The Arithmetic Mean: Calculating Central Tendency for Grouped and Ungrouped Data

Every time a researcher reports an “average” score – whether it’s the mean anxiety level of therapy participants, the typical reaction time in a cognition experiment, or the average IQ across a school – they are using the arithmetic mean. It is the single most widely used measure of central tendency in psychology and statistics, and for good reason: it condenses an entire dataset into one representative number. But calculating the mean isn’t always as simple as adding numbers and dividing. Depending on how your data is organized – raw individual scores or grouped into intervals – the method you use changes. This post walks you through each approach, step by step.

What is the arithmetic mean? 🔗

According to research published in medical and statistical literature, the arithmetic mean is computed by adding all values in a dataset and dividing by the number of observations. It is the most commonly used form of average – so much so that when someone simply says “mean,” they almost always refer to the arithmetic mean. Introductory statistics in psychology describe the mean as the balancing point of a distribution – the point where the total distance to all scores above it equals the total distance to all scores below it. This property makes it mathematically precise and ideal for datasets that don’t contain extreme outliers.

In psychological research, the mean is used extensively for group comparisons – comparing treatment versus control groups, tracking a group’s progress from intake to discharge, or measuring how students develop over different grade levels. It also forms the foundation for more advanced statistics like standard deviation, t-tests, and ANOVA.

The general formula for the arithmetic mean is:

M = ΣX / N

Where ΣX is the sum of all values and N is the total number of observations.

Mean from ungrouped data 🔗

Ungrouped data refers to raw, individual data points – a list of numbers that haven’t been categorized or placed into intervals. In educational psychology contexts, the formula for ungrouped data is expressed as M = ΣX/N, where each score is summed directly and divided by the count of scores.

Steps for ungrouped data 🔗

The process is straightforward:

1. List all individual data values (X₁, X₂, X₃ … Xₙ).
2. Sum all values: ΣX.
3. Count the total number of values: N.
4. Divide the sum by N: M = ΣX / N.

Worked example 🔗

Suppose a psychologist records the number of hours per week five participants spend on mindfulness practice: 4, 6, 5, 8, 7.

• ΣX = 4 + 6 + 5 + 8 + 7 = 30
• N = 5
• M = 30 / 5 = 6 hours

The mean mindfulness time is 6 hours per week. Every data point is included in the calculation, which is both a strength and a limitation – one extreme value can shift the mean noticeably.

Mean from grouped data: the long method 🔗

When a dataset is large, listing every individual score becomes impractical. In such cases, data is organized into a frequency distribution table – grouping scores into class intervals and recording how many observations fall within each interval. With grouped data, individual values are no longer available, so the mean is estimated by working with the midpoint of each class interval.

The long method (also called the direct method) uses these midpoints multiplied by their corresponding frequencies to estimate the sum of all values. The formula is:

M = Σ(f × x) / N

Where f is the frequency of each class interval, x is the midpoint (class mark) of each interval, and N is the total number of frequencies (Σf).

Steps for grouped data (long method) 🔗

1. Find the midpoint (x) of each class interval: x = (lower limit + upper limit) / 2.
2. Multiply each midpoint by its frequency: f × x.
3. Sum all f × x values: Σ(f × x).
4. Sum all frequencies: N = Σf.
5. Apply the formula: M = Σ(f × x) / N.

Worked example 🔗

A researcher collects stress scores from 40 participants and groups them as follows:

M = 1300.0 / 40 = 32.5

The estimated mean stress score is 32.5. Note that this is an estimate – because we don’t know each individual score, we assume all values within each interval are centered at the midpoint.

Mean from grouped data: the short method 🔗

The long method works well, but when class midpoints are large numbers, multiplying them by frequencies produces unwieldy figures that are prone to arithmetic errors. The short method (also called the assumed mean method or shortcut method) solves this by working with smaller deviation values rather than raw midpoints.

The short-cut approach is used whenever data values are large and direct calculation becomes time-consuming. Instead of multiplying each midpoint by its frequency, you select one midpoint as an assumed mean (A), calculate how much each other midpoint deviates from it, then adjust the assumed mean using those deviations. The assumed mean method works by choosing a value close to the actual mean, and then calculating deviations from this assumed value to find the true mean.

