. 2021 Jan 18;19(1):63–90. doi: 10.1007/s11150-020-09540-9

Table 2.

Predictors of parent hours helping children with schoolwork

	Probability that parent has positive hours		Number of parent hours
	Coeff.	s.e.	Coeff.	s.e.
Household and respondent characteristics
Household type (married couple omitted)
Single female	−0.006	0.007	−0.701^b	0.348
Single male	−0.030^a	0.008	−0.999	0.626
Number of children <18	0.017^a	0.002	1.251^a	0.099
R age	−0.003^a	0.000	−0.097^a	0.011
R is working female	−0.039^a	0.004	−1.489^a	0.178
Respondent education (<HS degree omitted)
HS degree or GED	0.011	0.015	1.071^b	0.448
Some college or AA degree	0.034^a	0.012	1.830^a	0.465
4-year college	0.057^a	0.012	2.243^a	0.427
R race/ethnicity (white non-Hispanic omitted)
Black non-Hispanic	0.061^a	0.011	1.747^a	0.322
Hispanic	0.018^a	0.007	−0.130	0.362
Asian	−0.001	0.008	−0.644^c	0.340
Other	0.028^b	0.014	1.021^a	0.393
Covid-19 Shocks
Lost employment income	0.003	0.007	0.714^a	0.208
Food insufficiency	0.038^a	0.005	1.982^a	0.199
Rent or mortgage late last month	0.011	0.008	0.167	0.277
Not at work due to Covid-19 illness	−0.055^a	0.020	0.600	1.343
R-reported school delivery mode(s), live teacher hours, and adequacy of HH technology access (canceled entirely is omitted)
Online only with:
No teacher hours, some tech access problems	0.014	0.016	0.349	0.429
No teacher hours, no tech access problems	−0.051^a	0.013	−1.869^a	0.368
Live teacher hours, some tech access problems	0.344^a	0.013	8.678^a	0.546
Live teacher hours, no tech access problems	0.316^a	0.013	8.294^a	0.446
Online & paper and other modes with:
No teacher hours, some tech access problems	0.025	0.017	0.555	0.592
No teacher hours, no tech access problems	0.043^a	0.016	1.083^a	0.397
Live teacher hours, some tech access problems	0.321^a	0.014	7.265^a	0.566
Live teacher hours, no tech access problems	0.351^a	0.013	8.716^a	0.375
State and week fixed effects	Yes		Yes
Adjusted R2 or pseudo-R2	0.279		0.038
Number of observations	199,014		199,014

All regressions are run using sample weights. Standard errors are clustered at the state level.

^aStatistically significant at 1% level

^bStatistically significant at 5% level

^cStatistically significant at 10% level