Standard Error

Standard Error

Discussion: Please discuss, elaborate and give example on the topic. Be careful with grammar and spelling. No running head please. Please Use only the reference I will attach as the professor will not be able to give grade.

Author: (Jackson, S. L. (2017). Statistics plain and simple. (4th ed.). Boston, MA: Cengage Learning.)

Topic

What level of measurement can be used for this test for the independent and dependent variables?

Reference:

Module 9: The Single-Sample z Test

The z Test: What It Is and What It Does

The Sampling Distribution

The Standard Error of the Mean

Calculations for the One-Tailed z Test

Interpreting the One-Tailed z Test

Calculations for the Two-Tailed z Test

Interpreting the Two-Tailed z Test

Statistical Power

Assumptions and Appropriate Use of the z Test

Confidence Intervals Based on the z Distribution

Review of Key Terms

Module Exercises

Critical Thinking Check Answers

Module 10: The Single-Sample t Test

The t Test: What It Is and What It Does

Student’s t Distribution

Calculations for the One-Tailed t Test

The Estimated Standard Error of the Mean

Interpreting the One-Tailed t Test

Calculations for the Two-Tailed t Test

Interpreting the Two-Tailed t Test

Assumptions and Appropriate Use of the Single-Sample t Test

Confidence Intervals Based on the t Distribution

Review of Key Terms

Module Exercises

Critical Thinking Check Answers

Chapter 5 Summary and Review

Chapter 5 Statistical Software Resources

In this chapter, we continue our discussion of inferential statistics—procedures for drawing conclusions about a population based on data collected from a sample. We will address two different statistical tests: the z test and t test. After reading this chapter, engaging in the Critical Thinking checks, and working through the problems at the end of each module and at the end of the chapter, you should understand the differences between the two tests covered in this chapter, when to use each test, how to use each to test a hypothesis, and the assumptions of each test.

MODULE 9

The Single-Sample z Test

Learning Objectives

•Explain what a z test is and what it does.

•Calculate a z test.

•Explain what statistical power is and how to make statistical tests more powerful.

•List the assumptions of the z test.

•Calculate confidence intervals using the z distribution.

The z Test: What It Is and What It Does

The z test is a parametric statistical test that allows you to test the null hypothesis for a single sample when the population variance is known. This procedure allows us to compare a sample to a population in order to assess whether the sample differs significantly from the population. If the sample was drawn randomly from a certain population (children in academic after-school programs) and we observe a difference between the sample and a broader population (all children), we can then conclude that the population represented by the sample differs significantly from the comparison population.

z test A parametric inferential statistical test of the null hypothesis for a single sample where the population variance is known.

Let’s return to our example from the previous module and assume that we have actually collected IQ scores from 75 students enrolled in academic after-school programs. We want to determine whether the sample of children in academic after-school programs represents a population with a mean IQ greater than the mean IQ of the general population of children. As stated previously, we already know μ (100) and σ (15) for the general population of children. The null and alternative hypotheses for a one-tailed test are:

H0:μ0≤μ1,orμacademicprogram≤μgeneralpopulationH0:μ0≤μ1, or μacademic program ≤μgeneral population

H0:μ0>μ1,orμacademicprogram>μgeneralpopulationH0:μ0>μ1, or μacademic program >μgeneral population

In Module 6 we learned how to calculate a z score for a single data point (or a single individual’s score). To review, the formula for a z score is:

z=X−μσz=X−μσ

where

X = each individual score

μ = the population mean

σ = the population standard deviation

Remember that a z score tells us how many standard deviations above or below the mean of the distribution an individual score falls. When using the z test, however, we are not comparing an individual score to the population mean. Instead, we are comparing a sample mean to the population mean. We therefore cannot compare the sample mean to a population distribution of individual scores. We must compare it instead to a distribution of sample means, known as the sampling distribution.

The Sampling Distribution

If you are becoming confused, think about it this way. A sampling distribution is a distribution of sample means based on random samples of a fixed size from a population. Imagine that we have drawn many different samples of some size (say 75) from the population (children whose IQ can be measured). For each sample that we draw, we calculate the mean; then we plot the means of all the samples. What do you think the distribution will look like? Well, most of the sample means will probably be similar to the population mean of 100. Some of the sample means will be slightly lower than 100; some will be slightly higher than 100; and others will be right at 100. A few of the sample means, however, will not be similar to the population mean. Why? Based on chance, some samples will contain some of the rare individuals with either very high IQ scores or very low IQ scores. Thus, the means for those samples will be much lower than 100 or much higher than 100. Such samples, however, will be few in number. Hence, the sampling distribution (the distribution of sample means) will be normal (bell-shaped), with most of the sample means clustered around 100 and a few sample means in the tails or the extremes. Therefore, the mean for the sampling distribution will be the same as the mean for the distribution of individual scores (100).

sampling distribution A distribution of sample means based on random samples of a fixed size from a population.

The Standard Error of the Mean

Here is a more difficult question: Would the standard deviation of the sampling distribution, known as the standard error of the mean , be the same as that for a distribution of individual scores? We know that σ = 15 for the distribution of individual IQ test scores. Would the variability in the sampling distribution be as great as it is in a distribution of individual scores? Let’s think about it. The sampling distribution is a distribution of sample means. In our example, each sample has 75 people in it. Now, the mean for a sample of 75 people could never be as low or as high as the lowest or highest individual score. Why? Most people have IQ scores around 100. This means that in each of the samples, most people will have scores around 100. A few people will have very low scores, and when they are included in the sample, they will pull the mean for that sample down. A few others will have very high scores, and these scores will raise the mean for the sample in which they are included. A few people in a sample of 75, however, can never pull the mean for the sample as low as a single individual’s score might be or as high as a single individual’s score might be. For this reason, the standard error of the mean (the standard deviation for the sampling distribution) can never be as large as σ (the standard deviation for the distribution of individual scores).

standard error of the mean The standard deviation of the sampling distribution.

How does this relate to the z test? A z test uses the mean and standard deviation for the sampling distribution to determine whether the sample mean is significantly different from the population mean. Thus, we need to know the mean (μ) and the standard error of the mean(σ¯¯¯X)(σX¯)for the sampling distribution. We have already said that μ for the sampling distribution is the same as μ for the distribution of individual scores—100. How will we determine what σ¯¯¯XσX¯ is?

To find the standard error of the mean, we would need to draw a number of samples from the population, determine the mean for each sample, and then calculate the standard deviation for this distribution of sample means. This is hardly feasible. Luckily for us, there is a method of finding the standard error of the mean without doing all of this. This is based on the central limit theorem. The central limit theorem is a precise description of the distribution that would be obtained if you selected every possible sample, calculated every sample mean, and constructed the distribution of sample means. The central limit theorem states that for any population with mean μ and standard deviation σ, the distribution of sample means for sample size N will have a mean of μ and a standard deviation of σ/√Nσ/N and will approach a normal distribution as N approaches infinity. Thus, according to the central limit theorem, in order to determine the standard error of the mean (the standard deviation for the sampling distribution) we take the standard deviation for the population (σ) and divide by the square root of the sample size√N:N:

σ¯¯¯X=σ√NσX¯=σN

central limit theorem A theorem which states that for any population with mean μ and standard deviation σ, the distribution of sample means for sample size N will have a mean of μ and a standard deviation of σ/√Nσ/N and will approach a normal distribution as N approaches infinity.

