p denotes the probability or proportion of either yes or no responses. So if p corresponds to yes, then p=460/2300=0.2 or 20% and 1-p=1840/2300=0.8 or 80% represents no.
X represents each individual response (yes or no) to the survey. So the number of X=yes comes to 460 out of 2300 surveyed; there must be 1840 cases where X=no. Alternatively, the number of X=no comes to 1840 out of 2300, and there were 460 cases where X=yes. It all depends on whether p is measuring the probability or proportion that there were 460 yesses or 460 nos.
If 2300 represents a random sample of people out of the population (probably consisting of millions of people) then p̂ (p hat) is the proportion of people within the random sample. p (without the hat) is usually used to denote the proportion of people in the entire population. Surveying the whole population could be expensive and time-consuming so the usual practice is to take samples of a reasonable size and use the various p̂ values to estimate as accurately as possible what p actually is. Usually in statistics, a "confidence interval" for p is established, that is, a very safe guarantee (for example, 99% certainty) that p lies between two calculated values. You can never be 100% certain unless you survey the whole population and even then p can change as people change their minds over time, and it takes time to complete the survey.
p is used to predict the result of taking a random sample of people. If n is the size of the sample then you can predict that n×p will be the number of people in the sample responding in a particular way (yes or no). This product, np, is called the mean (or average). On this basis, statistics was developed.
