DIFFERENCES BETWEEN

PARAMETRIC AND
NON-PARAMETRIC TESTS
For hypothesis testing
Apply both and compare the results
References and other suggested readings on the topic:
Books:
• Chen, P. Y., & Popovich, P. M. (2002). Correlation: Parametric and nonparametric measures. Thousand Oaks, CA: Sage.
• Field, A. (2009). Discovering statistics using SPSS. Sage publications.
• Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., & Tatham, R. L. (1998). Multivariate data analysis (Vol. 5, No. 3, pp. 207-219). Upper Saddle
River, NJ: Prentice hall.
• Kothari, C. R. (2004). Research methodology: Methods and techniques. New Age International.
• Pallant, J. (2001). SPSS Survival Manual: A Step by Step Guide to Data Analysis Using SPSS for Windows (Versions 10 and 11): SPSS Student Version
11.0 for Windows. Open University Press.
• Siegel, S., & Castellan, N. J. (1988). Nonparametric statistics for the behavioral sciences (2nd ed.). New York: McGraw-Hill.

Research Journals:
• Conover, W. J., & Iman, R. L. (1981). Rank transformations as a bridge between parametric and nonparametric statistics. The American
Statistician, 35(3), 124-129.
• Harwell, M. R. (1988). Choosing between parametric and nonparametric tests. Journal of Counseling & Development, 67(1), 35-38.
• Havlicek, L. L., & Peterson, N. L. (1976). Robustness of the Pearson correlation against violations of assumptions. Perceptual and Motor
Skills, 43(3_suppl), 1319-1334.
• Hunter, M. A., & May, R. B. (1993). Some myths concerning parametric and nonparametric tests. Canadian Psychology/Psychologie canadienne, 34(4),
384.
• Hunter, J. E., & Schmidt, F. L. (1990). Dichotomization of continuous variables: The implications for meta-analysis. Journal of Applied
Psychology, 75(3), 334.
• Norman, G. (2010). Likert scales, levels of measurement and the “laws” of statistics. Advances in health sciences education, 15(5), 625-632.
• Pearson, E. S. (1931). The analysis of variance in cases of non-normal variation. Biometrika, 114-133.

Video files:
• Field A., (2012, Feb 2) Nonparametric Tests on SPSS [Video File]. Retrieved from https://www.youtube.com/watch?v=esNb6RFIXvw
• Grande T., (2015, July 30) Parametric and Nonparametric Statistical Tests [Video File]. Retrieved from https://www.youtube.com/watch?
v=pWEWHKnwg_0
• Vidya-mitra (2016, Oct 16) Parametric & Non-Parametric Tests (PSY) [Video File]. Retrieved from https://www.youtube.com/watch?v=MOy2gcCkcRY
• Parameters are descriptions to describe the entire population.
• Population mean
• Statistic is a description that describes a sample
• Sample mean
Introduction

• Prediction or educated guess

• Hypothesis is proposition or predictive statement about the possible outcome of
a research study based on a particular property of a population.
• Hypothesis testing is a method of testing a claim or hypothesis about a parameter
in a population, using data measured in a sample.
• Why tested using sample??

Marks of students in class Salary of employees in an organization in

97 INR
96 4,22,30,500
95 40,000
94 35,000
99 30,000
98 25,000
5,000
Types of data-different types of tests
• Nominal
• Ordinal
• Interval
• Ratio
• Skewed
• Scattered
 Parametric Test Non-Parametric Test
• Method of measurement that rely on • Distribution free tests.
assumptions that data are drawn from • Based on ranked data.
normal distribution • When you want comparative rather than
• Assumptions - data are drawn from a absolute results.
specific distribution. • But leads to loss of information related
• Based on measured actual data. to magnitude of differences (Rank 1- 99
and rank 2 98 or 2)
• Show example of correlation • When median and not mean to represent
• Good to test group means. Use a the centre of your distribution better.
mean value for the central tendency. Example
• When the standard error is low • Pareto's principle, outliers,
(Standard deviation of sample mean) heterogeneous population
• Software used are SPSS, R, STATA • Software- STATA and R are better for
Non-Parametric tests than SPSS
(Wilcox, 2005).
WINSORIZING OUTLIERS (Good or
bad?)
 Parametric Test Non-Parametric Test
• Assumptions need to be met • Assumption free test****
• Assumptions (Field, 2009)*: • Less rigid assumptions
 Normally distributed ** (bell, systematic, • Assumptions (Field, 2009):
asymptotic, unimodal, mean=median=mode) KS
test (0.05 above), pictorial representation,  Don’t require data to follow normal distribution.
Skewness, Kurtosis (between plus 2, minus 2)  Independent random samples
 Homogeneity of variances*** (variance within  Measures at least on ordinal scale in most of the
the sample not fluctuating) Levene test (0.05 tests*****
more)  Expected frequency in 2X2 table should be at
 Data measured at least at Interval scale. least 10 and in other tables at least 5 (Explain)
 Random independent sample  Test dependent assumptions******
 No Outliers
*parametric tests are extremely robust with respect to violations of assumptions (Norman, 2010; Hunter & Schmidt, 1990; Pearson, 1931; Havlicek &
Peterson, 1976).
**can be applied even with non-normal data but sample size should be suitably large. (Central Limit Theorem)
***can work even when groups have different spreads, assuming unequal variances, Welch Test is also there.
****Myth- Assumptions are less rigid and fewer but assumptions are there (Hunter & May, 1993; Field, 2009)
***** With nominal data in case of Chi-square test and McNemar's test
****** Mann Whitney- two sample having same shape; Wilcoxon Signed Rank Test- Systematic Distribution; Kruskal Wallis and Friedman’s ANOVA- Same
shape and equal variances
 Advantages and disadvantages

Parametric Test Non-Parametric Test

• More statistical Power: More likely to detect a • Have less statistical power*. Larger sample size can
significant effect when one truly exists. (because on be required to draw conclusions with the same degree
actual data) of confidence
• More powerful when assumptions are not fulfilled.
• Less likely than nonparametric tests to make type II • Have more applicability as have milder assumptions.
error (More power, it is chances or probability of • When data is nominal or ordinal or ranked or when
rejecting the truly false null hypothesis) outliers can’t be removed.
• More generalizability • More applicability, Is more robust in nature
• Doesn’t require previous knowledge about the
• Do require previous knowledge about the population. population.
• The computations on nonparametric statistics are
• Parametric tests are usually more widely usually less complicated than those for parametric
available and more well known. statistics, particularly for small samples.

*Myth: with large populations and when assumptions of parametric tests are failed, nonparametric tests become nearly as powerful as parametric tests and in few cases even more than
them (Hunter & May, 1993).
 Non-Parametric substitute (Pallant, 2001; Field,
Parametric Tests
2009).
Spearman’s rho Correlation or Kendall tau Correlation
Pearson Correlation (r)
(when sample size is low).
Partial Correlation X
Multiple Regression X (Non linear regression method which works on Medians)
X Chi-square
Independent Sample t-test Mann-Whitney Test or Kolmogorov-Smirnov Test.
Wilcoxon Signed-Rank Test
Paired Samples t-test
McNemar (when data is nominal).
Kruskal-Wallis
One-way between groups ANOVA
(Jonckheere–Terpstra test when the treatments are ordered)
One way Repeated Measures ANOVA Friedman’s test.
Two-way between groups ANOVA
Multivariate ANOVA (MANOVA) X
Analysis of covariance (ANCOVA)
Factor Analysis X
Thank You