We can now use this information to calculate the z test. The formula for the z test is

z=¯¯¯X−μσ¯¯¯Xz=X¯−μσX¯

where

¯¯¯XX¯ = sample mean

μ = mean of the sampling distribution

σ¯¯¯XσX¯ = standard deviation of the sampling distribution, or standard error of the mean

THE z TEST (PART I)

Concept

Description

use

Sampling Distribution

A distribution of sample means where each sample is the same size (N)

Used for comparative purposes for z tests—a sample mean is compared with the sampling distribution to assess the likelihood that the sample is part of the sampling distribution

Standard Error of the Mean (σ¯¯¯X)(σX¯)

The standard deviation of a sampling distribution, determined by dividing σ by √NN

Used in the calculation of z

z Test

Indication of the number of standard deviation units the sample mean is from the mean of the sampling distribution

An inferential test that compares a sample mean with the sampling distribution in order to determine the likelihood that the sample is part of the sampling distribution

1.Explain how a sampling distribution differs from a distribution of individual scores.

2.Explain the difference between σ¯¯¯XσX¯ and σ.

3.How is a z test different from a z score?

Calculations for the One-Tailed z Test

You can see that the formula for a z test represents finding the difference between the sample mean (¯¯¯X)(X¯) and the population mean (μ) and then dividing by the standard error of the mean (σ¯¯¯X)(σX¯). This will tell us how many standard deviation units a sample mean is from the population mean, or the likelihood that the sample is from that population. We already know μ and σ, so all we need is to find the mean for the sample (¯¯¯X)(X¯) and to calculate σ¯¯¯XσX¯ based on a sample size of 75.

Suppose we find that the mean IQ score for the sample of 75 children enrolled in academic after-school programs is 103.5. We can calculate σ¯¯¯XσX¯ based on knowing the sample size and σ.

σ¯¯¯X=σ√N=15√75=158.66=1.73σX¯=σN=1575=158.66=1.73

We now use σ¯¯¯XσX¯ (1.73) in the z test formula.

z=¯¯¯X−μσ¯¯¯X=103.5−1001.73=3.51.73=+2.02z=X¯−μσX¯=103.5−1001.73=3.51.73=+2.02

Instructions on using the TI-84 calculator to conduct this one-tailed z test appear in the Statistical Software Resources section at the end of this chapter.

Interpreting the One-Tailed z Test

Figure 9.1 represents where the sample mean of 103.5 lies with respect to the population mean of 100. The z test score of 2.02 can be used to test our hypothesis that the sample of children in the academic after-school program represents a population with a mean IQ greater than the mean IQ for the general population. To do this, we need to determine whether the probability is high or low that a sample mean as large as 103.5 would be chosen from this sampling distribution. In other words, is a sample mean IQ score of 103.5 far enough away from, or different enough from, the population mean of 100 for us to say that it represents a significant difference with an alpha level of .05 or less?

How do we determine whether a z score of 2.02 is statistically significant? Because the sampling distribution is normally distributed, we can use the area under the normal curve (Table A.1 in Appendix A). When we discussed z scores in Module 6, we saw that Table A.1 provides information on the proportion of scores falling between μ and the z score and the proportion of scores beyond the z score. To determine whether a z test is significant, we can use the area under the curve to determine whether the chance of a given score’s occurring is 5% or less. In other words, is the score far enough away from (above or below) the mean that only 5% or less of the scores are as far or farther away?

Using Table A.1, we find that the z score that marks off the top 5% of the distribution is 1.645. This is referred to as the z critical value , or zcv. For us to conclude that the sample mean is significantly different from the population mean, then, the sample mean must be at least ±1.645 standard deviations (z units) from the mean. The critical value of 1.645 is illustrated in Figure 9.2. The z we obtained for our sample mean (zobt) is 2.02, and this value falls within the region of rejection for the null hypothesis. We therefore reject H0 which states that the sample mean represents the general population mean and support our alternative hypothesis that the sample mean represents a population of children in academic after-school programs whose mean IQ is greater than 100. We make this decision because the z test score for the sample is larger than (further out in the tail than) the critical value of ±1.645. In APA style, it would be reported as follows: z (N = 75) = 2.02, p < .05 (one-tailed).

critical value The value of a test statistic that marks the edge of the region of rejection in a sampling distribution, where values equal to it or beyond it fall in the region of rejection.

FIGURE 9.1 The obtained mean in relation to the population meanFIGURE 9.2 The z critical value and the z obtained for the z test example

Keep in mind that when a result is significant, the p value (the a level, or probability of a Type I error) is reported as less than (<) .05 (or some smaller probability) not greater than (>)—an error commonly made by students. Remember the p value, or alpha level, indicates the probability of a Type I error. We want this probability to be small, meaning we are confident that there is only a small probability that our results were due to chance. This means it is highly probable that the observed difference between the sample mean and the population mean is truly a meaningful difference.

The test just conducted was a one-tailed test, because we predicted that the sample would score higher than the population. What if this were reversed? For example, imagine I am conducting a study to see whether children in athletic after-school programs weigh less than children in the general population. Can you determine what H0 and Ha are for this example?

H0:μ0≥μ1,orμweightofchildreninathleticporgrams≥μweightofchildreningeneralpopulationH0:μ0≥μ1, or μweight of children in athletic porgrams≥μweight of children in general population

H0:μ0<μ1,orμweightofchildreninathleticporgrams<μweightofchildreningeneralpopulationH0:μ0<μ1, or μweight of children in athletic porgrams<μweight of children in general population

Assume that the mean weight of children in the general population (μ) is 90 pounds, with a standard deviation (σ) of 17 pounds. You take a random sample (N = 50) of children in athletic after-school programs and find a mean weight(¯¯¯X)(X¯) of 86 pounds Given this information, you can now test the hypothesis that the sample of children in the athletic after-school program represents a population with a mean weight that is less than the mean weight for the general population of children.

First, we calculate the standard error of the mean (σ¯¯¯X)(σX¯).

σ¯¯¯X=σ√N=17√50=177.07=2.40σX¯=σN=1750=177.07=2.40

Now, we enter σ¯¯¯XσX¯ into the z test formula.

z=¯¯¯X−μσ¯¯¯X=86−902.40=−42.40=−1.67z=X¯−μσX¯=86−902.40=−42.40=−1.67

The z score for this sample mean is −1.67, meaning that it falls 1.67 standard deviations below the mean. The critical value for a one-tailed test was 1.645 standard deviations. This means the z score has to be at least 1.645 standard deviations away from (above or below) the mean in order to fall in the region of rejection. In other words, the critical value for a one-tailed z test is ±1.645.

Is our z score at least that far away from the mean? It is, but just barely. Therefore, we reject H0 and support Ha—that children in the athletic after-school programs weigh significantly less than children in the general population and hence represent a population of children who weigh less. In APA style, this would be written as z (N = 50) = −1.67, p <.05 (one-tailed). Instructions on using the TI-84 calculator to conduct this one-tailed z test appear in the Statistical Software Resources section at the end of this chapter.

Calculations for the Two-Tailed z Test

So far, we have completed two z tests, both one-tailed. Let’s turn now to a two-tailed z test. Remember that a two-tailed test is also known as a nondirectional test—a test in which the prediction is simply that the sample will perform differently from the population, with no prediction as to whether the sample mean will be lower or higher than the population mean.

Suppose that in the previous example we used a two-tailed rather than a one-tailed test. We expect the weight of the children in the athletic after-school program to differ from that of children in the general population, but we are not sure whether they will weigh less (because of the activity) or more (because of greater muscle mass). H0 and Ha for this two-tailed test appear next. See if you can determine what they would be before you continue reading.

