Table 1
The data set compared with the Gemius estimated total population of Facebook users in Denmark
| Digital Footprints data set (August 2014) | Gemius Facebook user data set (August 2014) | Facebook users in DK (Gemius, August 2014) | Danish population (August 2014) | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Age 15–29 | 239 | 23.9 | 616 | 15.5 | 926,384 | 29.1 | −5.2 | 1,042,898 | 24.7 |
| Age 30–44 | 191 | 19.1 | 835 | 21.0 | 797,608 | 25.1 | −6 | 1,013,786 | 24.0 |
| Age 45–59 | 282 | 28.2 | 1,428 | 35.8 | 827,729 | 26.0 | 2.2 | 1,153,027 | 27.3 |
| Age 60+ | 201 | 20.1 | 1,108 | 27.8 | 628,619 | 19.8 | 0,3 | 1,010,763 | 23.9 |
| Female | 518 | 51.8 | 1,858 | 46.6 | 1,604,057 | 50.4 | 1.4 | 2,052,919 | 48.6 |
| Male | 482 | 48.2 | 2,129 | 53.4 | 1,576,282 | 49.6 | −1.4 | 2,167,555 | 51.4 |
| Short education | 88 | 8.8 | 2,172 | 54.5 | 1,793,297 | 56.4 | −47.6 | 2,282,398 | 54.1 |
| High school, colleges and equivalent education | 381 | 38.1 | 1,157 | 29.0 | 884,085 | 27.8 | 10.3 | 1,202,341 | 28.5 |
| Higher education | 531 | 53.1 | 658 | 16.5 | 502,958 | 15.8 | 37.3 | 735,735 | 17.4 |
| Rural | 543 | 54.3 | 3,022 | 75.8 | 232,1514 | 73.0 | −18.7 | 3,146,063 | 74.5 |
| Urban | 457 | 45.7 | 965 | 24.2 | 858,826 | 27.0 | 18.7 | 1,074,411 | 25.5 |
| Total | 1,000 | 100 | 3,984 | 100 | 3,180,340 | 100 | - | 4,220,474 | 100 |
Figure 1
Monthly profile posts before normalization
Table 2
Poisson regression of monthly wallposts over time (2007–2014) by gender, age, education and area of residence. Robust standard errors and exponentiated coefficients (incidence rate ratios) are used
| Model 1 Coef./Std err. | Model 2 Coef./Std err. | Model 3 Coef./Std err. | Model 4 Coef./Std err. | Model 5 Coef./Std err. | |
|---|---|---|---|---|---|
| Date (monthly) | 1.006*** (0.001) | 1.010*** (0.002) | 1.003* (0.001) | 1.000 (0.005) | 1.007*** (0.001) |
| Gender (1=female) | 1.471*** (0.159) | 92.883*** (121.536) | 1.447*** (0.153) | 1.463*** (0.157) | 1.445*** (0.156) |
| 15–29 | 1.000 (.) | 1.000 (.) | 1.000 (.) | 1.000 (.) | 1.000 (.) |
| 30–44 | 0.957 (0.126) | 0.959 (0.119) | 0.105 (0.161) | 0.947 (0.129) | 0.960 (0.123) |
| 45–59 | 1.052 (0.195) | 1.023 (0.177) | 0.095 (0.199) | 1.028 (0.196) | 1.020 (0.182) |
| 60+ | 1.239 (0.388) | 1.100 (0.334) | 0.001*** (0.002) | 1.231 (0.381) | 1.155 (0.350) |
| Short | 1.000 (.) | 1.000 (.) | 1.000 (.) | 1.000 (.) | 1.000 (.) |
| Medium | 1.469* (0.229) | 1.496* (0.232) | 1.459* (0.229) | 0.058 (0.201) | 1.481* (0.232) |
| Long | 1.547** (0.247) | 1.563** (0.246) | 1.513** (0.243) | 0.026 (0.087) | 1.553** (0.248) |
| Urban (1=urban) | 1.166 (0.116) | 1.152 (0.113) | 1.165 (0.115) | 1.160 (0.116) | 10.214 (13.109) |
| Gender: | |||||
| Male × Date | 1.000 (.) | ||||
| Female × Date | 0.993** (0.002) | ||||
| Age: | |||||
| 15–29 × Date | 1.000 (.) | ||||
| 30–44 × Date | 1.004 (0.002) | ||||
| 45–59 × Date | 1.004 (0.003) | ||||
| 60+ × Date | 1.011*** (0.003) | ||||
| Education: | |||||
| Short × Date | 1.000 (.) | ||||
| Medium × Date | 1.005 (0.006) | ||||
| Long × Date | 1.007 (0.005) | ||||
| Urbanization: | |||||
| Rural × Date | 1.000 (.) | ||||
| Urban × Date | 0.997 (0.002) | ||||
| Constant | 0.078*** (0.053) | 0.006*** (0.006) | 0.423 (0.403) | 3.070 (10.050) | 0.030*** (0.027) |
| N. of obs. | 48,300 | 48,300 | 48,300 | 48,300 | 48,300 |
Comment: The models are based on random-effects Poisson regression models to accentuate trends within time-invariant predictor variables (e.g. gender). Hausman tests comparing the consistency of random-effects to fixed effects estimates indicate discrepancies in random-effects estimates. Robust standard errors are used to account for overdispersion. N = 922.
