Marketing Research的课件- Preliminary_Data_Analysis_s.ppt

Preliminary Data Analysis and Data Preparation Session: January 22; 2010 Steps prior to data entry Data editing: First ‘look’ at data to identify potential problems/correct them, in the field by interviewer and field supervisor Errors, Inconsistencies, Ineligible Respondents, Systematic Missings - early remedy? Coding Coding closed-ended questions involves specifying how the responses are to be entered Open-ended questions are difficult to code Set up Code Book with category labels and values? Depends on type of MVA In this session, we focus on steps after data entry… Preliminary inspection Identification of outliers Missing data Checking assumptions for MVA Graphical inspection and simple analyses 1 variable: frequency table simple statistics: central tendency: mean, median, mode dispersion: variance (stand.dev.), range histogram ‘time series’ plot 2 variables: scatterplot correlation, cross-tab Histogram Histogram: Skewness of distribution? Bivariate analysis Metric versus metric Scatterplot, Pearson correlation Non-metric versus non-metric Cross-tab Spearman correlation (Rho) or Kendall’s Tau Example: Simple Scatterplot Cross-tab (1) Cross-tabs (2): Stockout Reactions per Brand Type Rho and Tau Useful to assess link between two ordinal variables Examples: Education (highest obtained) Swimming certificate (highest obtained) Categorically measured variables (e.g. shopping frequency, income class, age category) Example: Income and Shopping Frequency (Categories) Preliminary Data analysis: Summary Obtain preliminary insights using univariate en bivariate analyses First impression concerning missings, outliers and distributional properties ? crucial before using MVA? Outliers Outliers Types ‘good’: true value (probably) ‘bad’: someting is wrong? Causes Procedural error Exceptional circumstances (cause known or unknown) ‘Regular’ levels, yet unique in combination with other variables (bivariate en multivariate outliers) Identifying outliers Univariat