*********************************************************************************** * * * Code for "Weapon of the Strong? Participatory Inequality a * and the Internet" June 14, 2010. * * START WITH ORIGINAL PEW DATA * * * ####### User Must Insert a "Get File" statement for the raw pew data. *. GET FILE = 'USER FILE'. ************************************************************************************* * * NOTE: There are some variables in this code that were not used * in the final paper. *. *********************************************************************************** * * Create a new weight. *. compute nweight = weight*(2251/6271). weight by nweight. variable labels nweight 'Weight to 2251'. ********************************************************************************** * * VARIOUS LEVELS OF WEB USE * * WEB-USER *. compute webuser = 0. if (q6a = 1 or q6b =1) webuser = 1. variable labels webuser 'On the Internet'. freq webuser. * * FASTHOMEUSER *. compute fasthome = 0. if (modem ge 2 and modem le 6) fasthome = 1. variable labels fasthome 'Uses Broadband at Home'. freq fasthome. * * FASTHOM3--Three Level Use of Internet *. compute fasthom3 = 0. if (webuser eq 1) fasthom3 = 1. if (fasthome eq 1) fasthom3 = 2. variable labels fasthom3 'Level of Connection'. value labels fasthom3 0 'Not on Internet' 1 'On Web-Not Broadband at Home' 2 'Broadband at Home'. * * WEBFOUR -- Four Level Use of Internet *. compute modem@ = modem. recode modem@ (sysmis = -99). *. compute webfour = 0. if (webuser eq 0) webfour = 0. if (webuser eq 1 and modem@ eq -99 and fasthom3 eq 1) webfour = 1. if (webuser eq 1 and (modem@ eq 1 or modem@ eq 7 or modem@ eq 9) and fasthom3 eq 1) webfour = 2. if (webuser eq 1 and fasthom3 eq 2) webfour = 3. value labels webfour 0 'Not on Internet' 1 'On Internet-Not at Home' 2 'Slow Internet at Home' 3 'Home Broadband'. freq webuser fasthome fasthom3 webfour. * * SOCIAL NETWORKING SITES *. compute socnet = activ87. recode socnet (1=1) (2=1) (else = 0). variable labels socnet 'R Uses Social Networking Site'. execute. ******************************************************************************************************************* * * COMPARABLE WEB AND NON-WEB ACTS * * CREATE WEBACTS AND OFFACTS FROM FIVE DIFFERENT ACTS ON AND OFF LINE *. * For these acts, a value of one means doing it. *. compute conemail = q11b. compute conoff = q11a. compute petweb = q11d. compute petoff = q11c. compute letemail = q11f. compute letoff = q11e. recode conemail conoff petweb petoff letemail letoff (1=1) (else = 0). value labels conemail conoff petweb petoff letemail letoff 1 'Yes' 0 'No'. freq conemail conoff petweb petoff letemail letoff. * * For Money, Q25=1 (internet alone), Q25=2 (off alone), Q25=3 (both). * Hence it is recoded to make 1 represent doing it. *. compute moneyweb = q25. compute moneyoff = q25. recode moneyweb (1=1) (3=1) (else=0). recode moneyoff (2=1) (3=1) (else = 0). value labels moneyweb moneyoff 1 'Yes' 0 'No'. freq moneyweb moneyoff q25. * * Communicating with fellow members of a community or political group. *. * Four ways to communicate with fellow group members on-line: * email (b), instant messaging (f), social network (g), or group's website (h); * GRPWEB is yes to any one of them. *. compute grpweb = 0. if (grp01b = 1 or grp01f = 1 or grp01g = 1 or grp01h = 1) grpweb=1. * * Three ways of communicating to group members off-line: * face to face (a), telephone (d), or letter or other print (c). * GRPOFF is yes to any one of them. *. compute grpoff = 0. if (grp01a = 1 or grp01c = 1 or grp01d = 1) grpoff = 1. value labels grpweb grpoff 1 'Yes' 0 'No'. execute. freq grpweb grpoff. * * Now we have all the pieces for WEBACTS and OFFACTS * * ### This is coded in such a way that missing values are zeros. *. count webacts5 = conemail petweb letemail moneyweb grpweb (1). count offacts5 = conoff petoff letoff moneyoff grpoff (1). compute allacts = webacts5 + offacts5. freq allacts webacts5 offacts5. * * Instead of counts, do the "yes-no" version. *. compute webact5@ = webacts5. compute offact5@ = offacts5. recode webact5@ offact5@ (0=0) (1 thru 5 = 1). * * Eleven Act Allact11 scale. *. compute contact = 0. if (q11a = 1 or q11b =1) contact = 1. *. compute petition = 0. if (q11c = 1 or q11d = 1) petition = 1. *. compute lettered = 0. if (q11e = 1 or q11f = 1) lettered = 1. *. count allact11 = q10a q10b q10c q10d q10e q10f q10h q24a contact petition lettered (1). * * Turn Into "Yes-No" version. *. compute all11@ = allact11. recode all11@ (0=0) (else = 1). execute. * * Political Blogging * -- Other's blog q16a * -- Your blog q16d *. compute polblog = q16a. recode polblog (1=1) (else = 0). if (q16d eq 1) polblog = 1. execute. * * Political Social Networking *. compute polsocnt = 0. if (q17a eq 1 or q17b eq 1 or q17c eq 1 or q17d eq 1) polsocnt = 1. freq q17a q17b q17c q17d polsocnt polblog. * * Contributions *. compute off$ = q27. compute web$ = q26. missing values off$ web$ (8,9). value labels off$ web$ 1 '$50 or Less' 2 '$51 to $100' 3 '$101 to $250' 4 '$251 to $500' 5 '$501 to $1000' 6 '$1001 to $2500' 7 'More than $2500'. execute. ***************************************************************************************************** * * CREATE SOCIO-ECONOMIC VARIABLES *. * Education-- * ### 21 refusals are lost with this. *. missing values educ (9). * * Indicator for When Education Available. *. compute educind = 0. if (educ ge 1 and educ le 8) educind = 1. freq educ educind. * * Recode of Education Into Five Categories *. compute newedu = educ. recode newedu (2=1) (3=2) (4=2) (5=3) (6=4) (7=5). value labels newedu 1 'LT HS Grad' 2 'HS Grad/Tech' 3 'Some Coll' 4 'Coll Grad' 5 'Some Grad'. freq newedu. * * Income * ### 494 (21.9% are lost here) * * Indicator for when Income Available *. missing values inc (). compute incind = 1. if (inc eq 10) incind = 0. freq incind. crosstabs incind by educind. missing values inc (10). execute. * * -- Income Categories *. compute inc6 = inc. recode inc6 (1=1) (2=1) (3=2) (4=2) (5=3) (6=3) (7=4) (8=5) (9=6). value labels inc6 1 '<20K' 2 '20-40K' 3 '40-75K' 4 '75-100K' 5 '100-150K' 6 '>150'. freq inc6. * * Income Quintiles * * This creates one variable * NNEWINC@ - Quintiles (1-5) of income *. set seed = 100000. execute. compute inc@ = inc + RV.Uniform(0,.2). RANK VARIABLES=inc@ (A) /NTILES (5) /PRINT=YES /TIES=MEAN . descriptives ninc@. * * Now we want to create best SES measure. * To do this we use rankings for the cases which * have both INC and EDUC. * * We create rankings for just the subset. Hence we must * create variables for just the subsets. * *. compute educ@ = educ + RV.Uniform(0,.2). compute educ@@ = educ@. if (incind eq 0 or educind eq 0) educ@@ = -999. missing values educ@@ (-999). * * Note that we use variable that already has a "random amount" * added to distinguish income values. *. compute inc@@ = inc@. if (incind eq 0 or educind eq 0) inc@@ = -999. missing values inc@@ (-999). *. RANK VARIABLES=educ@@ inc@@ (A) /RANK /PRINT=YES /TIES=MEAN. EXECUTE. * * Now we get the combined ranking number. *. means reduc@@ rinc@@. compute sesrank = reduc@@ + rinc@@. RANK VARIABLES= sesrank (A) /NTILES (5) /PRINT=YES /TIES=MEAN . freq nsesrank. * * Interest in Presidential Campaign -- I think that ELSE should be None or 1. * ### "Else" amounts to 4 DK and 13 Refused. *. compute presint = q19. recode presint (1=4) (2=3) (3=2) (4=1) (else = sysmis). value labels presint 4 'Quite a Lot' 3 'Some' 2 'Little' 1 'None'. freq presint q19. * * Impact on Community (Efficacy) -- Once again, I think that ELSE should be None or 1. * ### "Else" amounts to 44 DK and 20 Refused. *. compute commeff = q3. recode commeff (1=4) (2=3) (3=2) (4=1) (else = sysmis). value labels commeff 4 'Big Impact' 3 'Moderate' 2 'Small' 1 'No Impact'. freq commeff q3. * * Party Strength--If you immediately name Dem or Rep party, then you have strong strength (3) * If when asked about leaning you mention Dem or Rep party, then you have moderate strength (2). * If you mention independent then you have low Party Strength (1). * Otherwise you are zero (0). [This probably gets it wrong for third party supporters, but * their number is mall and otherwise it seems to be OK] * ### This is coded so that missing or refused are at zero. *. compute partstrn = 0. if (party eq 1 or party eq 2) partstrn = 3. if (partyln eq 1 or partyln eq 2) partstrn = 2. if (party eq 3) partstrn = 1. value labels partstrn 3 'Strong' 2 'Moderate' 1 'Weak' 0 'Nonexistent'. * * Party *. freq