STATISTICAL TECHNIQUES FOR research ROMMEL S DE GRACIA

STATISTICAL TECHNIQUES FOR research ROMMEL S. DE GRACIA

What options you have? • BE LAME – “Usual Thinking Syndrome” – soothed with what you have and “think no more”; • BE RIDICULOUS - Do things badly; or • BE STRATEGIC – be an agent of positive change; Do the right things right.

Considerations for Choosing a Statistical Technique THE RESEARCH QUESTION. What is it that we want to know? That is, do we want to test differences, to explore relationships between variables, or to simply describe our data? THE NATURE AND LEVEL OF MEASUREMENT OF THE DATA COLLECTED. That is, is our data numerical or categorical? Is our data of the nominal, ordinal, or interval/ratio level? NUMBER OF VARIABLES INVOLVED IN THE INVESTIGATION. That is, do we need to use

Scale of Measurement Nominal – qualitative and categorical (for identification) Ordinal – qualitative and ranked data – (converted numerically) Interval-Ratio – quantitative with meaning

Guide for Choosing Statistical Techniques 1 Univariate techniques are used to investigate one variable at a time. 1 Bivariate techniques are used to investigate two variables at time. 1 Multivariate techniques are used to analyze more than two variables simultaneously.

Some Examples: Example of Research Questions What is the compliance in the implementation of the Grade 2 English curriculum relative to a. Percent of LC’s implemented b. Number of teaching -learning delivered c. Percent of learning competencies mastered Independent Variable Dependent Variable Essential Feature Level/ Extent of Compliance Descriptive a. % of LCs implemented. b. No of teachinglearning delivered c. % of learning competence Descriptive Statistics Mean % Median Standard Deviation

Guide for Choosing Statistical Techniques for Exploring Differences between Groups - Testing whethere is a significant difference among a number of groups. Most of these analyses involve comparing the mean/median scores of groups on one or more dependent variables. Techniques for Exploring Relationships among Variables - Testing if the relationships (causal and non-causal) among variables are statistically significant. The following are considered: 1) relationship between two variables (e. g. , Pearson r) and 2) causal relationships of variable/s on another variable (e. g. , linear regression analysis).

A Comparison Nonparametric tests are distribution free and do not require testing for assumptions. When assumptions are met, parametric tests are more powerful.

YES NO YES

Some Examples: Example of Research Questions Is there a change in the participant’s level of understanding action research before and after the seminar? Independent Variable Time (Before and After) Categorical IV: Two Levels Dependent Variable Level of Understanding Essential Feature Same people/ participants on two different occasions Parametric Statistics Non-Parametric (Alternative) Paired. Sample t-Test (Dependent Wilcoxon Signed. Rank Test Samples)

Some Examples: Example of Independent Research Variable Questions GENDER Is there a (Male and significant Female) difference in the Spelling Test Categorical Scores between IV: males and Two Levels females? Dependent Variable Essential Feature Parametric Statistics Non-Parametric (Alternative) Spelling Test Scores Two Groups: Different people in each group T-Test for Mann Whitney test Independent Samples Wilcoxon Rank sum test

Some Examples: Example of Independent Research Variable Questions Is there a significant difference in the Spelling Test Scores for the three-different curricula (RBEC, STEM, SPA)? Dependent Variable Essential Feature Parametric Statistics Non-Parametric (Alternative)

Some Examples: Example of Research Questions Is there a significant difference in the Spelling Test Scores for the three-different curricula (RBEC, STEM, SPA)? Independent Variable Dependent Variable Essential Feature Parametric Statistics Non-Parametric (Alternative) Curricula (RBEC, STEM, SPA) Spelling Test Scores More than two groups: Different people in each group One-Way ANOVA using F-Test Kruskal Wallis ANOVA by ranks Categorical IV: Three Levels/Groups

Some Examples: Example of Independent Research Variable Questions Is there a significant difference in the performance in English 2 when grouped according to the strategies of teaching (traditional and non-traditional) used? Dependent Variable Essential Feature Parametric Statistics Non-Parametric (Alternative)

Some Examples: Example of Research Questions Is there a significant difference in the performance in English 2 when grouped according to the strategies of teaching (traditional and non-traditional) used? Parametric Statistics Independent Variable Dependent Variable Essential Feature Strategies of teaching (Traditional and nontraditional) Performance in English 2 Two groups of T-Test for Mann different Independent samples Whitney test Samples Categorical IV: Two levels Non-Parametric (Alternative) Wilcoxon Rank sum test

Some Examples:

Example of Independent Research Variable Questions Is there a relationship between educational attainment and percent of LCs implemented? Dependent Variable Essential Feature Parametric Statistics Non-Parametric (Alternative)

Some Examples: Example of Research Questions Is there a relationship between educational attainment and percent of LCs implemented? Independent Variable Dependent Variable Essential Feature Educational attainment Percent of implemented LCs Ordinal data versus Interval Ratio Ordinal Data Parametric Statistics Non-Parametric (Alternative) - - - - Spearman- Rho

WORKSHOP 1