ࡱ> a bjbj 4 hGbhGbs& 66666JJJ8. JXZ:"YYYYYYY$X\_Y6Y66Zc7c7c766Yc7Yc7c7RWTWPD$0KUY(Z0XZiU`5*`<WW`6'Wc7YY-66XZ` B : Sociology 7706 Longitudinal Data Analysis Instructor: Natasha Sarkisian Introduction to Stata Preparation for Using Stata on Citrix STEP 1. Use your browser to go to BC Apps on bcapps.bc.edu, login using your BC credentials, and open Stata 16SE app. For more information on troubleshooting, see  HYPERLINK "https://www.bc.edu/offices/help/teaching/app_server.html" https://www.bc.edu/offices/help/teaching/app_server.html Note: There are times when too many people use Stata 16, and the system doesnt have enough licenses to add one more person; you can use Stata 15 or even 14 instead in these cases. STEP 2. To be able to store files on BC server, map the drive for BC apps storage; the detailed instructions for each operating system are available here:  HYPERLINK "http://www.bc.edu/offices/help/teaching/app_server/apps-files/files-map.html" http://www.bc.edu/offices/help/teaching/app_server/apps-files/files-map.html. You have to be at BC or connect using BC VPN ( HYPERLINK "https://www.bc.edu/offices/help/getstarted/network/vpn.html" https://www.bc.edu/offices/help/getstarted/network/vpn.html) in order to have access to BC apps storage. Note: On Macs, you might need to repeat mapping the drive steps from time to time if BC Apps storage does not show up in a list of available folders. So keep the link and remember how to do this step for potential future use if you have a Mac. If you run into problems, contact the BC technology help center. If they cannot resolve your issue or you need further help, please come see me. STEP 3. Open your BC apps storage folder and create a subfolder for this class; call your subfolder stats2020. STEP 4. Download GSS 2012 dataset from the course webpage and place it into your stats2020 subfolder in your folder on BC apps storage server. STEP 5. Once again, go to BC Apps on bcapps.bc.edu and open Stata 16SE app, or go back to that window if its still open. Open GSS 2012 file from inside Stata program by typing the following commands in the command line on the bottom of Stata screen (press enter after each command line): cd L:\stats2020\ log using stataprep.log, replace use gss2012.dta, clear log close All of these commands should run and generate no red error messages on your Stata screen if all went well. If thats the case, you can close Stata and BC Apps once you are done. STEP 6. Open a file explorer program and navigate to the BC Apps storage folder. Find stataprep.log file in your list and open it using any text editor. Make sure it contains a few lines of text, including your use gss2012.dta, clear command. Close it again. STEP 7. Open your email, attach the stataprep.log file (make sure to navigate to the apps storage folder and then to stats2020 subfolder) and email it to yourself. Opening data files Download GSS 2012 dataset from the course webpage; place it into your folder on BC apps storage server. Open Stata 16SE on BCApps. To check the current working directory, type pwd in the Command window immediately after starting Stata (without running a cd command). . pwd L:\ Lets change the working directory for easier file access using cd (c=change d=directory) command (alternatively, you can specify the path each time, or open and save files using Stata menus): . cd L:\stats2020\ Now that we have the correct working directory set up, we can open the data file from inside Stata program using the following command: . use gss2012.dta, clear You can also open files from the web, e.g.: . use https://www.sarkisian.net/socy7706/gss2012.dta, clear Keeping a record of your work Opening the log file (we include replace option so that you dont get an error if you already used that log file name before it will get replaced by the new one; if you want to add to the existing log, use append instead of replace): log using learn_stata.log, replace To see the log, you can at any time press the button and view a snapshot of the log. (You can also close or suspend log using that same button.)  Two types of log -- .log and .scml. I choose .log rather than .scml type of file so it can be read in any text editor or word processor. I would recommend that you always use .log format for now. But you can also easily convert .scml type log into the text format log: translate mylog.smcl mylog.log You can also use translate command to recover a log when you have forgotten to start one: translate @Results mylog.txt Note that if you are opening a Stata log file in a Word processor, you should change the font to a fixed width font, such as Courier New (otherwise the output looks misaligned). Courier New 10 or 9 point usually works the best. Otherwise things wont be aligned. Basic syntax of Stata commands: Command What do you want to do? Names of variables, files, etc. Which variables or files do you want to use? Qualifier on observations -- Which observations do you want to use? Options Do you have any other preferences regarding this command? Help and installation Help in Stata help and search commands: . help tabulate . search logistic Keyword search Keywords: logistic Search: (1) Official help files, FAQs, Examples, SJs, and STBs Search of official help files, FAQs, Examples, SJs, and STBs [U] Chapter 26 . . . . . . . . . . Overview of Stata estimation commands (help estcom) [R] clogit . . . . . . . Conditional (fixed-effects) logistic regression (help clogit) [R] cloglog . . . . . . . . . . . . . . . Complementary log-log regression (help cloglog) [R] constraint . . . . . . . . . . . . . . . Define and list constraints (help constraint) [R] fracpoly . . . . . . . . . . . . . . Fractional polynomial regression (help fracpoly) [R] glogit . . . . . . . . . . . . . . Logit and probit for grouped data (help glogit) [R] logistic . . . . . . . . . Logistic regression, reporting odds ratios (help logistic) [R] logistic postestimation . . . . . . Postestimation tools for logistic (help logistic postestimation) [R] logit . . . . . . . . . . logistic regression, reporting coefficients (help logit) [R] logit postestimation . . . . . . . . . Postestimation tools for logit (help logit postestimation) [R] mfp . . . . . . . . . . . . Multivariable fractional polynomial models (help mfp) [R] mlogit . . . . . . . . . Multinomial (polytomous) logistic regression (help mlogit) [R] nlogit . . . . . . . . . . . . . . . . . . . Nested logit regression (help nlogit) [R] ologit . . . . . . . . . . . . . . . . . Ordered logistic regression (help ologit) --Break-- r(1); You can also use net search command that will search Stata resources online in addition to local resources: . net search spost (contacting http://www.stata.com) 16 packages found (Stata Journal and STB listed first) ------------------------------------------------------ st0094 from http://www.stata-journal.com/software/sj5-4 SJ5-4 st0094. Confidence intervals for predicted outcomes... / Confidence intervals for predicted outcomes in regression / models for categorical outcomes / by Jun Xu and J. Scott Long, Indiana University / Support: spostsup@indiana.edu / After installation, type help prvalue and prgen spost9_ado from http://www.indiana.edu/~jslsoc/stata spost9_ado | Stata 9-13 commands for the post-estimation interpretation / Distribution-date: 05Aug2013 / of regression models. Use package spostado.pkg for Stata 8. / Based on Long & Freese - Regression Models for Categorical Dependent / Variables Using Stata. Second Edition. / Support spost9_do from http://www.indiana.edu/~jslsoc/stata spost9_do | SPost9 example do files. / Distribution-date: 27Jul2005 / Long & Freese 2005 Regression for Categorical Dependent Variables / using Stata. Second Edition. Stata Version 9. / Support www.indiana.edu/~jslsoc/spost.htm / Scott Long & Jeremy Freese spostado from http://www.indiana.edu/~jslsoc/stata spostado: Stata 8 commands for the post-estimation interpretation of / regression models. Based on Long's Regression Models for Categorical / and Limited Dependent Variables. / Support: www.indiana.edu/~jslsoc/spost.htm / Scott Long & Jeremy Freese (spostsup@indiana.edu) spostrm7 from http://www.indiana.edu/~jslsoc/stata spostrm7: Stata 7 do & data files to reproduce RM4CLDVs results using SPost. / Files correspond to chapters of Long: Regression Models for Categorical / & Limited Dependent Variables. / Support: www.indiana.edu/~jslsoc/spost.htm / Scott Long & Jeremy Freese spostst8 from http://www.indiana.edu/~jslsoc/stata spostst8: Stata 8 do & data files to reproduce RM4STATA results using SPost. / Files correspond to chapters of Long & Freese-Regression Models for Categorical / Dependent Variables Using Stata (Stata 8 Revised Edition). / Support: www.indiana.edu/~jslsoc/spost.htm / Scott Long & spost13_ado from http://www.indiana.edu/~jslsoc/stata Distribution-date: 15Jul2015 / spost13_ado | SPost13 commands from Long and Freese (2014) / Regression Models for Categorical Outcomes using Stata, 3rd Edition. / Support www.indiana.edu/~jslsoc/spost.htm / Scott Long (jslong@indiana.edu) & Jeremy Freese (jfreese@northwestern.edu) spost9_legacy from http://www.indiana.edu/~jslsoc/stata Distribution-date: 18Feb2014 / spost9_legacy | SPost9 commands not included in spost13_ado. / From Long and Freese, 2014, Regression Models for Categorical Outcomes / using Stata, 3rd Edition. / Support www.indiana.edu/~jslsoc/spost.htm / Scott Long (jslong@indiana.edu) & spost13_do from http://www.indiana.edu/~jslsoc/stata Distribution-date: 05Aug2014 / spost13_do | SPost13 examples from Long and Freese, 