H0:μ0=μ1,orμathleticprograms=μgeneralpopulationH0:μ0=μ1, or μathletic programs=μgeneral population

H0:μ0≠μ1,orμathleticprograms≠μgeneralpopulationH0:μ0≠μ1, or μathletic programs≠μgeneral population

Let’s use the same data as before: The mean weight of children in the general population (μ) is 90 pounds, with a standard deviation (σ) of 17 pounds; for children in the sample (N = 50), the mean weight (¯¯¯X)(X¯) is 86 pounds. Using this information, you can now test the hypothesis that children in athletic after-school programs differ in weight from those in the general population. Notice that the calculations will be exactly the same for this z test. That is, σ¯¯¯XσX¯ and the z score will be exactly the same as before. Why? All of the measurements are exactly the same. To review:

σ¯¯¯X=σ√N=17√50=177.07=2.40σX¯=σN=1750=177.07=2.40

z=¯¯¯X−μσ¯¯¯X=86−902.40=−42.40=−1.67z=X¯−μσX¯=86−902.40=−42.40=−1.67

Interpreting the Two-Tailed z Test

If we end up with the same z score, how does a two-tailed test differ from a one-tailed test? The difference is in the z critical value (zcv). In a two-tailed test, both halves of the normal distribution have to be taken into account. Remember that with a one-tailed test, the zcv was ±1.645; this z score was so far away from the mean (either above or below) that only 5% of the scores were beyond it. How will the zcv for a two-tailed test differ?

With a two-tailed test, the zcv has to be so far away from the mean that a total of only 5% of the scores are beyond it (both above and below the mean). A zcv of ±1.645 leaves 5% of the scores above the positive zcv and 5% below the negative zcv. If we take both sides of the normal distribution into account (which we do with a two-tailed test because we do not predict whether the sample mean will be above or below the population mean), then 10% of the distribution will fall beyond the two critical values. Thus, ±1.645 cannot be the critical value for a two-tailed test because this leaves too much chance (10%) operating.

To determine the zcv for a two-tailed test, then, we need to find the z score that is far enough away from the population mean that only 5% of the distribution—taking into account both halves of the distribution—is beyond the score. Because Table A.1 represents only half of the distribution, we need to look for the z score that leaves only 2.5% of the distribution beyond it. Then, when we take into account both halves of the distribution, 5% of the distribution will be accounted for (2.5% + 2.5% = 5%). Can you determine what zscore this would be, using Table A.1?

If you concluded that it would be ±1.96, then you are correct. This is the z score that is far enough away from the population mean (using both halves of the distribution) that only 5% of the distribution is beyond it. The critical values for both one- and two-tailed tests are illustrated in Figure 9.3.

FIGURE 9.3 Regions of rejection and critical values for one-tailed versus two-tailed testsFIGURE 9.4 The z critical value and the z obtained for the two-tailed z test example

Okay, what do we do with this critical value? We use it exactly the same way as we did the zcv for a one-tailed test. In other words, the zobt has to be as large as or larger than the zcv in order for us to reject H0. Is our zobt as large as or larger than ±1.96? No (this is illustrated in Figure 9.4). Our zobt was −1.67 and not in the region of rejection. We therefore fail to reject H0 and conclude that the weight of children in the athletic after-school program does not differ significantly from the weight of children in the general population. Instructions on using the TI-84 calculator to conduct this two-tailed z test appear in the Statistical Software Resources section at the end of this chapter.

With exactly the same data (sample size, μ, σ, ¯¯¯XX¯, and σ¯¯¯XσX¯), we rejected H0 using a one-tailed test and failed to reject H0 with a two-tailed test. How can this be? The answer is that a one-tailed test is statistically a more powerful test than a two-tailed test. Statistical power refers to the probability of correctly rejecting a false H0. With a one-tailed test, you are more likely to reject H0 because the zobt does not have to be as large (as far away from the population mean) to be considered significantly different from the population mean. (Remember, the zcv for a one-tailed test is ±1.645, but for a two-tailed test, it is ±1.96.)

statistical power The probability of correctly rejecting a false H0.

Statistical Power

Let’s think back to the discussion of Type I and Type II errors in the previous module. We said that in order to reduce your risk of a Type I error, you need to lower the alpha level—for example, from .05 to .01. We also noted, however, that lowering the alpha level increases the risk of a Type II error. How, then, can we reduce our risk of a Type I error but not increase our risk of a Type II error? As we just noted, a one-tailed test is more powerful—you do not need as large a zcv in order to reject H0. Here, then, is one way to maintain an alpha level of .05 but increase your chances of rejecting H0. Of course, ethically you cannot simply choose to adopt a one-tailed test for this reason. The one-tailed test should be adopted because you truly believe that the sample will perform above (or below) the mean.

By what other means can we increase statistical power? Look back at the z test formula. We know that the larger the zobt, the greater the chance that it will be significant (as large as or larger than the zcv) and that we can therefore reject H0. What could we change in our study that might increase the size of the zobt? Well, if the denominator in the z formula were a smaller number, then the zobt would be larger and more likely to fall in the region of rejection. How can we make the denominator smaller? The denominator is σ¯¯¯XσX¯. Do you remember the formula for σ¯¯¯XσX¯?

σ¯¯¯X=σ√NσX¯=σN

It is very unlikely that we can change or influence the standard deviation for the population (σ). What part of the σ¯¯¯XσX¯ formula can we influence? The sample size (N).

If we increase sample size, what will happen to σ¯¯¯XσX¯? Let’s see. We’ll use the same example as before, a two-tailed test with all of the same measurements. The only difference will be in sample size. Thus, the null and alternative hypotheses will be

H0:μ0=μ1,orμweightofchildreninathleticafter-schoolprograms=μweightofchildreningeneralpopulationH0:μ0≠μ1,orμweightofchildreninathleticafter-schoolprograms≠μweightofchildreningeneralpopulationH0:μ0=μ1, or    μweight of children in athletic after-school programs = μweight of children in general populationH0:μ0≠μ1, or    μweight of children in athletic after-school programs ≠ μweight of children in general population

The mean weight of children in the general population (μ) is once again 90 pounds, with a standard deviation (σ) of 17 pounds, and the sample of children in the after-school program again has a mean (¯¯¯X)(X¯) weight of 86 pounds. The only difference will be in sample size. In this case, our sample has 100 children in it. Let’s test the hypothesis (conduct the z test) for these data.

σ¯¯¯X=17√100=1710=1.70σX¯=17100=1710=1.70

z=86−901.70=−41.70=−2.35z=86−901.70=−41.70=−2.35

Do you see what happened when we increased sample size? The standard error of the mean (σ¯¯¯X)(σX¯) decreased (we will discuss why in a minute), and the zobt increased—in fact, it increased to the extent that we can now reject H0 with this two-tailed test because our zobt of −2.35 is further away from the mean than the zcv of −1.96. Therefore, another way to increase statistical power is to increase sample size.

Why does increasing sample size decrease (σ¯¯¯X)(σX¯)? Well, you can see why based on the formula, but let’s think back to our earlier discussion about σ¯¯¯XσX¯ We said that it was the standard deviation for a sampling distribution—a distribution of sample means of a set size. If you recall the IQ example we used in our discussion of σ¯¯¯XσX¯ and the sampling distribution, we said that μ = 100 and σ = 15. We discussed what σ¯¯¯XσX¯ would be for a sampling distribution in which each sample mean was based on a sample size of 75. We further noted that σ¯¯¯XσX¯ would always be smaller (have less variability) than σ because it represents the standard deviation of a distribution of sample means, not a distribution of individual scores. What, then, will increasing sample size do to σ¯¯¯XσX¯? If each sample in the sampling distribution had 100 people in it rather than 75, what do you think this would do to the distribution of sample means?

As we noted earlier, most people in a sample will be close to the mean (100), with only a few people in each sample representing the tails of the distribution. If we increase sample size to 100, we will have 25 more people in each sample. Most of them will probably be close to the population mean of 100; therefore, each sample mean will probably be closer to the population mean of 100. Thus, a sampling distribution based on samples of N = 100 rather than N = 75 will have less variability, which means that σ¯¯¯XσX¯ will be smaller.

