In: Statistics and Probability
Before going to the market, a pregnancy test has been applied to a population of 10,000 women from which 300 were actually pregnant. Suppose this new product returns a “yes” result 99.5 % of the cases in which the person was actually pregnant and 98% of the cases returns a “no” result when the person was not pregnant.
Now that the test is in the market: what is the probability that a random user (a women) of the test is not pregnant if the test returns a “yes “ result? Draw the appropriated probability tree
P[ Women is pregnant ] = 300/10000
P[ Women is pregnant ] = 0.03
P[ Women is not pregnant ] = 1 - P[ Women is pregnant ]
P[ Women is not pregnant ] = 1 - 0.03
P[ Women is not pregnant ] = 0.97
P[ new product returns a “yes” | she is actually pregant ] = 99.5% = 0.995
P[ new product returns a “no” | she is not actually pregnant ] = 98% 0.98
P[ new product returns a “yes” | she is not actually pregnant ] = 1 - P[ new product returns a “no” | she is not actually pregnant ]
P[ new product returns a “yes” | she is not actually pregnant ] = 1 - 0.98
P[ new product returns a “yes” | she is not actually pregnant ] = 0.02
P[ new product returns a “yes” ] = P[ new product returns a “yes” | she is not actually pregnant ]*P[ Women is not pregnant ] + P[ new product returns a “yes” | she is actually pregant ]*P[ Women is pregnant ]
P[ new product returns a “yes” ] = 0.02*0.97 + 0.995*0.03
P[ new product returns a “yes” ] = 0.0194 + 0.02985
P[ new product returns a “yes” ] = 0.04925
We need to find P[ she is not actually pregnant | new product returns a “yes” ]
P[ she is not actually pregnant | new product returns a “yes” ] = P[ new product returns a “yes” | she is not actually pregnant ]*P[ Women is not pregnant ] / P[ new product returns a “yes” ]
P[ she is not actually pregnant | new product returns a “yes” ] = 0.02*0.97 / 0.04925
P[ she is not actually pregnant | new product returns a “yes” ] = 0.0194/0.04925
P[ she is not actually pregnant | new product returns a “yes” ] = 0.3939