class 8 log

--------------------------------------------------------------------------------------------

log: C:\AAA Miker Files\newer web pages\soc_388_notes\soc_388_2003\class 8.log

log type: text

opened on: 22 Oct 2003, 11:14:20

. use "C:\AAA Miker Files\newer web pages\soc_388_notes\70-80-90 MR updated cl7.dta", clear

. keep meth feth mgen fgen year count

. describe

Contains data from C:\AAA Miker Files\newer web pages\soc_388_notes\70-80-90 MR updated cl

> 7.dta

obs: 225

vars: 6 17 Oct 2001 12:19

size: 6,300 (99.9% of memory free)

-------------------------------------------------------------------------------

storage display value

variable name type format label variable label

-------------------------------------------------------------------------------

meth str8 %9s

feth str9 %9s

mgen byte %8.0g

fgen byte %8.0g

year byte %8.0g

count long %12.0g Frequency

-------------------------------------------------------------------------------

Sorted by: year mgen fgen

Note: dataset has changed since last saved

. tabulate meth

meth | Freq. Percent Cum.

------------+-----------------------------------

Blk_NH | 45 20.00 20.00

Mex_Am | 45 20.00 40.00

Oth_H | 45 20.00 60.00

Oth_NH | 45 20.00 80.00

Wht_NH | 45 20.00 100.00

------------+-----------------------------------

Total | 225 100.00

. desmat: poisson count meth feth mgen fgen year

------------------------------------------------------------------------------------------

Poisson regression

------------------------------------------------------------------------------------------

Dependent variable count

Optimization: ml

Number of observations: 225

Initial log likelihood: -2252613.647

Log likelihood: -253644.595

LR chi square: 3997938.105

Model degrees of freedom: 12

Pseudo R-squared: 0.887

Prob: 0.000

------------------------------------------------------------------------------------------

nr Effect Coeff s.e.

------------------------------------------------------------------------------------------

count

meth

1 Mex_Am -0.591** 0.008

2 Oth_H -1.370** 0.010

3 Oth_NH -1.730** 0.012

4 Wht_NH 2.505** 0.005

feth

5 Mex_Am -0.571** 0.008

6 Oth_H -1.330** 0.010

7 Oth_NH -1.464** 0.011

8 Wht_NH 2.552** 0.005

mgen

9 2 3.428** 0.007

fgen

10 2 3.462** 0.007

year

11 80 1.680** 0.004

12 90 1.294** 0.004

13 _cons -1.226** 0.013

------------------------------------------------------------------------------------------

* p < .05

** p < .01

. poisgof

Goodness-of-fit chi2 = 505957.2

Prob > chi2(212) = 0.0000

. *This is the "everything is independent from everything model"

* Note the number of degrees of freedom as (5-1)+(5-1) +(2-1)+(2-1)+(3-1) +1

. predict P_1

(option n assumed; predicted number of events)

. table meth, contents (sum count)

----------------------

meth | sum(count)

----------+-----------

Blk_NH | 45681

Mex_Am | 25294

Oth_H | 11609

Oth_NH | 8100

Wht_NH | 559137

----------------------

. table meth, contents (sum P_1)

----------------------

meth | sum(P_1)

----------+-----------

Blk_NH | 45681

Mex_Am | 25294

Oth_H | 11609

Oth_NH | 8100

Wht_NH | 559137

----------------------

. table meth year, contents (sum count) row col

------------------------------------------

| year

meth | 70 80 90 Total

----------+-------------------------------

Blk_NH | 4759 26972 13950 45681

Mex_Am | 1239 13236 10819 25294

Oth_H | 1445 5459 4705 11609

Oth_NH | 463 4097 3540 8100

Wht_NH | 56997 298483 203657 559137

Total | 64903 348247 236671 649821

------------------------------------------

. table meth year, contents (sum P_1) row col

--------------------------------------------------

| year

meth | 70 80 90 Total

----------+---------------------------------------

Blk_NH | 4562.539 24481.01 16637.46 45681

Mex_Am | 2526.321 13555.36 9212.316 25294

Oth_H | 1159.487 6221.405 4228.108 11609

Oth_NH | 809.014 4340.889 2950.097 8100

Wht_NH | 55845.64 299648.3 203643 559137

Total | 64903 348247 236671 649821

--------------------------------------------------

. desmat: poisson count meth*feth

------------------------------------------------------------------------------------------

Poisson regression

------------------------------------------------------------------------------------------

Dependent variable count

Optimization: ml

Number of observations: 225

Initial log likelihood: -2252613.647

Log likelihood: -676246.840

LR chi square: 3152733.615

Model degrees of freedom: 24

Pseudo R-squared: 0.700

Prob: 0.000

------------------------------------------------------------------------------------------

nr Effect Coeff s.e.