2014, / Regression Models for Categorical Outcomes using Stata, 3rd Edition. / Support www.indiana.edu/~jslsoc/spost.htm / Scott Long (jslong@indiana.edu) & Jeremy Freese (jfreese@northwestern.edu) spost13_do12 from http://www.indiana.edu/~jslsoc/stata Distribution-date: 11Aug2014 / spost13_do12 | SPost13 examples for Stata 12 from Long and Freese, 2014, / Regression Models for Categorical Outcomes using Stata, 3rd Edition. / Support www.indiana.edu/~jslsoc/spost.htm / Scott Long (jslong@indiana.edu) & difd from http://fmwww.bc.edu/RePEc/bocode/d 'DIFD': module to evaluate test items for differential item functioning (DIF) / DIF detection is a first step in assessing bias in test items. / difd detects DIF in test items between groups, conditional on / the trait that the test is measuring, using logistic / regression. The criteria for difdetect from http://fmwww.bc.edu/RePEc/bocode/d 'DIFDETECT': module to detect and adjust for differential item functioning (DIF) / Detection of and adjustment for differential item functioning / (DIF): Identifies differential item functioning, creates / dummy/virtual items to be used to adjust ability (trait) / estimates, and calculates the difwithpar from http://fmwww.bc.edu/RePEc/bocode/d 'DIFWITHPAR': module for detection of and adjustment for differential item functioning (DIF) / Identifies differential item functioning, creates / dummy/virtual items to be used to adjust ability (trait) / estimates in PARSCALE, writes the code and data file needed to / process the updated grcompare from http://fmwww.bc.edu/RePEc/bocode/g 'GRCOMPARE': module to make group comparisons in binary regression models / This is a Stata module to make group comparisons in binary / regression models using alternative measures, including gradip: / average difference in predicted probabilities over a range; / grdiame:difference in group prepar from http://fmwww.bc.edu/RePEc/bocode/p 'PREPAR': module to write code and data file needed to process variables in PARSCALE / This program writes the input code and data file for PARSCALE, / which is a real time-saver if you aren't familiar with / PARSCALE. / KW: PARSCALE / Requires: Stata version 8.2, PARSCALE and runparscale from http://fmwww.bc.edu/RePEc/bocode/r 'RUNPARSCALE': module to run PARSCALE from Stata / Builds a PARSCALE data file and command file, executes the / command file, displays the PARSCALE log file in Stata results / window, and merges the PARSCALE theta estimates and their / standard errors back into the original data set. / 1 reference found in tables of contents --------------------------------------- http://www.indiana.edu/~jslsoc/stata/ 2014-08-10 / SPost: Interpreting regression models. Scott Long & Jeremy Freese / Workflow: Workflow of data analysis. Scott Long / Teaching: Teaching files. Scott Long / Research: Research examples & commands. Scott Long / Support: www.indiana.edu/~jslsoc/spost.htm / Note that some of the things we found are user-written programs that implement user-written commands that can be quite helpful; to install, click on the package and click to install, or type . net install spost13_ado, from( HYPERLINK "http://www.indiana.edu/~jslsoc/stata" http://www.indiana.edu/~jslsoc/stata) Also, if you have Stata on your own computer, do not forget to do Stata updates on a regular basis, including updating all installed programs (ado files). . update all Good resource for learning Stata:  HYPERLINK "http://www.ats.ucla.edu/stat/stata/" http://www.ats.ucla.edu/stat/stata/ Forum to ask questions about Stata (but search for answers first!):  HYPERLINK "http://www.statalist.org/" http://www.statalist.org/ Examining and editing the data Describing the dataset: . des Contains data from L:\stats2020\gss2012.dta obs: 1,974 vars: 800 11 Sep 2013 06:50 size: 1,717,380 ------------------------------------------------------------------------------- storage display value variable name type format label variable label ------------------------------------------------------------------------------- year int %8.0g GSS YEAR FOR THIS RESPONDENT id int %8.0g RESPONDNT ID NUMBER wtss double %12.0g WTSS WEIGHT VARIABLE vpsu byte %8.0g LABA Variance primary sampling unit vstrat int %8.0g LABA Variance stratum abany byte %8.0g LABB ABORTION IF WOMAN WANTS FOR ANY REASON abdefect byte %8.0g LABB STRONG CHANCE OF SERIOUS DEFECT abhlth byte %8.0g LABB WOMANS HEALTH SERIOUSLY ENDANGERED abnomore byte %8.0g LABB MARRIED--WANTS NO MORE CHILDREN abpoor byte %8.0g LABB LOW INCOME--CANT AFFORD MORE CHILDREN abrape byte %8.0g LABB PREGNANT AS RESULT OF RAPE absingle byte %8.0g LABB NOT MARRIED accntsci byte %8.0g LABC HOW SCIENTIFIC: ACCOUNTING accptoth byte %8.0g LABD R ACCEPT OTHERS EVEN WHEN THEY DO THINGS WRONG acqntsex byte %8.0g ACQNTSEX R HAD SEX WITH ACQUAINTANCE LAST YEAR actupset byte %8.0g LABE PPL AT WORK THROW THINGS WHEN UPSET WITH R --Break-- r(1); I used Break button to stop Stata from producing more output. If you do want to see all the output, either click on the more link on the bottom of the output viewer, or press space key. For some of you, more link doesnt appear and the output appears all at once rather than one page at a time. That is regulated with set more off and set more on commands in Stata (you can add perm option to make Stata remember your preference).  Get codebook info: . codebook class ------------------------------------------------------------------------------- class SUBJECTIVE CLASS IDENTIFICATION ------------------------------------------------------------------------------- type: numeric (byte) label: CLASS range: [1,4] units: 1 unique values: 4 missing .: 17/1974 tabulation: Freq. Numeric Label 200 1 LOWER CLASS 853 2 WORKING CLASS 839 3 MIDDLE CLASS 65 4 UPPER CLASS 17 . List values of selected variables for each observation: . list wrkstat hrs1 wrkslf +----------------------------+ | wrkstat hrs1 wrkslf | |----------------------------| 1. | WORKING 15 SOMEONE | 2. | WORKING 30 SOMEONE | 3. | WORKING 60 SOMEONE | 4. | other . SOMEONE | 5. | retired . SOMEONE | |----------------------------| 6. | other . SOMEONE | 7. | KEEPING . SOMEONE | --Break-- r(1); Using data browser to look at the data and data editor to change data:  There is no UNDO button!!! That applies to all data management commands, too. But the changes are made only to the dataset thats in Statas memory. So if you close it without saving, you can always start again with your original dataset. If, however, you want to save your changes, you should save the data file with any changes you made typically with a different filename: . save gss2012changed.dta, replace file L:\stats2020\gss2012changed.dta saved If you are not sure you want to keep your changes, use preserve command in the beginning to save a copy of the dataset in Stata memory; restore in the end will return the data to that saved version. Using do-files You should keep a do-file with all your analysis steps that way, if you make a mistake, you can easily rerun things. To have that, we can save all the commands that we did interactively into a do-file, or we can right away write a do-file and then execute it. Open do-file editor, create and save your file (.do) or use doedit command. You can execute that file from the do-file editor or using the command line: . do mydofile.do But be careful to specify the location of your file or make sure it is in the working directory specified in the last cd command. It is often convenient to create and edit do-files in another text editor in Windows, I prefer TextPad:  HYPERLINK "http://www.textpad.com" http://www.textpad.com; another good option is Notepad++. For a Mac OS, you can use Sublime Text. And if you want to save all commands youve done so far, right click on the command window and select Save Review Contents. If some of your commands had errors (highlighted in red), you can right click on each of them and delete them from the Review window before copying. Or you can select some commands and send them to do editor by right clicking and selecting Send to Do File Editor. You can also keep the log of just the commands: cmdlog using filename Then you can use that log as a do-file. Its a good idea to specify Stata version in the beginning of each do file, e.g. version 16 Whether in do files or when entering commands interactively, it is useful to include comments on what you are doing: Everything typed after a star (*) or after // is treated as a comment and not executed; same with any text between /* and */ In addition, people often use /// as a line break tool to better format do-files: use gss2002.dta, clear sum age /// here I am summarizing age wrkstat /// here I am summarizing work status, and next sex sex Note that you cant include /// on the last line of a command (or in the end of a one-line command) because otherwise it doesnt see a carriage return and doesnt execute that command at all. Use star to create comments on a separate line in such cases. Lines in do files can be either separated with line breaks (CR=carriage return) or a delimiter ; To change, you specify the following command in the beginning of a do file. #delimit ; To restore the carriage return delimiter inside a file, use #delimit cr. When a do-file begins execution, the delimiter is automatically set to carriage return, even if it was called from another do-file that set the delimiter to semicolon. Also, the current do-file need not worry about restoring the delimiter to what it was because Stata will do that automatically. Closing log and exiting Stata . log close . exit, clear