A researcher uses the 30 students in her psychology class as a sample to represent all young adults. She measures their scores on a memory task and calculates a 95% confidence interval for the mean score. The most significant flaw that invalidates the calculation of the standard error for this interval is the violation of which assumption?
Answer: The independence assumption, because students in the same class are not independent observations.
The formula for the standard error of the mean, s/n, is derived under the assumption that the n observations are independent. Students in the same class are not independent; they may share common characteristics (e.g., major, motivation) or influence one another. This non-independence invalidates the standard error calculation, making the resulting confidence interval untrustworthy. While it is also not a random sample (Choice D), the lack of independence is the issue that most directly corrupts the standard error formula itself.
A researcher fits a simple linear regression model to a dataset. To check the assumptions for inference, a plot of the residuals versus the fitted values is created. The plot shows a distinct fan shape, with the vertical spread of the residuals increasing as the fitted values increase. Which assumption is most clearly violated, and what is the primary consequence for a confidence interval for the slope?
Answer: The equal variance (homoscedasticity) assumption is violated, which makes the standard error of the slope unreliable.
A fan-shaped pattern in a residual plot, where the spread of residuals changes with the fitted values, is the classic sign of heteroscedasticity, or unequal variance. This violation means that the standard error of the slope, which assumes constant variance, is incorrect. This, in turn, makes any resulting confidence intervals or p-values unreliable.
A market researcher constructs a contingency table from survey data to test for an association between age group and preferred social media platform. To check the 'Large Counts' condition for a chi-square test of independence, one must first calculate the expected count for each cell. How is the expected count for a specific cell calculated?
Answer: By multiplying the row total for that cell by the column total for that cell, then dividing by the grand total.
The formula for the expected count in any cell of a two-way table for a chi-square test of independence is (Row Total × Column Total) / Grand Total. This calculation is based on the null hypothesis that there is no association between the two categorical variables. Choice A describes the expected count for a goodness-of-fit test with a uniform null hypothesis. Choice B confuses the condition with the observed count.
A company claims that customer support calls are uniformly distributed throughout the 7 days of the week. A sample of 105 calls is taken to perform a χ2 goodness-of-fit test. Which of the following statements correctly describes the 'Large Counts' condition for this test?
Answer: The expected number of calls for each of the 7 days must be sufficiently large.
The 'Large Counts' condition for a chi-square goodness-of-fit test requires that all expected counts be sufficiently large (a common rule of thumb is at least 5). In this case, the null hypothesis is a uniform distribution, so the expected count for each day would be 105 / 7 = 15. Since 15 is greater than 5, the condition is met. The condition is not on the observed counts (Choice A), as random variation could cause a single day's observed count to be small even if the condition is met.
A nutritionist recruits 50 pairs of siblings to compare two diets. Within each pair, one sibling is randomly assigned to Diet A and the other to Diet B. After 12 weeks, the weight loss for each individual is recorded. To determine if there is a significant difference in mean weight loss, what is the critical condition regarding normality that must be checked for the most appropriate statistical test?
Answer: The distribution of the pairwise differences in weight loss (Diet A - Diet B) must be approximately normal.
This is a matched pairs experimental design because siblings are paired and treatments are randomized within pairs. The appropriate procedure is a paired t-test. The condition for a paired t-test is not on the individual groups, but on the distribution of the differences between the paired observations. A two-sample t-test, which would check normality on both groups separately (Choice B), is inappropriate because the samples are not independent.
A study is based on a random sample of 100 professional athletes' salaries, taken from a population known to be extremely right-skewed with high outliers. A researcher constructs a 95% t-interval for the mean salary. Which statement best describes the validity of the 'Nearly Normal' condition?
Answer: The condition is likely not met, as the extreme skewness of the population requires a much larger sample size for the CLT to apply effectively.
The Central Limit Theorem (CLT) states that the sampling distribution of the sample mean will be approximately normal for a large enough sample size. However, 'large enough' depends on the shape of the population distribution. For a population that is extremely skewed and has high outliers, a sample size of 100 may not be sufficient for the sampling distribution to be approximately normal. The rule of thumb 'n>30' (Choice A) is an oversimplification and does not hold for severely non-normal populations. The shape of the sample will reflect the population (so C is wrong), and random sampling doesn't fix this issue (so D is wrong).
