Data Analysis
CSIR NET Chemical Sciences data analysis: mean and standard deviation, absolute and relative errors, linear regression, covariance and correlation.
- 3
- Topics
- 59
- Min read
- 01 Mean and Standard DeviationStart here CSIR NET Chemical Sciences data analysis: mean, median, sample vs population standard deviation, RSD, standard error, Gaussian curve, confidence limits, Q-test.
- 02 Absolute and Relative Errors CSIR NET Chemical Sciences: absolute and relative error, accuracy vs precision, systematic and random errors, significant figures and propagation of errors.
- 03 Linear Regression, Covariance and Correlation CSIR NET Chemical Sciences: least-squares straight lines, slope and intercept errors, linearised plots in kinetics and thermodynamics, covariance and r.
- Data Analysis Formula Sheet and Short Notes Revise The whole chapter on one page.
Chapter Overview
The CSIR NET Chemical Sciences syllabus lists a short physical chemistry unit, Data Analysis, in four topic lines: mean and standard deviation, absolute and relative errors, linear regression, and covariance and correlation coefficient. It is the statistics every practising chemist uses on every experiment: reporting replicate titres, deciding whether a result is accurate, fitting a calibration line, and pulling an activation energy out of an Arrhenius plot.
The unit is small and almost entirely numerical, which makes it one of the most reliable scoring areas in the paper. The formulas are few; the marks are lost to a handful of predictable slips (dividing by \( n \) instead of \( n - 1 \), adding errors in the wrong form, using log₁₀ with an ln formula). This chapter is built around those slips.
What You’ll Learn
By the end of this chapter you will be able to:
- Compute the mean, median, sample and population standard deviation, variance, RSD and standard error of replicate data, and say what shifting or scaling the data does to each
- Use the Gaussian distribution and Student’s \( t \) to put confidence limits on a mean, and apply the Q-test to a suspect value
- Express absolute and relative errors, separate accuracy from precision and systematic from random error, and handle significant figures, including in logs
- Propagate errors through sums, products, powers, logs and exponentials, by both the maximum-error and the standard-deviation rules
- Fit a least-squares straight line from raw data or from given sums, find the uncertainty of its slope and intercept, and read an unknown off it
- Linearise kinetic, thermodynamic, electrochemical and adsorption laws and extract the physical constant from the slope or intercept
- Compute the covariance and Pearson’s \( r \), relate \( r \) to both regression slopes, and know what \( r \) and \( r^2 \) do and do not show
Prerequisites
- Averages and percentages: Percentages, Ratios, Averages and Mixtures
- Reading spread and scatter off graphs: Histograms, Distributions, Scatter and Box Plots
- Logarithms, and BSc-level rate laws and the Arrhenius equation
Learning Path
graph TD
A[Averages, logarithms,
rate laws] --> B[1. Mean and Standard Deviation
s, RSD, standard error,
Gaussian, t, Q-test]
B --> C[2. Absolute and Relative Errors
accuracy vs precision,
sig figs, propagation]
B --> D[3. Linear Regression,
Covariance and Correlation
least squares, linear plots, r]
C --> D
D --> E[Quick Revision]Page 1 defines the standard deviation used everywhere else. Page 2 needs it for propagation. Page 3 uses both: the uncertainty of a slope is a standard deviation, and the constant extracted from it carries a propagated error.
Topics in This Chapter
| # | Topic | What it covers | Time |
|---|---|---|---|
| 1 | Mean and Standard Deviation | Mean, median, range; sample vs population SD and \( n - 1 \); computational and coded forms; variance, RSD, CV; shift and scale rules; pooled SD; standard error; Gaussian bands; confidence intervals with \( t \); t- and F-tests in brief; Q-test | 25 min |
| 2 | Absolute and Relative Errors | Absolute and relative error and uncertainty; accuracy vs precision; systematic (instrumental, method, personal; constant vs proportional), random and gross errors; significant figures and logs; propagation rules (maximum and quadrature), general formula; pH, molarity, equilibrium-constant and Arrhenius examples | 30 min |
| 3 | Linear Regression, Covariance and Correlation | Least squares, normal equations, slope and intercept, line through origin, residuals; \( s_{y/x} \), \( s_b \), \( s_a \), unknown from a line; linearised laws table; covariance and its properties; Pearson’s \( r \), \( r^2 \), regression of \( y \) on \( x \) and \( x \) on \( y \); limits of \( r \) | 35 min |
| — | Quick Revision | Every formula and trap on one page | 5 min |
Syllabus Coverage
| Official topic line | Covered in |
|---|---|
| Mean and standard deviation | Mean and Standard Deviation |
| Absolute and relative errors | Absolute and Relative Errors (propagation of errors included) |
| Linear regression | Linear Regression, Covariance and Correlation (least squares, slope and intercept errors, linearised plots) |
| Covariance and correlation coefficient | Linear Regression, Covariance and Correlation (covariance, Pearson’s \( r \), \( r^2 \); covariance in error propagation) |
How CSIR NET Asks It
Chemical Sciences is marked as follows, per the NTA information bulletin for December 2026:
| Part | Questions given | Attempt | Correct | Wrong | Maximum |
|---|---|---|---|---|---|
| B | 40 | 35 | +2 | −0.5 | 70 |
| C | 60 | 25 | +4 | −1 | 100 |
If more than the permitted number of questions is answered in a part, only the first ones answered (35 in Part B, 25 in Part C) are evaluated. The bulletin gives no count of questions per syllabus line. The kinds of question this chapter supports:
- Part B: the mean, SD or RSD of four to six numbers; the effect of adding a constant or scaling; the standard error of a mean; the percentage error of a reported value; the error in an area, volume or density from a measured length; which error is systematic; the error in [H⁺] from a pH uncertainty; the slope and intercept from given sums; which plot is linear for a law, and its slope; the value of \( r \) after a change of variables.
- Part C: confidence intervals and the number of replicates needed; outlier rejection and recomputation; propagation through a formula with several measured quantities; the temperature sensitivity of a rate constant; extracting \( E_a \), \( k \) or a concentration from tabulated data; means, \( r \) and SDs from two given regression lines; statement sets on accuracy, precision and the limits of \( r \).
Quick Reference
| Area | Key facts |
|---|---|
| Spread | \( s = \sqrt{\sum (x - \bar{x})^2/(n - 1)} \); RSD = \( s/\bar{x} \); \( s_{\bar{x}} = s/\sqrt{n} \) |
| Gaussian | ±1σ 68.3%, ±2σ 95.4%, ±3σ 99.7% |
| Confidence | \( \mu = \bar{x} \pm t s/\sqrt{n} \) |
| Errors | \( E_a = x - x_t \); \( E_r = E_a/x_t \) |
| Propagation | sums: absolute; products: relative; \( a^n \): × \( n \); \( \log a \): \( 0.434\,s_a/a \) |
| Regression | \( b = S_{xy}/S_{xx} \); \( a = \bar{y} - b\bar{x} \); \( s_{y/x} \) with \( n - 2 \) |
| Correlation | \( r = S_{xy}/\sqrt{S_{xx}S_{yy}} \); \( b_{yx} b_{xy} = r^2 \) |
| Arrhenius | \( \ln k \) vs \( 1/T \): slope \( -E_a/R \) |
This chapter supports almost every numerical unit of the paper: the calibration lines of Analytical Chemistry, the Kohlrausch extrapolation in Electrochemistry, the counting statistics in Nuclear Chemistry, and the linear plots of Chemical Thermodynamics. For the subject overview and marking, see Chemical Sciences. Found an error? See About for how to report it.