------------------------------------------------------------------------------------------

count

meth

1 Mex_Am -6.114** 0.103

2 Oth_H -4.921** 0.057

3 Oth_NH -6.043** 0.100

4 Wht_NH -4.236** 0.041

feth

5 Mex_Am -4.984** 0.059

6 Oth_H -4.637** 0.050

7 Oth_NH -4.684** 0.051

8 Wht_NH -3.025** 0.023

meth.feth

9 Mex_Am.Mex_Am 10.244** 0.119

10 Mex_Am.Oth_H 6.510** 0.121

11 Mex_Am.Oth_NH 6.211** 0.125

12 Mex_Am.Wht_NH 7.193** 0.106

13 Oth_H.Mex_Am 5.698** 0.091

14 Oth_H.Oth_H 7.583** 0.076

15 Oth_H.Oth_NH 4.500** 0.098

16 Oth_H.Wht_NH 5.702** 0.063

17 Oth_NH.Mex_Am 6.128** 0.129

18 Oth_NH.Oth_H 5.388** 0.130

19 Oth_NH.Oth_NH 8.232** 0.113

20 Oth_NH.Wht_NH 6.694** 0.103

21 Wht_NH.Mex_Am 7.145** 0.073

22 Wht_NH.Oth_H 6.605** 0.066

23 Wht_NH.Oth_NH 6.876** 0.066

24 Wht_NH.Wht_NH 9.809** 0.046

25 _cons 8.461** 0.005

------------------------------------------------------------------------------------------

* p < .05

** p < .01

. poisgof

Goodness-of-fit chi2 = 1351162

Prob > chi2(200) = 0.0000

*Note in the above model the number of degrees of freedom is 5*5=25. For saturated models,

you just need to know what sized table it fits exactly, and that's how many cells it has.

. desmat meth*feth

Desmat generated the following design matrix:

nr Variables Term Parameterization

First Last

1 _x_1 _x_4 meth ind(1)

2 _x_5 _x_8 feth ind(1)

3 _x_9 _x_24 meth.feth ind(1).ind(1)

. gen ethintdum=0

. replace ethnintdum=1 if meth==feth

variable ethnintdum not found

r(111);

. replace ethintdum=1 if meth==feth

(45 real changes made)

. desmat: poisson count meth*mgen*year feth*fgen*year ethintdum

------------------------------------------------------------------------------------------

Poisson regression

------------------------------------------------------------------------------------------

Dependent variable count

Optimization: ml

Number of observations: 225

Initial log likelihood: -2252613.647

Log likelihood: -14085.822

LR chi square: 4477055.651

Model degrees of freedom: 57

Pseudo R-squared: 0.994

Prob: 0.000

------------------------------------------------------------------------------------------

nr Effect Coeff s.e.

------------------------------------------------------------------------------------------

