Introduction to statistics for editors: understanding and reporting descriptive and bivariate analyses

Facilitator: Darko Hren, PhD, Split, Croatia

Most research, especially in biomedical, behavioral and social sciences, involves statistical analysis of data. Manuscript editors rarely have appropriate training in statistics and data presentation and often have trouble figuring out what the results of statistical analyses mean. Most people who try to learn statistics from a book give up after the introduction or, at best, after first couple of chapter – learning statistics from books is both difficult and boring.

The aim of this workshop is to offer participants a more digestible opportunity to develop understanding of basic statistical concepts.

Purpose:To help manuscript editors understand basic statistical concepts and terminology and provide them with elementary skills in reporting analysed data.

Approach: The workshop will be a mixture of input and practical work. Each concept will first be explained and examples discussed. Practical exercises will be given along the way to put the knowledge into practice.

Structure:The workshop will consist of three modules with two 15 minute breaks.

Module I* (70 minutes) Descriptives, normal distribution and hypothesis testing: Participants will become familiar with basic descriptive indicators (mean, standard deviation, standard error, confidence intervals, median, interquartile range, frequencies, percentages). They will learn when it is appropriate to use these indicators and what they mean. Elementary logic of sampling procedures, normal distribution and hypothesis testing will be explained.
Module II (50 minutes) Group comparisons: What are independent and paired samples? Methods for comparing groups: • for categorical data – chi square test; • for continuous normaly distributed – t-test and ANOVA; • for ordinal or continuous non-normally distributed – Mann-Whitney test, Wilcoxon matched pairs test, Kruskal-Wallis test, Friedman's test. Post hoc tests – what, why, how and which? Practice in reporting results of testing group comparisons.
Module III (30 minutes) Bivariate association: The difference between group comparisons and testing for associations. Methods for testing associations: • for categorical data – chi-square test of association; • for continuous normally distributed – Pearson's r coefficient of correlation; • for ordinal or continuous non-normally distributed – Spearman's ρ. Association between risk factors and and outcome: absolute risk, relative risk, odds ratio. Practice in reporting results of testing bivariate assiciations.

Outcome skills:After the workshop, participants will understand different approaches for data description and the rationale for using them. They will understand methods for testing differences between two or more groups and for testing association between two variables. They will also learn how the results of these analyses can be meaningfully presented.

Pre-meeting information: This workshop will be taught with a presumption that participants know and understand different scales of measurement and study designs.
For basic understanding (enough to follow the workshop) of scales of measurement read:
http://allpsych.com/researchmethods/measurementscales.htmlor http://en.wikipedia.org/wiki/Level_of_measurement
For a more detailed explanation read:
http://www.wadsworth.com/psychology_d/templates/student_resources/workshops/stat_workshp/scale/scale_01.html;
For study design start from: http://en.wikipedia.org/wiki/Study_designand then follow the links to: Randomized controlled trial; Cohort study, Case-control study, Cross-sectional study.