Assumptions and Appropriate Use of the z Test

As noted earlier in the module, the z test is a parametric inferential statistical test for hypothesis testing. Parametric tests involve the use of parameters or population characteristics. With a z test, the parameters, such as m and s, are known. If they are not known, the z test is not appropriate. Because the z test involves the calculation and use of a sample mean, it is appropriate for use with interval or ratio data. In addition, because we use the area under the normal curve (Table A.1), we are assuming that the distribution of random samples is normal. Small samples often fail to form a normal distribution. Therefore, if the sample size is small (N < 30), the z test may not be appropriate. In cases where the sample size is small, or where s is not known, the appropriate test would be the t test, discussed in the next module.

THE z TEST (PART II)

Concept

Description

Examples

One-Tailed z Test

A directional inferential test in which a prediction is made that the population represented by the sample will be either above or below the general population

Ha:μ0<μ1orHa:μ0>μ1Ha:μ0<μ1orHa:μ0>μ1

Two-Tailed z Test

A nondirectional inferential test in which the prediction is made that the population represented by the sample will differ from the general population, but the direction of the difference is not predicted

Ha:μ0≠μ1Ha:μ0≠μ1

Statistical Power

The probability of correctly rejecting a false H0

One-tailed tests are more powerful; increasing sample size increases power

1.Imagine that I want to compare the intelligence level of psychology majors to the intelligence level of the general population of college students. I predict that psychology majors will have higher IQ scores. Is this a one- or two-tailed test? Identify H0 and Ha.

2.Conduct the z test for the previous example. Assume that μ = 100, σ = 15, ¯¯¯XX¯ = 102.75, and N = 60. Should we reject H0 or fail to reject H0?

Confidence Intervals Based on the z Distribution

In this text, hypothesis tests such as the previously described z test are the main focus. However, sometimes social and behavioral scientists use estimation of population means based on confidence intervals rather than statistical hypothesis tests. For example, imagine that you want to estimate a population mean based on sample data (a sample mean). This differs from the previously described z test in that we are not determining whether the sample mean differs significantly from the population mean; rather, we are estimating the population mean based on knowing the sample mean. We can still use the area under the normal curve to accomplish this—we simply use it in a slightly different way.

Let’s use the previous example in which we know the sample mean weight of children enrolled in athletic after-school programs (¯¯¯XX¯ = 86), σ (17), and the sample size (N = 100). However, imagine that we do not know the population mean (μ). In this case, we can calculate a confidence interval based on knowing the sample mean and s. A confidence interval is an interval of a certain width, which we feel “confident” will contain μ. We want a confidence interval wide enough that we feel fairly certain it contains the population mean. For example, if we want to be 95% confident, we want a 95% confidence interval.

confidence interval An interval of a certain width that we feel confident will contain μ.

How can we use the area under the standard normal curve to determine a confidence interval of 95%? We use the area under the normal curve to determine the z scores that mark off the area representing 95% of the scores under the curve. If you consult Table A.1, you will find that 95% of the scores will fall between ±1.96 standard deviations above and below the mean. Thus, we could determine which scores represent ±1.96 standard deviations from the mean of 86. This seems fairly simple, but we must remember that we are dealing with a distribution of sample means (the sampling distribution) and not with a distribution of individual scores. Thus, we must convert the standard deviation (σ) to the standard error of the mean (σ¯¯¯XσX¯, the standard deviation for a sampling distribution) and use the standard error of the mean in the calculation of a confidence interval. Remember we calculate σ¯¯¯XσX¯ by dividing σ by the square root of N.

σ¯¯¯X=17√100=1710=1.70σX¯=17100=1710=1.70

We can now calculate the 95% confidence interval using the following formula:

CI=¯¯¯X±z(σ¯¯¯X)CI=X¯ ± z(σX¯)

where

¯¯¯X=thesamplemeanσ¯¯¯X=thestandarderrorofthemean,andz=thezscorerepresentingthedisiredconfidenceintervalX¯=the sample meanσX¯=the standard​ error of the ​mean, andz=the z score representing the disired confidence interval

Thus:

Cl=86±1.96(1.70)=86±3.332=82.668−89.332Cl=86 ±1.96(1.70)=86±3.332=82.668−89.332

Thus, the 95% confidence interval ranges from 82.67 to 89.33. We would conclude, based on this calculation, that we are 95% confident that the population mean lies within this interval.

What if we wanted to have greater confidence that our population mean is contained in the confidence interval? In other words, what if we want to be 99% confident? We would have to construct a 99% confidence interval. How would we go about doing this?

We would do exactly what we did for the 95% confidence interval. First, we would consult Table A.1 to determine what z scores mark off 99% of the area under the normal curve. We find that z scores of ±2.58 mark off 99% of the area under the curve. We then apply the same formula for a confidence interval used previously.

Cl=¯¯¯X±z(σ¯¯x)Cl=86±2.58(1.70)=86±4.386=81.614−90.386Cl=X¯±z(σx¯)Cl=86±2.58(1.70)=86±4.386=81.614−90.386

Thus, the 99% confidence interval ranges from 81.61 to 90.39. We would conclude, based on this calculation, that we are 99% confident that the population mean lies within this interval.

Typically, statisticians recommend using a 95% or a 99% confidence interval. However, using Table A.1 (the area under the normal curve), you could construct a confidence interval of 55%, 70%, or any percentage you desire.

It is also possible to do hypothesis testing with confidence intervals. For example, if you construct a 95% confidence interval based on knowing a sample mean and then determine that the population mean is not in the confidence interval, the result is significant. For example, the 95% confidence interval we constructed earlier of 82.67−89.33 did not include the actual population mean reported earlier in the module (μ = 90). Thus, there is less than a 5% chance that this sample mean could have come from this population—the same conclusion we reached when using the z test earlier in the module.

REVIEW OF KEY TERMS

central limit theorem (p. 142)

confidence interval (p. 151)

critical value (p. 144)

sampling distribution (p. 141)

standard error of the mean (p. 141)

statistical power (p. 148)

z test (p. 140)

MODULE EXERCISES

(Answers to odd-numbered questions appear in Appendix B.)

1.What is a sampling distribution?

2.Explain why the mean of a sampling distribution is the same as the mean for a distribution of individual scores, but the standard deviation for a sampling distribution differs from the standard deviation of a distribution of individual scores.

3.Explain why the zcv for a one-tailed test differs from that for a two-tailed test.

4.Explain what statistical power is and how we can increase power.

5.A researcher is interested in whether students who attend private high schools have higher average SAT scores than students in the general population of high school students. A random sample of 90 students at a private high school is tested and has a mean SAT score of 1,050. The average for public high school students is 1,000 (σ = 200).

a.Is this a one- or two-tailed test?

b.What are H0 and Ha for this study?

c.Compute zobt.

d.What is zcv?

e.Should H0 be rejected? What should the researcher conclude?

f.Determine the 95% confidence interval for the population mean, based on the sample mean.

6.The producers of a new toothpaste claim that it prevents more cavities than other brands of toothpaste. In other words, those who use the new toothpaste should have fewer cavities than those who use other brands. A random sample of 60 people uses the new toothpaste for six months. The mean number of cavities at their next checkup is 1.5. In the general population, the mean number of cavities at a six-month checkup is 1.73 (σ = 1.12).

a.Is this a one- or two-tailed test?

b.What are H0 and Ha for this study?

c.Compute zobt.

d.What is zcv?

e.Should H0 be rejected? What should the researcher conclude?

f.Determine the 95% confidence interval for the population mean, based on the sample mean.

CRITICAL THINKING CHECK ANSWERS

Critical Thinking Check 9.1

1.A sampling distribution is a distribution of sample means. Thus, rather than representing scores for individuals, the sampling distribution plots the means of samples of a set size.

2.σ¯¯¯XσX¯ is the standard deviation for a sampling distribution. It therefore represents the standard deviation for a distribution of sample means. σ is the standard deviation for a population of individual scores rather than sample means.

3.A z test compares the performance of a sample to the performance of the population by indicating the number of standard deviation units the sample mean is from the mean of the sampling distribution. A z score indicates how many standard deviation units an individual score is from the population mean.