Descriptive Statistics in Stata Lets reopen the data file and continue our log: . use gss2012.dta, clear . log using learn_stata.log, append Frequency tables -- tabulate command: . tab class SUBJECTIVE | CLASS | IDENTIFICATIO | N | Freq. Percent Cum. --------------+----------------------------------- LOWER CLASS | 200 10.22 10.22 WORKING CLASS | 853 43.59 53.81 MIDDLE CLASS | 839 42.87 96.68 UPPER CLASS | 65 3.32 100.00 --------------+----------------------------------- Total | 1,957 100.00 This also allows us to identify the mode here, WORKING CLASS is the mode. Including missing values: . tab class, miss SUBJECTIVE | CLASS | IDENTIFICATIO | N | Freq. Percent Cum. --------------+----------------------------------- LOWER CLASS | 200 10.13 10.13 WORKING CLASS | 853 43.21 53.34 MIDDLE CLASS | 839 42.50 95.85 UPPER CLASS | 65 3.29 99.14 . | 17 0.86 100.00 --------------+----------------------------------- Total | 1,974 100.00 To suppress labels and see numeric values: . tab class, nol SUBJECTIVE | CLASS | IDENTIFICAT | ION | Freq. Percent Cum. ------------+----------------------------------- 1 | 200 10.22 10.22 2 | 853 43.59 53.81 3 | 839 42.87 96.68 4 | 65 3.32 100.00 ------------+----------------------------------- Total | 1,957 100.00 Multiple univariate tables of frequencies are obtained using tab1 command: . tab1 marital class -> tabulation of marital MARITAL | STATUS | Freq. Percent Cum. --------------+----------------------------------- married | 900 45.59 45.59 widowed | 163 8.26 53.85 divorced | 317 16.06 69.91 separated | 68 3.44 73.35 NEVER MARRIED | 526 26.65 100.00 --------------+----------------------------------- Total | 1,974 100.00 -> tabulation of class SUBJECTIVE | CLASS | IDENTIFICATIO | N | Freq. Percent Cum. --------------+----------------------------------- LOWER CLASS | 200 10.22 10.22 WORKING CLASS | 853 43.59 53.81 MIDDLE CLASS | 839 42.87 96.68 UPPER CLASS | 65 3.32 100.00 --------------+----------------------------------- Total | 1,957 100.00 Measures of central tendency and variability: . sum tvhours Variable | Obs Mean Std. Dev. Min Max -------------+-------------------------------------------------------- tvhours | 1298 3.088598 2.8651 0 24 . sum tvhours, detail HOURS PER DAY WATCHING TV ------------------------------------------------------------- Percentiles Smallest 1% 0 0 5% 0 0 10% 1 0 Obs 1298 25% 1 0 Sum of Wgt. 1298 50% 2 Mean 3.088598 Largest Std. Dev. 2.8651 75% 4 24 90% 6 24 Variance 8.208798 95% 8 24 Skewness 3.123997 99% 15 24 Kurtosis 18.48296 . tabstat tvhours, stats(mean median min max p25 p75 range iqr sd variance) variable | mean p50 min max p25 p75 -------------+------------------------------------------------------------ tvhours | 3.088598 2 0 24 1 4 -------------------------------------------------------------------------- variable | range iqr sd variance -------------+----------------------------------------- tvhours | 24 3 2.8651 8.208798 ------------------------------------------------------- Cross-tabulation: . tab wrkslf wrkgovt R SELF-EMP OR | GOVT OR PRIVATE WORKS FOR | EMPLOYEE SOMEBODY | governmen private | Total --------------+----------------------+---------- SELF-EMPLOYED | 6 161 | 167 SOMEONE ELSE | 365 1,324 | 1,689 --------------+----------------------+---------- Total | 371 1,485 | 1,856 With column percentages: . tab wrkslf wrkgovt, col +-------------------+ | Key | |-------------------| | frequency | | column percentage | +-------------------+ R SELF-EMP OR | GOVT OR PRIVATE WORKS FOR | EMPLOYEE SOMEBODY | governmen private | Total --------------+----------------------+---------- SELF-EMPLOYED | 6 161 | 167 | 1.62 10.84 | 9.00 --------------+----------------------+---------- SOMEONE ELSE | 365 1,324 | 1,689 | 98.38 89.16 | 91.00 --------------+----------------------+---------- Total | 371 1,485 | 1,856 | 100.00 100.00 | 100.00 With three types of percentages and chi-square test: . tab wrkslf wrkgovt, col row cell chi2 +-------------------+ | Key | |-------------------| | frequency | | row percentage | | column percentage | | cell percentage | +-------------------+ R SELF-EMP OR | GOVT OR PRIVATE WORKS FOR | EMPLOYEE SOMEBODY | governmen private | Total --------------+----------------------+---------- SELF-EMPLOYED | 6 161 | 167 | 3.59 96.41 | 100.00 | 1.62 10.84 | 9.00 | 0.32 8.67 | 9.00 --------------+----------------------+---------- SOMEONE ELSE | 365 1,324 | 1,689 | 21.61 78.39 | 100.00 | 98.38 89.16 | 91.00 | 19.67 71.34 | 91.00 --------------+----------------------+---------- Total | 371 1,485 | 1,856 | 19.99 80.01 | 100.00 | 100.00 100.00 | 100.00 | 19.99 80.01 | 100.00 Pearson chi2(1) = 30.8474 Pr = 0.000 Data Management in Stata Creating new variables: typically done using gen (often followed by replace), recode, or egen commands; egen is more advanced and we wont cover it for now. For example, I want to create a variable that is 0 for those people who work less than 40 hours a week and 1 for those who work 40 hours a week or more (we call this a dichotomy or a dummy variable, where 0 means absence of some characteristic and 1 means presence). Lets first examine the variable that exists in the data set: . codebook hrs1 ----------------------------------------------------------------------------------------- hrs1 NUMBER OF HOURS WORKED LAST WEEK ----------------------------------------------------------------------------------------- type: numeric (byte) label: LABAD, but 68 nonmissing values are not labeled range: [1,89] units: 1 unique values: 68 missing .: 808/1,974 examples: 40 45 . . . sum hrs1, det NUMBER OF HOURS WORKED LAST WEEK ------------------------------------------------------------- Percentiles Smallest 1% 5 1 5% 9 1 10% 19 2 Obs 1,166 25% 34 4 Sum of Wgt. 1,166 50% 40 Mean 40.27358 Largest Std. Dev. 15.54011 75% 48 89 90% 60 89 Variance 241.495 95% 65 89 Skewness .0575558 99% 80 89 Kurtosis 3.649988 We start by generating a new variable called hrs40 with all missing values. Then we will first fill in zeroes for those who work less than 40 hours for pay, and finally fill in 1s for those who work 40+ hours. To do so, we need to use conditions expressed as an if statement. To express conditions, we can use the following: < less > more == equal <= less or equal >= more or equal ~= or != not equal Can connect them with & (and) and | (or). Can also use parentheses to combine conditions. So in our case, we do: . gen hrs40=. (1,974 missing values generated) . replace hrs40 = 0 if hrs1<40 (377 real changes made) . replace hrs40 = 1 if hrs1>=40 & hrs1~=. (789 real changes made) . tab hrs40, missing hrs40 | Freq. Percent Cum. ------------+----------------------------------- 0 | 377 19.10 19.10 1 | 789 39.97 59.07 . | 808 40.93 100.00 ------------+----------------------------------- Total | 1,974 100.00 Label the variable: . label variable hrs40 "R works 40 hours a week or more" Label its values: two steps, first define a set of labels, then apply this set to a variable: . label define hrs40label 0 "less than 40" 1 "40 or more" . label values hrs40 hrs40label . tab hrs40, missing R works 40 | hours a week | or more | Freq. Percent Cum. -------------+----------------------------------- less than 40 | 377 19.10 19.10 40 or more | 789 39.97 59.07 . | 808 40.93 100.00 -------------+----------------------------------- Total | 1,974 100.00 . codebook hrs40 ----------------------------------------------------------------------------------------- hrs40 R works 40 hours a week or more ----------------------------------------------------------------------------------------- type: numeric (float) label: hrs40label range: [0,1] units: 1 unique values: 2 missing .: 808/1,974 tabulation: Freq. Numeric Label 377 0 less than 40 789 1 40 or more 808 . To rename a variable: . rename hrs40 hours40 There is a simpler way to generate a dummy variable that only uses one step; for example, to generate a dichotomy indicating married respondents (0=not married, 1=married): . codebook marital ----------------------------------------------------------------------------------------- marital MARITAL STATUS ----------------------------------------------------------------------------------------- type: numeric (byte) label: MARITAL range: [1,5] units: 1 unique values: 5 missing .: 0/1,974 tabulation: Freq. Numeric Label 900 1 married 163 2 widowed 317 3 divorced 68 4 separated 526 5 NEVER MARRIED . gen married=(marital==1) if marital<. . tab married married | Freq. Percent Cum. ------------+----------------------------------- 0 | 1,074 54.41 54.41 1 | 900 45.59 100.00 ------------+----------------------------------- Total | 1,974 100.00 The additional condition, if marital<. , is included in order to preserve missing data without that condition, anyone with marital variable missing would be coded as 0. Here, however, we dont have any missing values on marital status. Another command we can use to create variables is recode command. For example, to generate marital status with 3 categories, married, previously married, never married: . recode marital (1=1) (2/4=2) (5=3), gen(marital3) (1249 differences between marital and marital3) . tab marital3 RECODE of | marital | (MARITAL | STATUS) | Freq. Percent Cum. ------------+----------------------------------- 1 | 900 45.59 45.59 2 | 548 27.76 73.35 3 | 526 26.65 100.00 ------------+----------------------------------- Total | 1,974 100.00 Label the new variable: . label variable marital3 "marital status 3 categories" Label values of the new variable: . label define marital3label 1 "married" 2 "previously married" 3 "never married" . label