To test a new fertilizer, a researcher divides 40 identical plots of land into two groups of 20. One group is randomly assigned to receive the new fertilizer, and the other receives a standard fertilizer. The satisfaction of which condition is most essential for drawing a conclusion that the new fertilizer causes a change in mean yield?
Answer: Random assignment of the fertilizers to the plots.
Random assignment is the key element of a well-designed experiment that allows for causal inference. By randomly assigning treatments, a researcher can balance out the effects of confounding variables, making it possible to attribute differences in the response variable (yield) to the explanatory variable (fertilizer type). Random sampling (Choice A) is crucial for generalizing results to a larger population, but not for establishing causation. Normality (C) and equal variances (D) are conditions for the validity of the t-test's p-value, not for making causal claims.
An analyst is examining the relationship between advertising spending and sales using a simple linear regression model. The two plots below are produced: a scatterplot of the data with the regression line and a plot of residuals versus fitted values. Based on the plots provided, which condition for inference appears to be violated?
Answer: The linearity condition.
The plot of residuals versus fitted values shows a clear, non-random curved (inverted U) pattern. This indicates that the relationship between the variables is not linear, and a simple linear model is inappropriate. The linearity assumption is therefore violated. The vertical spread of the residuals appears relatively constant, so there is no evidence against the equal variance condition. The other plots would be needed to assess normality or independence.
In an experiment on a new fuel additive, the gas mileage of 12 cars is measured before and after the additive is used. The researchers plan to test if the mean gas mileage has increased. What is the most important consideration when checking the 'Nearly Normal' condition for the appropriate hypothesis test?
Answer: The distribution of the 12 differences in mileage (after - before) for each car must be approximately normal.
This is a matched pairs design, as the same 12 cars are measured twice. The correct procedure is a paired t-test. The paired t-test analyzes the differences between the paired measurements. Therefore, the 'Nearly Normal' condition applies to the distribution of these differences, not the original 'before' or 'after' distributions. The lack of independence (Choice C) is the reason for using a paired test, not a barrier to checking its conditions.
A survey asks 500 people, recruited from a single large shopping mall on a Tuesday afternoon, about their weekly grocery budget. The goal is to construct a 95% confidence interval for the mean budget of all adults in the city. The most important statistical assumption violated by this sampling plan is which of the following?
Answer: The random sampling condition, as the sample is not representative of all adults in the city.
When constructing confidence intervals, you need to verify several key assumptions. The most fundamental is that your sample must be representative of the population you're trying to make inferences about.
The correct answer is A because this sampling method creates a massive representativeness problem. Recruiting people only from one shopping mall on a Tuesday afternoon introduces severe selection bias. You're missing entire demographics: people who work traditional hours, those who shop at different stores, people who avoid malls, and those who do their grocery shopping on other days. This convenience sample systematically excludes large portions of the city's adult population, making it impossible to generalize results to "all adults in the city."
Option B is incorrect because the Central Limit Theorem actually works very well with financial data, and with n=500, the sampling distribution of the mean will be approximately normal regardless of the population distribution's shape.
Option C misapplies the 10% condition. This rule requires that your sample be less than 10% of the total population you're studying (all adults in the city), not 10% of the mall's daily shoppers. Since 500 is almost certainly less than 10% of all city adults, this condition is satisfied.
Option D confuses individual variability with sampling requirements. The "equal variance condition" isn't a standard requirement for confidence intervals about means, and budget variability across individuals doesn't violate any assumptions.
Study tip: Always ask "Does this sample fairly represent the target population?" before worrying about technical conditions. Biased sampling trumps all other statistical concerns.
To test if the proportion of voters favoring Candidate A differs between two districts, a pollster samples 50 voters from District 1 and 40 from District 2. They find 5 supporters in District 1 and 6 in District 2. In checking conditions for a two-sample z-test for a difference in proportions, which condition is not met?
Answer: The Large Counts condition, because the calculated expected number of successes for one group is less than 10.