CRITICAL THINKING CHECK 9.2

1.Predicting that psychology majors will have higher IQ scores makes this a one-tailed test.

H0:μpsychologymajors≤μgeneralpopulationHa:μpsychologymajors>μgeneralpopulationH0:μpsychology majors ≤μgeneral populationHa:μpsychology majors >μgeneral population

2.σ¯¯¯X=15√60=157.75=1.94σX¯=1560=157.75=1.94

z=102.75−1001.94=2.751.94=+1.42z=102.75−1001.94=2.751.94=+1.42

Because this is a one-tailed test, zcv = ±1.645. The zobt = +1.42. We therefore fail to reject H0; psychology majors do not differ significantly on IQ scores in comparison to the general population of college students.

MODULE 10

The Single-Sample t Test

Learning Objectives

•Explain what a t test is and what it does.

•Calculate a t test.

•List the assumptions of the t test.

•Calculate confidence intervals using the t distribution.

The t Test: What It Is and What It Does

The t test for a single sample is similar to the z test in that it is also a parametric statistical test of the null hypothesis for a single sample. As such, it is a means of determining the number of standard deviation units a score is from the mean (μ) of a distribution. With a t test, however, the population variance is not known. Another difference is that tdistributions, although symmetrical and bell-shaped, are not normally distributed. This means that the areas under the normal curve that apply for the z test do not apply for the t test.

t test A parametric inferential statistical test of the null hypothesis for a single sample where the population variance is not known.

Student’s t Distribution

The t distribution, known as Student’s t distribution , was developed by William Sealey Gosset, a chemist, who worked for the Guinness Brewing Company of Dublin, Ireland, at the beginning of the 20th century. Gosset noticed that when working with small samples of beer (N < 30) chosen for quality-control testing, the sampling distribution of the means was symmetrical and bell-shaped, but not normal. Therefore, the proportions under the standard normal curve did not apply. In other words, with small sample sizes, the curve was symmetrical, but it was not the standard normal curve. As the size of the samples in the sampling distribution increased, the sampling distribution approached the normal distribution and the proportions under the curve became more similar to those under the standard normal curve. He eventually published his findings under the pseudonym “Student”; and with the help of Karl Pearson, a mathematician, he developed a general formula for the t distributions (Peters, 1987; Stigler, 1986; Tankard, 1984).

Student’s t distribution A set of distributions that, although symmetrical and bell-shaped, are not normally distributed.

We refer to t distributions in the plural because unlike the z distribution, of which there is only one, the t distributions are a family of symmetric distributions that differ for each sample size. As a result, the critical value indicating the region of rejection changes for samples of different sizes. As the size of the samples increases, the t distribution approaches the z or normal distribution. Table A.2 in Appendix A at the back of your book provides the t critical values (t cv) for both one- and two-tailed tests for various sample sizes and alpha levels. Notice, however, that although we have said that the critical value depends on sample size, there is no column in the table labeled N for sample size. Instead, there is a column labeled df, which stands for degrees of freedom . The degrees of freedom are related to sample size. For example, assume that you are given six numbers, 2, 5, 6, 9, 11, and 15. The mean of these numbers is 8. If you are told that you can change the numbers as you like, but that the mean of the distribution must remain at 8, how many numbers can you change arbitrarily? You can change five of the six numbers arbitrarily. Once you have changed five of the numbers arbitrarily, the sixth number is determined by the qualification that the mean of the distribution must equal 8. Therefore, in this distribution of six numbers, five are free to vary. Thus, there are five degrees of freedom. For any single distribution, then, df= N − 1.

degrees of freedom (df) The number of scores in a sample that are free to vary.

Look again at Table A.2, and notice what happens to the critical values as the degrees of freedom increase. Look at the column for a one-tailed test with alpha equal to .05 and degrees of freedom equal to 10. The critical value is ±1.812. This is larger than the critical value for a one-tailed z test, which was ±1.645. Because we are dealing with smaller, nonnormal distributions when using the t test, the t score must be farther away from the mean in order for us to conclude that it is significantly different from the mean. What happens as the degrees of freedom increase? Look in the same column—one-tailed test, alpha = .05—for 20 degrees of freedom. The critical value is ±1.725, smaller than the critical value for 10 degrees of freedom. Continue to scan down the same column, one-tailed test and alpha=.05, until you reach the bottom, where df = ∞. Notice that the critical value is ±1.645, the same as it is for a one-tailed z test. Thus, when the sample size is large, the t distribution is the same as the z distribution.

Calculations for the One-Tailed t Test

Let’s illustrate the use of the single-sample t test to test a hypothesis. Assume the mean SAT score of students admitted to General University is 1,090. Thus, the university mean of 1,090 is the population mean (μ). The population standard deviation is unknown. The members of the Biology Department believe that students who decide to major in biology have higher SAT scores than the general population of students at the university. The null and alternative hypotheses are thus

H0:μ0≤μ1,orμbiologystudents≤μgeneralpopulationHa:μ0≤μ1,orμbiologystudents>μgeneralpopulationH0:μ0≤μ1, or μbiology students ≤μgeneral populationHa:μ0≤μ1, or μbiology students >μgeneral population

TABLE 10-1 SAT scores for a sample of 10 biology majors

x

1,010

1,200

1,310

1,075

1,149

1,078

1,129

1,069

1,350

1,390

ΣX = 11,760

¯¯¯X=ΣXN=11,76010=1,176X¯=ΣXN=11,76010=1,176

Notice that this is a one-tailed test because the researchers predict that the biology students will perform higher than the general population of students at the university. The researchers now need to obtain the SAT scores for a sample of biology majors. This information is provided in Table 10.1, which shows that the mean SAT score for the sample is 1,176. This sample mean represents our estimate of the population mean SAT score for biology majors.

The Estimated Standard Error of the Mean

The t test will tell us whether this mean differs significantly from the university mean of 1,090. Because we have a small sample (N = 10) and because we do not know σ, we must conduct a t test rather than a z test. The formula for the t test is

t=¯¯¯X−μS¯¯¯Xt=X¯−μSX¯

This looks very similar to the formula for the z test that we used in Module 9. The only difference is in the denominator, where S¯¯¯XSX¯ (the estimated standard error of the mean of the sampling distribution) has been substituted forσ¯¯¯XσX¯. We useS¯¯¯XSX¯ rather than σ¯¯¯XσX¯ because we do not know σ (the standard deviation for the population) and thus cannot calculate σ¯¯¯XσX¯. We can, however, determine s (the unbiased estimator of the population standard deviation) and, based on this, we can determine S¯¯¯XSX¯. The formula for S¯¯¯XSX¯ is

estimated standard error of the mean An estimate of the standard deviation of the sampling distribution.

S¯¯¯X=s√NSX¯=sN

We must first calculate s (the estimated standard deviation for a population, based on sample data) and then use this to calculate the estimated standard error of the mean (S¯¯¯X)(SX¯). The formula for s, which we learned in Module 3, is

S=√Σ(X−¯¯¯X)2N−1S=Σ(X−X¯)2N−1

Using the information in Table 10.1, we can use this formula to calculate s.

S=√156,3529=√17,372.44=131.80S=156,3529=17,372.44=131.80

Thus, the unbiased estimator of the standard deviation (s) is 131.80. We can now use this value to calculate S¯¯¯XSX¯, the estimated standard error of the sampling distribution.