party. *. * * Sex * ### No missing values. *. compute male = sex. recode male (1=1) (2=0). * * Suburban-Rural * ### No missing values. *. compute urban = usr1. compute suburban = usr1. compute rural = usr1. recode urban (1=1) (else = 0). recode suburban (2=1) (else = 0). recode rural (3=1) (else = 0). *. * * Race--### This is coded so that DK, Refused, or Missing are not black. *. compute black = race. recode black (2=1) (else = 0). freq black race. * * Latino. LATINO and LATINO# are the same. * ### This is coded so that DK, Refused, or Missing are not hispanic. *. compute latino = hisp. recode latino (1=1) (else = 0). execute. * * Religious Attendance * ### This is coded so that 12 DK and 48 refused are missing. *. compute relatt = attend. recode relatt (6=0) (4=6) (5=6) (3=18) (2=52) (1=85) (else = sysmis). execute. * * Age variable -- Note that I don't go to 98 or 99 because they * are missing values. *. compute age7new = age. recode age7new (18 thru 24 = 1824) (25 thru 30=2530) (31 thru 40 = 3140) (41 thru 50 = 4150) (51 thru 60 = 5160) (61 thru 70 = 6170) (71 thru 96 = 7196). missing values age7new (98, 99). freq age7new. compute age6new = age. recode age6new (18 thru 24 = 1824) (25 thru 34 = 2534) (35 thru 44 = 3544) (45 thru 54 = 4554) (55 thru 64 = 5564) (65 thru 96 = 6596). missing values age6new (98,99). freq age6new. compute age5new = age. recode age5new (18 thru 30 = 1830) (31 thru 40 = 3140) (41 thru 50 = 4150) (51 thru 60 = 5160) (61 thru 96 = 6196). missing values age5new (98, 99). freq age5new. * * Age Dummy Variables *. compute age1825 = 0. compute age2630 = 0. compute age3140 = 0. compute age4150 = 0. compute age5160 = 0. compute age6175 = 0. compute age7696 = 0. if (age ge 18 and age le 25) age1825 = 1. if (age ge 26 and age le 30) age2630 = 1. if (age ge 31 and age le 40) age3140 = 1. if (age ge 41 and age le 50) age4150 = 1. if (age ge 51 and age le 60) age5160 = 1. if (age ge 61 and age le 75) age6175 = 1. if (age ge 76 and age le 96) age7696 = 1. compute ageind = age1825 + age2630 + age3140 + age4150 + age5160 + age6175 + age7696. execute. ********************************************************************************************************************* * * Internet Utilization by SES Quintile -- FIGURE 1A -- page 491 *. weight by nweight. means webuser fasthome by nsesrank. * * Internet Utilization by Age -- FIGURE 1B in paper -- Page 491 *. weight by nweight. means webuser fasthome by age7new. ********************************************************************************************************************** * Political Activity by SES Quintile-- FIGURE 2 in paper --- Page 492 *. weight by nweight. means webact5@ offact5@ all11@ by nsesrank. * * -- This provides the line for "among web users". *. weight by nweight. temp. select if (webuser eq 1). means webact5@ by nsesrank. *********************************************************************************************************************** * Political Activity by Age -- FIGURE 3 in paper -- Page 494 *. weight by nweight. means all11@ offact5@ webact5@ by age7new. * * -- This adds data for activity by age for internet users. *. temp. select if (webuser eq 1). means webact5@ by age7new. ********************************************************************************************************************** * Political Activity by Age and SES Quintile -- * FIGURE 4 in the paper -- Page 494 -- * * NOTE: The paper inadvertently uses the data for the entire * sample (the first line of code below) instead of for just web * users (the next three lines of code) as it says in the text and * on the label of the figure. The results are essentially the same. *. means webact5@ by age5new by nsesrank. temp. select if (webuser eq 1). means webact5@ by age5new by nsesrank. ************************************************************************************************************************* * Percent Making a Campaign Contribution by * Family Income--FIGURE 5A in paper -- page 495. * *. weight by nweight. means moneyoff moneyweb by inc6. * * Percent Making a Campaign Contribution by Age -- * FIGURE 5B in paper -- page 495. *. weight by nweight. means moneyoff moneyweb by