count

meth

1 Mex_Am 2.065** 0.203

2 Oth_H 1.780** 0.206

3 Oth_NH 1.173** 0.242

4 Wht_NH 2.070** 0.184

mgen

5 2 5.017** 0.180

meth.mgen

6 Mex_Am.2 -2.942** 0.203

7 Oth_H.2 -2.561** 0.206

8 Oth_NH.2 -2.815** 0.245

9 Wht_NH.2 -0.787** 0.184

year

10 80 4.180** 0.330

11 90 4.502** 0.332

meth.year

12 Mex_Am.80 -0.153 0.210

13 Mex_Am.90 -0.212 0.210

14 Oth_H.80 0.304 0.213

15 Oth_H.90 0.066 0.214

16 Oth_NH.80 -0.172 0.250

17 Oth_NH.90 -0.296 0.251

18 Wht_NH.80 -0.833** 0.191

19 Wht_NH.90 -1.138** 0.192

mgen.year

20 2.80 -0.946** 0.186

21 2.90 -1.514** 0.187

meth.mgen.year

22 Mex_Am.2.80 0.430* 0.210

23 Mex_Am.2.90 0.625** 0.211

24 Oth_H.2.80 -0.900** 0.214

25 Oth_H.2.90 -0.431* 0.215

26 Oth_NH.2.80 0.235 0.254

27 Oth_NH.2.90 0.619* 0.254

28 Wht_NH.2.80 0.719** 0.190

29 Wht_NH.2.90 1.184** 0.191

feth

30 Mex_Am 2.788** 0.284

31 Oth_H 2.470** 0.288

32 Oth_NH 3.204** 0.284

33 Wht_NH 3.243** 0.270

fgen

34 2 5.802** 0.268

feth.fgen

35 Mex_Am.2 -3.691** 0.285

36 Oth_H.2 -3.195** 0.288

37 Oth_NH.2 -4.590** 0.287

38 Wht_NH.2 -1.902** 0.269

feth.year

39 Mex_Am.80 -1.089** 0.291

40 Mex_Am.90 -0.940** 0.292

41 Oth_H.80 -0.250 0.295

42 Oth_H.90 -0.182 0.296

43 Oth_NH.80 -0.672* 0.291

44 Oth_NH.90 -0.770** 0.292

45 Wht_NH.80 -1.539** 0.276

46 Wht_NH.90 -1.801** 0.277

fgen.year

47 2.80 -1.541** 0.273

48 2.90 -2.023** 0.274

feth.fgen.year

49 Mex_Am.2.80 1.530** 0.291

50 Mex_Am.2.90 1.717** 0.292

51 Oth_H.2.80 -0.158 0.295

52 Oth_H.2.90 0.120 0.297

53 Oth_NH.2.80 1.057** 0.294

54 Oth_NH.2.90 1.473** 0.295

55 Wht_NH.2.80 1.611** 0.275

56 Wht_NH.2.90 2.055** 0.277

ethintdum

57 1 3.218** 0.006

58 _cons -5.767** 0.322

------------------------------------------------------------------------------------------

* p < .05

** p < .01

. poisgof

Goodness-of-fit chi2 = 26839.64

Prob > chi2(167) = 0.0000

. desmat: poisson count meth*mgen*year feth*fgen*year ethintdum*year

------------------------------------------------------------------------------------------

Poisson regression

------------------------------------------------------------------------------------------

Dependent variable count

Optimization: ml

Number of observations: 225

Initial log likelihood: -2252613.647

Log likelihood: -13301.451

LR chi square: 4478624.393

Model degrees of freedom: 59

Pseudo R-squared: 0.994

Prob: 0.000

------------------------------------------------------------------------------------------

nr Effect Coeff s.e.

------------------------------------------------------------------------------------------