For a hypothesis test comparing two proportions, the Large Counts condition is checked using the pooled sample proportion, p^pool=n1+n2x1+x2. Here, p^pool=50+405+6=9011≈0.122. We then check if n1p^pool, n1(1−p^pool), n2p^pool, and n2(1−p^pool) are all at least 10. The expected successes for District 2 is 40×(11/90)≈4.89. Since 4.89 is less than 10 (and also less than the stricter rule of 5), the Large Counts condition is not met. Choice C is incorrect because it uses the observed count, not the expected count based on the pooled proportion.
An analyst suspects a violation of an assumption for linear regression inference. The primary consequence is that, while the slope estimate remains unbiased, the p-values and confidence intervals are unreliable because the formula for the standard error of the slope is no longer appropriate. The violation of which assumption leads directly to this specific set of consequences?
Answer: Homoscedasticity (Equal Variance)
Violation of the homoscedasticity assumption (i.e., heteroscedasticity) means the variance of the errors is not constant across all levels of the predictor variable. The standard formula for the standard error of the slope uses a pooled estimate of this variance. If the variance is not constant, this pooled estimate is incorrect, leading to unreliable standard errors, p-values, and confidence intervals. However, heteroscedasticity does not introduce bias into the slope estimate itself. In contrast, violating linearity (Choice A) would cause the slope estimate to be biased.
A quality control inspector at a factory that produces 50,000 microchips per day selects a simple random sample of 200 microchips without replacement to estimate the proportion of defective chips. Why is it important to check the '10% condition' before constructing a confidence interval?
Answer: To ensure that the individual observations can be treated as approximately independent when calculating the standard error.
When sampling without replacement, the selections are not truly independent. However, if the sample size is no more than 10% of the population size, the effect of sampling without replacement on the variance of the sampling distribution is negligible. Checking this condition allows us to use the simpler standard error formula (e.g., p(1−p)/n) that assumes independence. Here, 200 is much less than 10% of 50,000, so the condition is met.
A biologist models the relationship between the wingspan (in cm) and weight (in g) of a species of bird using a least-squares regression line. To perform statistical inference on the slope of this line, several conditions must be met. To specifically check the normality assumption, which graphical display is most appropriate?
Answer: A histogram or normal probability plot of the residuals from the regression.
For inference in linear regression, the assumption is that the residuals (the differences between observed and predicted y-values) are normally distributed for any given x-value. This is checked by examining the distribution of the collected residuals, typically with a histogram or a normal probability plot. It is a common misconception that the x or y variables themselves must be normally distributed (Choices B and D). A scatterplot (Choice A) is used to check the linearity assumption, not normality.
A researcher is conducting a one-sample t-test for a population mean with a small sample size (n=15). A histogram of the sample data is constructed to check the 'Nearly Normal' condition. In which of the following scenarios would the validity of the t-test be most compromised?
Answer: The histogram shows strong skew to the right with a high outlier.
The t-procedures are known to be robust to moderate departures from normality, especially symmetry violations like bimodality or uniform distributions. However, they are not robust to outliers or strong skew, particularly with small sample sizes. An outlier can have a large effect on both the sample mean and the sample standard deviation, severely compromising the validity of the p-value and any confidence interval.
A researcher compares the mean effectiveness of three different pain relievers. They randomly assign 15 subjects to each drug (total n=45). The sample standard deviations of the pain scores for the three groups were 10.5, 11.2, and 25.8. Based on this information, which condition for a one-way ANOVA is most likely violated?
Answer: The homogeneity of variances condition.
When you encounter ANOVA problems, you need to check whether the data meets four key assumptions before the test is valid: independence of observations, normality within each group, equal variances across groups (homogeneity of variances), and adequate sample sizes.
The critical clue here lies in the sample standard deviations: 10.5, 11.2, and 25.8. Notice how the third group's standard deviation (25.8) is more than twice as large as the others. Since variance equals the square of standard deviation, the variances are approximately 110, 125, and 665. The largest variance is about six times bigger than the smallest ones! This dramatic difference strongly suggests the homogeneity of variances condition is violated.
Looking at the wrong answers: Choice B is incorrect because we have no information about the distribution shapes within each group—we only know the standard deviations. Choice C is wrong because the problem states subjects were "randomly assigned," which typically ensures independence. Choice D is factually incorrect since all three groups have exactly 15 subjects each, making the sample sizes perfectly equal.