values marital3 marital3label Check the results: . codebook marital3 ----------------------------------------------------------------------------------------- marital3 marital status 3 categories ----------------------------------------------------------------------------------------- type: numeric (byte) label: marital3label range: [1,3] units: 1 unique values: 3 missing .: 0/1,974 tabulation: Freq. Numeric Label 900 1 married 548 2 previously married 526 3 never married These are various recoding techniques to create and deal with categorical and dichotomous variables; but recoding is also very useful for continuous variables. For example, they often have extreme values (outliers) that could impact our results unduly, and to reduce their impact, we can topcode or bottomcode the variable, that is, recode the very high or very low values to bring them more into the regular range. For example: . tab agekdbrn, miss R'S AGE | WHEN 1ST | CHILD BORN | Freq. Percent Cum. ------------+----------------------------------- 13 | 1 0.05 0.05 14 | 7 0.35 0.41 15 | 20 1.01 1.42 16 | 30 1.52 2.94 17 | 68 3.44 6.38 18 | 101 5.12 11.50 19 | 102 5.17 16.67 20 | 110 5.57 22.24 21 | 134 6.79 29.03 22 | 78 3.95 32.98 23 | 95 4.81 37.79 24 | 82 4.15 41.95 25 | 94 4.76 46.71 26 | 70 3.55 50.25 27 | 78 3.95 54.20 28 | 58 2.94 57.14 29 | 52 2.63 59.78 30 | 55 2.79 62.56 31 | 32 1.62 64.18 32 | 32 1.62 65.81 33 | 28 1.42 67.22 34 | 29 1.47 68.69 35 | 21 1.06 69.76 36 | 13 0.66 70.42 37 | 11 0.56 70.97 38 | 10 0.51 71.48 39 | 7 0.35 71.83 40 | 6 0.30 72.14 41 | 3 0.15 72.29 42 | 1 0.05 72.34 46 | 1 0.05 72.39 50 | 1 0.05 72.44 . | 544 27.56 100.00 ------------+----------------------------------- Total | 1,974 100.00 There are many missing values here as well; those are people without children. The extreme values are probably legitimate values, but in many cases, if using this variable, you might still want to bottomcode and topcode it (i.e., assign all values below 13 to have the value of 13 and all values above 44, you would assign value 44) in order to avoid the undue influence of extreme values on your results. Here, I will be generating a top-coded and bottom-coded version of agekdbrn; only the most extreme outliers (<1% of distribution) are typically top and bottom coded: . . gen agekdbrn_tb=clip(agekdbrn, 13, 44) (544 missing values generated) . sum agekdbrn_tb Variable | Obs Mean Std. Dev. Min Max -------------+--------------------------------------------------------- agekdbrn_tb | 1,430 24.07413 5.589537 13 44 . gen agekdbrn_t=clip(agekdbrn, ., 44) (544 missing values generated) . sum agekdbrn_t Variable | Obs Mean Std. Dev. Min Max -------------+--------------------------------------------------------- agekdbrn_t | 1,430 24.07413 5.589537 13 44 Note that missing values are still coded as missing, they are not topcoded into the top value. You can also create new variables by transforming existing continuous variables. The mathematical operators can be used: addition +, subtraction -, multiplication *, division /, and powers ^. You can also use parentheses when calculations are complex and the order of operations matters. Age squared: . gen age2=age^2 (5 missing values generated) Square root of age: . gen agesqrt=sqrt(age) (5 missing values generated) Log of age: . gen agelg=log(age) (5 missing values generated) Saving the dataset with the newly created variables: . save gss2012_2.dta file L:\stats2020\gss2012_2.dta saved Entering your own data We can do that in data editor, but we need to clear the memory first: clear all Then you can open the editor and start entering numbers. If you have variables that contain text, you can just enter text; the resulting variables will be non-numeric (string). If, however, that text only includes a few categories, then you should assign numbers to those categories and enter numbers. For example, if you have a variable year in college, dont type junior or senior but enter numbers 1, 2, 3, or 4 and then label the values using label define and label values commands, as discussed above. In addition to entering the data themselves, you should also name the variables either in the Variables window (bottom right) or using commands, e.g. rename var1 income When done, save the new dataset: . save enter_example.dta file L:\stats2020\enter_example.dta saved Graphics in Stata . scatter hrs1 prestg80 . graph matrix hrs1 hrs2 prestg80 sphrs1 sppres80 . graph bar, over(class) blabel(bar) . histogram hrs1, normal (bin=32, start=1, width=2.75) We can save graphs for future use (in Stata-specific graph format; can only be opened in Stata): graph save mygraph.gph To then display that graph, we type: graph use mygraph.gph You can also export them into different, non-Stata formats: . graph export mygraph.wmf, replace The output format is determined by the suffix of the file name (see help graph export): Implied suffix option Output format --------------------------------------------------------------------- .ps as(ps) PostScript .eps as(eps) EPS (Encapsulated PostScript) .wmf as(wmf) Windows Metafile .emf as(emf) Windows Enhanced Metafile .pdf as(pdf) PDF .png as(png) PNG (Portable Network Graphics) .tif as(tif) TIFF WMF works best for Windows computers, PNG works best for Macs. You can also edit graphs in interactive mode using Graph Editor in Stata; however, I would recommend using options to generate the graphs you need because that allows for reproduceability. There is, however, also an option to use Recorder together with Graph Editor its not quite the same level of reproducibility and readability as using syntax, but at least then you can reapply all the changes to another graph (or another version of the same graph) after starting Graph Editor, press Record prior to editing the graph, then press stop when you are done and give your recording a name:     You can then create another graph and apply a recording to it: scatter hrs1 occ80, play(myrecord) To further explore the options available for graphics, use: . help graph Stata versions and settings There are different versions of Stata: Variable number limits are 2,047 for Stata/IC, and 99 for Small Stata. When using Stata/MP and Stata/SE, the maximum number of variables in your dataset can be changed by using set maxvar command. The default value of maxvar is 5,000 for Stata/MP and Stata/SE. Here, we are using Stata/IC; the version on the apps server is Stata/SE. Besides set maxvar, to make it easier for you to work with Stata, you can change some of other default settings using set command, e.g.: set logtype text set more off Some Stata settings can be made permanent -- for example, if you want Stata to never pause output with a --more-- in the Results window, you could type . set more off, perm Another useful set command that you will likely encounter once you start running statistical models on large data is set matsize (can also be used with permanently option). set matsize sets the maximum number of rows/columns that can be placed in a matrix used by Stata's data analysis commands. For Stata/IC, the initial value is 400, but it may be changed upward or downward. The upper limit is 800. For Stata/MP and Stata/SE, the default value is 400, but it may be changed upward or downward. The upper limit is 11,000. This command may not be used with Small Stata; matsize is permanently frozen at 100. Another useful set command has to do with graphs. . set scheme schemename [, permanently] set scheme allows you to set the graphics scheme to be used. The default setting is s2color. You can use point and click to explore graphics schemes.     PAGE  PAGE 19  ),Ja|+ , r s 8 d e l  [ \ # $ _ ` y ººwkjh]70JU]h09h]70J]h09h]7]hnh]70J\jh]70JU\h4kRh]70J\h]7jh]7U h]7\hnh]7\hnh]75\ h>^A5hNXh]75 h]75hGgh%5\ hGg5\ h3u|5\hGgh( 5\*J`a d e <OpEFgd]7$a$gd]7gdGgy  |"?_:;<@Lfo>BDEFM9JKLSlyh<CJOJQJ\^JaJ#hCh<CJOJQJ\^JaJh"bh<\ h<\hvVhvV\ h#v5\ hJE\hnh]75\ h]75\h]7 h]7\hnh]7\ h]7];FKL~ !gdKgdZgd<gd#vgd]7%+/@5<Aa}~뽸{iiW#hj\hZCJOJQJ\^JaJ#h+}hZCJOJQJ\^JaJ hZ\hZCJOJQJ\^JaJh#vCJOJQJ\^JaJ#hj\h#vCJOJQJ\^JaJhvVh\ h\ hvV\h]7CJOJQJ\^JaJ#h7`/h<CJOJQJ\^JaJh]yh<\ h<\h<CJOJQJ\^JaJ !53452Rͻojje`Q?`#hLLhZCJOJQJ\^JaJhZCJOJQJ\^JaJ hZ\ h\ h#v\#h7`/h#vCJOJQJ\^JaJ&jh#vCJOJQJU\^JaJh<h<\h<CJOJQJ\^JaJhKCJOJQJ\^JaJ#h7`/hKCJOJQJ\^JaJh]yh<\ hK\ hYQg\h]yhK\ h;N\h#vhz15\h#vhA5\45Qnwx NgdYQg & Fh^hgd>gd>gdZgdKQmnRvwxr"s""":;;ºrr`X`hA#h>\#hA#h>CJOJQJ\^JaJh>CJOJQJ\^JaJ#hj\h>CJOJQJ\^JaJh]yh>\ h>\h}EhYQg5\h> h>5\ h<\h]yhK\ hPJh]yhKPJhZCJOJQJ\^JaJ#hLLhZCJOJQJ\^JaJ hZ\h]yhZ\#mn`vw-GHe}~gd>D Y Z !0!1!!!!!!!L"b"l"r"s""""##O### $gd> $Y$$$$$%r%%&S&T&&& 'V''''(g((((!)k)))3*4*g*gd>g***E++++,[,,,,)-p--.J.K.../e////.0u0000gd>01j112U2V222#3q3333C444'5(5Z555D6666 7S777gd>778i888K9L9t999:Y::::;;<<<<<<<E=F===gd\gd>;;;;;;<<<<<<<<<<== =C=D=E=쥝wok`YK`=`Yh+Qh\0JCJ\aJjFUh+Qh\U h+Qh\jh+Qh\Uh\h+Qh\5h\CJOJQJ\^JaJh>CJOJQJ\^JaJh]yh>\hA#h>\#hA#h>CJOJQJ\^JaJ'hjhJE0JCJOJQJ\^JaJ#hJEhJECJOJQJ\^JaJhJECJOJQJ\^JaJ&jhJECJOJQJU\^JaJE=F=b=h==============>>>>#>%>.>/>:>EEFJFͿyjy[yLyhh\0JCJaJjh\Uh\h+Qh\5 h\5h\5OJQJ\^J====>>>;>j>>>)?R???*@l@@@6AAA BOBBB#C\CCgdCJOJQJ\^JaJhJECJOJQJ\^JaJ#hJEhJECJOJQJ\^JaJh]yh>\ h>\#h[eh[eCJOJQJ\^JaJ hz1\h]yhz1\&jKVhqgCCJOJQJU\^JaJ hA\ h6\ h"\ hK\h"hK>*\ JAJwJJJJJJKBKgKKKKK LELjLLLLLLLfNgNNgdYQggdKgdJEgd>gd[eLLMMeNfNgNtNNNNNNNNNN~OOOOOOOPPPP.Q/QǹǨը~oe^WSWSKKh]yhYQg\hYQg h:hYQg hYQg5\h"hYQg5\h>CJOJQJ\^JaJ h>\h]yh>\hYQgCJOJQJ\^JaJh]7CJOJQJ^JaJ hChYQgCJOJQJ^JaJh>CJOJQJ^JaJhYQgCJOJQJ^JaJh6CJOJQJ^JaJ hK\ hYQg\ h#v\ hz1\h]yh#v\NNNOOO0QAQBQQQRRDTETvTTTTUUUVVWVXVoVVVgdZgd>gdYQg/Q0Q@QAQBQRR1R2R>RTRVRWRmRnRRRRRRSSSDTETvTTTUUUUVXV˯˪yg` *hZ\#h!hZCJOJQJ\^JaJhZCJOJQJ\^JaJ#h+vhZCJOJQJ\^JaJh]yhZ\h:hZ\ hZ\hKhYQg\hGLhYQg0J\jhYQgU\ hYQg\h]yhYQg\hYQgCJOJQJ\^JaJ#h_ZhYQgCJOJQJ\^JaJ h}hYQg!XV\VgVoVzVVVVVVVV5XXXZZ!Z-Z:Z;Z\ hCh>CJOJQJ^JaJ#hCh>CJOJQJ\^JaJ#hOh>CJOJQJ\^JaJ h>5\ hYQg\hZCJaJhPhZCJaJ hZ\h>CJOJQJ\^JaJhZCJOJQJ\^JaJ#hWWhZCJOJQJ\^JaJhZCJOJQJ\^JaJ#hj\hZCJOJQJ\^JaJVVVW5XXXXZZ Z!Z-Z;ZgdYQggdZgdZ]ZcZ|ZZZZZZZZZ\]]] ]!]"]