The formula is:

M = A + (Σfd / N)

Where A is the assumed mean, d is the deviation of each midpoint from A (d = x − A), f is the frequency, and N is the total frequency.

Steps for grouped data (short method) 🔗

1. Find the midpoint (x) for each class interval.
2. Select one midpoint as the assumed mean (A) – ideally near the center of the distribution.
3. Calculate deviations: d = x − A for each class.
4. Multiply each deviation by its frequency: f × d.
5. Sum all f × d values: Σfd.
6. Apply the formula: M = A + (Σfd / N).

Worked example 🔗

Using the same stress score dataset from above, we apply the short method. Let’s take A = 34.5 (midpoint of the middle interval) as the assumed mean:

M = 34.5 + (−80 / 40) = 34.5 − 2 = 32.5

The result matches the long method exactly – 32.5. The short method simply gets there with smaller numbers and less risk of computational error.

Long method vs. short method: when to use which 🔗

Both methods always produce the same result, so the choice is purely about efficiency. The long method is straightforward and conceptually transparent – it makes explicit what the mean actually represents. It’s ideal when midpoints are small and calculations are simple.

The short method is the better choice when midpoints are large or the dataset contains many class intervals. The assumed mean approach is most useful when the dataset has many values or the numbers are large, making direct calculation slow and error-prone. By reducing computations to small deviation values, the method results in smaller numerical values that make calculations easier with no long computations.

In practice, the assumed mean you pick does not need to be perfectly central – while any value can technically serve as the assumed mean, choosing one close to the actual mean minimizes computation and reduces potential errors.

Why the mean matters – and when to be cautious 🔗

The arithmetic mean is the most commonly used measure of central tendency in psychology, computed by summing all scores and dividing by the number of participants. It underpins many inferential statistics and is often the first value reported in research papers. However, it comes with an important caveat.

The mean is sensitive to extreme values (outliers): an unusually high or low score can shift the mean considerably, making it less representative of the typical participant. For example, if most participants in a well-being study score between 40 and 60, but one outlier scores 200, the mean will be pulled upward in a way that misrepresents the group. In such cases, the median is often preferred. The mean is most appropriate when scores are distributed symmetrically around a central point, and when further statistics like standard deviation or correlation coefficients are to be computed – since many of these rely directly on the mean as their foundation.

The arithmetic mean is also only meaningful for quantitative data measured on an interval or ratio scale. Applying it to categorical or nominal data produces numbers that carry no interpretable meaning.

A note on notation 🔗

You will encounter two notations depending on context. When referring to a sample, the mean is typically written as x̄ (x-bar) or M. When referring to an entire population, it is written as μ (the Greek letter mu). The calculation is identical in both cases – the distinction simply signals whether you’ve measured everyone in the group or just a representative subset. The sample mean is usually indicated by M or x̄, while population parameters use Greek notation.

What do you think? When would you choose the short method over the long method in a real research scenario – and does the size of the dataset alone determine that choice, or are there other factors worth considering? If a dataset on therapy outcomes includes one extreme outlier, how might that change your interpretation of the mean you’ve calculated?

References

1. https://pmc.ncbi.nlm.nih.gov/articles/PMC3127352/
2. https://open.maricopa.edu/psy230mm/chapter/chapter-4-measures-of-central-tendency/
3. https://stats.libretexts.org/Courses/Adler_University/Graduate-Level_Statistics_in_Psychology/05:_Measures_of_Central_Tendency/5.01:_What_is_Central_Tendency_Why_Do_We_Use_It
4. https://www.psychologydiscussion.net/educational-psychology/statistics/measures-of-central-tendency-and-how-it-helps-in-educational-psychology/2741
5. https://www.riosalado.edu/web/oer/WRKDEV100-20011_INTER_0000_v1/lessons/Mod05_MeanMedianMode.shtml
6. https://www.geeksforgeeks.org/maths/shortcut-method-for-arithmetic-mean/
7. https://www.geeksforgeeks.org/maths/assumed-mean-method/
8. https://testbook.com/maths/assumed-mean-method
9. https://collegedunia.com/exams/assumed-mean-method-mathematics-articleid-7366
10. https://www.vedantu.com/maths/assumed-mean-method
11. https://pressbooks.bccampus.ca/psychcapilano/chapter/descriptive-statistics/