UNIVERSIDAD JUREZ AUTNOMA DE TABASCO

Divisin Acadmica de Ciencias Econmico Administrativas

Nivel Lic. Contadura Pblica Asignatura

Introduccin a la Estadstica

Tpico
Modelos de densidad normal Distribucin de probabilidad normal Promedio general: 8.17 Telfono: 9933096528

Expositor Jimnez Peralta Mara Guadalupe Facilitador M.A y M.I Fabio Garca Martnez

Villahermosa, Tabasco 26 de Agosto de 2013

CONTINUOUS PROBABILITY DISTRIBUTIONS

THE NORMAL DISTRIBUTION

A variable is said to be normally distributed or to have a normal distribution if its distribution has the shape of a normal curve. Two items of importance relative to normal distributions are as follows. If a variable of a population is normally distributed and is the only variable under consideration, then it has become common statistical practice to say that the population is normally distributed or that we have a normally distributed population. In practice it is unusual for a distribution to have a exactly the shape of a normal curve. If a variables distribution is shaped roughly like a normal curve, then we say that the variable is is approximately normally distributed or has approximately a normal distribution.

A normal distribution (and hence a normal curve) is completely determined by the mean and standard deviation; that is, two normally distributed variables having the same mean and standard deviation must have the same distribution. We often identify normal curve by stating the corresponding mean and standard deviation and calling those the parameters of the normal curve. The equation of the normal curve with parameters and is y= e e 2.718 and 3.142
( )

, where

A normal distribution is symetric about and centered at the mean of the variable and its spread depends on the standars of the variable the larger the standard deviation, the flatter and more spread out the distribution. The three-standard-deviations rule, when applied to a variable, states that almost all of the possible observations of the variable lie within three standard deviations to either side of the mean. One important difference between discrete and continuous probability distributions involves the calculation of probabilities associated with specific values of the random variable. For instance, in the market research example in which 100 people were surveyed, we could use the binomial distribution to find the probability of any specific number of positive reviews, such as P(x 75) or P(x 76). While these individual

probabilities may be small values, they can be computed because the random variable is discrete. However, if the random variable is continuous as in the Pepsi-Cola example, there are an uncountably infinite number of possible outcomes for the random variable. Theoretically, the probability of any one of these individual outcomes is zero. That is, P(x 11.92) 0 or P(x 12.05) 0. Thus, when you are working with continuous distributions, you will find the probability for a range of possible values such as P(x 11.92) or P(11.92 x 12.0). Likewise, you can conclude that P(x 11.92) because we assume that P(x 11.92) 0. P(x 11.92)

There are many different continuous probability distributions, but the most important of these is the normal distribution. The Normal Distribution You will encounter many business situations in which the random variable of interest will be treated as a continuous variable. There are several continuous distributions that are frequently used to describe physical situations. The most useful continuous probability distribution is the normal distribution. The reason is that the output from a great many processes (both man-made and natural) are normally distributed. Figure 6.1 illustrates a typical normal distribution and highlights the normal distributions characteristics. All normal distributions have the same general shape as the one shown in Figure 6.1. However, they can differ in their mean value and their variation, depending on the situation being considered. The process being represented determines the scale of the horizontal axis. It may be pounds, inches, dollars, or any other attribute with a continuous measurement. Figure 6.2 shows several normal distributions with different centers and different spreads. Note that the total area (probability) under each normal curve equals 1. The normal distribution is described by the rather-complicated-looking probability density function, shown in Equation 6.1.

Fig. 6.1

To graph the normal distribution, we need to know the mean, , and the standard deviation, .
Fig. 6.2

Placing

, ,

and a value of the variable, x, into the probability density function, we can calculate a height, f(x), of the density function. If we try enough x-values, we will get a curve like those shown in Figures 6.1 and 6.2. The area under the normal curve corresponds to probability. The probability, P(x), is equal to 0 for any particular x. However, we can find the probability for a range of values between x 1 and x 2 by finding the area under the curve between these two values. A special normal distribution called the standard normal distribution is used to find areas (probabilities) for a normal distribution. Normal distribution a continuous probability distribution Unlike the binomial and Poisson distributions, which describe the probabilities for the occurrence of random variables that are discrete events, the normal distribution describes random variables that are continuous (i.e. any value within a given range might occur). The normal distribution is often described as the classic bell shaped curve. The typical shape of this distribution is displayed below. This picture illustrates that normal distributions are always symmetrical about their mean value. This distribution is of interest to statisticians because a vast number of populations and their data sets follow this bell shaped pattern. Because this distribution is central to the concepts of sampling distributions, statistical inference, estimation, and hypothesis testing, it is important that students of statistics thoroughly understand the normal distribution and be adept at making computations involving this distribution. Because the normal distribution is a continuous distribution, it cannot provide a probability for a single, unique value within the interval of values covered by the distribution. (Incidentally, this is the case with all other continuous probability distributions.) It can only provide a probability for a range of values within the interval. There is a cumbersome formula that describes the height of the curve at any point along the horizontal axis (the range of possible values for the random variable, and in theory this range extends from to + ). The height of the curve at any point (x) along the