S¯¯¯X=S√N=131.80√10=131.803.16=41.71SX¯=SN=131.8010=131.803.16=41.71

Finally, we can use this value for S¯¯¯XSX¯ to calculate t.

t=¯¯¯X−μS¯¯¯X=1,176−1,09041.71=8641.71=+2.06t=X¯−μSX¯=1,176−1,09041.71=8641.71=+2.06

Interpreting the One-Tailed t Test

Our sample mean falls 2.06 standard deviations above the population mean of 1,090. We must now determine whether this is far enough away from the population mean to be considered significantly different. In other words, is our sample mean far enough away from the population mean that it lies in the region of rejection? Because this is a one-tailed alternative hypothesis, the region of rejection is in one tail of the sampling distribution. Consulting Table A.2 for a one-tailed test with alpha =.05 and df = N − 1 = 9, we see that tcv = ±1.833. The tobt of 2.06 is therefore within the region of rejection. We reject H0 and support Ha. In other words, we have sufficient evidence to allow us to conclude that biology majors have significantly higher SAT scores than the rest of the students at General University. In APA style, this would be reported as t(9) = 2.06, p < .05 (one-tailed). Figure 10.1illustrates the obtained t with respect to the region of rejection. Instructions on using Excel, SPSS, or the TI-84 calculator to conduct this one-tailed single sample t test appear in the Statistical Software Resources section at the end of this chapter.

FIGURE 10.1 The t critical value and the t obtained for the single-sample one-tailed t test example

Calculations for the Two-Tailed t Test

What if the Biology Department had made no directional prediction concerning the SAT scores of its students? In other words, suppose the members of the department were unsure whether their students’ scores would be higher or lower than those of the general population of students and were simply interested in whether biology students differed from the population. In this case, the test of the alternative hypothesis would be two-tailed, and the null and alternative hypotheses would be

H0:μ0=μ1,orμbiologystudents=μgeneralpopulationHa:μ0=μ1,orμbiologystudents≠μgeneralpopulationH0:μ0=μ1, or μbiology students=μgeneral populationHa:μ0=μ1, or μbiology students≠μgeneral population

Assuming that the sample of biology students is the same, ¯¯¯XX¯, s, and S¯¯¯XSX¯ would all be the same. The population at General University is also the same, so μ would still be 1,090. Using all of this information to conduct the t test, we end up with exactly the same t test score of + 2.06. What, then, is the difference for the two-tailed t test? It is the same as the difference between the one- and two-tailed z test—the critical values differ.

Interpreting the Two-Tailed t Test

Remember that with a two-tailed alternative hypothesis, the region of rejection is divided evenly between the two tails (the positive and negative ends) of the sampling distribution. Consulting Table A.2 for a two-tailed test with alpha = .05 and df= N − 1 = 9, we see that tcv = ±2.262. The tobt of 2.06 is therefore not within the region of rejection. We do not reject H0 and thus cannot support Ha. In other words, we do not have sufficient evidence to allow us to conclude that the population of biology majors differs significantly on SAT scores from the rest of the students at General University. Thus, with exactly the same data, we rejected H0 with a one-tailed test, but failed to reject H0 with a two-tailed test, illustrating once again that one-tailed tests are more powerful than two-tailed tests. Figure 10.2 illustrates the obtained t for the two-tailed test in relation to the regions of rejection. Instructions on using Excel, SPSS, or the TI-84 calculator to conduct this two-tailed single-sample t test appear in the Statistical Software Resources section at the end of this chapter.

FIGURE 10.2 The t critical value and the t obtained for the single-sample two-tailed test example

Assumptions and Appropriate Use of the Single-Sample t Test

The t test is a parametric test, as is the z test. As a parametric test, the t test must meet certain assumptions. These assumptions include that the data are interval or ratio and that the population distribution of scores is symmetrical. The t test is used in situations that meet these assumptions and in which the population mean is known but the population standard deviation (σ) is not known. In cases where these criteria are not met, a nonparametric test is more appropriate. Nonparametric tests are covered in Chapter 10.

THE t TEST

Concept

Description

Use/Examples

Estimated standard error of the mean (S¯¯¯X)(SX¯)

The estimated standard deviation of a sampling distribution, calculated by dividing s by √NN

Used in the calculation of a t test

t test

Indicator of the number of standard deviation units the sample mean is from the mean of the sampling distribution

An inferential statistical test that differs from the z test in that the sample size is small (usually <30) and σ is not known

One-tailed t test

A directional inferential test in which a prediction is made that the population represented by the sample will be either above or below the general population

Ha:μ0<μ1orHa:μ0>μ1Ha:μ0<μ1orHa:μ0>μ1

Two-tailed t test

A nondirectional inferential test in which the prediction is made that the population represented by the sample will differ from the general population, but the direction of the difference is not predicted

Ha:μ0≠μ1Ha:μ0≠μ1

1.Explain the difference in use and computation between the z test and the t test.

2.Test the following hypothesis using the t test: Researchers are interested in whether the pulse of long-distance runners differs from that of other athletes. They suspect that the runners’ pulses will be lower. They obtain a random sample (N = 8) of long-distance runners, measure their resting pulse, and obtain the following data: 45, 42, 64, 54, 58, 49, 47, 55. The average resting pulse of athletes in the general population is 60 beats per minute.

Confidence Intervals Based on the t Distribution

You might remember from our discussion of confidence intervals in Module 9 that they allow us to estimate population means based on sample data (a sample mean). Thus, when using confidence intervals, rather than determining whether the sample mean differs significantly from the population mean, we are estimating the population mean based on knowing the sample mean. We can use confidence intervals with the t distribution just as we did with the z distribution (the area under the normal curve).

Let’s use the previous example in which we know the sample mean SAT score for the biology students (¯¯¯XX¯= 1,176), the estimated standard error of the mean (S¯¯¯XSX¯= 41.71), and the sample size (N = 10). We can calculate a confidence interval based on knowing the sample mean and S¯¯¯XSX¯. Remember that a confidence interval is an interval of a certain width, which we feel “confident” will contain μ. We are going to calculate a 95% confidence interval—in other words, an interval that we feel 95% confident contains the population mean. In order to calculate a 95% confidence interval using the t distribution, we use Table A.2 (Critical Values for the Student’s t Distribution) to determine the critical value of t at the .05 level. We use the .05 level because 1 minus alpha tells us how confident we are, and in this case 1 − alpha is 1 − .05 = 95%.

For a one-sample t test, the confidence interval is determined with the following formula:

CI=¯¯¯X±tCV(S¯¯¯X)CI=X¯ ± tCV(SX¯)

We already know ¯¯¯XX¯ (1,176) and S¯¯¯XSX¯ (41.71), so all we have left to determine is tcv. We use Table A.2 to determine the tcv for the .05 level and a two-tailed test. We always use the tcvfor a two-tailed test because we are describing values both above and below the mean of the distribution. Using Table A.2, we find that the tcv for 9 degrees of freedom (remember df= N − 1) is 2.262. We now have all of the values we need to determine the confidence interval.

Let’s begin by calculating the lower limit of the confidence interval:

CI=1,176−2.262(41.71)=1,176−94.35=1,081.65CI=1,176−2.262(41.71)=1,176−94.35=1,081.65

The upper limit of the confidence interval is

CI=1,176+2.262(41.71)=1,176+94.35=1,270.35CI=1,176+2.262(41.71)=1,176+94.35=1,270.35

Thus, we can conclude that we are 95% confident that the interval of SAT scores from 1,081.65 to 1,270.35 contains the population mean (μ).

As with the z distribution, we can calculate confidence intervals for the t distribution that give us greater or less confidence (for example a 99% confidence interval or a 90% confidence interval). Typically, statisticians recommend using either the 95% or 99% confidence interval (the intervals corresponding to the .05 and .01 alpha levels in hypothesis testing). You have likely encountered such intervals in real life. They are usually phrased in terms of “plus or minus” some amount, called the margin of error. For example, when a newspaper reports that a sample survey showed that 53% of the viewers support a particular candidate, the margin of error is typically also reported—for example, “with a ±3% margin of error.” This means that the researchers who conducted the survey created a confidence interval around the 53% and that if they actually surveyed the entire population, μ would be within ±3% of the 53%. In other words, they believe that between 50% and 56% of the viewers support this particular candidate.