age7new. ************************************************************************************************************************* * FIGURE 6 in the paper -- page 497 * * BarCharts of income and age for three groups: all respondents, those who made * donations of $50 or less offline; and those who made * donations of $50 or less over the Internet. * * -- All respondents *. weight by nweight. freq inc6 age7new. * * -- $50 or less on Internet. *. weight by nweight. temp. select if (q26 eq 1). freq inc6 age7new. * * -- $50 or less offline *. temp. select if (q27 eq 1). freq inc6 age7new. ****************************************************************************************************************************** *. FIGURE 7 in paper -- Page 499 * Gotten a request to participate in politics via e-mail or phone. * Note that answer (6) which is "occasionally" is a volunteered response. * The answers here for those contacted monthly or more so (6) omitted. * *. weight by nweight. compute reqemail = q23a. compute reqphone = q23c. * * -- This omits those contacted "occasionally". It is only * for those contacted monthly or more often. *. recode reqemail reqphone (1=1) (2=1) (3=1) (4=1) (else=0). execute. means reqemail reqphone by nsesrank age7new. ***************************************************************************************************************************** * FIGURE 8 in paper -- Page 500 * * Taking Public Positions by SES Quintile -- FIGURE 8 top -- page 500 in paper. *. weight by nweight. means polblog letemail letoff by nsesrank. * * Taking Public Positions by Age -- FIGURE 8 bottom -- page 500 in paper. *. weight by nweight. means polblog letemail letoff by age7new. ***************************************************************************************************************************** * FIGURE 9 -top in the paper--Page 502. * * Percent Politically Engaged on Social Networking Sites by SES Quintile-- *. means polsocnt by nsesrank. temp. select if (socnet eq 1). means polsocnt by nsesrank. * * Percent Politically Engaged On Social Networking Sites by Age * FIGURE 9-bottom in the paper -- Page 502. *. means polsocnt by age7new. temp. select if (socnet eq 1). means polsocnt by age7new. ******************************************************************************************************************* ********************************************************************************************************************************* * NOW CODE FOR IN-TEXT DATA *. * * PAGES 489-490--Webuse by income and education * * Roughly half ofthose in lowest income category (below 20K) are online * Nearly universal for those in highest category (150K or more) * 38% for non high school grads * 95% for those with some graduate education *. weight by nweight. crosstabs webuser by inc6 educ6/cells=counts cols. * * PAGE 490-- Use of High speed internet * 56.5% of all respondents * Three quarters of those online (webuser). * 30% for incomes less than 20K * 88% for income 150K above * 22% for non high school grads * 81% for education beyond college *. * PAGE 490 -- Using Internet * 90% of those 18-24 * 32% of those 71+ * * PAGE 490 -- Using High Speed internet * 70% of those 18-24 * 19% of those 71+. *. weight by nweight. mean fasthome. *. temp. select if (webuser eq 1). mean fasthome. *. crosstabs fasthome by webuser/cells=counts cols. mean fasthome by inc6 educ6. mean webuser fasthome by age7new. * * PAGE 490. * Scales with Number of Acts *. weight by nweight. means allact11 all11@ webacts5 webact5@ offacts5 offact5@. * * PAGE 490 -- Active online and offline *. correlation webacts5 offacts5. crosstabs webact5@ by offact5@/cells=counts cols rows. * * PAGE 494 -- Frequency of Acts * 6% Online * 15% Offline *. weight by nweight. freq moneyweb moneyoff. * * PAGE 494 -- Dollars on and offline [?????