Quick Answer: What Does a Z Score Show?

A Z Score tells how far a value is from the mean in standard deviation units. A Z Score of 0 means the value is exactly at the mean. A Z Score of +1 means the value is one standard deviation above the mean. A Z Score of -1 means the value is one standard deviation below the mean. A large absolute Z Score, such as |z| > 2 or |z| > 3, can flag unusual or extreme observations for review.

In this worked example, the main variable is G3 final grade. The verified SPSS output shows N = 649, mean = 11.91, standard deviation = 3.231, and the standardized variable z_G3 has mean approximately 0.00000 and standard deviation exactly 1.000000. The lowest G3 Z Score is -3.685, and the highest G3 Z Score is 2.196. This means the most distant G3 observations are on the lower-score side.

What Is a Z Score?

A Z Score is a standardized score that shows how many standard deviations a raw value is above or below the mean. Instead of reading a raw score only in its original unit, a Z Score expresses the same value on a common scale where the mean is 0 and the standard deviation is 1.

For example, a raw G3 final grade of 15 is easier to interpret after standardization. If the G3 mean is about 11.91 and the standard deviation is about 3.23, then a score of 15 is roughly one standard deviation above the mean. The Z Score makes that relationship explicit. A score of 0, on the other hand, is far below the mean and receives a large negative Z Score.

The Z Score is useful because it allows different variables to be compared on the same scale. G3 is measured in grade points from 0 to 19, age is measured in years from 15 to 22, and absences are measured as count values from 0 to 32. Their raw units are different, but their Z Score values are directly comparable as standardized distances from each variable’s own mean.

Z Score Formula

The standard Z Score formula is:

In symbols, the formula is often written as:

z = (x - x̄) / s

Where x is the raw value, x̄ is the sample mean, and s is the sample standard deviation. For a population version, the formula is written as:

z = (x - μ) / σ

Where μ is the population mean and σ is the population standard deviation. In most practical data-analysis projects, the sample formula is used because the mean and standard deviation are calculated from the sample data.

Worked Z Score Example for G3

Using the verified SPSS values for G3, the mean is 11.91 and the standard deviation is 3.231. A student with G3 = 19 has the following approximate Z Score:

z = (19 - 11.91) / 3.231
z ≈ 2.196

This means a G3 score of 19 is about 2.20 standard deviations above the mean. A student with G3 = 0 has the following approximate Z Score:

z = (0 - 11.91) / 3.231
z ≈ -3.685

This means a G3 score of 0 is about 3.69 standard deviations below the mean. That is why the most distant G3 cases in this analysis appear on the negative side.

Hypothesis-Style Screening Rule for Z Score

A Z Score is not a full inferential test by itself, but it can be used with a hypothesis-style screening decision. This is useful when you want clear reporting language for unusual or extreme observations.

Dataset and Variables Used

This Z Score guide uses the student performance dataset. The main outcome variable is G3 final grade. Additional numeric variables were standardized to show how Z Score values can compare unusual and extreme values across variables with different raw scales.

The analysis uses Z Score values for descriptive screening and standardization. It does not claim that every extreme Z Score is an error. Instead, the goal is to locate observations far from the mean and interpret them with charts, frequency tables, and context.

Verified SPSS Output Interpretation for Z Score

The SPSS output confirms that the Z Score variables were created correctly. The raw variables have their original means and standard deviations, while the standardized variables have mean approximately 0 and standard deviation 1.

SPSS Descriptive Statistics

SPSS Standardized Variables

G3 Z Score Flag Counts

G3 Z Score Band Table

SPSS Explore Output for z_G3

Group Z Score Summary by Sex

Python Chart-by-Chart Interpretation

The Python charts explain the Z Score result visually. The first two charts compare raw and standardized distributions. The next charts show standardized bands, unusual and extreme counts, most distant cases, and group mean Z Score differences.

Python Chart 1: Raw Distribution with Standard Deviation Bands

This chart begins with the raw G3 scale. The center line marks the mean, which is about 11.91. The dashed lines show approximately one standard deviation below and above the mean, and the dotted lines show approximately two standard deviations below and above the mean. The chart makes the Z Score idea easier to understand because each band corresponds to a standardized distance. A G3 value near the mean has a Z Score close to 0. A value around one dashed-line distance above the mean has a Z Score near +1, and a value around one dashed-line distance below the mean has a Z Score near -1.

The raw histogram shows that most G3 scores fall between the -1 SD and +1 SD region. The low score values near 0 are far to the left and fall beyond the -2 SD band. This explains why the minimum G3 Z Score is -3.685. The upper side extends to 19, but it does not reach beyond +3 standard deviations. Therefore, the raw distribution chart supports the SPSS finding that the most extreme G3 values are on the lower side.

