Table 2.
Mean differences in marital satisfaction across different meeting venues
| Source | Weighted n | Unadjusted mean marital satisfaction scores | Coefficients from regression with covariates (SE) |
| On-line sources | |||
| Instant messaging | 279 | 5.66a | −0.04 (0.08) |
| 133 | 5.67ab | −0.02 (0.12) | |
| Chat room | 596 | 5.42bde | −0.25 (0.08) |
| Discussion group | 113 | 5.57ade | −0.12 (0.13) |
| Social network | 1,301 | 5.72a | 0.02 (0.05) |
| Virtual world | 125 | 5.65ab | −0.03 (0.11) |
| Multiplayer game | 222 | 5.72a | 0.05 (0.09) |
| On-line community | 393 | 5.29e | −0.37 (0.08) |
| Message on blog | 102 | 5.59ab | −0.07 (0.13) |
| Other (on-line) | 158 | 5.55d | 0.12 (0.12) |
| On-line dating | 2,782 | 5.69a | — |
| On-line dating sites | |||
| eHarmony | 714 | 5.86a | 0.34 (0.09) |
| Match | 663 | 5.70c | 0.15 (0.09) |
| Yahoo | 201 | 5.29d | −0.23 (0.15) |
| Plenty of Fish | 151 | 5.65abc | 0.07 (0.14) |
| Small sites | 691 | 5.71abc | 0.17 (0.11) |
| Other (on-line dating) | 361 | 5.52bd | — |
| Off-line sources | |||
| Work | 2,474 | 5.38de | −0.04 (0.06) |
| Friends | 2,135 | 5.47bc | 0.03 (0.06) |
| School | 1,277 | 5.59a | 0.12 (0.07) |
| Family | 769 | 5.43bcd | 0.01 (0.08) |
| Bar/club | 988 | 5.39cd | −0.03 (0.07) |
| Place of worship | 466 | 5.58ab | 0.10 (0.08) |
| Social gathering | 1,133 | 5.56ab | 0.12 (0.07) |
| Grew up together | 873 | 5.67a | 0.21 (0.07) |
| Blind date | 299 | 5.31ce | −0.15 (0.12) |
| Other (off-line) | 944 | 5.42cd | — |
Weighted cell size is listed in the second columns. Post hoc analyses are expressed in superscripts in the third column and were done using least-significant differences tests. Means under “On-line sources,” “On-line dating sources,” or “Off-line sources” that do not share a superscript differ at P < 0.05. The sample sizes differ across various pairwise comparisons, and the effect size required for statistical significance differs accordingly. In some cases, a given mean difference in a pairwise comparison based on a relatively large sample size (e.g., eHarmony vs. Match) reaches statistical significance even though a nominally larger mean difference in a pairwise comparison involving fewer observations (e.g., eHarmony vs. Plenty of Fish) does not reach statistical significance. The final column is regression coefficient effect estimates adjusting for year of marriage, sex, age, ethnicity, income, education, religion, and employment. Survey weights can bias estimates of SDs, so we report SEs in accordance with standard statistical practice for survey weighted data.