hand and that these metrics are interpretted correctly.
In medicine, this would be similar to examining a patient with high potassium levels. The interpretation of this value is very different in a patient with end-stage renal disease requiring dialysis versus an otherwise healthy patient with hyperkalemic emergency. One cannot simply look at the potassium level and act unilaterally. Statisticians often feel similiarly about these metrics.
Philip Cochetti could you tell me about chi square p-value in R software? it should be sig then the model will be fit or in-sig p-vlaue represent the model fit?
Arsalan Haider, can you please give a little more information about your use-case? What are the data that you are attempting to fit? What type of model are you using to fit these data? Which R command are you using to fit your model?
In some model-selection contexts, the likelihood ratio tests can be used to compare the fit of two models. For example for nested models, the -2 log(Λ) — the log-likelihood ratio test statistic — is asymptotically chi-squared (χ² ) distributed with degrees freedom equal to the difference in dimension (per Wilks's theorem provided the assumptions for this type of modeling are satisfied). In this context, if the distribution is as likely under the null hypothesis as the alternative hypothesis -2 log(Λ) is closer to 0 and thus fails the test that the two models are statistically different from one another.
If your model is of this type, a significant model is one that rejects the null hypothesis i.e. that the p > χ² meets or exceeds the significance threshold. However, that doesn't mean that each individual predictor is significant and there are diagnostics such as confirming that the model has "converged" for one. Here is an example output based on multinomial logistic regression and annotates the R inputs and outputs showing explicitly what each item means. https://stats.idre.ucla.edu/r/dae/multinomial-logistic-regression/ The "things to consider" section at the bottom is very good and talks about diagnostics, assumptions and concerns with this type of model.
If this makes little to no sense and doesn't look familiar to you, I might suggest consulting a local statistician at your institution who might be able to better assist you.
Thank you very much for sharing valuable information, Philip Cochetti.
Philip Cochetti thank you for responding, I am using SEM in R for statistical analysis, where my p=0.567 and CFI， ＧＦＩａｒｅ １.０００， ａｎｄ ＲＭＳＥＡ＝．０００ ｗｈａｔ ｄｏ ｙｏｕ ｔｈｉｎｋ ｏｆ ｍｙ ｒｅｓｕｌｔ？
Sorry, I haven't done structural equation modeling.
My (limited) understanding is that the causal assumptions used here can be sensitive to the input model used and the size of your sample. I would contact someone who has expertise in SEM before drawing conclusions.
- Badges
- Science topic