Python Chart 2: Standardized Z Score Distribution

This chart shows the same G3 values after standardization. The x-axis is no longer raw grade points; it is the G3 Z Score. The solid line at z = 0 is the standardized mean. The dashed lines at -1 and +1 show one standard deviation below and above the mean. The dotted lines at -2 and +2 show two standard deviations below and above the mean.

The standardized distribution confirms that the Z Score transformation worked correctly. The mean is approximately 0 and the standard deviation is approximately 1. However, the shape of the distribution does not become normal simply because the values were standardized. The left tail remains visible, and a small group of cases appears below z = -3. This is an important lesson: Z Score standardization changes the scale, but it does not remove skewness, kurtosis, or outliers.

Python Chart 3: Cases by Standardized Band

This chart summarizes how many G3 cases fall into each Z Score band. The largest groups are from -1 to 0 and 0 to 1. Specifically, 36.4% of cases fall between -1 and 0, while 41.0% fall between 0 and 1. Together, most students fall within one standard deviation of the G3 mean.

The chart also shows the unusual and extreme regions. About 7.4% of cases fall between -2 and -1, 2.5% fall below -3, 0.2% fall between -3 and -2, and 0.3% fall between 2 and 3. No cases appear above +3. The visual result agrees with the SPSS frequency table. The main standardized distribution is centered, but the lower tail contains the most extreme observations.

Python Chart 4: Unusual and Extreme Values by Variable

This chart compares unusual and extreme values across several variables. The blue bars represent |z| > 2, while the orange bars represent |z| > 3. This is one of the biggest strengths of Z Score standardization: it allows variables with different raw scales to be compared on the same standardized distance scale.

The chart shows that absences has the largest number of unusual values beyond |z| > 2, which makes sense because absences is a count variable with a long right tail. G1 and G2 also show many unusual values, mostly because low grade values are far below their means. G3 has 19 cases beyond |z| > 2 and 16 cases beyond |z| > 3. Age has fewer unusual cases, but the high end of age can still be distant because the age standard deviation is small.

Python Chart 5: Most Distant G3 Cases

This chart lists the most distant G3 cases by absolute Z Score. Most of the top cases are on the negative side, meaning they are far below the overall G3 mean. The vertical reference lines help separate ordinary, unusual, and extreme standardized values. Cases beyond -3 should be reviewed carefully because they are more than three standard deviations below the mean.

In practical reporting, the chart should not be used to automatically remove these cases. Instead, it should guide the analyst toward case review. A low G3 value may represent a real poor performance outcome, a withdrawal, a failed final grade, or a data-entry problem. The Z Score tells us that the case is distant; the dataset context tells us what that distance means.

Python Chart 6: Group Mean Standardized Score

This chart compares the mean standardized G3 score by sex group. The female group has a mean Z Score of about 0.107, while the male group has a mean Z Score of about -0.155. The total standardized mean remains 0 because the full variable was standardized across all 649 cases.

This does not mean that the group difference is automatically statistically significant. It means the female group mean is slightly above the overall G3 mean, and the male group mean is slightly below it. A separate two-group mean test would be needed for a formal hypothesis decision. For descriptive interpretation, the Z Score chart gives a simple standardized effect-style view of how far each group mean sits from the overall mean.

R Chart-by-Chart Validation

The R charts validate the same Z Score results using an independent workflow. R uses the same standardization logic: subtract the mean and divide by the standard deviation. The R charts confirm that the standardized G3 distribution has mean about 0, standard deviation about 1, and a stronger lower-tail pattern than upper-tail pattern.

R Chart 1: Raw Distribution with Standard Deviation Bands

The R raw distribution chart confirms the same grade pattern shown in Python. Most G3 values are concentrated around the middle grades, while a small group of very low values sits far left of the mean. The standard deviation bands show why those values receive large negative Z Score values. The R chart therefore validates the Python interpretation that the most extreme G3 cases are low-score observations.

R Chart 2: Standardized Distribution

The R standardized distribution shows the same result as SPSS and Python: the mean is approximately 0, and the standard deviation is approximately 1. The left tail remains present after standardization, proving again that Z Score conversion does not make a variable normal. It only places the variable on a standard scale.

R Chart 3: Cases by Standardized Band

The R band chart confirms that most G3 cases fall between -1 and +1. It also confirms the low-tail problem: 2.5% of cases are below -3, and only 0.3% are between 2 and 3. There are no cases above +3. This pattern supports the interpretation that the extreme G3 values are concentrated on the lower side.

R Chart 4: Unusual and Extreme Values by Variable

The R comparison chart confirms that the Z Score outlier pattern differs by variable. Absences has many high standardized values because some students have very high absence counts compared with the average. G1, G2, and G3 have distant low-score observations. Age has fewer unusual values, but high age values can still appear distant because age has a small standard deviation.