Figure 2
Monthly posts per gender (top left), age (top right), education (bottom left) and area of residence (bottom right)
Figure 3
The percentage of posters in age groups
| Model 1 Coef./Std err. | Model 2 Coef./Std err. | Model 3 Coef./Std err. | Model 4 Coef./Std err. | Model 5 Coef./Std err. | |
|---|---|---|---|---|---|
| Date (monthly) | 1.000 (0.003) | 0.996 (0.006) | 0.993 (0.005) | 1.002 (0.013) | 1.002 (0.004) |
| Gender (1=female) | 1.103 (0.120) | 0.017 (0.072) | 1.085 (0.119) | 1.105 (0.120) | 1.105 (0.121) |
| 15–29 | 1.000 (.) | 1.000 (.) | 1.000 (.) | 1.000 (.) | 1.000 (.) |
| 30–44 | 0.840* (0.069) | 0.847* (0.070) | 0.481 (2.411) | 0.842* (0.069) | 0.839* (0.069) |
| 45–59 | 0.846 (0.107) | 0.850 (0.106) | 0.000 (0.002) | 0.085 (0.106) | 0.845 (0.106) |
| 60+ | 0.620*** (0.086) | 0.623*** (0.086) | 0.000* (0.000) | 0.625*** (0.088) | 0.624*** (0.087) |
| Short | 1.000 (.) | 1.000 (.) | 1.000 (.) | 1.000 (.) | 1.000 (.) |
| Medium | 1.666** (0.290) | 1.663** (0.290) | 1.670** (0.294) | 0.344 (3.241) | 1.665** (0.290) |
| Long | 1.817*** (0.309) | 1.813*** (0.307) | 1.795*** (0.307) | 45.064 (407.550) | 1.819*** (0.309) |
| Urban (1=urban) | 0.982 (0.110) | 0.983 (0.110) | 0.987 (0.113) | 0.982 (0.110) | 32.477 (147.544) |
| Gender: | |||||
| Male × Date | 1.000 (.) | ||||
| Female × Date | 1.007 (0.007) | ||||
| Age: | |||||
| 15–29 × Date | 1.000 (.) | ||||
| 30–44 × Date | 1.001 (0.008) | ||||
| 45–59 × Date | 1.012 (0.008) | ||||
| 60+ × Date | 1.031* (0.013) | ||||
| Education: | |||||
| Short × Date | 1.000 (.) | ||||
| Medium × Date | 1.002 (0.015) | ||||
| Long × Date | 0.995 (0.014) | ||||
| Urbanization: | |||||
| Rural × Date | 1.000 (.) | ||||
| Urban × Date | 0.995 (0.007) | ||||
| Constant | 5.050 (10.650) | 57.825 (211.587) | 489.468 (1689.259) | 1.328 (11.516) | 1.382 (3.296) |
| N. of obs. | 12,276 | 12,276 | 12,276 | 12,276 | 12,276 |
Comment: The models are based on random-effects Poisson regression models to accentuate trends within time-invariant predictor variables (e.g. gender). Hausman tests comparing the consistency of random effects to fixed effects estimates indicate discrepancies in random effects estimates. Robust standard errors are used to account for overdispersion. N = 922.