%_7_N_O_P_aQaSaaaaUcVcǿp^Oh(%CJOJQJ\^JaJ#h:2h(%CJOJQJ\^JaJh:2CJOJQJ\^JaJ#h:2h:2CJOJQJ\^JaJ h}\h]yhA\ h \ hA\ h) \#h[ehACJOJQJ\^JaJhhA\ hz1\h]yhz1\#hOhOCJOJQJ\^JaJh>CJOJQJ\^JaJ h>\ hO\/[b[[[[.\a\\\]]"]4]5]E]U]e]]]]1^d^^^^$_%_P_a_b_gdAb_p_~____`P````aSaTaiajaaaaaa/bbbbbb.cUcVcgd:2gdz1gdAVcocpcccccd9dlddde,e-e[e\ejeee?f@fAfWffffg!ggd}EgdKgd:2Vc,e-eZe[e\e@fAfjj@kOkPkQkckllopstttutt̮̮̽vnvnv\MChh]75\h>CJOJQJ\^JaJ#hj\h>CJOJQJ\^JaJhJEhJE\#hJEhJECJOJQJ\^JaJhJEh>\hOCJOJQJ\^JaJh}ECJOJQJ\^JaJhzrCJOJQJ\^JaJh#vCJOJQJ\^JaJ#h[eh}ECJOJQJ\^JaJ hK\ h#v\ h:2\#h:2h:2CJOJQJ\^JaJ!g_ggggh:hxhhhhAiii"jmjnjjjkPkQkckykzkkkklgdJEgd>gd}ElPllllllmm1mGm]msmmmmmmnDnvnnn o=onoooopgdJEp1p2pHp^ptpppppppq%qVqqqqrOrrrrsHsyssstAtgdJEAtBtttuttt-u.uwvxvvvh]7CJOJQJ^JaJh]7CJOJQJ^JaJh]7 hCh]7h]7CJOJQJ\^JaJ&yyy(zfzgzzz{A{{{{{} }}}.}?}S}}}}}}}}}~gd]7~0~1~[~s~t~~~~M~!"ڀۀ@rց:gd]7:`ar̂&̓΃VW5LMh†Degd]7gd]7567LMh݅Jg!;IJqrsދ q GiEbÑőrֹֹƦƦƦওh]7h]7CJOJPJQJ^JaJ$hefgd]7gd]7Ođőrs֓8i˔-^"SHgd]7rsM6QRSՠ֠נ'0̜̉ys`SISy`Ih]7CJPJaJh> }h]7CJPJaJ$h> }h]7CJOJPJQJ^JaJ h]7PJh]7CJOJPJQJ^JaJ$huh]7CJOJPJQJ^JaJ h9(h]7 hCJOJQJ^JaJ hCh]7CJOJQJ^JaJh]7CJOJQJ^JaJ h9h]7h]7$h> }h]7CJOJPJQJ^JaJh]7CJPJaJh#^Fh]7CJPJaJ' $%lvwyz$%FG`+^gd\gd>gd]7"*IJkտzkYQLQ; hh>CJOJQJ^JaJ h\\h]yh>\#h8h>CJOJQJ\^JaJh\CJOJQJ\^JaJ hmf&h\CJOJQJ^JaJh\CJOJQJ^JaJ%h\CJOJQJ\^JaJmH sH %h>CJOJQJ\^JaJmH sH +h+Qh>CJOJQJ\^JaJmH sH h>CJOJQJ\^JaJ#hj\h>CJOJQJ\^JaJh"h>5\+IJ¦æ;<`aԧN{*PMOPgd>¦æ;V_aMNPQSTVWȹ|rh^Vh"h>\jh>U\jD`h>U\j3h>U\jh>U\ h>\ h\\#hTPh>CJOJQJ\^JaJh\CJOJQJ\^JaJh]yh>\h>CJOJQJ\^JaJ#h8h>CJOJQJ\^JaJ h]yh>h>CJOJQJ^JaJ h8h>CJOJQJ^JaJPRSUVXY %&,EZ[ 9:xgd> %&'UV$+,M[  .ڲqrstvwyz{mf^Z^Z^hhjhhU heChz'h>CJOJQJ^JaJ#h6dh>CJOJQJ\^JaJh>CJOJQJ\^JaJ#hCh>CJOJQJ\^JaJ#hNh>CJOJQJ\^JaJh]yh>\ h>\h"h>5\ h>5\h>CJOJQJ\^JaJ#hj\h>CJOJQJ\^JaJ#xyٲڲqrsuvxy{|~h]h&`#$gdi8gd]7gd>z|} heChz'h3u|0JmHnHuh^ h^0Jjh^0JUjhhUhh21h:pNx]/ =!"#8$8% FUDd~ X  C 4Astata_buttonsRTW||"E&ߡevTD[FnTW||"E&ߡeJFIF``C      C  " }!1AQa"q2#BR$3br %&'()*456789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz w!1AQaq"2B #3Rbr $4%&'()*56789:CDEFGHIJSTUVWXYZcdefghijstuvwxyz ?ࢿV*Sƛχ>/oh:Nn}ϘʷcsIk`?xx'Ey|_UwI|y/s|߿ ~lvi%7s %&V|#Լ' #k&7%O]Z_Pq}%xGVo=n+gkkY&EUA+)pGZ:i?п /bU^߷~8GZISަWMW_7' 2uä÷wM!YN׺کi ,@FY뀋 9J.x~:/'~k1j,>Cb{)lեI Ęxm0C[ I>/[&s=/"В٦r S2y7 rs~7W{yNp~*Qs^}?U//-uk+MjncSnt먳qL`0Aτ<EҼIZ޷-IܕY-B ϒ G[|_OC?k?|2~iNo]mjQ V%^ /|)xI]Bog6W?m'#V_H7Ǿ:5ƕ~ *^e^oOvGrLJ"@Թr;nM5K[Wreo^X^>wkw?>/ĝ.(ܶuq^Oozx,DßsqE >"tEב P:8  0t wLsƾ3/bo7ZW?g_w3m;jpK4Ewg4w=dYEx]_,|S[¶'j:uk$s[S"H[,@Y=y,|潮j,W7YgKjYmȱgU [#F@ڦ/<xbҮDe}-[݋;1ⶎ!0t?|O^~,'}fDҍ:ygu,ʹ0Ap+BY? 'ÿ lUQѼ iZhEj7IO,{20rǙAso׮oszƭ:'|G֚z 3XN=^\)|%W?IUGἷ6 Px^oۋ/_ MAj#PKXm"+m|DY?2_]ջ鶽%ߋ-޵kMsSy_[1ܠ k¼!'C.&x6ZmݓuwzLCKqLu mZ!$ѫ" +,V}kwΣV+[Ե(YdEVV(Y!YN3:CoYϤ|x]x&yΟW *EMV07|YkgYOk"SxK/DD."hi s=xa5?x2^&;{S7/mIQ|̮W駯,h_-ZL|Ey:t}wQ>7 )$mMsl-ʹ寜|Xk|CCGKNfie,&1\27?oO}-lzy/kR#}+PYOb!5qpLV7wZ,eXmx&H`Uc|uOi࿊<)Үt.#]m$QݔEDK+nZ=I'k~-ϝ/(Z2:xI jyF]˿ԭyLFNXP 4OֵMGU-;bq^{VWދ,IV 1|3C'}VƗw?-Cu>8%/ZKT7GUWGo%_nɽ?<7RPۿl}bjn]G Q'/ ai~&ox OŬ铇Y>ЬBla_'7l~ YZ|;񾯤$3jww|pkKkyO%_fz)?񳽛U,Qwi?jy~ }K@G,BeNGz_z)rC$_ |GV>:𦙭s3XV2*F(2X0w^>S8 ա63hoGXQCY!C}ekkp‹0DETP0w }?j>Svin%|o6q͕mxJIi.'OLUp`$7^y/+IYzծoپNia|݅ǙnUQr+oGkQy?'g߄o!xgOmR,4-̐G:XZ4&#>T9ܫ| *𶅡\[ưWZ^KUl]U㷵Y%X2Ht2cyF-;DoA}Ś|x: |1\Ga,CwbmK ɞ`da|-/?c߈ j|C׮|AxB}SJWմԲ(>fA fmC%kVM|ioE:}~mAT]7/s_4oYYi'Pb a8ߞYR0PvWTwzAy 1Fc OudPH(#uf_ㆿ#hw_Mg[ 纈DȧtK,L%IE s'kikZw^U?g=oޟg"H?3/??Mhw` Mx(W_~2xbd{8F]ob2+ bƗ7f^•|38? W  vZ[B5]>X]52+Fd,g=2:/zUj'fBI7?Q߃?OY[hZGuD@Yԓ^MK~֐| j6c Rf F*zʵB&puB$w=[F 52z/z+7?QcG]օy%}SNf1{RA=Fr0yO +׷d"kRɻvx;j1rj? ?zC<ƿ 4kS L[WPI)S`Gh;*r@>__EMKPuip4k8o?+;HH:qp,u ,{͗#nM_gdM;)`:R)%mޚ+}zC? Kް|_⻭3> vEorA#WRFˎUfNTl57?G#SH3|YŪ?- {2b1 ) 9)jQ[^5[cXl@8-bD$0#|Lka2̫YRnQmJc ]aw7{XC? ?zC|ccᧈhMu~=;? ?zC𯖾>~NAECTlHID[Yb ~'soOxwᇈ )|.M&e8 bJU;k[ngŽ1JTo=Uml=M=!W7U/ _.Z[\,^m1!gz+Ux~'}\Jb@dk1͵y3`u]XǚUO[{ͯ,t9K߫GM=!Q|=Ɵ~VZwOo>Ԯmsms(s I1 W+~n"hd[sdEm8Uq׼nצ;0ս9\|f#SH357?\?_w~KDWQ,rǜqۃ\mſx W/0aȮhY+E;M=!Qx-WO k,ma#^㿊0uKjZ'TDV(| s*vG57?P_i2i0')_?:σ^"eaF`/ePCpA.wk?ΩH'MhQ+p (:֟P)矴W q~$ k_}P5_ }9XU5B@-E{O:&mo;xlI^X8f)>s1 ?]lM|opw"tZ]VkgɹSoOO{㧄moښ|>*k % :|nI,Q[[oz+mn,ndإvETVf='s'k-2φٿ/!)QW s7'ͳYc**Il''dmGy5 cR4[%lX%>"+ֿ w]Ԓ4,)YRgwrK;1ff$I'&̟o&\> fk fh̟o&Q/h |}jqxvuV(m/çEoo YaX#1tB"eoc4]~v77c> xImgçk[Fp4TE\P? >!9>|S7C~ >&xbɂ[]'-o!nQ C3rԝAGZ~xs@7>tSqv\cv͌}k{2φٿuNstZ~ Ӝm'__Ǘa6&Ϧi/WgVx;g]->X8c|[]5]ZU=6`˺7BU##\?)F'<f, &qG6濳Mrk~8yT maB۴.RC4]@Z<㑹/ 3|_uuYXKp$eLbfq:V渎nc iVE5kME!ܥleʳ)1̟o&3 aRd{ta8ZTqaƯ7ny/7>WAK2φٿ<;/HKR]~9Dmh-u HngZEٴgO'Gkw7`Pn%qarR=ɤekBAQ7~^)^i<1kvp-lv!@N٠OxK_7H]^W ?]lM|52W>2TҚ{+hF2Z鉟>@Nz |/o|Ex̎Woq4zVyeVζă̍"\h_?LCoO_ҭKƽ[{9eȮJYN~aְg7_,@ xBm>oP[.g8)ƍҞ3u]ux߲$"o]?ֻM2OwDYmIld* DJ*9̝PÔöFửIYi#[ùY%kw;\'m/:}S]ϱs$E2p3ޛܛvM׿gW1˧Ix}\.=}wHoBXd[]nڧ"TuiKPuQU`*+{ Fѭf 4\ee␘Rv_NO[7Q:8iMVRVVNh^n{>ĺFuUI$/`?Ҡ?4fm-^K7_||iWO?. C?)k+蟔EOPO_"k,kN$JKShEK g^)%ED(!,GĝK -"%H!#:T4; E\S-I S*8%A:3Ưb[+*&8؜0/9 !3gM_į=gcW4.o0hۤ)X"rV.C?D޾Ǿ JOZuisګ7dXImǓZ/QEjfQEQEQEQEQEQEx'*xq?]i>o/s9oVLl|E9m}6y/塭tK/ >wos_m.պ1l(?!eBɻ ψ$8ּQOB ǥ=H ʏːLx%F6\ڏg4s_ڟؿ_gL7|*)Ly+*?uon4{+MΥc%PFDXe_>5jGV}}Ί _*hWhvf,jNXOҶ>>|?hZFϧYڍDdz·v !e$0<@kIcG,mv<*+ 8zQáo[i7x&KZܝ&?KdA_Qw+~ ණ>^]=Ǚx-v)(LÖ3nU7|^񖕩\kvAɤ}Su$3,|d@Uj ~ |t ;Q/Kkap.M(vZCc'p eGt{{L/GKת߈wZω4/wkUFG1Cu$(_t_P9&㿊qm5atڭ[b3ʟ>{)? ~!7MMNIA+IO ֮LqFQ HXWoPTg'ng%Zw~&?7 (((([> ,aF ,q;UF|/( T]za1)g׭fOK{Y`vO {m`l~^ ѼES|ǂ<x_Mx_X6mkqu=RpHpHdXsWJ_Ø1W}6q>1*<ªAQgݏC;Q@?Gx?>}(?PTy( T]@Suax| >{jPTy( T]@Suax| >{jPTy( T]@Suax| >{jPTy( T]@Suax| >{jPTy( T]@Suax| >{jPTy( T]@Suax| >{jPT|5~._! &?톏ifqoڭaQQr]$ge7_x񠱿d ŮGimo#pRi6|FpyQW6 66q>1*<ªAQgݏC;Q@( ~6^Hw|ss3k:Uko QI#ʼn  dOgC?xE[thڥ7 2::I" ; =Ox{(vG)Mqk_xt_>&Ӯ4JQ}x)cޒ+vV 3A"Q? Cn5jpM eV oR)`؆c2P Oh;~:.ΏZ,/R]{I,2଑3++1k?<#K>g|9 igL̑.Q7He=Ox{(vG= ;?G( T]za1lVQ dd~>IJݖi[s8}m.@{xF =ڧ?=?5'͔ncU&Gsf2pp ilt%)ͥ>so\9TgC>dɐ|EhѱEW.s1| _xy${ 'IO:C˛˺`xEqj>鷋9s20* u?( i_(k|M xBtzt+~/n 祼Gt`VG2΃oUw Y?GtVVooMq#r_/|'*o?4ׇ_ m7<5[|wIoQEQEQEQEQEQEQEQEQEMßU~oQRW:?य़nS} ?ߌJQEQEQEQEQEQEQE|RoW?>S~2*gGoJo%^<ӿ(s)Mdw}EP ~D? }_x'PQ#ԃQEQEQEMßU~oWMßU~oPRW:s)Mdw}@Q@Q@Q@Q@Q@Q@Q@Q@|+ׂ5K n^0՞&S%-n,I:>`9̟~ȟ?bO|'|j;6kGIWJm LSpWjq]Ӎr{윒~QjWk5}}<g-9'ľri ӯ}bLo" {I"x`Z7V{{ٯ^!|4!ۨ|iŝ>$jۮ&I"Ib[yxhRE{&Q^,wmYxMuo Gi)ҡʫ=Zv7:K[S,zj筴{4JQWGy!kW?w߲W j>W5eݝsEA{qM,Ms!]H7fEK|D$~>#/j {y4˛{gOZ LȖrfy\ɉ<; ?`Gj|4Sς|_7OQ{] |PҮC4!Ki]d׌S掭%ipRAIғjm⥮+*sNطzf? iZ֑Sn-iT\K%wm6JyiTCn)ӗ4T`* ( ( ( +R|2?| Ww_kJ^]b7 r឵G>*xŞndxtR JWFB̄0hZk<_OĿ$E'}];|Mi)zuHʆRBz[&> Ux 1| j=c6sZރ6eOKD.3l""C3G>*xŞndxtR JWFB̄0k[4 (Q@Q@xw?g</Y?7T6kzXy ]>ib#ʻq]뺳Z4sR/OvuzQE`hQEQEQEQEQEQEn/+Ox~.W~&G%0[jk"bg5{teh#@2kګŸM?O//-aS3ծ KfUBsy[Eȹ$=&Vkvݣz <M]KN{wB Oҏ=6MO_jv X" *)ڼ>x▵} Oum%W-.uM-"eI{dFtV.ieѿ<#Q~A7<_^\h_&9q&2ܒp9$xπ'gUI[R UJY#EK-nO.&DيqRr(O~׻m͇U_V(}|m>Ƅhڏh5;E0^#,.ɽw.W9U_//?bs?giz&Λ{ˬih5mUE $HdhuϖfM)JP撳hQEdhQEQEQEQE| W__~$4//jZ6oK1ŒFAbuO$mO_ |u'Ox?DӼsi66EW||DD],oX_ b(|+'}O_ژtCWV>~|A i&ivk_ ~ZmSUiWm}gKaz$كP\gN>J擾(PM_NiEw-QIw%o6OTw DFu?/=m6Ico mWUd;;y٠i'b\<) {o W7 xl_4[o h>)ƟiUɺKkk8Qb *D ۸218߈;OA.'Ou=>)e221 Ġ_.Ez-t+VwfՒ~_#ݧi|"KF)wbܳo1fW-e)~3\g%j3H(J h%DI~RꟴI=LڟO1~xHm-Y Em{si%lbI\8X߳_/5> lگ,qO tM_B4[HyDf WC/i5딡MœaGKvK}IuޮR1R~ٿ~'kſbxkAO4;MM^+[Y‹IR$hLFƹVvV^{%'hzQEAAEPwŞ{ttO {>Vkmz6GK_㆓^34񣮡eL&V7y7I,|4#],|Xj$,5 m|g@%դ2*XB_ˌ*2$UmR~?ӷC矅?J>-/^ƻx&HFHh!S!ڳ+lxǗߡS'ۻmz˲%dKEQR0((((((g*ni/o;\ks[^{3Ha]".UXU]x*|[yWP\%{-*O}:=QfM";ռsRJ^ҟϯ잞ㅞ9'rS :4WU%)|Aq -!؆}2 cPs7 Ӿ?)1>VO7.eXK{- ߞO__߇>Þ?𯄾h <_Kwpb #YEuaE )G5vH"/?_h>b69x' M<;~;τ$lҼ! ;R wմԷR<h ؄9.J%*Ro-cVnIl58[%Ss[k~!~Gt'~_kLM;Y̰׭Q/ܙL,c$FCt4@7"B]v__|h{bmSCռgԊ+u_gkEF*~,i<藞忮wu }?!^|Ŀ~2_ |B M>MIg욄I4Ҥ]ӳhp6Fž|}׌ I?`/{UΣx[~ׁiYG񗃵ijeӓKXO&;Z4i$kƟ4RQ9"+K䤜IY ˫n\Ͻ~MFuSVM%vPVW|G|K X\m'^Er"[kkUm⏃Nz4 գxbgl5/Z>4S=Ɠ\_1IvɺX$/t6ZE%ە-:sokVw);+[CA_5xh?N4Ū:uf>`NXcEdZ 'ڧ]xH/oh:Nn}ϘʷcsIk`?xx'Ey|_UwI|y/s|߿ ~lvi%7s %&V|#Լ' #k&7%O]Z_Pq}%xGVo=n+gkkY&EUA+)pGZ:i?п /bU^߷~8GZISަWMW_7' 2uä÷wM!YN׺کi ,@FY뀋 9J.x~:/'~k1j,>Cb{)lեI Ęxm0C[ I>/[&s=/"В٦r S2y7 rs~7W{yNp~*Qs^}?U//-uk+MjncSnt먳qL`0Aτ<EҼIZ޷-IܕY-B ϒ G[|_OC?k?|2~iNo]mjQ V%^ /|)xI]Bog6W?m'#V_H7Ǿ:5ƕ~ *^e^oOvGrLJ"@Թr;nM5K[Wreo^X^>wkw?>/ĝ.(ܶuq^Oozx,DßsqE >"tEב P:8  0t wLsƾ3/bo7ZW?g_w3m;jpK4Ewg4w=dYEx]_,|S[¶'j:uk$s[S"H[,@Y=y,|潮j,W7YgKjYmȱgU [#F@ڦ/<xbҮDe}-[݋;1ⶎ!0t?|O^~,'}fDҍ:ygu,ʹ0Ap+BY? 'ÿ lUQѼ iZhEj7IO,{20rǙAso׮oszƭ:'|G֚z 3XN=^\)|%W?IUGἷ6 Px^oۋ/_ MAj#PKXm"+m|DY?2_]ջ鶽%ߋ-޵kMsSy_[1ܠ k¼!'C.&x6ZmݓuwzLCKqLu mZ!$ѫ" +,V}kwΣV+[Ե(YdEVV(Y!YN3:CoYϤ|x]x&yΟW *EMV07|YkgYOk"SxK/DD."hi s=xa5?x2^&;{S7/mIQ|̮W駯,h_-ZL|Ey:t}wQ>7 )$mMsl-ʹ寜|Xk|CCGKNfie,&1\27?oO}-lzy/kR#}+PYOb!5qpLV7wZ,eXmx&H`Uc|uOi࿊<)Үt.#]m$QݔEDK+nZ=I'k~-ϝ/(Z2:xI jyF]˿ԭyLFNXP 4OֵMGU-;bq^{VWދ,IV 1|3C'}VƗw?-Cu>8%/ZKT7GUWGo%_nɽ?<7RPۿl}bjn]G Q'/ ai~&ox OŬ铇Y>ЬBla_'7l~ YZ|;񾯤$3jww|pkKkyO%_fz)?񳽛U,Qwi?jy~ }K@G,BeNGz_z)rC$_ |GV>:𦙭s3XV2*F(2X0w^>S8 ա63hoGXQCY!C}ekkp‹0DETP0w }?j>Svin%|o6q͕mxJIi.'OLUp`$7^y/+IYzծoپNia|݅ǙnUQr+oGkQy?'g߄o!xgOmR,4-̐G:XZ4&#>T9ܫ| *𶅡\[ưWZ^KUl]U㷵Y%X2Ht2cyF-;DoA}Ś|x: |1\Ga,CwbmK ɞ`da|-/?c߈ j|C׮|AxB}SJWմԲ(>fA fmC%kVM|ioE:}~mAT]7/s_4oYYi'Pb a8ߞYR0PvWTwzAy 1Fc OudPH(#uf_ㆿ#hw_Mg[ 纈DȧtK,L%IE s'kikZw^U?g=oޟg"H?3/??Mhw` Mx(W_~2xbd{8F]ob2+ bƗ7f^•|38? W  vZ[B5]>X]52+Fd,g=2:/zUj'fBI7?Q߃?OY[hZGuD@Yԓ^MK~֐| j6c Rf F*zʵB&puB$w=[F 52z/z+7?QcG]օy%}SNf1{RA=Fr0yO +׷d"kRɻvx;j1rj? ?zC<ƿ 4kS L[WPI)S`Gh;*r@>__EMKPuip4k8o?+;HH:qp,u ,{͗#nM_gdM;)`:R)%mޚ+}zC? Kް|_⻭3> vEorA#WRFˎUfNTl57?G#SH3|YŪ?- {2b1 ) 9)jQ[^5[cXl@8-bD$0#|Lka2̫YRnQmJc ]aw7{XC? ?zC|ccᧈhMu~=;? ?zC𯖾>~NAECTlHID[Yb ~'soOxwᇈ )|.M&e8 bJU;k[ngŽ1JTo=Uml=M=!W7U/ _.Z[\,^m1!gz+Ux~'}\Jb@dk1͵y3`u]XǚUO[{ͯ,t9K߫GM=!Q|=Ɵ~VZwOo>Ԯmsms(s I1 W+~n"hd[sdEm8Uq׼nצ;0ս9\|f#SH357?\?_w~KDWQ,rǜqۃ\mſx W/0aȮhY+E;M=!Qx-WO k,ma#^㿊0uKjZ'TDV(| s*vG57?P_i2i0')_?:σ^"eaF`/ePCpA.wk?