How do I interpret the result of the prediction interval and point-prediction?

71 Views Asked by At

Consider a simple linear model: I have obtained a prediction interval of $(37, 66)$ and a point prediction of $52$. (the problem behind is not a concern I reckon, for the sake of the question).

Now I know that if the actual response will be $\geq 50$, then a plant will grow, whereas if it is less than $50$, it will fail to grow.

The question is to interpret the prediction interval (and the point prediction I guess), to understand whether there is evidence that the plant will not grow.

How do I go about interpreting this?

For example the upper end of the interval of prediction is bigger than $50$, and also the predicted value is slightly bigger than $50$, however the lower end of the prediction interval goes as far down as $37$.. so how does a statistician interpret this? Is there evidence? And if I need to use some mathematical tools, what techniques should I use to see if there is evidence?

1

There are 1 best solutions below

0
On BEST ANSWER

This is the problem with intervals...you lose a lot of probabilistic information. You should look at the likelihood ratio of response $\geq 50$ vs $<50$. If you are using a Bayesian model, then the posterior odds ratio or predictive probability of $\geq 50$ may be useful.