} * 74% of political dollars donated offline * 49% of political dollars donatedo n the Internet. temp. recode web$ off$ (1=25) (2=75) (3=175) (4=375) (5=750) (6=1750) (7=3750). weight web$. freq web$. weight off$. freq off$. weight by nweight. freq web$ off$. * * PAGE 494 -- Proportion of Donations $50 and Less and between $51 and $100 * 38% Online for $50 and less * 39% Offline for $50 and less * * 28% Online for $51 to $100 * 29% Offline for $51 to $100 * * 1% Online over $1000 * 5% Offline over $1000 *. weight by nweight. freq web$ off$. * * PAGE 496 -- Political Requests by Phone or E-mail *. freq reqemail reqphone. * * PAGE 501: Social Networkers * 78% of those under 25 are social networkers; 3% for those over 60. *. weight by nweight. mean socnet by age7new. temp. recode age7new (6170=6196) (7196=6196). crosstabs socnet by age7new/cells=counts cols. * * PAGE 503-- * Chracteristics of those Politically Engaged on Social Networking Sites. *. weight by nweight. compute stud@ = -99. if (stud eq 1 or stud eq 2) stud@ = 1. if (stud eq 3) stud@ = 0. missing values stud@ (-99). value labels stud@ 1 'Student'. execute. weight by nweight. temp. select if (polsocnt eq 1). freq stud@. temp. select if (polsocnt eq 1). crosstabs stud@ by age7new/cells=counts total. ************************************************************************************************************** * * Code for F-tests--See Supplementary Material * Start with data from above. *. ******************************************************************************************************************************* * * Creating Stacked Data * * * First create additional indicators for missing data so that we can exclude those cases later. *. compute pintind = 0. if (presint ge 1 and presint le 4) pintind = 1. compute comind = 0. if (commeff ge 1 and commeff le 4) comind = 1. compute relind = 0. if (relatt ge 0 and relatt le 85) relind = 1. *. compute missreg = 0. if ((educind ne 1) or (incind ne 1) or (ageind ne 1) or (pintind ne 1) or (comind ne 1) or (relind ne 1) or (ageind ne 1)) missreg = 1. freq missreg. * * Save a Subset of the Data to a file *. SAVE OUTFILE = 'USER FILE\CutDownPew1.sav' /keep = nweight webacts5 offacts5 educ inc presint commeff partstrn male sex suburban rural relatt relattsq black latino age age1825 age2630 age3140 age4150 age5160 age6175 age7696 educind incind ageind pintind comind relatt ageind missreg weight nweight webfour. * * Get the Saved Subset *. GET FILE = 'USER FILE\CutDownPew1.sav'. execute. * * "Double" the number of cases by stacking: Simply take the original data and put a * copy "beneath" it so that the number of cases is increased. *. ADD FILES /FILE=* /FILE='USER FILE\CutDownPew1.sav'. EXECUTE. * * Compute a Dummy Variable for Those Acts Not on Internet *. compute offgrp = 0. if ($casenum ge 2252 and $casenum le 4502) offgrp = 1. freq offgrp. * * Compute the Dependent variable which stacks answers to questions * about doing things on the web versus doing them offline. For the first * half of the dataset, we have answers to webacts5 questions; for the * second half we have answers to offacts5 questions. *. compute bothacts = webacts5. if (offgrp eq 1) bothacts = offacts5. temp. select if (offgrp = 1). correlation offacts5 webacts5 bothacts. temp. select if (offgrp = 0). correlation offacts5 webacts5 bothacts. * * Save the File. *. SAVE OUTFILE = 'USER FILE\CutDownPew2.sav' /keep = nweight webacts5 offacts5 educ inc presint commeff partstrn male sex suburban rural relatt relattsq black latino age age1825 age2630 age3140 age4150 age5160 age6175 age7696 educind incind ageind pintind comind relatt ageind missreg weight nweight webfour bothacts offgrp. execute. * * * REGRESSIONS--FOR F-TESTS *. GET FILE = 'USER FILE\CutDownPew2.sav'. * * FIRST WE DO HIGHLY CONSTRAINED TESTS FOR THREE SAMPLES * We have overall regression in which everything is constrained (except * for allowing for varying intercept with each of two groups). *. * * -- This is for WHOLE SAMPLE--Note OFFGRP in last regression. *. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 0). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696/dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 1). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696/dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 /dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 offgrp /dep=bothacts/meth=enter. * * Just Those on INTERNET *. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 0 and webfour ne 0). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696/dep=bothacts/meth=enter. temp. select if (missreg ne 1 and offgrp eq 1 and webfour ne 0). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696/dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and webfour ne 0). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696/dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and webfour ne 0). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 offgrp/dep=bothacts/meth=enter. * * Just Those on HIGH SPEED INTERNET *. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 0 and webfour eq 3). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696/dep=bothacts/meth=enter. temp. select if (missreg ne 1 and offgrp eq 1 and webfour eq 3). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696/dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and webfour eq 3). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696/dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and webfour eq 3). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 offgrp/dep=bothacts/meth=enter. * * SECOND WE DO SOMEWHAT CONSTRAINED TESTS FOR THREE SAMPLES * In the overall regression, only the coefficients for eduation, income, interest, * efficacy, and partisan strength are constrained. *. * * Getting Interaction Terms to Impose Constraints *. compute educ# = educ*offgrp. compute inc# = inc*offgrp. compute presint# = presint*offgrp. compute commeff# = commeff*offgrp. compute partstr# = partstrn*offgrp. compute male# = male*offgrp. compute suburb# = suburban*offgrp. compute rural# = rural*offgrp. compute relatt# = relatt*offgrp. compute black# = black*offgrp. compute latino# = latino*offgrp. compute age# = age*offgrp. compute age1825# = age1825*offgrp. compute age2630# = age2630*offgrp. compute age3140# = age3140*offgrp. compute age4150# = age4150*offgrp. compute age5160# = age5160*offgrp. compute age6175# = age6175*offgrp. compute age7696# = age7696*offgrp. * * -- This is for WHOLE SAMPLE--Note OFFGRP in last regression. *. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 0). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 /dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 1). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 /dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1). reg vars = bothacts educ inc presint commeff partstrn male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age1825 age1825# age2630 age2630# age3140 age3140# age5160 age5160# age6175 age6175# age7696 age7696# /dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1). reg vars = bothacts educ inc presint commeff partstrn male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age1825 age1825# age2630 age2630# age3140 age3140# age5160 age5160# age6175 age6175# age7696 age7696# offgrp /dep=bothacts/meth=enter. * * Just Those on INTERNET *. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 0 and webfour ne 0). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 /dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 1 and webfour ne 0). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 /dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and webfour ne 0). reg vars = bothacts educ inc presint commeff partstrn male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age1825 age1825# age2630 age2630# age3140 age3140# age5160 age5160# age6175 age6175# age7696 age7696# /dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and webfour ne 0). reg vars = bothacts educ inc presint commeff partstrn male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age1825 age1825# age2630 age2630# age3140 age3140# age5160 age5160# age6175 age6175# age7696 age7696# offgrp/dep=bothacts/meth=enter. * * Just Those on HIGH SPEED INTERNET *. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 0 and webfour eq 3). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 /dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 1 and webfour eq 3). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 /dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and webfour eq 3). reg vars = bothacts educ inc presint commeff partstrn male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age1825 age1825# age2630 age2630# age3140 age3140# age5160 age5160# age6175 age6175# age7696 age7696# /dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and webfour eq 3). reg vars = bothacts educ inc presint commeff partstrn male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age1825 age1825# age2630 age2630# age3140 age3140# age5160 age5160# age6175 age6175# age7696 age7696# offgrp /dep=bothacts/meth=enter. * * THIRD WE DO MINIMALLY CONSTRAINED TESTS FOR THREE SAMPLES * We have overall regression in which only the coefficients for education and income * are constrained (except for allowing for varying intercept with each of two groups). *. * * -- This is for WHOLE SAMPLE--Note OFFGRP in last regression. *. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 0). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696/dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 1). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696/dep=bothacts/meth=enter. temp. select if (missreg ne 1). reg vars = bothacts educ inc presint presint# commeff commeff# partstrn partstr# male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age1825 age1825# age2630 age2630# age3140 age3140# age5160 age5160# age6175 age6175# age7696 age7696# /dep=bothacts/meth=enter. temp. select if (missreg ne 1). reg vars = bothacts educ inc presint presint# commeff commeff# partstrn partstr# male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age1825 age1825# age2630 age2630# age3140 age3140# age5160 age5160# age6175 age6175# age7696 age7696# offgrp /dep=bothacts/meth=enter. * * Just Those on INTERNET *. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 0 and webfour ne 0). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 /dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 1 and webfour ne 0). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 /dep=bothacts/meth=enter. temp. select if (missreg ne 1 and webfour ne 0). reg vars = bothacts educ inc presint presint# commeff commeff# partstrn partstr# male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age1825 age1825# age2630 age2630# age3140 age3140# age5160 age5160# age6175 age6175# age7696 age7696# /dep=bothacts/meth=enter. temp. select if (missreg ne 1 and webfour ne 0). reg vars = bothacts educ inc presint presint# commeff commeff# partstrn partstr# male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age1825 age1825# age2630 age2630# age3140 age3140# age5160 age5160# age6175 age6175# age7696 age7696# offgrp /dep=bothacts/meth=enter. * * Just Those on HIGH SPEED INTERNET *. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 0 and webfour eq 3). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 /dep=bothacts/meth=enter. weight by nweight. temp. select if (missreg ne 1 and offgrp eq 1 and webfour eq 3). reg vars = bothacts educ inc presint commeff partstrn male suburban rural relatt black latino age1825 age2630 age3140 age5160 age6175 age7696 /dep=bothacts/meth=enter. temp. select if (missreg ne 1 and webfour eq 3). reg vars = bothacts educ inc presint presint# commeff commeff# partstrn partstr# male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age1825 age1825# age2630 age2630# age3140 age3140# age5160 age5160# age6175 age6175# age7696 age7696# /dep=bothacts/meth=enter. temp. select if (missreg ne 1 and webfour eq 3). reg vars = bothacts educ inc presint presint# commeff commeff# partstrn partstr# male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age1825 age1825# age2630 age2630# age3140 age3140# age5160 age5160# age6175 age6175# age7696 age7696# offgrp /dep=bothacts/meth=enter. * * NOW CONSTRAIN AGE *. * * Now constrain income, education, and AGE for each Internet Group * . *. * * -- WHOLE SAMPLE *. weight by nweight. temp. select if (missreg ne 1). reg vars = bothacts educ inc presint presint# commeff commeff# partstrn partstr# male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age age# /dep=bothacts/meth=enter. * * IINTERNET USERS *. weight by nweight. temp. select if (missreg ne 1 and webfour ne 0). reg vars = bothacts educ inc presint presint# commeff commeff# partstrn partstr# male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age age# /dep=bothacts/meth=enter. * * HIGH SPEED INTERNET USERS *. weight by nweight. temp. select if (missreg ne 1 and webfour eq 3). reg vars = bothacts educ inc presint presint# commeff commeff# partstrn partstr# male male# suburban suburb# rural rural# relatt relatt# black black# latino latino# age age# /dep=bothacts/meth=enter.