count

meth

1 Mex_Am 2.135** 0.206

2 Oth_H 1.810** 0.209

3 Oth_NH 1.191** 0.245

4 Wht_NH 2.009** 0.186

mgen

5 2 5.020** 0.180

meth.mgen

6 Mex_Am.2 -2.958** 0.203

7 Oth_H.2 -2.570** 0.206

8 Oth_NH.2 -2.866** 0.245

9 Wht_NH.2 -0.790** 0.184

year

10 80 4.682** 0.331

11 90 5.259** 0.333

meth.year

12 Mex_Am.80 -0.221 0.212

13 Mex_Am.90 -0.273 0.213

14 Oth_H.80 0.287 0.216

15 Oth_H.90 0.017 0.217

16 Oth_NH.80 -0.195 0.253

17 Oth_NH.90 -0.304 0.254

18 Wht_NH.80 -0.787** 0.193

19 Wht_NH.90 -1.032** 0.194

mgen.year

20 2.80 -0.948** 0.186

21 2.90 -1.518** 0.187

meth.mgen.year

22 Mex_Am.2.80 0.444* 0.210

23 Mex_Am.2.90 0.648** 0.211

24 Oth_H.2.80 -0.901** 0.214

25 Oth_H.2.90 -0.393 0.215

26 Oth_NH.2.80 0.275 0.254

27 Oth_NH.2.90 0.701** 0.255

28 Wht_NH.2.80 0.721** 0.190

29 Wht_NH.2.90 1.187** 0.191

feth

30 Mex_Am 2.846** 0.286

31 Oth_H 2.541** 0.290

32 Oth_NH 3.451** 0.287

33 Wht_NH 3.234** 0.271

fgen

34 2 5.804** 0.268

feth.fgen

35 Mex_Am.2 -3.708** 0.285

36 Oth_H.2 -3.205** 0.288

37 Oth_NH.2 -4.608** 0.287

38 Wht_NH.2 -1.904** 0.269

feth.year

39 Mex_Am.80 -1.137** 0.293

40 Mex_Am.90 -1.034** 0.294

41 Oth_H.80 -0.294 0.297

42 Oth_H.90 -0.326 0.298

43 Oth_NH.80 -0.877** 0.293

44 Oth_NH.90 -1.115** 0.295

45 Wht_NH.80 -1.526** 0.277

46 Wht_NH.90 -1.793** 0.278

fgen.year

47 2.80 -1.542** 0.273

48 2.90 -2.028** 0.274

feth.fgen.year

49 Mex_Am.2.80 1.543** 0.291

50 Mex_Am.2.90 1.747** 0.292

51 Oth_H.2.80 -0.160 0.295

52 Oth_H.2.90 0.164 0.297

53 Oth_NH.2.80 1.070** 0.294

54 Oth_NH.2.90 1.506** 0.295

55 Wht_NH.2.80 1.612** 0.275

56 Wht_NH.2.90 2.057** 0.277

ethintdum

57 1 3.874** 0.025

ethintdum.year

58 1.80 -0.568** 0.026

59 1.90 -0.884** 0.026

60 _cons -6.344** 0.323

------------------------------------------------------------------------------------------

* p < .05

** p < .01

. poisgof

Goodness-of-fit chi2 = 25270.9

Prob > chi2(165) = 0.0000

. *

. *That previous model treats year as a nominal categorical variable everywhere

. *Now let's treat year as a continuous variable for the purposes of interaction with ethnic intermarriage

. desmat: poisson count meth*mgen*year feth*fgen*year ethintdum*@year

------------------------------------------------------------------------------------------

Poisson regression

------------------------------------------------------------------------------------------

Dependent variable count

Optimization: ml

Number of observations: 225

Initial log likelihood: -2252613.647

Log likelihood: -13334.989

LR chi square: 4478557.317

Model degrees of freedom: 58

Pseudo R-squared: 0.994

Prob: 0.000

------------------------------------------------------------------------------------------

nr Effect Coeff s.e.

------------------------------------------------------------------------------------------