REVIEW OF KEY TERMS

degrees of freedom (df) (p. 156)

estimated standard error of the mean (p. 157)

Student’s t distribution (p. 155)

t test (p. 155)

MODULE EXERCISES

(Answers to odd-numbered questions appear in Appendix B.)

1.Explain how a t test differs from a z test.

2.Explain how S¯¯¯XSX¯ differs from σ¯¯¯XσX¯.

3.Why does tcv change when sample size changes? What must be computed in order to determine tcv?

4.Henry performed a two-tailed test for an experiment in which N = 24. He could not find his t table, but he remembered the tcv at df = 13. He decided to compare his tobt to this tcv . Is he more likely to make a Type I or a Type II error in this situation?

5.A researcher hypothesizes that people who listen to music via headphones have greater hearing loss and will thus score lower on a hearing test than those in the general population. On a standard hearing test, μ = 22.5. The researcher gives this same test to a random sample of 12 individuals who regularly use headphones. Their scores on the test are 16, 14, 20, 12, 25, 22, 23, 19, 17, 17, 21, 20.

a.Is this a one- or two-tailed test?

b.What are H0 and Ha for this study?

c.Compute tobt.

d.What is tcv?

e.Should H0 be rejected? What should the researcher conclude?

f.Determine the 95% confidence interval for the population mean, based on the sample mean.

6.A researcher hypothesizes that individuals who listen to classical music will score differently from the general population on a test of spatial ability. On a standardized test of spatial ability, μ = 58. A random sample of 14 individuals who listen to classical music is given the same test. Their scores on the test are 52, 59, 63, 65, 58, 55, 62, 63, 53, 59, 57, 61, 60, 59.

a.Is this a one- or two-tailed test?

b.What are H0 and Ha for this study?

c.Compute t obt.

d.What is tcv?

e.Should H0 be rejected? What should the researcher conclude?

f.Determine the 95% confidence interval for the population mean, based on the sample mean.

CRITICAL THINKING CHECK ANSWERS

CRITICAL THINKING CHECK 10.1

1.The z test is used when the sample size is greater than 30 and thus normally distributed, and σ

is known. The t test, on the other hand, is used when the sample is smaller than 30 and bell-shaped but not normal, and s is not known.

2.For this sample,

H0:μrunners≥μotherathletesH0:μrunners<μotherathletesH0:μrunners≥μother athletesH0:μrunners<μother athletes

¯¯¯X=51.75s=7.32μ=60s¯¯¯X=7.32√8=7.322.83=2.59t=51.75−602.59=−8.252.59=−3.19df=8−1=7tCV=±1.895tobt=−3.19X¯=51.75s=7.32μ=60sX¯=7.328=7.322.83=2.59t=51.75−602.59=−8.252.59=−3.19df=8−1=7tCV=±1.895tobt=−3.19

Reject H0. The runners’ pulses are significantly lower than the pulses of athletes in general.

CHAPTER FIVE SUMMARY AND REVIEW

The z and t Tests

CHAPTER SUMMARY

Two parametric statistical tests were described in this chapter—the z test and the t test. Each compares a sample mean to the general population. Because both are parametric tests, the distributions should be bell-shaped and certain parameters should be known (in the case of the z test, μ and σ must be known; for the t test, only μ is needed). In addition, because they are parametric tests, the data should be interval or ratio in scale. These tests use the sampling distribution (the distribution of sample means). They also use the standard error of the mean (or estimated standard error of the mean for the t test), which is the standard deviation of the sampling distribution. Both z and t tests can test one- or two-tailed alternative hypotheses, but one-tailed tests are more powerful statistically.

CHAPTER 5 REVIEW EXERCISES

(Answers to exercises appear in Appendix B.)

Fill-in Self-Test

Answer the following questions. If you have trouble answering any of the questions, restudy the relevant material before going on to the multiple-choice self-test.

1.A ______________ is a distribution of sample means based on random samples of a fixed size from a population.

2.The______________ is the standard deviation of the sampling distribution.

3.A ______________ test is used when σ and μ are known and the sample is 30 or larger.

4.The set of distributions that, although symmetrical and bell-shaped, are not normally distributed is called the ______________.

5.The ______________ is a parametric statistical test of the null hypothesis for a single sample where the population variance is not known.

6.A ______________ test is used when μ is known but not σ and the sample is 30 or less.

Multiple-Choice Self-Test

Select the single best answer for each of the following questions. If you have trouble answering any of the questions, restudy the relevant material.

1.The sampling distribution is a distribution of

a.sample means.

b.population means.

c.sample standard deviations.

d.population standard deviations.

2.A one-tailed z test, p = .05, is to ________ and a two-tailed z test, p = .05, is to ________.

a.±1.645; ±1.96

b.±1.96; ±1.645

c.Type I error; Type II error

d.Type II error; Type I error

3.Which of the following is an assumption of the z test?

a.The data should be ordinal or nominal.

b.The population distribution of scores should be normal.

c.The population mean (μ) is known, but not the standard deviation (σ).

d.The sample size is typically less than 30.

4.Which type of test is more powerful?

a.A directional test

b.A nondirectional test

c.A two-tailed test

d.All of the alternatives are equally powerful.

5.For a one-tailed test with df = 11, what is tcv?

a.1.796

b.2.201

c.1.363

d.1.895

6.Which of the following is an assumption of the t test?

a.The data should be ordinal or nominal.

b.The population distribution of scores should be skewed.

c.The population mean (μ) and standard deviation (σ) are known.

d.The sample size is typically less than 30.

Self-Test Problems

1.A researcher is interested in whether students who play chess have higher average SAT scores than students in the general population. A random sample of 75 students who play chess is tested and has a mean SAT score of 1,070. The population average is 1,000 (σ = 200).

a.Is this a one- or two-tailed test?

b.What are H0 and Ha for this study?

c.Compute zobt.

d.What is zcv?

e.Should H0 be rejected? What should the researcher conclude?

f.Determine the 95% confidence interval for the population mean, based on the sample mean.

2.A researcher hypothesizes that people who listen to classical music have higher concentration skills than those in the general population. On a standard concentration test, the overall mean is 15.5. The researcher gave this same test to a random sample of 12 individuals who regularly listen to classical music. Their scores on the test were as follows:

16, 14, 20, 12, 25, 22, 23, 19, 17, 17, 21, 20

a.Is this a one- or two-tailed test?

b.What are H0 and Ha for this study?

c.Compute tobt.

d.What is tcv?

e.Should H0 be rejected? What should the researcher conclude?

f.Determine the 95% confidence interval for the population mean, based on the sample mean.

CHAPTER FIVE

Statistical Software Resources

If you need help getting started with Excel or SPSS, please see Appendix C: Getting Started with Excel and SPSS.

MODULE 9 Single-Sample z Test

The problems we’ll be using to illustrate how to calculate the single-sample z test appear in Module 9.

Using the TI-84 for the One-tailed z Test from Module 9

The z test will be used to test the hypothesis that the sample of children in the academic after-school programs represents a population with a mean IQ greater than the mean IQ for the general population. To do this, we need to determine whether the probability is high or low that a sample mean as large as 103.5 would be chosen from this sampling distribution. In other words, is a sample mean IQ score of 103.5 far enough away from, or different enough from, the population mean of 100 for us to say that it represents a significant difference with an alpha level of .05 or less?

1.With the calculator on, press the STAT key.

2.Highlight TESTS.

3.1: Z-Test will be highlighted. Press ENTER.

4.Highlight STATS. Press ENTER.

5.Scroll down to μ0: and enter the mean for the population (100).

6.Scroll down to σ: and enter the standard deviation for the population (15).

7.Scroll down to X: and enter the mean for the sample (103.5).

