01 / 20

Basic Statistics for
Biochem Data

Master the essential statistical tools: p-values, error bars, and significance testing used to interpret biochemistry experiments.

Biochemistry

Why Statistics Matter in Biochemistry

Every biochemistry experiment produces variable data. Enzyme assays, protein concentrations, and cell counts never give the exact same number twice. Statistics give us formal tools to separate real biological effects from random noise.

Core Statistical Terms

Mean \(\bar{x}\)
The arithmetic average of your data: \(\bar{x} = \frac{1}{n}\sum_{i=1}^{n} x_i\)
Standard Deviation \(s\)
Measures how spread out values are from the mean: \(s = \sqrt{\frac{\sum(x_i - \bar{x})^2}{n-1}}\)
SEM
Standard Error of the Mean: \(\text{SEM} = \frac{s}{\sqrt{n}}\). How precisely you know the mean.

Visualizing Data Spread

0 20 40 60 Activity (U/mg) Control Drug A Drug B Error bars = SD

The Null Hypothesis \(H_0\)

In hypothesis testing, we start by assuming there is no real difference between groups. This assumption is called the null hypothesis (\(H_0\)). The alternative hypothesis (\(H_a\)) states that a real difference exists.

\(H_0\): Null
"The drug has no effect on enzyme activity." Any observed difference is due to chance.
\(H_a\): Alternative
"The drug does change enzyme activity." The observed difference is real and reproducible.

What Is a p-value?

The p-value is the probability of observing data at least as extreme as your results, assuming the null hypothesis is true.

0 +t -t
Distribution under \(H_0\)
 p-value region

Interpreting \(p\)-values

\(p < 0.05\)
Statistically significant. Less than 5% chance the result is due to random variation. Reject \(H_0\).
\(p < 0.01\)
Highly significant. Less than 1% chance. Stronger evidence against \(H_0\).
\(p \geq 0.05\)
Not significant. Cannot reject \(H_0\). The observed difference may be due to chance.

SD vs SEM: A Key Distinction

Standard Deviation (SD) Standard Error (SEM)
Formula \(s = \sqrt{\frac{\sum(x_i - \bar{x})^2}{n-1}}\) \(\text{SEM} = \frac{s}{\sqrt{n}}\)
Measures Spread of individual data points Precision of the estimated mean
As \(n\) grows Stays roughly the same Gets smaller
Use for Showing biological variability Comparing group means
Error bars Wider: shows data range Narrower: shows mean certainty

Reading Error Bars Correctly

0 50 100 Group A (SD bars) Group A (SEM bars) Group B (SD bars) Group B (SEM bars) SD: wide SEM: narrow

How to Analyze Biochem Data

1
Collect replicates. Perform at least \(n = 3\) independent trials per condition to enable meaningful statistics.
2
Calculate mean and SD. Compute \(\bar{x}\) and \(s\) for each group. Check for obvious outliers.
3
Choose error bar type. Use SD to show variability; use SEM to compare means between groups.
4
Run a statistical test. For two groups, use a t-test. For three or more groups, use ANOVA.
5
Report the p-value. State the exact value (e.g., \(p = 0.003\)) and whether it meets your \(\alpha\) threshold.

The t-test: Comparing Two Means

The independent two-sample t-test asks whether two group means differ more than expected by chance. The test statistic is:

Numerator
The difference between the two group means: how far apart they are.
Denominator
The pooled uncertainty in those means (based on SD and sample size).

Example 1: Enzyme Inhibition

An enzyme assay measures activity in control vs inhibitor-treated samples (\(n = 5\) each):

Control Inhibitor
Mean (\(\bar{x}\)) 42.0 U/mg 28.4 U/mg
SD (\(s\)) 3.6 4.1
SEM 1.61 1.83

Example 2: Overlapping Error Bars

0 25 50 Diet A Diet B overlap zone

Example 3: Sample Size and Power

A researcher tests whether a new substrate increases reaction rate. Same means, same SD, but different sample sizes:

\(n = 3\) \(n = 10\) \(n = 30\)
Difference in means 5.0 5.0 5.0
Pooled SD 6.0 6.0 6.0
SEM of difference 4.90 2.68 1.55
t-statistic 1.02 1.86 3.23
p-value \(p = 0.37\) \(p = 0.08\) \(p = 0.002\)

Common Mistakes

✕ Confusing SD and SEM
Using SEM bars to claim "low variability" when SD is actually large. SEM shrinks with more replicates; SD does not.
✕ Eyeballing error bars
Claiming "no difference" because bars overlap. Overlapping SEM bars can still yield \(p < 0.05\). Always run the test.
✕ p = proof
Thinking \(p < 0.05\) proves the hypothesis is true. It only means the data are unlikely under \(H_0\). It says nothing about effect size.

Summary Reference Table

Concept What It Tells You Key Formula
Mean Central value of data \(\bar{x} = \frac{\sum x_i}{n}\)
SD Spread of individual points \(s = \sqrt{\frac{\sum(x_i-\bar{x})^2}{n-1}}\)
SEM Precision of estimated mean \(\frac{s}{\sqrt{n}}\)
t-statistic Signal-to-noise for group comparison \(t = \frac{\bar{x}_1 - \bar{x}_2}{SE_{\text{diff}}}\)
p-value Probability of data under \(H_0\) From t-distribution table

Question 1 of 5

A researcher reports \(p = 0.03\) for the difference in protein yield between two purification methods. At \(\alpha = 0.05\), what is the correct conclusion?

Question 2 of 5

A dataset has \(s = 12\) and \(n = 9\). What is the standard error of the mean (SEM)?

Question 3 of 5

Two groups have identical means and SDs. If you increase the sample size from \(n = 5\) to \(n = 50\) per group, what happens?

Question 4 of 5

A bar graph shows two groups with SEM error bars that clearly overlap. Which statement is most accurate?

Question 5 of 5

Which error bar type would you choose to display how precisely your experiment estimates the true mean enzyme activity?