Sample Size and Statistical Analysis

The sample, drawn from the 2004 General Social Survey databases, for the final regression analysis was a total of 10,818 respondents.  All of the respondents were women.  The large sample size permitted for testing for statistically significant differences, although, as noted above, the differences for the independent variable of working for pay outside the home (or not) were not significant, so they were excluded from the final analysis. 
The low R2 values obtained show significant variance.  This means that the differences, it can be said with great confidence, are real and not merely the results of random variation (as might be the case with a much smaller sample size, in which sampling error would probably be greater). 
A four-point scale was used to measure respondents’ reactions to the statements:
1. Agree strongly
2. Agree somewhat
3. Disagree somewhat
4. Disagree strongly

Thus a mean of 2.776 for the total sample would indicate that overall, on average, respondents were more likely to disagree with the stated opinion than to agree with it.