horizontal axis is computed from the following (which is called the probability density function):

x=
It is the area under this curve that depicts the probability. The total area under the curve equals 1.0 (since this represents all possible events, and the sum of the probabilities for all events must be equal to 1.0). Finding the area under segments of the curve requires the use of integral calculus. Fortunately that exercise is not necessary, for tables have been constructed to provide probabilities for a wide variety of events. Such tables typically appear in virtually every math and statistic textbook, and are readily available at a number of other sources (e.g. Internet). Furthermore, in our current state of technological development, you will find functions in standard spreadsheet software packages that compute the normal probabilities quite handily. Typical Normal Distribution The typical normal distribution is pictured below. The distribution may be more peaked and narrow, or more squat and wide, but it will always be symmetrical about its mean. How narrow vs. wide or peaked vs. squat a particular normal distribution is will be dependent upon its mean and standard deviation.

 Observations about the Normal Probability Distribution There are several generalizations that are made regarding the normal probability distribution. These generalizations include the following: 1. There is an unlimited family of normal probability distributions. They are differentiated by their mean, , and standard deviation, . 2. The highest point on the normal curve is at the mean, which is also the median and the mode of the distribution. 3. The mean of the distribution can be any numerical value: negative, zero, or positive. 4. The normal probability distribution is symmetrical, with the shape of the curve to the left of the mean a mirror image of the shape of the curve to the right of the mean. The tails of the curve extend to infinity in both directions and theoretically never touch the horizontal axis. 5. The standard deviation determines the width of the curve. Larger values of the standard deviation result in wider, flatter curves, showing more variability in the data. 6. The total area under the curve for the normal probability distribution is 1 (which is true for all continuous probability distributions).

7. The inflection points (called knees) of the curve occur 1 standard deviation above and below the mean (these are the points where the change in slope changes in direction). 8. Probabilities for the normal random variable are given by areas under the curve for intervals of values. Technically, the probability for a specific value in a normal probability distribution is 0. These generalizations suggest that many data sets that we treat as normal probability distributions are in fact approximations of a normal distribution. For example, we often look at the exam grades in a large class, analyze them, and then claim (for example) that the grades are normally distributed with a mean of 75 and a standard deviation of 10. If this grade distribution situation were truly reflective of a normal distribution, than individual grades on the exam would be able to range from to + . Of course, assuming the exam is graded on a 100 point scale, individual grades could only range from a low of 0 to a high of 100. However in situations like these, the normal distribution can serve as a very close approximation to reality. Therefore, while many of the applications we examine in this set of notes may not be precisely represented by normal probability distributions, they will be a very close approximation. Normal Probability Tables The number of normal probability distributions, each uniquely described by its combination of mean and standard deviation, is infinite. It would be an impossible task to develop a separate table of normal probabilities for each of these distinct distributions. There is only one normal distribution for which the probabilities have been accumulated in a table, and that distribution is called the standard normal distribution. This is the normal distribution whose mean value is 0 and whose standard deviation is 1, as displayed below.

=1

z=3 z=2 z=1

=0 z=+1 z=+2 z=+3

The z values are a representation of the number of standard deviations away from the mean. For example, the probability associated with being one or fewer standard deviations to the right of the mean is simply the area under the normal curve up to the point where z = +1. It is depicted by the shaded area on the picture below. =1

z=3 z=2 z=1

=0 z=+1 z=+2 z=+3

Regardless of what the parameters ( and ) are for a particular normal deviation, one can always find the probability associated with a particular value by converting to a z score (i.e., finding how many standard deviations from the mean the value in question is). The conversion is z = (x )/ The probability associated with this z score can be read from the standard normal table of probabilities. That probability will represent the probability associated with the original x value.