Useful bits-and-pieces: Social & Behavioral Research

Black Swan

Just thought I would start a thread that would list things that might be useful to those in social, behavioural, business, education, survey, evaluation, etc., researches. Not intended to address specific research problems, rather just to be used to simulate ideas and thoughts about research approaches. Please feel free to contribute and comment.

Although a bit generic, and that which may or may not be in agreement with specific conceptual frameworks, research questions, design, methodologies, etc., here's something that may be useful to increasing stages of survey data analysis.

Description:
Percentage tables, and tabled measures of central tendency (means, medians, and modes), and dispersion (ranges and standard deviations), including bar and line graphs, may be used to describe data.

Survey Reliability Measure:
Cronbach coefficient alpha tests of internal consistency will be used to assess the reliability of survey questions. Alphas close to .75 or higher will be considered reliable.

Group Differences:
Differences between groups at the nominal data level may be analyzed using chi square; ordinal level with Mann-Whitney U; and interval or ratio levels t-test between 2 groups, or ANOVA for 3 or more groups. Results at the p<.05 confidence level will be considered significant.

Correlations:
Nominal data level associations may be analyzed using Pearson phi; at the ordinal level Spearman rho; and at the interval or ratio levels Pearson r. Values obtained will be displayed in a correlation matrix.

Multicollinearity Test:
A correlation matrix of independent predictor variables will be shown in a matrix, and when a pair of independents exceeds Pearson r of .40, one may be eliminated in the regression analysis as a solution to reduce the chance of multicollinearity (which threatens to inflate the regression R and invalidate the results). Note: Only one of several measures of multicollinearity.

Regression:
Multiple regression will be used to assess the combined interrelationships specified in the conceptual framework, and to identify optimum subsets of variables that effect the dependent outcomes.

Tom Dunne

I'll take a slightly different approach and list some books that I have found useful during my research:

Cohen and Manion, 2011: Research methods in education

• Publisher: Routledge; 7 edition (15 Mar. 2011)
• Language: English
• ISBN-10: 0415583365
• ISBN-13: 978-0415583367

Bryman, 2012: Social Research methods

• Publisher: OUP Oxford; 4 edition (19 Jan. 2012)
• Language: English
• ISBN-10: 0199588058
• ISBN-13: 978-0199588053

Maxwell, 2012: Qualitative Research Design

• Publisher: SAGE Publications, Inc; 3rd Revised edition edition (9 Aug. 2012)
• Language: English
• ISBN-10: 1412981190
• ISBN-13: 978-1412981194

Miles and Huberman, Saldana, 2013: Qualitative Data Analysis

• Publisher: SAGE Publications, Inc; 3rd Revised edition edition (11 Jun. 2013)
• Language: English
• ISBN-10: 1452257876
• ISBN-13: 978-1452257877

Black Swan

Surveys are taken so often in the modern world that saturation occurs, resulting in falling survey response rates. It's sometimes difficult to find a credible source that will suggest a minimum response threshold before analysis can proceed. Earl Babbie (2012) in The Practice of Social Research, 13th edition, suggests that at 50% response rate was sufficient.