8.Scroll down to n: and enter the sample size (75).

9.Lastly, scroll down to μ: and select the type of test (one-tailed), indicating that we expect the sample mean to be greater than the population mean (select >μ0). Press ENTER.

10.Highlight CALCULATE and press ENTER.

The z score of 2.02 should be displayed followed by the significance level of .02. If you would like to see where the z score falls on the normal distribution, repeat Steps 1-9, then highlight DRAW, and press ENTER.

The z test score is 2.02, and the alpha level or significance level is p = .02. Thus, the alpha level is less than .05, and the mean IQ score of children in the sample differs significantly from that of children in the general population. In other words, children in academic after-school programs score significantly higher on IQ tests than children in the general population. In APA style, it would be reported as follows: z(N = 75) = 2.02, p < .05 (one-tailed).

Using the TI-84 for the Two-Tailed z Test from Module 9

The previous example illustrated a one-tailed z test; however, some hypotheses are two-tailed and thus the z test would also be two-tailed. As an example, refer back to the study from Module 9 in which the researcher examined whether children in athletic after-school programs weighed a different amount than children in the general population. In other words, the researcher expected the weight of the children in the athletic after-school programs to differ from that of children in the general population, but he was not sure whether they would weigh less (because of the activity) or more (because of greater muscle mass). Let’s use the following data from Module 9: The mean weight of children in the general population (μ) is 90 pounds, with a standard deviation (σ) of 17 pounds; for children in the sample (N = 50), the mean weight (¯¯¯X)(X¯) is 86 pounds. Using this information, we can now test the hypothesis that children in athletic after-school programs differ in weight from those in the general population using the TI-84 calculator.

1.With the calculator on, press the STAT key.

2.Highlight TESTS.

3.1: Z-Test will be highlighted. Press ENTER.

4.Highlight STATS. Press ENTER.

5.Scroll down to μ0: and enter the mean for the population (90).

6.Scroll down to σ: and enter the standard deviation for the population (17).

7.Scroll down to ¯¯¯XX¯: and enter the mean for the sample (86).

8.Scroll down to n: and enter the sample size (50).

9.Lastly, scroll down to μ: and select the type of test (two-tailed), indicating that we expect the sample mean to differ from the population mean (select 2μ0). Press ENTER.

10.Highlight CALCULATE and press ENTER.

The z score of −1.66 should be displayed followed by the alpha level of .096, indicating that this test was not significant. We can therefore conclude that the weight of children in the athletic after-school programs did not differ significantly from the weight of children in the general population.

If you would like to see where the z score falls on the normal distribution, repeat Steps 1−9, then highlight DRAW, and press ENTER.

MODULE 10 The Single-Sample t Test

Let’s illustrate the use of the single-sample t test to test a hypothesis using the example from Module 10. Assume the mean SAT score of students admitted to General University is 1,090. Thus, the university mean of 1,090 is the population mean (μ). The population standard deviation is unknown. The members of the Biology Department believe that students who decide to major in biology have higher SAT scores than the general population of students at the university.

Notice that this is a one-tailed test because the researchers predict that the biology students will perform higher than the general population of students at the university. The researchers now need to obtain the SAT scores for a sample of biology majors. This information is provided in Table 10.1 in Module 10, which shows that the mean SAT score for a sample of 10 biology majors is 1,176.

Using Excel

To demonstrate how to use Excel to calculate a single-sample t test, we’ll use the data from Table 10.1, which represent SAT scores for 10 biology majors at General University. We are testing whether biology majors have higher average SAT scores than the population of students at General University. We begin by entering the data into Excel. We enter the sample data into the A column and the population mean of 1,090 into the B column. We enter the population mean next to the score for each individual in the sample. Thus, you can see that I’ve entered 1,090 in Column B ten times, one time for each individual in our sample of 10 biology majors.

Next highlight the Data ribbon, and then click on Data Analysis in the top right-hand corner. You should now have the following pop-up window:

Scroll down to t-Test: Paired Two Sample for Means, which is the procedure we’ll be using to determine the single-sample t test. Then click OK. You’ll be presented with the following dialog box:

With the cursor in the Variable 1 Range: box, highlight the data from Column A in the Excel spreadsheet so that they appear in the input range box. Move the cursor to the Variable 2 Range: box and enter the data from Column B in the spreadsheet into this box by highlighting the data. The dialog box should now appear as follows:

Click OK to execute the problem. You will be presented with the following output:

We can see the t test score of 2.063 and the one-tailed significance level of p = .035. Thus, our sample mean falls 2.06 standard deviations above the population mean of 1,090. We must now determine whether this is far enough away from the population mean to be considered significantly different. Because our obtained alpha level (significance level) is .035 and is less than .05, the result is significant. We reject H0 and support Ha. In other words, we have sufficient evidence to allow us to conclude that biology majors have significantly higher SAT scores than the rest of the students at General University. In APA style, this would be reported as t(9) = 2.06, p = .035 (one-tailed).

Using SPSS

To demonstrate how to use SPSS to calculate a single-sample t test, we’ll use the data from Table 10.1 in Module 10, which represent SAT scores for 10 biology majors at General University. We are testing whether biology majors have higher average SAT scores than the population of students at General University. We begin by entering the data into SPSS and naming the variable (if you’ve forgotten how to name a variable, please refer back to Appendix C).

Once the data are entered and the variable named, select the Analyze tab and, from the drop-down menu, Compare Means followed by One-Sample T Test. The following dialog box will appear:

Place the SATscore variable into the Test Variable box by utilizing the arrow in the middle of the window. Then let SPSS know what the population mean SAT score is (1,090). We enter this population mean in the Test Value box as in the following window.

Then click OK, and the output for the single-sample t test will be produced in an output window as follows:

We can see the t test score of 2.063 and the two-tailed significance level. Thus, our sample mean falls 2.06 standard deviations above the population mean of 1,090. We must now determine whether this is far enough away from the population mean to be considered significantly different. This was a one-tailed test; thus when using SPSS you will need to divide the significance level in half to obtain a one-tailed significance level because SPSS reports only two-tailed significance levels. Our alpha level (significance level) is .035 and is less than .05, meaning the result is significant. We reject H0 and support Ha. In other words, we have sufficient evidence to allow us to conclude that biology majors have significantly higher SAT scores than the rest of the students at General University. In APA style, this would be reported as t(9) = 2.06, p = .035 (one-tailed).

Using the TI-84

Let’s use the data from Table 10.1 to conduct the test using the TI-84 calculator.

1.With the calculator on, press the STAT key.

2.EDIT will be highlighted. Press the ENTER key.

3.Under L1 enter the SAT data from Table 10.1.

4.Press the STAT key once again and highlight TESTS.

5.Scroll down to T-Test. Press the ENTER key.

6.Highlight DATA and press ENTER. Enter 1,090 (the mean for the population) next to μ0:. Enter L1 next to List (to do this press the 2nd key followed by the 1 key).

7.Scroll down to μ: and select >μ0 (for a one-tailed test in which we predict that the sample mean will be greater than the population mean). Press ENTER.

8.Scroll down to and highlight CALCULATE. Press ENTER.

The t score of 2.06 should be displayed, followed by the significance level of .035. In addition, descriptive statistics will be shown. If you would like to see where the t score falls on the distribution, repeat Steps 1−7, then highlight DRAW, and press ENTER.

Blog ArchiveCopyright © 2019 HomeworkMarket.com Read More
Applied SciencesArchitecture and DesignBiologyBusiness & FinanceChemistryComputer ScienceGeographyGeologyEducationEngineeringEnglishEnvironmental scienceSpanishGovernmentHistoryHuman Resource ManagementInformation SystemsLawLiteratureMathematicsNursingPhysicsPolitical SciencePsychology

"Order a similar paper and get 15% discount on your first order with us
Use the following coupon
"FIRST15"

Order Now