count

meth

1 Mex_Am 2.121** 0.205

2 Oth_H 1.807** 0.208

3 Oth_NH 1.201** 0.244

4 Wht_NH 2.023** 0.185

mgen

5 2 5.019** 0.180

meth.mgen

6 Mex_Am.2 -2.955** 0.203

7 Oth_H.2 -2.568** 0.206

8 Oth_NH.2 -2.855** 0.245

9 Wht_NH.2 -0.789** 0.184

year

10 80 1.942** 0.181

meth.year

11 Mex_Am.80 -0.206 0.212

12 Mex_Am.90 -0.258 0.212

13 Oth_H.80 0.295 0.215

14 Oth_H.90 0.018 0.216

15 Oth_NH.80 -0.208 0.252

16 Oth_NH.90 -0.314 0.253

17 Wht_NH.80 -0.808** 0.192

18 Wht_NH.90 -1.042** 0.193

mgen.year

19 2.80 -0.948** 0.186

20 2.90 -1.518** 0.187

meth.mgen.year

21 Mex_Am.2.80 0.440* 0.210

22 Mex_Am.2.90 0.645** 0.211

23 Oth_H.2.80 -0.907** 0.214

24 Oth_H.2.90 -0.392 0.215

25 Oth_NH.2.80 0.259 0.254

26 Oth_NH.2.90 0.693** 0.255

27 Wht_NH.2.80 0.720** 0.190

28 Wht_NH.2.90 1.187** 0.191

feth

29 Mex_Am 2.836** 0.286

30 Oth_H 2.526** 0.290

31 Oth_NH 3.389** 0.286

32 Wht_NH 3.232** 0.271

fgen

33 2 5.804** 0.268

feth.fgen

34 Mex_Am.2 -3.704** 0.285

35 Oth_H.2 -3.203** 0.288

36 Oth_NH.2 -4.604** 0.287

37 Wht_NH.2 -1.903** 0.269

feth.year

38 Mex_Am.80 -1.123** 0.292

39 Mex_Am.90 -1.027** 0.293

40 Oth_H.80 -0.267 0.296

41 Oth_H.90 -0.317 0.297

42 Oth_NH.80 -0.798** 0.292

43 Oth_NH.90 -1.062** 0.294

44 Wht_NH.80 -1.523** 0.276

45 Wht_NH.90 -1.792** 0.278

fgen.year

46 2.80 -1.541** 0.273

47 2.90 -2.027** 0.274

feth.fgen.year

48 Mex_Am.2.80 1.538** 0.291

49 Mex_Am.2.90 1.745** 0.292

50 Oth_H.2.80 -0.167 0.295

51 Oth_H.2.90 0.165 0.297

52 Oth_NH.2.80 1.064** 0.294

53 Oth_NH.2.90 1.504** 0.295

54 Wht_NH.2.80 1.611** 0.275

55 Wht_NH.2.90 2.057** 0.277

ethintdum

56 1 6.337** 0.082

57 year 0.257** 0.017

ethintdum.year

58 1.year -0.037** 0.001

59 _cons -24.166** 1.478

------------------------------------------------------------------------------------------

* p < .05

** p < .01

. *That's the version that uses a continuous variable year to interract with ethnic intermarriage.

. poisgof

Goodness-of-fit chi2 = 25337.98

Prob > chi2(166) = 0.0000

. *the change over time per year, -0.037 is roughly equivalent to 1/10 of the 1980 change

. *or 1/20 of the 1990 change from 1970 that we got in the previous model.

. *The baseline value of ethnintdum is larger because it's figured at year=0, in this case

> 1900

. *In other words, the model extrapolates back to 1900 and gives a very high ethnic endoga

> my for that year

. desmat: poisson meth*mgen

, invalid name

r(198);

. desmat: poisson count meth*mgen

------------------------------------------------------------------------------------------

Poisson regression

------------------------------------------------------------------------------------------

Dependent variable count

Optimization: ml

Number of observations: 225

Initial log likelihood: -2252613.647

Log likelihood: -1375789.526

LR chi square: 1753648.242

Model degrees of freedom: 9

Pseudo R-squared: 0.389

Prob: 0.000

------------------------------------------------------------------------------------------

nr Effect Coeff s.e.

------------------------------------------------------------------------------------------

count

meth

1 Mex_Am 1.654** 0.037

2 Oth_H 1.316** 0.038

3 Oth_NH 0.354** 0.044

4 Wht_NH 2.448** 0.035

mgen

5 2 3.256** 0.034

meth.mgen

6 Mex_Am.2 -2.423** 0.038

7 Oth_H.2 -2.992** 0.040

8 Oth_NH.2 -2.230** 0.046

9 Wht_NH.2 0.058 0.036

10 _cons 4.054** 0.034

------------------------------------------------------------------------------------------

* p < .05

** p < .01

. *10 terms for a saturated meth*mgen, which is just 5*2

. desmat: poisson count meth*mgen feth*fgen

------------------------------------------------------------------------------------------

Poisson regression

------------------------------------------------------------------------------------------

Dependent variable count

Optimization: ml

Number of observations: 225

Initial log likelihood: -2252613.647

Log likelihood: -338721.041

LR chi square: 3827785.214

Model degrees of freedom: 18

Pseudo R-squared: 0.850

Prob: 0.000

------------------------------------------------------------------------------------------

nr Effect Coeff s.e.

------------------------------------------------------------------------------------------

count

meth

1 Mex_Am 1.654** 0.037

2 Oth_H 1.316** 0.038

3 Oth_NH 0.354** 0.044

4 Wht_NH 2.448** 0.035

mgen

5 2 3.918** 0.034

meth.mgen

6 Mex_Am.2 -2.423** 0.038

7 Oth_H.2 -2.992** 0.040

8 Oth_NH.2 -2.230** 0.046

9 Wht_NH.2 0.058 0.036

feth

10 Mex_Am 1.386** 0.044

11 Oth_H 1.393** 0.044

12 Oth_NH 1.530** 0.043

13 Wht_NH 2.777** 0.041

fgen

14 2 4.168** 0.040

feth.fgen

15 Mex_Am.2 -2.052** 0.045

16 Oth_H.2 -2.963** 0.046

17 Oth_NH.2 -3.329** 0.045

18 Wht_NH.2 -0.229** 0.041

19 _cons -1.190** 0.052

------------------------------------------------------------------------------------------

* p < .05

** p < .01

. *19 terms here because the second 5*2 table has one redundant piece of information, the

> total count

. log close

log: C:\AAA Miker Files\newer web pages\soc_388_notes\soc_388_2003\class 8.log

log type: text

closed on: 22 Oct 2003, 12:22:36

------------------------------------------------------------------------------------------