ΩH'MhQ+p (:֟P)矴W q~$ k_}P5_ }9XU5B@-E{O:&mo;xlI^X8f)>s1 ?]lM|opw"tZ]VkgɹSoOO{㧄moښ|>*k % :|nI,Q[[oz+mn,ndإvETVf='s'k-2φٿ/!)QW s7'ͳYc**Il''dmGy5 cR4[%lX%>"+ֿ w]Ԓ4,)YRgwrK;1ff$I'&̟o&\> fk fh̟o&Q/h |}jqxvuV(m/çEoo YaX#1tB"eoc4]~v77c> xImgçk[Fp4TE\P? >!9>|S7C~ >&xbɂ[]'-o!nQ C3rԝAGZ~xs@7>tSqv\cv͌}k{2φٿuNstZ~ Ӝm'__Ǘa6&Ϧi/WgVx;g]->X8c|[]5]ZU=6`˺7BU##\?)F'<f, &qG6濳Mrk~8yT maB۴.RC4]@Z<㑹/ 3|_uuYXKp$eLbfq:V渎nc iVE5kME!ܥleʳ)1̟o&3 aRd{ta8ZTqaƯ7ny/7>WAK2φٿ<;/HKR]~9Dmh-u HngZEٴgO'Gkw7`Pn%qarR=ɤekBAQ7~^)^i<1kvp-lv!@N٠OxK_7H]^W ?]lM|52W>2TҚ{+hF2Z鉟>@Nz |/o|Ex̎Woq4zVyeVζă̍"\h_?LCoO_ҭKƽ[{9eȮJYN~aְg7_,@ xBm>oP[.g8)ƍҞ3u]ux߲$"o]?ֻM2OwDYmIld* DJ*9̝PÔöFửIYi#[ùY%kw;\'m/:}S]ϱs$E2p3ޛܛvM׿gW1˧Ix}\.=}wHoBXd[]nڧ"TuiKPuQU`*+{ Fѭf 4\ee␘Rv_NO[7Q:8iMVRVVNh^n{>ĺFuUI$/`?Ҡ?4fm-^K7_||iWO?. C?)k+蟔EOPO_"k,kN$JKShEK g^)%ED(!,GĝK -"%H!#:T4; E\S-I S*8%A:3Ưb[+*&8؜0/9 !3gM_į=gcW4.o0hۤ)X"rV.C?D޾Ǿ JOZuisګ7dXImǓZ/QEjfQEQEQEQEQEQEx'*xq?]i>o/s9oVLl|E9m}6y/塭tK/ >wos_m.պ1l(?!eBɻ ψ$8ּQOB ǥ=H ʏːLx%F6\ڏg4s_ڟؿ_gL7|*)Ly+*?uon4{+MΥc%PFDXe_>5jGV}}Ί _*hWhvf,jNXOҶ>>|?hZFϧYڍDdz·v !e$0<@kIcG,mv<*+ 8zQáo[i7x&KZܝ&?KdA_Qw+~ ණ>^]=Ǚx-v)(LÖ3nU7|^񖕩\kvAɤ}Su$3,|d@Uj ~ |t ;Q/Kkap.M(vZCc'p eGt{{L/GKת߈wZω4/wkUFG1Cu$(_t_P9&㿊qm5atڭ[b3ʟ>{)? ~!7MMNIA+IO ֮LqFQ HXWoPTg'ng%Zw~&?7 (((([> ,aF ,q;UF|/( T]za1)g׭fOK{Y`vO {m`l~^ ѼES|ǂ<x_Mx_X6mkqu=RpHpHdXsWJ_Ø1W}6q>1*<ªAQgݏC;Q@?Gx?>}(?PTy( T]@Suax| >{jPTy( T]@Suax| >{jPTy( T]@Suax| >{jPTy( T]@Suax| >{jPTy( T]@Suax| >{jPTy( T]@Suax| >{jPT|5~._! &?톏ifqoڭaQQr]$ge7_x񠱿d ŮGimo#pRi6|FpyQW6 66q>1*<ªAQgݏC;Q@( ~6^Hw|ss3k:Uko QI#ʼn  dOgC?xE[thڥ7 2::I" ; =Ox{(vG)Mqk_xt_>&Ӯ4JQ}x)cޒ+vV 3A"Q? Cn5jpM eV oR)`؆c2P Oh;~:.ΏZ,/R]{I,2଑3++1k?<#K>g|9 igL̑.Q7He=Ox{(vG= ;?G( T]za1lVQ dd~>IJݖi[s8}m.@{xF =ڧ?=?5'͔ncU&Gsf2pp ilt%)ͥ>so\9TgC>dɐ|EhѱEW.s1| _xy${ 'IO:C˛˺`xEqj>鷋9s20* u?( i_(k|M xBtzt+~/n 祼Gt`VG2΃oUw Y?GtVVooMq#r_/|'*o?4ׇ_ m7<5[|wIoQEQEQEQEQEQEQEQEQEMßU~oQRW:?य़nS} ?ߌJQEQEQEQEQEQEQE|RoW?>S~2*gGoJo%^<ӿ(s)Mdw}EP ~D? }_x'PQ#ԃQEQEQEMßU~oWMßU~oPRW:s)Mdw}@Q@Q@Q@Q@Q@Q@Q@Q@|+ׂ5K n^0՞&S%-n,I:>`9̟~ȟ?bO|'|j;6kGIWJm LSpWjq]Ӎr{윒~QjWk5}}<g-9'ľri ӯ}bLo" {I"x`Z7V{{ٯ^!|4!ۨ|iŝ>$jۮ&I"Ib[yxhRE{&Q^,wmYxMuo Gi)ҡʫ=Zv7:K[S,zj筴{4JQWGy!kW?w߲W j>W5eݝsEA{qM,Ms!]H7fEK|D$~>#/j {y4˛{gOZ LȖrfy\ɉ<; ?`Gj|4Sς|_7OQ{] |PҮC4!Ki]d׌S掭%ipRAIғjm⥮+*sNطzf? iZ֑Sn-iT\K%wm6JyiTCn)ӗ4T`* ( ( ( +R|2?| Ww_kJ^]b7 r឵G>*xŞndxtR JWFB̄0hZk<_OĿ$E'}];|Mi)zuHʆRBz[&> Ux 1| j=c6sZރ6eOKD.3l""C3G>*xŞndxtR JWFB̄0k[4 (Q@Q@xw?g</Y?7T6kzXy ]>ib#ʻq]뺳Z4sR/OvuzQE`hQEQEQEQEQEQEn/+Ox~.W~&G%0[jk"bg5{teh#@2kګŸM?O//-aS3ծ KfUBsy[Eȹ$=&Vkvݣz <M]KN{wB Oҏ=6MO_jv X" *)ڼ>x▵} Oum%W-.uM-"eI{dFtV.ieѿ<#Q~A7<_^\h_&9q&2ܒp9$xπ'gUI[R UJY#EK-nO.&DيqRr(O~׻m͇U_V(}|m>Ƅhڏh5;E0^#,.ɽw.W9U_//?bs?giz&Λ{ˬih5mUE $HdhuϖfM)JP撳hQEdhQEQEQEQE| W__~$4//jZ6oK1ŒFAbuO$mO_ |u'Ox?DӼsi66EW||DD],oX_ b(|+'}O_ژtCWV>~|A i&ivk_ ~ZmSUiWm}gKaz$كP\gN>J擾(PM_NiEw-QIw%o6OTw DFu?/=m6Ico mWUd;;y٠i'b\<) {o W7 xl_4[o h>)ƟiUɺKkk8Qb *D ۸218߈;OA.'Ou=>)e221 Ġ_.Ez-t+VwfՒ~_#ݧi|"KF)wbܳo1fW-e)~3\g%j3H(J h%DI~RꟴI=LڟO1~xHm-Y Em{si%lbI\8X߳_/5> lگ,qO tM_B4[HyDf WC/i5딡MœaGKvK}IuޮR1R~ٿ~'kſbxkAO4;MM^+[Y‹IR$hLFƹVvV^{%'hzQEAAEPwŞ{ttO {>Vkmz6GK_㆓^34񣮡eL&V7y7I,|4#],|Xj$,5 m|g@%դ2*XB_ˌ*2$UmR~?ӷC矅?J>-/^ƻx&HFHh!S!ڳ+lxǗߡS'ۻmz˲%dKEQR0((((((g*ni/o;\ks[^{3Ha]".UXU]x*|[yWP\%{-*O}:=QfM";ռsRJ^ҟϯ잞ㅞ9'rS :4WU%)|Aq -!؆}2 cPs7 Ӿ?)1>VO7.eXK{- ߞO__߇>Þ?𯄾h <_Kwpb #YEuaE )G5vH"/?_h>b69x' M<;~;τ$lҼ! ;R wմԷR<h ؄9.J%*Ro-cVnIl58[%Ss[k~!~Gt'~_kLM;Y̰׭Q/ܙL,c$FCt4@7"B]v__|h{bmSCռgԊ+u_gkEF*~,i<藞忮wu }?!^|Ŀ~2_ |B M>MIg욄I4Ҥ]ӳhp6Fž|}׌ I?`/{UΣx[~ׁiYG񗃵ijeӓKXO&;Z4i$kƟ4RQ9"+K䤜IY ˫n\Ͻ~MFuSVM%vPVW|G|K X\m'^Er"[kkUm⏃Nz4 գxbgl5/Z>4S=Ɠ\_1IvɺX$/t6ZE%ە-:sokVw);+[CA_5xh?N4Ū:uf>`NXcEdZ 'ڧ]xH/oh:Nn}ϘʷcsIk`?xx'Ey|_UwI|y/s|߿ ~lvi%7s %&V|#Լ' #k&7%O]Z_Pq}%xGVo=n+gkkY&EUA+)pGZ:i?п /bU^߷~8GZISަWMW_7' 2uä÷wM!YN׺کi ,@FY뀋 9J.x~:/'~k1j,>Cb{)lեI Ęxm0C[ I>/[&s=/"В٦r S2y7 rs~7W{yNp~*Qs^}?U//-uk+MjncSnt먳qL`0Aτ<EҼIZ޷-IܕY-B ϒ G[|_OC?k?|2~iNo]mjQ V%^ /|)xI]Bog6W?m'#V_H7Ǿ:5ƕ~ *^e^oOvGrLJ"@Թr;nM5K[Wreo^X^>wkw?>/ĝ.(ܶuq^Oozx,DßsqE >"tEב P:8  0t wLsƾ3/bo7ZW?g_w3m;jpK4Ewg4w=dYEx]_,|S[¶'j:uk$s[S"H[,@Y=y,|潮j,W7YgKjYmȱgU [#F@ڦ/<xbҮDe}-[݋;1ⶎ!0t?|O^~,'}fDҍ:ygu,ʹ0Ap+BY? 'ÿ lUQѼ iZhEj7IO,{20rǙAso׮oszƭ:'|G֚z 3XN=^\)|%W?IUGἷ6 Px^oۋ/_ MAj#PKXm"+m|DY?2_]ջ鶽%ߋ-޵kMsSy_[1ܠ k¼!'C.&x6ZmݓuwzLCKqLu mZ!$ѫ" +,V}kwΣV+[Ե(YdEVV(Y!YN3:CoYϤ|x]x&yΟW *EMV07|YkgYOk"SxK/DD."hi s=xa5?x2^&;{S7/mIQ|̮W駯,h_-ZL|Ey:t}wQ>7 )$mMsl-ʹ寜|Xk|CCGKNfie,&1\27?oO}-lzy/kR#}+PYOb!5qpLV7wZ,eXmx&H`Uc|uOi࿊<)Үt.#]m$QݔEDK+nZ=I'k~-ϝ/(Z2:xI jyF]˿ԭyLFNXP 4OֵMGU-;bq^{VWދ,IV 1|3C'}VƗw?-Cu>8%/ZKT7GUWGo%_nɽ?<7RPۿl}bjn]G Q'/ ai~&ox OŬ铇Y>ЬBla_'7l~ YZ|;񾯤$3jww|pkKkyO%_fz)?񳽛U,Qwi?jy~ }K@G,BeNGz_z)rC$_ |GV>:𦙭s3XV2*F(2X0w^>S8 ա63hoGXQCY!C}ekkp‹0DETP0w }?j>Svin%|o6q͕mxJIi.'OLUp`$7^y/+IYzծoپNia|݅ǙnUQr+oGkQy?'g߄o!xgOmR,4-̐G:XZ4&#>T9ܫ| *𶅡\[ưWZ^KUl]U㷵Y%X2Ht2cyF-;DoA}Ś|x: |1\Ga,CwbmK ɞ`da|-/?c߈ j|C׮|AxB}SJWմԲ(>fA fmC%kVM|ioE:}~mAT]7/s_4oYYi'Pb a8ߞYR0PvWTwzAy 1Fc OudPH(#uf_ㆿ#hw_Mg[ 纈DȧtK,L%IE s'kikZw^U?g=oޟg"H?3/??Mhw` Mx(W_~2xbd{8F]ob2+ bƗ7f^•|38? W  vZ[B5]>X]52+Fd,g=2:/zUj'fBI7?Q߃?OY[hZGuD@Yԓ^MK~֐| j6c Rf F*zʵB&puB$w=[F 52z/z+7?QcG]օy%}SNf1{RA=Fr0yO +׷d"kRɻvx;j1rj? ?zC<ƿ 4kS L[WPI)S`Gh;*r@>__EMKPuip4k8o?+;HH:qp,u ,{͗#nM_gdM;)`:R)%mޚ+}zC? Kް|_⻭3> vEorA#WRFˎUfNTl57?G#SH3|YŪ?- {2b1 ) 9)jQ[^5[cXl@8-bD$0#|Lka2̫YRnQmJc ]aw7{XC? ?zC|ccᧈhMu~=;? ?zC𯖾>~NAECTlHID[Yb ~'soOxwᇈ )|.M&e8 bJU;k[ngŽ1JTo=Uml=M=!W7U/ _.Z[\,^m1!gz+Ux~'}\Jb@dk1͵y3`u]XǚUO[{ͯ,t9K߫GM=!Q|=Ɵ~VZwOo>Ԯmsms(s I1 W+~n"hd[sdEm8Uq׼nצ;0ս9\|f#SH357?\?_w~KDWQ,rǜqۃ\mſx W/0aȮhY+E;M=!Qx-WO k,ma#^㿊0uKjZ'TDV(| s*vG57?P_i2i0')_?:σ^"eaF`/ePCpA.wk?ΩH'MhQ+p (:֟P)矴W q~$ k_}P5_ }9XU5B@-E{O:&mo;xlI^X8f)>s1 ?]lM|opw"tZ]VkgɹSoOO{㧄moښ|>*k % :|nI,Q[[oz+mn,ndإvETVf='s'k-2φٿ/!)QW s7'ͳYc**Il''dmGy5 cR4[%lX%>"+ֿ w]Ԓ4,)YRgwrK;1ff$I'&̟o&\> fk fh̟o&Q/h |}jqxvuV(m/çEoo YaX#1tB"eoc4]~v77c> xImgçk[Fp4TE\P? >!9>|S7C~ >&xbɂ[]'-o!nQ C3rԝAGZ~xs@7>tSqv\cv͌}k{2φٿuNstZ~ Ӝm'__Ǘa6&Ϧi/WgVx;g]->X8c|[]5]ZU=6`˺7BU##\?)F'<f, &qG6濳Mrk~8yT maB۴.RC4]@Z<㑹/ 3|_uuYXKp$eLbfq:V渎nc iVE5kME!ܥleʳ)1̟o&3 aRd{ta8ZTqaƯ7ny/7>WAK2φٿ<;/HKR]~9Dmh-u HngZEٴgO'Gkw7`Pn%qarR=ɤekBAQ7~^)^i<1kvp-lv!@N٠OxK_7H]^W ?]lM|52W>2TҚ{+hF2Z鉟>@Nz |/o|Ex̎Woq4zVyeVζă̍"\h_?LCoO_ҭKƽ[{9eȮJYN~aְg7_,@ xBm>oP[.g8)ƍҞ3u]ux߲$"o]?ֻM2OwDYmIld* DJ*9̝PÔöFửIYi#[ùY%kw;\'m/:}S]ϱs$E2p3ޛܛvM׿gW1˧Ix}\.=}wHoBXd[]nڧ"TuiKPuQU`*+{ Fѭf 4\ee␘Rv_NO[7Q:8iMVRVVNh^n{>ĺFuUI$/`?Ҡ?4fm-^K7_||iWO?. C?)k+蟔EOPO_"k,kN$JKShEK g^)%ED(!,GĝK -"%H!#:T4; E\S-I S*8%A:3Ưb[+*&8؜0/9 !3gM_į=gcW4.o0hۤ)X"rV.C?D޾Ǿ JOZuisګ7dXImǓZ/QEjfQEQEQEQEQEQEx'*xq?]i>o/s9oVLl|E9m}6y/塭tK/ >wos_m.պ1l(?!eBɻ ψ$8ּQOB ǥ=H ʏːLx%F6\ڏg4s_ڟؿ_gL7|*)Ly+*?uon4{+MΥc%PFDXe_>5jGV}}Ί _*hWhvf,jNXOҶ>>|?hZFϧYڍDdz·v !e$0<@kIcG,mv<*+ 8zQáo[i7x&KZܝ&?KdA_Qw+~ ණ>^]=Ǚx-v)(LÖ3nU7|^񖕩\kvAɤ}Su$3,|d@Uj ~ |t ;Q/Kkap.M(vZCc'p eGt{{L/GKת߈wZω4/wkUFG1Cu$(_t_P9&㿊qm5atڭ[b3ʟ>{)? ~!7MMNIA+IO ֮LqFQ HXWoPTg'ng%Zw~&?7 (((([> ,aF ,q;UF|/( T]za1)g׭fOK{Y`vO {m`l~^ ѼES|ǂ<x_Mx_X6mkqu=RpHpHdXsWJ_Ø1W}6q>1*<ªAQgݏC;Q@?Gx?>}(?PTy( T]@Suax| >{jPTy( T]@Suax| >{jPTy( T]@Suax| >{jPTy( T]@Suax| >{jPTy( T]@Suax| >{jPTy( T]@Suax| >{jPT|5~._! &?톏ifqoڭaQQr]$ge7_x񠱿d ŮGimo#pRi6|FpyQW6 66q>1*<ªAQgݏC;Q@( ~6^Hw|ss3k:Uko QI#ʼn  dOgC?xE[thڥ7 2::I" ; =Ox{(vG)Mqk_xt_>&Ӯ4JQ}x)cޒ+vV 3A"Q? Cn5jpM eV oR)`؆c2P Oh;~:.ΏZ,/R]{I,2଑3++1k?<#K>g|9 igL̑.Q7He=Ox{(vG= ;?G( T]za1lVQ dd~>IJݖi[s8}m.@{xF =ڧ?=?5'͔ncU&Gsf2pp ilt%)ͥ>so\9TgC>dɐ|EhѱEW.s1| _xy${ 'IO:C˛˺`xEqj>鷋9s20* u?( i_(k|M xBtzt+~/n 祼Gt`VG2΃oUw Y?GtVVooMq#r_/|'*o?4ׇ_ m7<5[|wIoQEQEQEQEQEQEQEQEQEMßU~oQRW:?य़nS} ?ߌJQEQEQEQEQEQEQE|RoW?>S~2*gGoJo%^<ӿ(s)Mdw}EP ~D? }_x'PQ#ԃQEQEQEMßU~oWMßU~oPRW:s)Mdw}@Q@Q@Q@Q@Q@Q@Q@Q@|+ׂ5K n^0՞&S%-n,I:>`9̟~ȟ?bO|'|j;6kGIWJm LSpWjq]Ӎr{윒~QjWk5}}<g-9'ľri ӯ}bLo" {I"x`Z7V{{ٯ^!|4!ۨ|iŝ>$jۮ&I"Ib[yxhRE{&Q^,wmYxMuo Gi)ҡʫ=Zv7:K[S,zj筴{4JQWGy!kW?w߲W j>W5eݝsEA{qM,Ms!]H7fEK|D$~>#/j {y4˛{gOZ LȖrfy\ɉ<; ?`Gj|4Sς|_7OQ{] |PҮC4!Ki]d׌S掭%ipRAIғjm⥮+*sNطzf? iZ֑Sn-iT\K%wm6JyiTCn)ӗ4T`* ( ( ( +R|2?| Ww_kJ^]b7 r឵G>*xŞndxtR JWFB̄0hZk<_OĿ$E'}];|Mi)zuHʆRBz[&> Ux 1| j=c6sZރ6eOKD.3l""C3G>*xŞndxtR JWFB̄0k[4 (Q@Q@xw?g</Y?7T6kzXy ]>ib#ʻq]뺳Z4sR/OvuzQE`hQEQEQEQEQEQEn/+Ox~.W~&G%0[jk"bg5{teh#@2kګŸM?O//-aS3ծ KfUBsy[Eȹ$=&Vkvݣz <M]KN{wB Oҏ=6MO_jv X" *)ڼ>x▵} Oum%W-.uM-"eI{dFtV.ieѿ<#Q~A7<_^\h_&9q&2ܒp9$xπ'gUI[R UJY#EK-nO.&DيqRr(O~׻m͇U_V(}|m>Ƅhڏh5;E0^#,.ɽw.W9U_//?bs?giz&Λ{ˬih5mUE $HdhuϖfM)JP撳hQEdhQEQEQEQE| W__~$4//jZ6oK1ŒFAbuO$mO_ |u'Ox?DӼsi66EW||DD],oX_ b(|+'}O_ژtCWV>~|A i&ivk_ ~ZmSUiWm}gKaz$كP\gN>J擾(PM_NiEw-QIw%o6OTw DFu?/=m6Ico mWUd;;y٠i'b\<) {o W7 xl_4[o h>)ƟiUɺKkk8Qb *D ۸218߈;OA.'Ou=>)e221 Ġ_.Ez-t+VwfՒ~_#ݧi|"KF)wbܳo1fW-e)~3\g%j3H(J h%DI~RꟴI=LڟO1~xHm-Y Em{si%lbI\8X߳_/5> lگ,qO tM_B4[HyDf WC/i5딡MœaGKvK}IuޮR1R~ٿ~'kſbxkAO4;MM^+[Y‹IR$hLFƹVvV^{%'hzQEAAEPwŞ{ttO {>Vkmz6GK_㆓^34񣮡eL&V7y7I,|4#],|Xj$,5 m|g@%դ2*XB_ˌ*2$UmR~?ӷC矅?J>-/^ƻx&HFHh!S!ڳ+lxǗߡS'ۻmz˲%dKEQR0((((((g*ni/o;\ks[^{3Ha]".UXU]x*|[yWP\%{-*O}:=QfM";ռsRJ^ҟϯ잞ㅞ9'rS :4WU%)|Aq -!؆}2 cPs7 Ӿ?)1>VO7.eXK{- ߞO__߇>Þ?𯄾h <_Kwpb #YEuaE )G5vH"/?_h>b69x' M<;~;τ$lҼ! ;R wմԷR<h ؄9.J%*Ro-cVnIl58[%Ss[k~!~Gt'~_kLM;Y̰׭Q/ܙL,c$FCt4@7"B]v__|h{bmSCռgԊ+u_gkEF*~,i<藞忮wu }?!^|Ŀ~2_ |B M>MIg욄I4Ҥ]ӳhp6Fž|}׌ I?`/{UΣx[~ׁiYG񗃵ijeӓKXO&;Z4i$kƟ4RQ9"+K䤜IY ˫n\Ͻ~MFuSVM%vPVW|G|K X\m'^Er"[kkUm⏃Nz4 գxbgl5/Z>4S=Ɠ\_1IvɺX$/t6ZE%ە-:sokVw);+[CA_5xh?N4Ū:uf>`NXcEdZ 'ڧ]xH1&IDATx^ |Tda  H`EʦE‹khQqBՂmm,Ԩ%* %HkmL2>w;ɽ3OH9ssνsd~tOLH#xڷ[v`2uvBk$'i KUc@'Q /UWmhsctvq e0%&&45V-p1Gد#.De(f6`~sj ]{k5TM|[1oܭ\kH]!zsKǞ㪂'\zf_x߳/mcRQ/@$mFzh>lmmm%?']OoY7ffh'162ŬUq~.'nN住5$؈, '`[ɲX6[buNk scSh7_Ӧmk7r>{NDnD; @@ .\5BNSoOgkXY*fr}zOpk܈ҭݎt T  l+vȖ/Z/s-/WtuZw'?|ԤDiE^\\yB:ê ZdH ilZ[k15 G5ck>QQy @@@Utc V ۃ.O9;Ί s@@{\3X  ( AeggG^%A@e(++raNEّB @G@d')*PKqh@@vU:'*b g gezEvպ SfE;ٛQWl5H]%.riB֥D_+;NggU&yZrLDcn%S}lM-u/W*]z9 P0K *sZ*\V4K8_IQU8/?c0whT@@rZF+E^{Z+*t{TĔO@U[fL;ws+KJarD},]_R<&kd,C8#KYΕlߝhI~/ WM H~#Pc4U.NR|Y֡/w[`'"/jq-X:dgf[Z/cS%ou sd6#gil=˷"7{.(ٯ *u袋u a\>t(y'7I3>AC֠pUU7mܟ:⚢p3;}C3"1D!?,;o1ǶϷ˜f6V);w2k %.LʎSei\W7+medf ={ 82v T}RV~9F\66,,b(+å .KORY!% $WBZ=[x13OqRHT0J2!"b?`] v/]ΪXh{g]0Go2'[)v/gS:ԗ>"kϐ(4{֡ߋޭ|1-H SD "墝v'(f4Ξ= n$ga) iӱzݏbص.y/{Yz)״$}LÂ,Nld0uMD WK^X@,ri^.F9Yu[ʿ >-(<.