R Chart 5: Most Distant G3 Cases

The R most-distant-cases chart again shows that many of the farthest cases are below the mean, not above it. The standardized case list is useful for data cleaning, outlier review, and reporting. It can also be used to create a review table for the most extreme observations before deciding whether any action is needed.

R Chart 6: Group Mean Standardized Score

The R group chart validates the SPSS group summary. The female group sits slightly above the overall G3 mean, while the male group sits slightly below it. Because the units are standardized, the group means can be read directly as standardized distances from the overall mean. This makes the Z Score group comparison easier to interpret than raw grade differences alone.

SPSS, R, Python and Excel Workflows for Z Score

This section shows how to reproduce the Z Score analysis in SPSS, R, Python and Excel. The same logic is used in every tool: calculate the mean, calculate the standard deviation, subtract the mean from each value, and divide by the standard deviation.

SPSS Workflow

R Workflow

Python Workflow

Excel Workflow

Code Blocks for Z Score

SPSS Syntax for Z Score

* Z Score Analysis in SPSS.
* Create standardized variables and flag unusual/extreme cases.

DATASET ACTIVATE DataSet1.

TITLE "Z Score: Standardized Scores for Key Variables".

* Create z-score variables automatically.
DESCRIPTIVES VARIABLES=G1 G2 G3 age absences
  /SAVE
  /STATISTICS=MEAN STDDEV MIN MAX.

* Rename saved standardized variables if needed.
RENAME VARIABLES (ZG1=z_G1) (ZG2=z_G2) (ZG3=z_G3) (Zage=z_age) (Zabsences=z_absences).

* Flag G3 cases beyond absolute z-score cutoffs.
COMPUTE flag_abs_z_G3_gt2 = (ABS(z_G3) > 2).
COMPUTE flag_abs_z_G3_gt3 = (ABS(z_G3) > 3).
EXECUTE.

* Create a z-score band variable for G3.
RECODE z_G3
  (Lowest thru -3 = -4)
  (-3 thru -2 = -3)
  (-2 thru -1 = -2)
  (-1 thru 0 = -1)
  (0 thru 1 = 1)
  (1 thru 2 = 2)
  (2 thru 3 = 3)
  (3 thru Highest = 4)
  INTO z_band_G3.
EXECUTE.

VARIABLE LABELS
  z_G3 "Z score for G3"
  flag_abs_z_G3_gt2 "Flag: absolute z_G3 greater than 2"
  flag_abs_z_G3_gt3 "Flag: absolute z_G3 greater than 3"
  z_band_G3 "Z-score band for G3".

FREQUENCIES VARIABLES=flag_abs_z_G3_gt2 flag_abs_z_G3_gt3 z_band_G3
  /ORDER=ANALYSIS.

EXAMINE VARIABLES=z_G3
  /PLOT=BOXPLOT HISTOGRAM
  /STATISTICS=DESCRIPTIVES
  /CINTERVAL=95
  /MISSING=LISTWISE
  /NOTOTAL.

MEANS TABLES=z_G3 BY sex
  /CELLS=COUNT MEAN STDDEV.

OUTPUT EXPORT
  /CONTENTS EXPORT=VISIBLE
  /PDF DOCUMENTFILE="Z-Score-SPSS-Output.pdf".

Python Code for Z Score

import pandas as pd
import numpy as np

df = pd.read_csv("dataset.csv")

numeric_vars = ["G1", "G2", "G3", "age", "absences"]

for col in numeric_vars:
    df[col] = pd.to_numeric(df[col], errors="coerce")
    df[f"z_{col}"] = (df[col] - df[col].mean()) / df[col].std(ddof=1)

# G3 z-score flags
df["abs_z_G3"] = df["z_G3"].abs()
df["flag_abs_z_G3_gt2"] = df["abs_z_G3"] > 2
df["flag_abs_z_G3_gt3"] = df["abs_z_G3"] > 3

summary_rows = []

for col in numeric_vars:
    z_col = f"z_{col}"
    summary_rows.append({
        "variable": col,
        "n": df[col].notna().sum(),
        "raw_mean": df[col].mean(),
        "raw_sd": df[col].std(ddof=1),
        "z_min": df[z_col].min(),
        "z_max": df[z_col].max(),
        "z_mean": df[z_col].mean(),
        "z_sd": df[z_col].std(ddof=1),
        "count_abs_z_gt_2": (df[z_col].abs() > 2).sum(),
        "count_abs_z_gt_3": (df[z_col].abs() > 3).sum()
    })

z_summary = pd.DataFrame(summary_rows)
print(z_summary)