The homogeneity of variances assumption requires that population variances be approximately equal across all groups. When sample standard deviations differ dramatically (a common rule of thumb is when the largest is more than twice the smallest), this assumption is likely violated.
Study tip: Always examine the standard deviations or variances when checking ANOVA assumptions. If one group's standard deviation is much larger than the others, immediately suspect a homogeneity of variances violation.
A geneticist claims that a certain cross-breeding of plants will produce offspring with petal colors in the ratio 9:3:4 for Red:Yellow:White, respectively. To check the conditions for a χ2 goodness-of-fit test on a sample of 160 offspring, what is the correct expected count for white-petaled offspring?
Answer: 40
First, find the total number of parts in the ratio: 9 + 3 + 4 = 16. The proportion of white-petaled offspring expected under the null hypothesis is 4/16 = 1/4. To find the expected count, multiply this proportion by the total sample size: (1/4) * 160 = 40. This is the value that would be used to check the 'Large Counts' condition and to compute the chi-square test statistic. Choice A is the ratio number, not the count. Choice C is the expected count if a uniform distribution were assumed (160/3).
A researcher plans a two-sample t-test on two independent groups, each of size n=20. The sample data from both groups show moderate right skewness but no outliers. Which of the following is the most accurate assessment of the situation regarding the 'Nearly Normal' condition?
Answer: The t-test is reasonably robust to moderate skewness, so the p-value will be approximately correct, though the test's power may be reduced.
When evaluating whether a t-test is appropriate, you need to assess how violations of assumptions affect the test's validity and reliability. The "Nearly Normal" condition is one of the key assumptions, but understanding the t-test's robustness is crucial.
The t-test is remarkably robust to moderate departures from normality, especially with reasonable sample sizes like n=20 per group. Moderate right skewness without outliers typically won't severely compromise the test's validity. The Central Limit Theorem begins to take effect even with smaller samples, and the t-distribution can accommodate some skewness while still providing approximately correct p-values. However, skewness can reduce the test's power (ability to detect true differences), making answer A correct.
Answer B is incorrect because moderate skewness typically doesn't substantially inflate Type I error rates - the t-test maintains reasonable control over false positives even with mild normality violations. Answer C overstates the problem; the test isn't "completely invalid" with moderate skewness, though non-parametric alternatives could be considered. Answer D reflects the common misconception about the "magic number" of n=30 - while larger samples do help with robustness, n=20 per group is often sufficient for moderate skewness, and re-collecting data isn't automatically necessary.
Study tip: Remember that statistical tests exist on a spectrum of robustness. The t-test is quite forgiving of moderate normality violations, especially with decent sample sizes. Save non-parametric tests for severe skewness, heavy outliers, or very small samples where normality violations are pronounced.
A researcher tests if a coin is fair (H0:p=0.5) by flipping it 20 times and observing 4 heads. To verify that a one-proportion z-test is appropriate, they must check the 'Large Counts' condition. Which statement correctly assesses this condition for this hypothesis test?
Answer: The condition is met because the expected counts under the null hypothesis, np0 and n(1−p0), are both 10.
When conducting a one-proportion z-test, you must verify that the "Large Counts" condition is satisfied to ensure the sampling distribution is approximately normal. This condition requires that both expected counts under the null hypothesis are at least 10.
For this coin fairness test, you calculate the expected counts using the null hypothesis value p0=0.5. With n=20 flips:
- Expected successes: np0=20×0.5=10
- Expected failures: n(1−p0)=20×0.5=10
Since both expected counts equal 10 (which meets the ≥10 requirement), the Large Counts condition is satisfied, making answer A correct.
Answer B incorrectly applies the n≥30 rule, which is for t-tests or when you don't know the population standard deviation. For proportion tests, sample size alone doesn't determine appropriateness—the expected counts do.
Answer C makes a critical error by focusing on observed counts (4 heads observed) rather than expected counts. The Large Counts condition always uses expected values under the null hypothesis, not what actually happened in your sample.
Answer D also confuses observed counts (16 failures observed) with expected counts. While 16 > 10 is true, this isn't the right comparison for checking test conditions.
Study tip: Always remember that conditions for hypothesis tests depend on expected values under the null hypothesis, not your sample results. For proportion tests, ignore the actual sample size or observed counts—calculate np0 and n(1−p0) and check if both ≥10.