}6OQz\Y % jϥ^f_"?"eDFT=mmDP4@@\8!(sE- 2u*eS#!PCj1 @ 0ت|^zVb7\pADn@Cur}]D):TWWCբgB>}h4(\ ywgCqMrib߿>o'/^t%ΩvU\nm}5+&.+yگ^d`mdRjso'ӄ`!Kr} 6Ж;1Q;> JH/Ji`55z}?l7?Ê^Z~r!׉-8lڢE QsZftVY!Iɴ9,]áO641gfX>î|.Yvy *ϥې>G}TWW7sT&!!!)))ⲧNSSSo6uz;~xrrs=G^ GB駟W\qEϞ=#[D^bd4+ɖpw_w=xSC$f K-\B!|m81J#W2(E m q>+5d#e:C/xFѣG?^{o۟7|sǎ2FT UERT95A EeXz!.UTT}R0H9眉'v]M/)U y‹7|KY~ T6aC~PF b0KAO:Gtڵk׬YS\\8f̘ &L4)33bٲe555- 2S*Hũ*jrjə3g+ +Dĉ!illohhhw%xmd6%_p~2¦Td*rR}HJ-ZFЇR:[n)))YݻQV\Iy|elPA*NPUT!UKy j(\"M6mq6p/a ~'_|ESSSjEQhT~2~DĹ6275s[k] 3֎rQ1Z< TUX(vن]G׮]y=[ KBGP"}hZmٲ믿ꫯHzCo tޢ 2S*Hũ6*/$DfS^'-yl>rWc.jÕ9$.z4Khrk#+J|0X;2FR  w.d6fĈAz 6ӪVyH_H-w"E޽;Ӣoկ~u>|6W.[Qf*B8UBUk91t'('rB)DQhNNzᴵ;w (ijOҬdԽn 'U6sGGT(Q q\~gzf-!R+9dȐ;K/3ge]F|͠elPAw%#P.24wȑb$-&'b&$JnzM^=v=PnmK9\Y]Y`^߬}s9\?Ę'@_MeZ;2t'̤Ɂ}+úH|qEz2P6LE Jwq|[a?sɛN%/'[8p`XUu${HǻONwQz«#@YrU>?;|ת}\?}v>No %bBH}%嬳"qLCzI^"IEzD(3T*iW-5j*'?vIɓ'*Z,xy"׏+77+R,d'&+\19w s@}τ% I3%9,*Hot%7to])OFϋGI* ʋ/HJJf޽#+%G.+% JD_ғ[}]RzʮzSΐrEP0QW/N=@㑨\TGݖUE^+AA?þyW@9SG^Hoj}|{-Hrf(a)WAA9\#F\è~ M^פ>nfM_ϰK|^G4;VqDvP@i&IM%o"IlQEKZ 6#O% S\exT q…Ub̀ryt`P. riob w1 нE:ȁ7?l@@T2-b)o͟'X;@@D :&!7JE!v&^CCA(P~; eCmv{ŝ"[?guu p^~V(OTq䆐]^kcqO +7ި=~煶:v Qgwj&srFU˲2 \04r +bB~X?i|.t y;Qk>xFh`fL;Mv>+xvicUrj0~ nO-@z[ 0[]$U`R۰(\<^-(24恢[VWie9WcKO jV>|p~eUDR.v@ D0%Xʟdfp QQ~>6" "$rqA!qbPZi!{${f[jO؏!^ +0|?TjydEr MVȤ\-J%c|r2eUN}w ιeh }{PM~έt'ioq/x Q8ܑ͂x NoxB6n.ϛC[b1'8)C 6Đ#*(=e:lI+dg_riK7 Vhkoe-2\|Eה3'/ Q Ь`?) o 9BrуFG]??<Á?ѫÆl6vJ ]{\Hro'+V2H6 9m;/dmp 9rd1Z8U}O?[M5#q| ֓0pQ_XTϬ`?1 rx =繞|rSSml7۳k>.\m^Ga4xKh\A({|..gg8ıG+v}h i<>" /?ou sIc\3l$\o1Z=cK@@ ĸ"K=>$K=K@@ 8 =P.,rahK{cA\ #ޘb((\އ7kjr8yAʥ@a<;6+N_UX] (bV.F?ޡhi%"hG#6OY2f1,hM+dbq}C}V.,wv22J˖4ш͓")`\N/+9k"K +o 3I66S.f?vK?B~}9}3C<#q`MC5?nٲeڶ}5 +|Cv)ã<;J I-3~͜9WwQzwt-wȋp&JFL-fLTۙ {>xꞼu6)CB)@3q\ [0} #vbytvdos9(JdS1NSGk\U. YUb{ii7O4wIJp:ѫ@" (9SNzY:]Iw49n9rfk>C`jl^/ XЬW·Vڭo͆ ӻkv\eYKTe( jD*Г\[I{tM.I&cۙ:ttƇOl>lf}w@*V`/`mOzq qrR~>M;eTF\!;1W+ 9ޛVB X׽1?-"QJ?9jKΤDGbB[jMJf$:G5>|Q׃dgHz*3Ce]}VG7,Z:h/l|QYAJ^WewϬRqP[]e\y)m{kJt fj;Z\>x%VC}?.^_&ꗾ r2W{՝ٯZZqſh^9]?J? .ȮIVaƜﱇa{UuG]Yxǰ}zֵP*@8!׎yF<\\n6;NV[l9f0Yl;iLM~z~V8$'>i/oumџn6o"1qm9<6%(Qwj/pm& v*v0  U6%w5s^h]]<9c XC'P_=B2*fxmcc~^z>ܘ3[ s COWT"ZU~"X8wa{riر(5X[_w<{nMk֔DžGD 7dh[Ǐ|fp˽v}1n.Db~~ akOp:X-F\)}.`i6%)[Tѕ7m󶊷0q${2v# ٿ_{G#1eCT[}RZ[  oխ|~t=8\-pZw߷zm*Z<|OUʒ7b?JMWe?~y0}`{vJboY}o,W{pvn-NY(H4K?8o .)Z\> >g:H-tdi1JO.XY}RROp >xOMPIz \>.lݺ:#- x{`>II*D9EMbIc(_AG_`7ؽ*#\Wq\0mvu:բډrc9zCbsx"  sS۫qÇHZ>᪻?7߲gtݤ!_W5K޾iɧ 3ڴ|ϯ>xIEt_wO:s d7|.UN6|)ua g-sH zhͦΆ3QU5Wa +~_~WD5I$[H˾Y}7|3}Q6zlbtբƚ&dbϳ)\DK)V"(drrEatфs˳#A>WGʂt(WpG !=^\;L&_[s_$''uh5*̒>zٺyk&$7n~Ӎ}zhIxP.̄Q.G$s0e4Y׺ !O?;沋.֣{7ꢄʑE{\QewaA$(W[[kZڠL3 uיښztuA;5rf͟?kS63ù]AeS͘^Ţ]a1\3gtDAּ߯=UPfzѿOOՇECUK֥ 77"0pYQvthYy| f(ero>OWǑ/}&H 9ܗiף^PP.>ukSw˱+uDCxbٽGDUeἑ)3Jb2%{COh.疊\/c:-Ny}xAlL3 ڻ݋.޳gOᛵUU5~&aK:17}rnv־׳ Kg 17OCr4n͸wZ2 e(؏l޵kҥK;R[ZO\e1#s_r4nbB&)I`lMDWk̙:XUuxӌvQqV+;qѺ1vٵB%ydv,zu>S^6KW_|NsرcڵO߭[tȑ--vzn$wR~к)իm"!y.L8z+|5f̘왒BzUYyPycI{t4hЃ>?>mڴB \g}!C]0"ZS /u0z RZ%W;HEV.!ﴮa9`(Eyj1GC/$Z\J(Wa5Ꙩ@@*(TR P.,J@}.D:ɇH \ءWlT%ءWACK@d }. (rE8K@ \Q@d "2-$?Qk?H  EvOIw={:WCG3K| rk0j&%Z >O,P@KEaAt\F~1F*6l@]4[as DA"7%A@/+N<@~g$0H  @˽E[񥅳arPT$J)qʞ,b{8K=󹂌E|.X #\12(W\ 7: 1B#n@\rp #\12(W\ 7: 1B@o߾ !EL y toV(Mb@_sL\w5fF*F ;)*ALV\i%&@)ի[E @wRaf +r[.&  2Lx*BsFxraBhK{cA\ #ޘb(@7f@ʅ9 =P.,rahK{cA\ #ޘb(@7f@ʅ9 =P.,rahK{cA\ #ޘb(@7f@ʅ9 =P.,rahK{cA\ #ޘb(@7f@ʅ9 =P.,rahK{cA\ #ޘb(@7f@ʅ9 =P.,rahK{cA\ #ޘb(@7f@ʅ9 =P.,rahK{cA\  R+jX#^G?ڗwWHjn(TRM`_&<[Z+ͅ"ziJ楳쑖?|P'gLz?'v{Ŭ~0*I ʥ$] 1Fۯ,^ {>#r]Wi5㖱ir)@ȜlX󱾝ŭ/Z,ZKJ}Mpr)@0/ʊW_z^즖{}2GOV K1b@rlxf ^niԺbԗ]Z)cX~c-nݺu1 0eUvzkoey{o?unQ15.儑LIENDB`%-Dd r6   3 Ab,ϫʥ XC0քw,c3[no,ϫʥ XC0քPNG  IHDR]sRGB pHYs+,IDATx^] `TE} P0  "~b~@"h( $޷{u,:ݺuTSުt:K Ɵ(ZN(I!:2j<z`5C5r.jPQF LX2/ɡ{" K\ pqIs" -' T+T2YX HlڑѲQɧ_ pioIc# )[V+JTyAai3+p0<‹Q/&d8>D\۲.%AjX,Q*T\\/mSGe 11?1kvb>+woH?{m]T2ˋ+mf)qiuH$BP(ȕJDzmnW C^>JS.~];~5| yTŘ+)w9L\c@{Vb7AIzh! j5|CA}Jm#6\*r?q;' 0ae𰑈:BJA&+JLβ:.eTrec!jÓuu08 93<nuxIVŖ$2 2:չNSrn y25WWd'˲Fѕb;I[n@mb4A{},:Y3/RT/{⽢Ӄӫn]H,F$r; 5n6NtNYK P'RW/SiU5 uܫW,USzUQW[xլ# `Rolx+[}nuIp%SIA пo[I#\WM"/q"V\WM"/q"V\WM"/q"V\WM"/q"V\WM"/q"V\5/na*vQE;(b_4WiֆT L4Sf͚w4ЍJnj0H]lfVhfouN!|ݡl_J.2(- has <479,waH!&y–֨\HrUDElqDI Ɖ<~yҢS^s0yTFlj6;|xVQ3xыvWJzO274͢IhBcpT NYMZ<=6 69zC#S-,gjZvccV,1΋J55M6 Q4r{S7}j{RƶX0i5JNj0Y+۔E4|@ˁ48뙛Sh eƨf6Dz螛$#.&f6Od&`!d;SwǕCi~2[OK;$k:de!,*amfPOU29|Jfބ._(*qkuL"nucU] ҙ %|n4m;>ibX'j:4ˈEc['rR=M6)Gp:YԙdwtԐt<@?a OzYr00)C­ulxjm Dm Q% z\4\^:C؜Y]1"aJnfrJX!PLb(EDc7eS7F -o~ "^L.ذuȬY]LKBPgx.;HB)VdԨ xni7y-5*Cܴ#la5u ˣtlYw1Ò0`r~F "ibhkH؟.9@ۚ4W#|dv[`!ȩ;CI ry/8F;N=ıю;DL'A }Bdw(H!t&HљF /ÉH::h!@xND Й ЙF ͛7I)A5 qYd?ePGBk5Of[ݪi#[݆0rqmՎ hTw1/kBhO{^AۦUmFDmhVi+Ru9Dف&MtwY]vVXӱ/T*ՠA.`}Z9z1E+=zt~V7Q*蚧N9;ZH[G6yʕ+xDٲ`WYS֩T:ZkJBS*5JyU)*{l~D*?rȃ38&!}N,.魯GV4f4 l=ݽ4 ڣE=-)8~-:OBnߓ:t u `lpnAѥ?^ZU}me >u2gΜILLē[__gNu{݋Zɨ8iҤ3fqq`۷0"C TA!B9@Cx*+.ȑO#34u cB55Wh һݎ[sMzf\%o@K Y ءٳNJKKKII)<3˖-{g9͕.4`n#)j^FQ0uu*Z@UJ *qQxt2cZiRQ2Zl7穧Y}gn(P8rؠ9EO̝?){iH@&|`)jYp׸4"5~ X[?b\h> qy_8!uȿUV۽gwnnBw:.p%4YrPN< ,p:ؠ߿?eeel-@ ¨օBr48YNהJwPsߟ-^5)d~^=HPsUUx~N!H-+%4>*6ƃ,СC1l$t^dbN F>B'v4upp`*/+g(/[X:} V.{@a}inߩNMOJZ& O?￷S\uRL*~^g˓RNn3}=&*镚˥37j RJA]0a7 Y"9ծ?\ɢ=uo\(W_@(+K?w˗Ib,5ڢ`Gd2< )Y<8[6x،?o%[QQ2@!BMө,5ޔ^PPZ;jDquT5Uq(LTY--MHJ[u[0-n߾ "o 5"N{c .$RT)N/a_|8FQ^f~~^%uu|)BM$3PQf(L;v߾B77_i7S@}̓7Ք@Gayu{~}y{I+'n&*Dn2cRyaHa?ߛҭ0[: XF+y5-&]I?~ ?3؈9k.Hir*@s ^D&l 0s5"nAbFTDuveBZ(NjF*mE,)?6<0^@1f¿zk۟`ͤZl07;؁Ǔf!++ ׵kWF_tbmm}%;Rq&9\ )RՙHRYm$l۶-0Ţ\85c?b Tߎ^صߎѿfe)@maG e_^|6Էgh_`M^=:]=xA)H8!P_ʸ⧖Ot۞gsEiO21=1sSs3վz;~VGpˋX9v!0Hb^C?̙\-@ ¨%xm. ;zSO{,h\;W~E?xtE]&7 u ^=!gO%V*%JVՌ7ՖfSoBH5j}JPe=ĴEE7G7o~eZ5,g?4s,,Cԃ3Ֆk ׯg5̛5#ﮏCr|G{$biL)\սz9nBo DBJkX[  BQ0y?ƿ}"hGGG@X$u߀'fȘn2\,GB ¨%jѐq;8<ęA_}0"~¾tewwO/W  /4uݟyn7z#wHpsp4FO9us *JMI % ' XY#ƪMq:{D WF%){V*._RVxyZo/]哯҉5M~+?g4 _1yР7n\W1kϰaEǎDGGd {X8ύ\#vXAЫ gtZ?\fc^Dmټa;0GVBFL98<_S_J/>Yx_ ވ;0{搔E8Ϗry7,V-n {L]Z)A7;Dzye-*?˜\0 :ɀ(x D#[${$y$v"}a`RohG`뭷ِx#(S杰E||'֋dŗv!)TN1Noz{@,3r;@9wWa%/$ шFMjU#N=9سgx;MH,*xO=`ċ]/6+ 7Hnܸ7ߴ XE>w7:ny|IE^%IAK;i~G䤡/WbZH\PzQPnҡ}/X}f9eG@  [<QXٸ3.Ҽ`oٻj 4j^p6~k!\`F^~Ajqdn-2߳G7,᳃5dװʾ0 oqqqvk997=d oᅷz sBp-V}yؽKOrԛgBapO/o龍3[ck{`];;2/`R7l(\K:)/4n\O;ϚfHyƘx;~}M @h'^h'E$9n.2i pqAWMbqDAoIq6"OhH.v3H wl& @BBL@3!@x$j h0IẄ́f!t /t$]!4漀7y|qe">MqI\ķ:&,dR D _#='<3lUFW4xssڗlN:"aBKe|ZNy{0%Yh~.SצUfTHxh3NT ׻v4  ۼ:x}aB e27ǘzCaə;0t2,S{ꂉ8n_2ZdI!/6%BG1jc3n/>.Ej#V`x:rgk ًh%6ѹkG0g)61m$,YX b@v-qSD#p8$X|[[AڼM3W6>N晎M4-Z_hQx]9/Xc͏/^հb\zoOƛKERN}?[(BW?DF(dut(ĭ$dҐ3.'V;biVN1i$\zٿ|gm 8369XZҤuKP>Zu%J-,4^ m[\>am5N+7miъ^1m&+*[۷/##_/F5*)b̿sW :XOܺ8~)GQ48tŐG)Tq_fg#'h߂őS̔,]9p{SWqn"/0[66&4~ ▆:pG”ܺ4JP0]Vv1#pGnS~M/8:|;!Wh=q)`l J0|EcE#UKxD!0151d-};8*&.\J1q%!$7$Wo+[,kC2aLjTL\6#!5+nUi؏h?G-ڏh_=pZzcd1bQF>}p? Nj0$ZEGNKoi8 YU4ىӹ - 쓓zLؽ{7O"RPbРA#rG ͟_0FpIN:bC _Ȭ㥒дAE\7 y /bv6x1M<~\lϤ9=tZZԨk;3PB@ȃ/hV蓾5ZBwæNn {w~ʖ/ĈFգߒVKE e-xafr(al cb`ay-ȸyjcA ;%_}VuU7Cإ黄CH${{%{~PQw6|$j(6d*")@GFaoRE5t A:Hxx @'~P1N ^#0@G R7^9FN23H7;3t`)^RӞӒ$̓##?wЁƍVս; \%G: D$v3w 9rY 6ޛyQ$='t,l"AeB&ÿKPCǚ7slN6j 7`(BS>D Ep 7/̜b%/A(y|CN@p7/L{~]a0R$;1%7/ Rbެ6UW1dB{rb?A!6O)tscO4ӓH A Ёpt~:\/oo_:l ]#pWȼ!txBg @0"@xLA dN !вsOo~C`m&DCN$eTʋ /!"\>i &s`OKi&u|69Z(KB6Kx%Fm+G-z)Btt4 x56/␐$d 6E3_8FࣟRԴW~~f_"M۴iݻ|MSX^}]vmݺ522VWW+lA#ˢ|^zxyy_@Q*BCCWXR4~6tz18zE K^9s& 000 @ t&\ B;"tP|d8^œA),f_TWH$.:96= J?}twY@,ww|7~_W N+>A}#w/}u>#&-|0/4p0 oPN@i:~z | B@htǓ}Y~p BG @={%%y7?2֏ 6IENDB`7Dd { yy6   3 Ab7|uP,koC-?h6`[n6|uP,koC-?hPNG  IHDRs*ꇂsRGB pHYsj6IDATx^] @U>̳8ଈC<LK3{fK_/54˗ 980( ̝; =g>^k}qh4 %B 슀]0B XYh,ZLj'4}H4Tt,'' IE T$߷3?,hgg.%B #Pfi)-EHNǗ*' %B  `8RZrqq..V>,Ӳ3 `3p鉏ӬDKVsl2# , RJBVUr)Z%KJ%< ?WufµbӡZƪRuc#3S݋3eX!)$R1B|(׸eaN/ .ok_Dj\Zm-UF8>7ĥeȿK*T$ben?Y0K/Iz0Gyb!qrj ,\gҐ]XfAT\+ûT\ T$)DbiiUBq*bZ8FÚ0Q_J W9=51V?xhex2\*RB!P JeJXVqvvHrHP 2u)h_ΔVFeA%fʍ:g?s ̰QS0 )mS)vhDJ؄@y " 'FSV&AE,*J):2悻U}@ !jIh) qjM:RB`=!RqD.f-FD*"kw5R9͜K7ǍZ o pjaac"g~:$F%'e# >7uq%akukɎZ&*cۊ +c"be)7!PX,bsbll@/0}v`}L(̈=8uqu= CFNɺjFr0Y5֬V]鞔+)WMsd7:u\``y!O IF+1h0 w5Z} Mm]dLڬ9J*0mC;W #qgF}̊or%&Ul9%C5>-58ejnV)T*N`q[oMr|r LKk6TG@}d׭*uq CW.F^мEShudr#Pb뮚Oɜjˤɪ_AU(2ySvTY4*L?-Vjlr|V8fεY! Y53,d$>a}}Gt*,At\qVRTM,y?2T_I3Gl+h=ofm\C֗쯳=Tc,1&2ҽ ;먗R`ɓ PZJń\,)'x-idqkwKRf,-iXV&UT|"X$*KYhhB%W( L.GO+ʾ};w׆YQգGs ߭xBɇĀS&OR eb4OYt_g49)y@PQ 3.* C9O;7NX:ERYLmϞ=_~$U͌餛P*BuɴԴ.l#c>>x f2PjԻUl 6n~=z?[nh]4TU*Q*QG6c.{Zk7._и\/pָC\?47޺ r9*"2[>g WQ߾}njcx?&rw[,t$/Ll0Dvwu͒{SԤgQ2QQ̢65;m1rf)|@={ts*WH}zV+hfyw 6b1mlذa O?4F Z 6-- " U(Ĩ\r=z-881b& tʕ+1oH&f&1 )Ĩ55Dܼ_ǮF82y7~&ڵСCmw6Zi۶-BH=z޽;& ؄ EsO+8JU==W?"2A}o@SC/^2{gpb52K,]tɒ%ZBw܉ JfHҐD}NƝ[c㖍]{7nXZZ&T.X $"LU/U?ܭʽB$VRd\ze-ZHJ9s̮]tC,++%a8&~':t 80<<|ԨQ[n}g: !3 CDA B8@EVDJόtHӏ<;~^ᘈsXwrTՎ0xRSS3pc*ZޤIEVVٳg'NƂV̙3w^x`%D"Y-JrH\QFPRR(ItҨ"L$~b9:k^M:,qh\< fYfs }ңKNBBžtIO BPvB(w[L_U$k4n*7W7'QEӿ/A+clxMW t+r_?-[x5pqsv3lY\ϰmhQ@ ĕUhh(f撒".d;b#cFBdCfBDA B8@EyDCr8FtT4^M6!/,6O+Hh| 7p0 oh}Exnݺf[ AIVDb\5ƹ h3T:oY>~ rs///od6aM}zvo˲q'=ȵq={6fјLRurI$wlf/y_=h|ڵ T <ϖUtI,VHa(q+G?ϖ} ]{V?^9qӊoz٩4O[isiO*fU\\l\H ?X g0]ܱck8 Ȇ((G[ bUӢdM17ŗd|˼}"Q6f֒ O1<߿Q-`kd,nY0߿a j{rp@Ƌ&9lbVr\!+)4*@qq ,?>i 3oed|ne%n㬤DH1 ӱ}aGE COwqvqA{ ]\vb[˹J,1JyY[vmGN%WJE(=烡'Vq˟baLfSVzmo7Cö^;*;Oxu/Fv coCh L9c<:+_, Ql~'O jWq  QQBxi b /*.Pu8lj(!ExB_xR/B׻<ս!~Z3& n imD^'riٲhO |iKn>BW޾$T"U5Vi)R\ͫf>[5CsRoo8&͛dޜ'`;`ǔ)?~"xk~[i:n~YQ ۧcIJ|2qQO7gi(V@' rRSV9լmw+x{D'oϪU?xaJJ8ew!5|pqDw8- Ni2e ,ѣG߾}qYI)d@6dFDq(]ɨR5V\ f(ˬj4VyYxok]D1/"E'?i aB&L||<6V*ȤNط=:pDo-hd#biE`#AEɷnHׯ_E^FH5|%BН92lA{xyyzzU.Q.3Y[驩zqfNƔJQgu,.7"t K%bRRqjӣH/lab`S~g>{+5&P?axhr ~ C&7 2 2 8໮d..~ӶwfG<\<~|PDْYW \_'N*>[$;lY6 LS|O+A,E5I!m*K؉=줷zէ<<|d"JRJ:G32g¤>}s-(*V Z|cO5.12tԈжm38[6 cmDUo_S*оzUX"BTk~_}Z]VAr\܀aSqsa=TkIw>1lƌGFFb >}iӦ:ue\N6K͍j6spmP[1p!9##3.b&3G?!`G#lc8!` Y[y)7Jmj&EֆEqYht f?$ (b8Mn[#׎Z,6Lb|[RkULhmyCB{pL0 n~5[Ǒ273ctŅTQ-p,^Cb-ҁ26Khh+UT]A$ѻN 3j(E 3P) 3S(ԔFT͂7"|'UVQQ/nm9,-a>MyWt<Rlc-JB&wUY_ w#t驪R$;l ٴHCA^vSZ?*_d,1 Dl㖸5Qo+Ly,e*%'Dp5&nو [&nتj icg+endU>>nkE}xd>X LUJxBūߤ&5$aUǬq.9d0S4o4S{ʍ\P%bT&%=ZCBW.kBf1Vd읃CV~U^WzS.21ߤ&5!$༏NR:#lZ?KOA8ϹS9U S 3!7q7Uu6 w/e"8kw%YړYZ5Tn%i5qʠn毁:C s|Efܕ5Vg]6+p{ls/ffvXƹSyf^`~b1`!᪔k3,+LukX4LAZnxIAP̾#}-*CfĊ4$ rxe(kp , 5o^[Je mEf |fq0$b!