# Band table for G3
bins = [-np.inf, -3, -2, -1, 0, 1, 2, 3, np.inf]
labels = ["below -3", "-3 to -2", "-2 to -1", "-1 to 0",
          "0 to 1", "1 to 2", "2 to 3", "above 3"]

df["z_band_G3"] = pd.cut(df["z_G3"], bins=bins, labels=labels, right=False)
band_table = df["z_band_G3"].value_counts(sort=False).to_frame("frequency")
band_table["percent"] = 100 * band_table["frequency"] / len(df)

print(band_table)

# Most distant cases
top_cases = df.assign(case_number=np.arange(1, len(df) + 1)) \
              .sort_values("abs_z_G3", ascending=False) \
              [["case_number", "G3", "z_G3", "abs_z_G3"]] \
              .head(20)

print(top_cases)

R Code for Z Score

df <- read.csv("dataset.csv")

numeric_vars <- c("G1", "G2", "G3", "age", "absences")

for (v in numeric_vars) {
  df[[v]] <- as.numeric(df[[v]])
  df[[paste0("z_", v)]] <- as.numeric(scale(df[[v]]))
}

df$abs_z_G3 <- abs(df$z_G3)
df$flag_abs_z_G3_gt2 <- df$abs_z_G3 > 2
df$flag_abs_z_G3_gt3 <- df$abs_z_G3 > 3

summary_rows <- lapply(numeric_vars, function(v) {
  zvar <- paste0("z_", v)
  data.frame(
    variable = v,
    n = sum(!is.na(df[[v]])),
    raw_mean = mean(df[[v]], na.rm = TRUE),
    raw_sd = sd(df[[v]], na.rm = TRUE),
    z_min = min(df[[zvar]], na.rm = TRUE),
    z_max = max(df[[zvar]], na.rm = TRUE),
    z_mean = mean(df[[zvar]], na.rm = TRUE),
    z_sd = sd(df[[zvar]], na.rm = TRUE),
    count_abs_z_gt_2 = sum(abs(df[[zvar]]) > 2, na.rm = TRUE),
    count_abs_z_gt_3 = sum(abs(df[[zvar]]) > 3, na.rm = TRUE)
  )
})

z_summary <- do.call(rbind, summary_rows)
print(z_summary)

df$z_band_G3 <- cut(
  df$z_G3,
  breaks = c(-Inf, -3, -2, -1, 0, 1, 2, 3, Inf),
  labels = c("below -3", "-3 to -2", "-2 to -1", "-1 to 0",
             "0 to 1", "1 to 2", "2 to 3", "above 3"),
  right = FALSE
)

band_table <- as.data.frame(table(df$z_band_G3))
band_table$percent <- 100 * band_table$Freq / nrow(df)
print(band_table)

top_cases <- df[order(-df$abs_z_G3), c("G3", "z_G3", "abs_z_G3")]
head(top_cases, 20)

Excel Formulas for Z Score

Mean:
=AVERAGE(A2:A650)

Sample standard deviation:
=STDEV.S(A2:A650)

Manual Z Score:
=(A2-$B$1)/$B$2

Excel STANDARDIZE function:
=STANDARDIZE(A2,$B$1,$B$2)

Absolute Z Score:
=ABS(C2)

Unusual value flag:
=IF(ABS(C2)>2,"Unusual / Review","Ordinary")

Extreme value flag:
=IF(ABS(C2)>3,"Extreme / Review","Not extreme")

Z Score band:
=IFS(C2<-3,"below -3",C2<-2,"-3 to -2",C2<-1,"-2 to -1",C2<0,"-1 to 0",C2<1,"0 to 1",C2<2,"1 to 2",C2<3,"2 to 3",C2>=3,"above 3")

APA Reporting Wording for Z Score

When reporting Z Score results, include the raw mean and standard deviation, confirm that the standardized variable has mean 0 and standard deviation 1, and report the number of cases exceeding the selected screening cutoffs.

APA-Style Reporting Paragraph

Short Report Sentences

• Z Score standardization: G3 was standardized so that the transformed variable had M = 0.000 and SD = 1.000.
• Unusual values: A total of 19 G3 cases exceeded |z| > 2, indicating values more than two standard deviations from the mean.
• Extreme values: A total of 16 G3 cases exceeded |z| > 3, indicating extreme standardized values for review.
• Direction: The most distant G3 observations were below the mean, with a minimum Z Score of -3.685.

Common Mistakes in Z Score Interpretation

Z Score analysis is easy to calculate, but mistakes are common when interpreting the result. The main error is treating a large Z Score as automatic proof that a value is wrong. A large Z Score only tells us that the value is far from the mean.

When to Use Z Score

Use Z Score when you want to standardize values, compare variables with different scales, identify observations far from the mean, or create standardized scores for further analysis.

Downloads and Resources for Z Score

The following resources support this Z Score guide and can be used to verify the SPSS output and reproduce the analysis.

Related Statistical Guides

These related guides help readers connect Z Score with descriptive statistics, outlier detection, normality checks, spread, and reporting.