z scores
Z Scores
Importance of Z Scores
Z scores play a crucial role in statistics, especially when comparing across different samples, populations, or groups.
Example scenario: Two children take the same exam or two different exams at different times; comparison of their performance is facilitated by Z scores.
Definition of a Z Score
Formula:
Z = \frac{X - \mu}{\sigma}where:
Z = Z score
X = your score
μ = mean of the class
σ = standard deviation of the class
Calculation of Z Score
Example Calculation:
If a student scores 70, average (μ) is 80, and standard deviation (σ) is 10:
Z = \frac{70 - 80}{10} = \frac{-10}{10} = -1
Interpretation:
A Z score of -1 indicates how a score compares to the average of the class.
Visually, it places the score within a standard normal curve where:
Mean = 0
Z score of 1 = 34% area under the curve, Z score of -1 = 16% area under the curve.
Understanding Percentiles
Explanation of percentiles using Z scores:
A Z score of -1 corresponds to the 16th percentile, meaning the student performed better than 16% of peers.
A Z score of -2 with the same score changes the standing to the 2nd percentile (performed better than 1 out of 100).
When the standard deviation is adjusted (e.g., standard deviation changes from 10 to 5), the Z score also shifts, emphasizing the relationship between distribution spread and Z scores.
Positive Z Score Example
If a student scores 80 with mean 70 and standard deviation 10:
Z = \frac{80 - 70}{10} = \frac{10}{10} = 1
Interpretation gives a score indicating the 84th percentile, meaning the student outperformed 83 others out of 100.
Impact of Changing Distribution
Adjusting the standard deviation can lead to significant shifts in Z scores, affecting how students rank within percentiles.
Applications of Z Scores
Z scores are integral for admissions tests like LSAT, MCAT, GRE, where performance is assessed through percentiles.
Performance requirements for universities (e.g., UBC's minimum percentile for admission).
Statistical Judgments Based on Z Scores
The assessment of differences in scores and statistical significance occurs through hypothesis testing.
The research community often adopts an alpha level of 0.05 to denote acceptable risk levels for statistical significance.
Concept of statistical vs. practical significance.
Experimentation and Alpha Levels
Probability in Research
Alpha level ($\alpha = 0.05$):
Indicates a willingness to accept a 5% error in concluding that the observed differences did not occur by random chance.
Conversely, a 95% confidence indicates that observed differences are due to systematic variations, primarily the influence of independent variables (IV).
Statistical Significance
This highlights the importance of statistical significance rather than inherent importance, providing context for how data should be interpreted.
Difference Calculation for Experimental Groups
Visual representation of distributions for control vs. experimental groups and the identification of significant differences occurs through two-tail tests.
Critical values set at +/-1.96 for capturing 95% of data within a normal distribution.
Z Tables and Critical Values
Understanding Z Tables
Z tables provide critical values based on areas under the normal distribution curve, with areas representing probabilities associated with Z scores.
Example critical values include +/- 1.96 for 95% confidence, indicating significant differences outside these bounds.
Using Z Tables
Researchers retrieve probabilities or area under the curve for specified Z scores to evaluate significance in hypotheses.
Two distinct types of Z tables exist, showing either cumulative probabilities or tail probabilities.
Example Computation Using Z Tables
To find Z scores corresponding to specific percentages (e.g., the 90th percentile):
Look up values reflecting cumulative probabilities (e.g., for point 0.025, reference gives 1.645).
Application of Z Scores in Assessing Intelligence (IQ)
Example of IQ Score Calculation
A commonly cited example entails IQ scores where:
Average IQ = 100, with standard deviation = 15.
Z score interpretation shows performance relative to the mean, determining percentiles for classifications such as MENSA (Z score of 1.33 represents the 99th percentile).
Abnormality Measurement
Defining behaviors or characteristics as 'abnormal' using statistical standards of significance (e.g., Z scores indicating scores falling beyond a specified threshold according to normal distribution).
Percentiles and Areas Under the Curve
Transition to Measures of Spread
Following the discussion on Z scores, the next topic will be confidence intervals, which elaborate on the spread and reliability of statistical estimates and sample means.