+X)3 8lAN6Lf '|aqUTb:5m2 k5l\Z]D ^6 plFp-k |8+mCwίa`ݡjnOAGWU&5Q"7_X)zu, kf5IGUVqә+e=/3?f޷r"N%Q) V~uh ԰@`ӦMwNHH~m&8]XJ \5RUmvW | 3 ix3LE `xE`X' _^NB5v;ٵ |R%<,H^e;ِq Cs@6m:v숰mf,إ$h \ڻ+3[lq֖PyB#PCf_+YV,YlEߡVY`T @Cq"Z֣nUj-KHҠ7ي`Ǔp/vc٠Q[T[;kC,XZ999wilhtƋsMCj `P+fq8jE2˩j?b–ڈ,6 kt1KY/PJZeQ4v)jYjaVᄀ;v5Q5n<RK B@b!`Y)I$Aee%Ejx!abYw \JƎ fqrrzsok\x,GY; +]R NB" Ypqb~&#%AT)! fh\.պDX.D.4B - ʼne\&e )B00fpD\/c]8An-B⋱OEl\_v~r! E4B@a6 ǯivA{3dL2Ti\@%0 7wCZɩ;-|4g' !J w3iRqS&7mפKc,FB@0˜E6hD3ujש`¸ߚ{NPFBp 1I?9)ML8D)Ki3$B8eh$_?f^B6irV7IBQ,w=<<8;Lh+_v!(bT]\=rX`¸{x)! fO/o/o2,4B@k5aDBC  "` !@FF!@bcJ B!@#1333005 5 ֡JbnT ؙEQ vCilܵMnG1KcB!HEQvޭ}Pf 8߶[Ը-R YoTh(ⷿ8k K0C+@Ao<7$lЋۯ"B6KC. lS'~]G@Cne`^J\S{ 6MZ7>㏤4T<;ϼ2t׿Gv5GoM'6¦1~՝ڲjW߭h}cVͷ똷[?dW}wb!@Hvky|۷o7mu˗/o7m2kBQ"@/O[a7o^V===a e2X(AAAR(*o @"ж[VqJ-V0Pn\\e MD /۷~; @clDzSU6T~.q xfYpFJst^/FWWWsuyBpt]?f>صGSm7?V;1cv?K\ ,[ 2` a&$)Qu۶۠Qz;9eiKHTdm?gOΐ%;,+V@Kj$f1 epdrGqBջw?wdhZeY6TO]pAFf_9=y6g_7ӎ,סY86ڄ@#Ye) Znjl2 C6&o@aX42f!^_%  u.j4KuֆM\QGFq{CҢ̴iCXN!"˘E f1 epd'(n>"pgSVQ7yf=Ć|v_߻w6;%AV!Px ܻhr_,C~N͂fR)HҰoMlz 9 fQmĒS7cDѽ@\.Gnc\쟿9]Wc'>?W  lNA2 fqK F,OG>$Tz/ٹc͘Nz=3g'Y6 yWCYn !a-B0U3YA*Duﱴ tZA̞Ƭ=vz`{ߟSp,ub jGWS-x}Gl;y= ܂BS4B$,DDb "EeuN4X: sj1B:b1zTh¶ȫ Uٜbх?]#gsb4:O4z޸{oЩׇ])ʲ~[;׈r[`/|fO-YƖX t~H̨E}c61rxpZKcUy=w8%,,9OkCiw߮h_Dސy(!Шpk9tۗ~u (YGҥ^W%yCB fÙ}+\|z|uYzu"UJ\$ ՙ%&&իXk!iDqˊoX|9J!@XV>kfDOѵ 7,Y>ZU 0.Zx_~eRfBh)ZW_eXlhTqhL7y{W}oYP}ssl߾zY,E;YNn:s[_xhn1{l[!G/m|ǭeJ Yxg{WB6p("Иl{Cb.v?VV L>}?O4_~X1@(c"Ș#m=mǮGW oH{0y_=1iE!eod8eӭf_Ç;ֺ*!@84n~{snS(k1ÇF{Qc#}O[T\K!-G<2bS'N ±!"B M_=9/7-`rO =w̠A,Jc#3E\:rm-zfA-g:3`K/ϒJ)/!@4plb`s…>}^իWGYPѕ+aa ~%Bh؁Y7z|,fr;~BzmHnpD,<)))*qI'{2 !J1 B?,ǔ$1 B?,ǔ$1 B?,ǔ$1 B?E.WWN B"`Y%XƓE9RjPPBneDλ13]2b֨WhtcƌfI!@4DZG)/6\^󻄳N:O V!(QE>]afv㿻 BgW! ]VVGB 7ybwJkEp 4xA " `Yrի:i[Ju/+>*Bhp`B)7in~~~  j!@%G5dyyy]v $ ,JOKl?Bzim 5f瞝(;w S}QV]^NiJ!@!`Y\]]/n]s#f8΂Q1E@(p1Ys E^g2%S[א_P#D":}gvכR2ZRIE@H@}-F)wMi ("$w%73hvn],&bÀkZU"D ъ x2d'ŮFK2,a1u;bQE@P .GLXk~ mBd$PP+DmvSFRPhȈ4oޜ̾bΉ@U\|o|BgzVZvqg,O??֭[[nRK.)ŦMֿbÇ{RtϞ='|rJmv:3RԖROx>l)m۷oo`j.h|NmV0~#5[P 3m\~1!5ZxVz:ڡ4 ͚5 :#w^BfY,[x`.K>#ѤLqׯ_oF9.]ŬC3REh>u+ÒtVnܸh"Ν;~4̛7oӦM`:G쫭N}E ;|8#;(#Ҥڻwo{z2_Yiu߲epX5d"(49pĄ":?l?K,α"Xts 2_ڏ"dwe>:%}vI(L)3ԊN/FHPcˣ)@!#ТE E43eS/ 555P1W^y)ldެ.q"h ܬ(܃f'WǬak)ɾA#]K']dl~5GE<&cǎ.իWw̳LO?^ /&+K`+3 ~Y$tw&N0Z-/mսBd5C 2՗JE`V,Mjf9`:lR3){z{d5{GMUebEIY.Vڤ,tbBH:aϙɷ\-t5*n,/9ֵuJ;%$ndⷺǥd?tNLq;B/cK\0ױ(Z)bShHXQ҂}MEgZ?n3 5< ;i5O*K0{иa%C4yaܠ{ښ3`BV*0۹3!2I碡r} ۺcwˆE}ft$AaQT&㴥I:aYkHXgvk1)4G\WhRKnCT%E?˒!zssbya;!M#e 6cT&[]2N+A]rwdp}Y$)L6v֤( FfPmQm-i=Z^E dxe‚]2d.qD}ѱ{9z=Xm Zb`45@߶S4"PƦO-C4Ǜc BԒ~G&5C)i!H=J­x ZqY/I'͊jW1g*$N`nEebMgǬ,RlL,)w

)u. %%%Qjݺu޽3dUE %pٿ%6e" g̘2?PE CriΝ棏> ȃ^|E{ƾ}V\![U"(lܸXvmm咀VD,X+"}:ꨣp=466Bf"4a`R@:)J(FP0ҕAP#@mm4L2Z\;9G<",8eLWhٲ{yڅ ٳ.+hZr쳥W@w宪ڿ?e˶lْhv"4 4\m݂'4,cM(@`ufQs,AҞqnӫu덛&P 71cǎM(`UL(өeUꋀ쵠fbZXGa2aMiedVn[ϿMXTNP :} Ԧ*mAYܙsh/!%6ƑKݒqĸ3礓lKؒ:2eeQCn&+ug -OK:vYg۷O4ƕW^٫W=zt֭~.UѹݹNeTia`@d\;'CGQ5@Xd~ћN]nw;y`(K&^̂>"V{"*0Dbj] Vl"Pј}&٘&-ZP6zSڭڪHd޼e޴ȈCsg+wK]s.@V(U*"U΀CWb1&i?:Ԧo'B,#tzM~4)$6k[i6i]2\d ٱ?*($'\ɑһEڪX.fXĺb ཭د9w͋NpgqP5.XJ;i-*d 09sc1kl-O8$-H8=fnhMٛGsC!`;cZC=E :ٵ[T[ȳu ꗹnTUDP@u6ʖ#81%Ԍi" ̴:Cefc128fU"hiF)@"&+8A E@P {>3 8OMP ݘ9F+bmE_P@ъ+Bn5^-W~k֬ @%E@Yn~}2kFĊ"O|-UPPF'DPd47xpW Fs7Z>i;͸k4" N˜2K3yiYeө)-"F FO8|xq}oHdU2A^Rڜ wH)eUװ>p\iսD 5Oyod\Bџ5jΕ.ݴ>28d8YU5&EwE .\.Bvro^Pre˖mޜ,_F{gx 2rF]VǾ.kbxmE6mZJ|W^!ERh@;i1]o^1=%m;TEE+[eNj[ܝ4+mqJPr G{y饗mĈl`KJVTTdhNQڢg=pJLH\<'!}>cgw(+uXA2RwT"[zdcAy0J7|sȑ ￿ԢX.X5ˇ$SZڱ(EժBE@h]BJu;W2e e>ړO>izD9@gi\*@ؾ};tFvfuڵks׿qF߿߾rݻw(P"x}i/@W-է  xbBGBj|\ϛ7Þ107("(#d4E@p9Xo\ [ۯMD>1f̘("V;>ydc=ڬO=X7Ag}Hk4"( F3kX`4Elʤa]5M(@c5kYc6g[@X~ ;cUE Ӽys (^v#2 ^ܻwoUFP4"MN{o̘1/"{u;faD DU(@H RaAP?tcJhZ2 _;?x!*(@غu1sl4`1\I^u|XHkìXhQJj{葳a'=Nn9}r}eE줱&;V#jM_:5F{6dt/.0`Mj,sZb j…ɻF / ^rZ:(JT͙ELZE@(Ly{qƙ+*:Ν+NZiE Sxl!AM႖-[~1wBרQ>ڥsO݄N\{2rv-52g؆zv&BfC5i$q***2W٪YPrmr:yGÝ%0梁za1iS5A8 F*Uھ"207͜9Ӭq*O!i~>֬n׺8w>6Q-\I&䋕WNs1-u[|{dyox2qZ}'٭qnYc]OxvQ<όܿtDxpV($^.ň7׋gmզ d4\g`5^|f%C/]7[[]T`3&Rzc+!@šbU3oW/5y}5$6j%ŷu>9_ύi-ڸ`w݆2Y{݌8_< ;඾qc@‹a$݈n3P0'vIcX4m8t1[dxU'`zPu4֮j. CUt_Eg4\( jht/ULx_=[u==ӹx;eӆHwWmo/ci 6N罉Dպ#?8NARvoL a2W0tbZ9~򓟤Tuifե!9(6*ck Q:xT9@1}bb}yu =Uő)M-7{Ζ+aλ3sXV.yR/?BJU90>g:R[+CƉReyk)qH]bβVP<y꫺Gt::-KxI$LzP Uu/|WcmxOJIo2}Ci{;H;4CG<1Y`hdz;{Ŝ;0쌆lX/_n2яzSpϤخB<֑(K(Xw_o*N$ԱZN)mꪫcfcZLY;F[?'ŽP)U |h7M+NHn7.V^G+++sMڠl5bu u8z1f]f@5!E,k} 6m`i͚5V}_̽.]w40oK_җe:v ;Ȩ]"gкlժՅ^};88C.䒳:˳<޻ qDf7/{Ŵ!G FrB=gFL|~݄Jc6iJ2|k׮+Ď!oϟO V:#d>*(@20{/cb-b &c4&*9?\SE@&\ hk D̕lDRE8k2{Q7|"qbfYڄCkOP818*L`Dٳ ?("7!5z4%(:ą#O=Tfl8 rxj"0ҡC p PA6&'qF9e47 3{p֒*@`sz2I%P~gA9G@_/sfE@P0$"O2u)E"(@6H&2eL`4[Ns4 M7%ѼE@v죹hh֭XxMUE R'Zƞq|zeX !?~e|ҿfa4/E@(p}sVt p g~vP^0aBNJKK]FJGYNRc += $ YbE^XA*~S6XO,#)X )3Xaσ>ȃ2􊅤k2$MY{嗩SD8Up ; @']}kSn %)"շo_Ҳdól2iBᥗ^}#N~bfڵkb fo(b{ dxYb1c(Bl3f +(h6 Po}[Le k؏L?\?x SYPd@-eVJ)ƜDIö~1F yv9?%eǗxW\ rĀ%7l.4~2\r";w{衇=裼a$_G?m0 9"%w]g 7>|~R;"7C*oQ}<| f່ 1Wvͤ+M:2lI #nQ/:d撉?&~wʔdFXuW6sI'z,#6b?ӧ{OhPӟzW3px_pxD |I?0(CRHƒ!bd͍tP1N~ Z!b_)#f3q_@WyEI N/W '7'2VuпqA%}S1~v*%)β-OQFц$U |U<]m6VzjU:ÏNxiOFSK*aS޴[p_lKz=ھ>b8SdLy*6Lg̘a7uxS}Whc.Oaayz RcpaOрm.@ 145 ՘n}2q8>XK0z+)JFł pˑIU.1]Ír:0%6qD6_;CxseOʹ Aֈ ,J )7 8緺mmIzW~W-sѺlӦ gL풒)9ҢHfW8btbveJ[䢋.0tJE`͙3%7]IxL6 e?SrMJ`:;SJ{gɥa2ض Oe48(R8ˬsS&kY4n8>ER"U۷oe%mO=a." I*B08x1'_|+>(2v /qR6[ ?Y$)Q"l1(vB>z< FVEv>5m#IGab4w.\Bqo>Oh-Y'Ov)~\6{l\bPbgJKLZ!IrNE[]pidmIN_yzc+ GcsΕsUht+n:Cp"*)i Ya6/-`ömlxsl1WxhFWFsidm)jׄЅW_1^*ż$2<2 1=bф7b+gcSD`[f|#D.%mwQ=FQzͳ35&)G;FWn``}8~e@yvX=ӯ,gf>6jbx6Z5IENDB`s2 0@P`p2( 0@P`p 0@P`p 0@P`p 0@P`p 0@P`p 0@P`p8XV~ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@ 0@_HmH nH sH tH @`@ NormalCJ_HaJmH sH tH PP  Heading 1$$@&a$5CJOJQJ\^JVV +Q Heading 3$<@&5CJOJQJ\^JaJDA`D Default Paragraph FontViV  Table Normal :V 44 la (k (No List 4 @4 Footer  !.)@. Page NumberDZ@D _ Plain TextCJOJQJ^JaJ6U`!6 +Q Hyperlink >*B*phB^2B f Normal (Web)dd[$\$4B4 "bHeader  H$6/Q6 "b Header CharCJaJjcj [l1 Table Grid7:V0B' qB KComment ReferenceCJaJ<@< K Comment TextCJaJ:: KComment Text CharHH K Balloon TextCJOJQJ^JaJN/N KBalloon Text CharCJOJQJ^JaJT@T N6 List Paragraph5$7$8$H$^m$aJB/B ]7Plain Text Char OJQJ^JPK![Content_Types].xmlN0EH-J@%ǎǢ|ș$زULTB l,3;rØJB+$G]7O٭Vc:E3v@P~Ds |w<v %  %%%(y ;E=JFL/QXV]ZVctr0zZ\^_ahilnpqsw|F $g*07=C JNV/[b_Vc!glpAty~:eH!+Px[]`bcdefgjkmortuvxyz{}~+r[#_33445C55551JVJmJXXXXXXX !(!!8@0(  B S  ?!)    ntPX hv+.9?`j Z` o!t!,"2"""""$ $$$%%&&''C(I((())V*_*++(-1-'.-.s....//1122=6@6k6o677:8=8l8p888889 969;999 ::::::\;b;;;;;(<0<<<9=A=BBBB&C-C9C?CHH(J/JvL|LNN]W`Wb]i]]]]^G^N^G_J___```a0a3a4a6aaabbbbbbjcpcqcxccce e eeeehhhh5i>iilklllllooqqrrUrXr~~yt~ =H_bڕ(28@oy -459otN_ SUdfàѠԠUXfi *08?ȨOVȪsuvxy{|~I<>ORps     # QZS\HLim6:PX &nrMQbg&*9? UYln]eT] & 4"<"####$$f$i$$$t%|%%%K&U&'"'''(()) **V*_*u+z+++G,R,,,,,(-1--.H.J...////m0q0001 1L1M1113344 6 6=6@6k6o66677>7R7Z777*8,8l8p8888869;999 ::::::\;b;;;;;(<0<<<9=A===>>???@@@@@@@.A4AxAAAAAAAB1B:BhBpBBB#C-CmCwCCCCCD DpDzDDDEE:E=EEEiFmFFF2I4IBIEIJJKKvL|LLLM MMMXN[NoNrNNNNNNN)P4PPP;QEQQQ#R&R/R3R]RbRRRRRRRTTTU$U'URWUW2Y>YVYZYmYwYZ Z5ZnznnnnooooppQpWpppqqFrGrptrtttu uuuuuBuFuuuuuuv3v:vvvyvww]xaxxxxxxxxyyyryvyczkzzz{{{{{{||c|m|x||||||7}=}}}}~h~o~.2UZw|DNvۀ#&LO\c05҃Ӄsvkp†ck(,OTw|DNvux9<$'kn8F),@AknәיlqܛʜIM`d,0ÞҞ>Cs؟ޟadΠѠ=@cfޡLǤ47MPoziw-<ڪݪsuvxy{|~333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333333 )| w3445==B????BBBDDDOOPctlululoqqsuvxy{|~ )| w3445==B????BBBDDDOOPctlululoqqssuvvxy{|~ vfҜkL Ut;H.( )h)p0Nb FIh1?M<./MpFRQMN<IU.O-zq<^`o(. ^`hH. pL^p`LhH. @ ^@ `hH. ^`hH. L^`LhH. ^`hH. ^`hH. PL^P`LhH.^`o(.^`.pLp^p`L.@ @ ^@ `.^`.L^`L.^`.^`.PLP^P`L.h^`OJQJo(hHh^`OJQJ^Jo(hHohpp^p`OJQJo(hHh@ @ ^@ `OJQJo(hHh^`OJQJ^Jo(hHoh^`OJQJo(hHh^`OJQJo(hHh^`OJQJ^Jo(hHohPP^P`OJQJo(hH^`.^`.pL^p`L.@ ^@ `.^`.L^`L.^`.^`.PL^P`L.h^`OJQJo(hHh^`OJQJ^Jo(hHohpp^p`OJQJo(hHh@ @ ^@ `OJQJo(hHh^`OJQJ^Jo(hHoh^`OJQJo(hHh^`OJQJo(hHh^`OJQJ^Jo(hHohPP^P`OJQJo(hHh^`OJQJo(hHh^`OJQJ^Jo(hHohpp^p`OJQJo(hHh@ @ ^@ `OJQJo(hHh^`OJQJ^Jo(hHoh^`OJQJo(hHh^`OJQJo(hHh^`OJQJ^Jo(hHohPP^P`OJQJo(hH^`o(. ^`hH. pL^p`LhH. @ ^@ `hH. ^`hH. L^`LhH. ^`hH. ^`hH. PL^P`LhH.h^`OJQJo(hHh^`OJQJ^Jo(hHohpp^p`OJQJo(hHh@ @ ^@ `OJQJo(hHh^`OJQJ^Jo(hHoh^`OJQJo(hHh^`OJQJo(hHh^`OJQJ^Jo(hHohPP^P`OJQJo(hH^`o(. ^`hH. pL^p`LhH. @ ^@ `hH. ^`hH. L^`LhH. ^`hH. ^`hH. PL^P`LhH.^`o(.^`.pL^p`L.@ ^@ `.^`.L^`L.^`.^`.PL^P`L.h^`OJQJo(hHh^`OJQJ^Jo(hHohpp^p`OJQJo(hHh@ @ ^@ `OJQJo(hHh^`OJQJ^Jo(hHoh^`OJQJo(hHh^`OJQJo(hHh^`OJQJ^Jo(hHohPP^P`OJQJo(hH^`.^`.pp^p`.@ @ ^@ `.^`.^`.^`.^`.PP^P`. <IU )z1?M0QMt; F/Mvf                                                                                                    Md! K P|E c /gHT( G!(%f%(&z'Q\(f.*v,?-7`/=1[l1z1N2:23@4C6S6R7]7 h8i8k9G'?In?>^ANCqgCEJEUE}E*IiILLRM;NDmNqP+QsR~SZ09?i17Cy"h{_*Ik#vOA\"Gg87aS0^\[e"bs9Kf<_-7@uhL _%) 7\su@))))(@UnknownG.[x Times New Roman5Symbol3. .[x Arial?= .Cx Courier NewU  MS MinchoYu Gothic UI5. .[`)Tahoma;WingdingsA$BCambria Math"1h]{gd{gd{gݑW7ݑW7!84 2qHP ?}2!xx I K*increasing amount of memory for the data (default is 1Mb, our dataset is 2Natasha SarkisianNatalia Sarkisian<         Oh+'0 4 P\ |   L*increasing amount of memory for the data (default is 1Mb, our dataset is 2Natasha Sarkisian Normal.dotmNatalia Sarkisian5Microsoft Office Word@V@@@ݑ ՜.+,D՜.+,@ hp  Boston College7W L*increasing amount of memory for the data (default is 1Mb, our dataset is 2 Title8 8@ _PID_HLINKSA*a=http://www.textpad.com/Mhttp://www.statalist.org/%6 $http://www.ats.ucla.edu/stat/stata/X %http://www.indiana.edu/~jslsoc/stata-m<https://www.bc.edu/offices/help/getstarted/network/vpn.htmlXgMhttp://www.bc.edu/offices/help/teaching/app_server/apps-files/files-map.html`B9https://www.bc.edu/offices/help/teaching/app_server.html  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~      !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstvwxyz{|}~Root Entry FEData 1TableuN`WordDocument4 SummaryInformation(DocumentSummaryInformation8CompObjr  F Microsoft Word 97-2003 Document MSWordDocWord.Document.89q