Download Assumption about Research and more Assignments Research Methodology in PDF only on Docsity!
Missing Data : Statistical Problems
- (^) If you are missing much of your data, this can cause several
problems; e.g., can’t calculate the estimated model.
- (^) SEM requires a certain minimum number of data points in order to
compute estimates – each missing data point reduces your valid n by
- (^) Systematic missing data may indicate systematic bias (poor item
formulation, sensitivity, etc.).
- (^) If females are less likely to report gender than males, you will have “male- biased” data.
- (^) e.g., only 50% of the females report their gender, but 95% of the males report their gender.
- What then if you use gender as a moderator (or in some other critical
role)?
Missing Data : Logical Problems
Imputation Methods (Hair, table 2-2)
- (^) Option 1: Use only complete and valid data
- No imputation, just use valid cases or variables
- In SPSS: Exclude Pairwise (variable), Listwise (case)
- (^) Option 2: Use known replacement values
- (^) Match missing value with similar case’s value
- (^) Option 3: Use calculated replacement values
- (^) Use variable mean, median, or mode
- Predicted based on known relationships
Best Method – Prevention!
- (^) Shorter surveys (pre-testing critical!)
- (^) Easy to understand and to answer survey items (pre-testing critical)
- (^) Force completion
- (^) Bribe/motivate (iPad drawing)
- (^) Digital surveys (rather than paper)
- Put DVs at the beginning of the survey.
- (^) Put sensitive items at the end of the survey.
Distribution
7 To check distribution in SPSS: 1.Analyze, 2.Explore, 3.Plots: Histogram with normality plot
Outliers and Influentials
- (^) Outliers can influence your results, pulling the mean away from the
median.
- (^) Outliers also affect distributional assumptions and often reflect false
or mistaken responses
- (^) Two types of outliers:
- (^) outliers for individual variables (univariate)
- (^) Extreme values for a single variable
- (^) outliers for the model (multivariate)
- (^) Extreme (uncommon) values for a correlation
Handling Univariate Outliers
- (^) Univariate outliers should be examined on a case by case basis.
- (^) If the outlier is truly abnormal, and not representative of your
population, then it is okay to remove. But this requires careful
examination of the data points
- (^) e.g., you are studying dogs, but somehow a cat got ahold of your survey
- e.g., someone answered “3” for all 75 questions on the survey
- (^) However, just because a datapoint doesn’t fit comfortably with the
distributions does not nominate that datapoint for removal
- (^) ?Outliers on short ordinal scales (e.g., 5-point Likert)?
Multivariate AssumptionsMultivariate Assumptions:: (^) NormalityNormality (^) LinearityLinearity (^) HomogeneityHomogeneity (^) MulticollinearityMulticollinearity
Tests for Skewness and Kurtosis
- (^) Standard rule:
- (^) Skewness > 1 = positive (right) skewed
- (^) Skewness < -1 = negative (left) skewed
- (^) Skewness between -1 and 1 is fine
- Strict rule:
- Abs(Skewness) > 3*Std. error = Skewed (Hair)
- Same for Kurtosis
- (^) Practical purposes…
- (^) Problems arise outside of (+/-) 2.
- (^) Loose rule >10 Kline (2005)
1.Samples are significantly different (difference of means tests)
- (^) T-test: two samples, same sample at two times
- (^) ANOVA: multiple samples/ multiple times 2.Variables move together (covary) significantly
- (^) Correlation (not causation)
- (^) Regression analysis (implies causation)
Testing for significance
Key question in using statistics for hypothesis testing: Are findings statistically significant?
Confidence in findings
- (^) This means that we have a certain degree of confidence that the findings are not merely chance
- (^) 99% confidence in medical studies
- (^) 95% confidence is the standard in Soc. Sci.
- 90% confidence sometimes OK when exploratory
Using p-value (probability value) that results from statistical tests: p < 0.01 --- 99% confidence that results are significant p < 0.05 --- 95% confidence that results are significant p < 0.10 --- 90% confidence that results are significant Another way to think of it is 95% confident that we will get these results again if we do another test. We expect that 95% of the time, the results will be like this.
Reporting Significance
