chapter 7 : Hypothesis Text withZ Tests
Overview of Hypothesis Testing Using the Z-Test
Exam Review and Purpose
Testing is completed using z test
The goal is to understand hypothesis testing in the context of both sample statistics and population comparisons.
Key Concepts of Hypothesis Testing
Hypothesis Testing Procedure
Involves a series of steps to execute hypothesis testing effectively.
Steps create a logical framework for statistical tests.
Comparison to earlier experiences with hypothesis testing (e.g., science fair).
Shift from elementary hypothesis testing to inferential statistics.
Null Hypothesis
Definition: The null hypothesis (denoted as H ) posits that there are no differences between groups or no effect in the population.
The alternative hypothesis ( denoted as H ) posits that there is a difference between groups or that an effect exists in the population, suggesting that the observed data is not due to chance alone.
Importance of cautious reasoning to avoid overstepping conclusions and ensure accuracy in statistical claims.
Language surrounding null hypothesis: "fail to reject the null hypothesis" vs "retain the null hypothesis."
Assumptions in Hypothesis Testing
Key Assumptions in Statistical Tests
The majority of psychology tests involve three main assumptions:
Continuous Dependent Variables
Dependent variables must be measured on a continuous scale (e.g., ratio level measurement).
Calculating the mean is necessary; nominal or ordinal data cannot produce a mean calculation.
Example: Cannot average non-numerical categories like shoe colors.
Random Sampling
Participants must be selected randomly.
This assumption is often violated
Every member of the population of interest must have an equal chance of being selected for the study
True random samples are rare; convenience samples are more common in psychological studies.
Small convenience samples may limit generalizability, necessitating caution in interpretations.
Normal Population Distribution
The population from which the sample is drawn should be approximately normally distributed.
Each sample may differ, but if each sample has 30 participants, the Central Limit Theorem allows for normal approximation.
Central Limit Theorem: As sample sizes increase, the distribution of the sample means approximates normality.
The Role of Robust Tests
Robustness in Hypothesis Testing
Robust tests can accommodate assumptions that might be violated.
The Z-test is regarded as a robust test, providing confidence in results even amid assumption violations.
Robustness in hypothesis testing refers to the ability of a statistical test to produce valid results even when certain assumptions of the test are violated. In psychological research, many statistical tests, including the Z-test, are considered robust because they can still yield reliable statistical inferences under conditions where assumptions about the data are not entirely met. For example, while the Z-test assumes normality in population distribution, it can still provide accurate results when this assumption is partially violated, especially with larger sample sizes due to the Central Limit Theorem. This feature allows researchers to have confidence in their findings despite potential imperfections in sample data.
The Central Limit Theorem
The Central Limit Theorem (CLT) is a fundamental principle in statistics that explains how the distribution of sample means approaches a normal distribution as the sample size increases, regardless of the shape of the population distribution.
Key Points:
The theorem states that if you take sufficiently large random samples from a population, the means of these samples will form a normal distribution, even if the original population is not normally distributed.
istributed, provided the sample size is large enough.
This allows psychologists to make inferences about population means and conduct hypothesis testing with greater confidence.
Steps in the Hypothesis Testing Process
Steps Overview
Generally accepted steps may vary (5 or 6 steps recognized as necessary).
The six steps of hypothesis testing include:
Identify populations, comparison distributions, assumptions, and choose the appropriate test.
State null and research hypotheses in words and symbolic notation.
Determine characteristics of the comparison distribution.
Determine critical values (cutoffs) for hypothesis rejection.
Calculate the test statistic (e.g., z-score).
Make a final decision regarding the null hypothesis (reject or fail to reject).
Z-Test Details
Application of Z-Test
The Z-test is used when the population means and standard deviation are known or estimated.
Example scenario involving calorie counts at Starbucks:
Data collected from Starbucks to investigate whether menu labeling (calories posted) influences purchasing behavior.
Aim: Determine differences in calorie consumption between customers who see calorie counts and those who do not.
Step 1: Identify Populations and Comparison Distribution
Population 1: Customers at Starbucks locations with calorie counts posted.
Population 2: Customers at locations where calories are not displayed.
Comparison Distribution: Distribution of means from the two populations.
Step 2: State Hypotheses
Null Hypothesis (H ): No difference in average calories between the two populations.
Research Hypothesis (H ): A difference in average calories does exist.
Notation and language in hypotheses are critical to maintain clarity in interpretation.
Step 3: Determine Characteristics:
Example data specifics for the Starbucks case:
Population means for Non-Calorie Menu: 247 calories.
Sample Mean for Caloric Menu: 232 calories.
Standard Error calculation, SE = where population standard deviation is 201 and N (sample size) is 1000.
Critical Values and Decision-Making
Step 4: Determine Critical Values
The common Type 1 error threshold (alpha) is set at 0.05 for significance testing.
For a two-tailed test, critical values are set at 1.96 (each tail accounts for 2.5% of the error rate).
Step 5: Calculate Z Test Statistic
Formula for Z statistic: Z =
Example calculation based on the Starbucks data gives a Z of -2.36, indicating that the sample mean is statistically significant and falls in the rejection region.
Step 6: Conclusion
The decision to reject the null hypothesis indicates a statistically significant result in calorie consumption patterns related to menu labeling.
In summary, this analysis reinforces the importance of menu labeling in influencing consumer behavior, as evidenced by the significant difference in calorie consumption. This suggests that implementing clearer calorie counts may lead to healthier choices among consumers, ultimately fostering better public health outcomes.
Practical Implications
Research Integrity and Best Practices
Discuss problems of data mining and p-hacking influencing scientific reliability and outcomes.
Emphasizes caution against misrepresentation of statistical findings in research.
This convergence to normality typically occurs when the sample size is 30 or more, making the CLT particularly useful in psychology, where sample sizes can vary considerably.
The implication for psychological research is that researchers can use parametric tests (which assume normal distributions, such as the Z-test or t-test) even when the underlying data may not be perfectly normally d