#!/bin/bash

EXPECTED_ARGS=2
E_BADARGS=65
MYSQL=`which mysql`

Q1="CREATE USER '$1'@'localhost' IDENTIFIED BY '$2';"
Q2="GRANT ALL PRIVILEGES ON *.* TO '$1'@'localhost';"
Q3="CREATE USER '$1'@'%' IDENTIFIED BY '$2';"
Q4="GRANT ALL PRIVILEGES ON *.* TO '$1'@'%';"
Q5="FLUSH PRIVILEGES;"
SQL="${Q1}${Q2}${Q3}${Q4}${Q5}"

if [ $# -ne $EXPECTED_ARGS ]
then
  echo "Usage: $0 dbuser dbpass"
  exit $E_BADARGS
fi

$MYSQL -uroot -p -e "$SQL"

I want to be able to edit transcripts even after I’ve loaded them into the software (e.g., if I notice an error in the transcription). I also need to be able to share my work with colleagues who use a variety of computers including PCs, Macs, iPads, and Android phones.

Andy Begel and I relied on OneNote, and that seemed to work better than Atlas.ti, NVivo, or HyperRESEARCH, and it came with my version of Office. Trouble is, now I’m working on a Mac, and there’s no OneNote for Mac. I liked the idea of doing all my analysis in Office so that it would be easy for me to share with my collaborators regardless of the software they had available (since Google Docs can handle the basic functions of Office). I also liked that Office would give me direct control over the links between codes and data so that I could change either on the fly.

Daniel Meyer and Leanne Avery published an article in Field Methods a couple of years ago about using Excel for QDA. Sadly, I didn’t notice until today that Dr. Meyer is a colleague of mine at IIT. Anyway, starting with their article, and the assurance of some friends that yes, I could do this all in Excel, I set off to DIY myself a QDA software solution.

I’m dealing with interview data, and I had all the interviews transcribed by Scribie. Scribie transcripts contain info in the both the header and footer, have a cover page, and contain a rating page at the end. None of that matters for my analysis. What I want instead is a file of the form “timestamp TAB speaker TAB talk” that I can copy and paste into a corresponding Excel sheet with 3 columns. Here are the steps for using Word to prepare a Scribie transcript for use in Excel:

1. Save DOCX file as TXT
2. Remove the header, footer, scribie cover, scribie rating sheet
3. Replace “S1:” and “S2:” with correct names – use “^tNAME^t” to add tabs before and after
4. Replace scribie line breaks (“^l^l”) with paragraph breaks (“^p”)
5. Remove line breaks between end of paragraphs and scribie notes such as “[laughter]” (Find: “^p[“, Replace: “[“)
6. Select all, copy, paste into Excel

You should now have 3 full columns in Excel – time, speaker, and their comment. This approach treats all turns Scribie identifies as a speech event. Sometimes that produces long paragraphs, sometimes just a single word. At this stage in my coding, I do not need to code line by line, so this approach works for me. If you want to code each line, you can do some Find and Replace operations to make “time TAB speaker TAB sentence” lines based on the sentence’s punctuation. For instance, replacing “.” with “.^p” would put each sentence on a new line. Getting the first two columns on that same line will be tricky though.

Now that I have all the comments in Excel, I can easily attach codes to specific sections of speech. To do this, I have created a separate sheet for my Codebook where column 1 is the code’s label, and column 2 is the code’s definition. In my main sheet, with the transcript data, I add three columns for “Primary,” “Secondary,” and “Tertiary” codes. Excel’s Data Validation feature enables me to select codes from a drop-down in each of those columns. As I add codes and definitions, the drop-down list updates to include them.

Once I have all my transcripts and codes into Excel, I can go merrily along, choosing codes from the drop-down list without having to navigate to a new window. If I need to edit a comment (as often happens when interviewees mumble), I can do so without impacting my codes. I can easily go back to the recording, listen for the part that Scribie missed, update it in Excel, and continue on my merry way. All the QDA software tools require that you do all your data cleaning and editing before you start coding, and that’s just not the way I work. I code and clean, clean and code, as I go along. This Excel approach lacks some of QDA’s more advanced features, but I find that thoughtful Pivot Table use can make up for most of those. The upside is that I already know how to work with Pivot Tables so I don’t have to sit through those webcasts on how to use NVivo and HyperRESEARCH. Data analysis is hard enough without software getting in the way. If you were looking for a straightforward DIY solution, I hope this post helps.

First, let’s update yum

sudo yum -y update

Then we install PHP and Git (Git will come in handy later, and I don’t need MySQL so won’t be installing it):

sudo yum install php
sudo yum install php-mysql
sudo yum install git

Apache’s already installed but isn’t started. To start it

sudo service httpd start

By default, the DirectoryRoot is /var/www/html, and I’m leaving it that way. Put a dummy index.html file in that location. Then, make sure you have port 80 open on your instance (see below) and go to your elastic IP in your browser. Here’s mine:

http://50.19.246.137

Opening Port 80

1. In the AWS Management Console, under Security Groups, select the group that’s associated with the instance you’re running.
2. Click on “Inbound” in the properties window.
3. Select “HTTP” from the “Create a new rule” dropdown.
4. Enter 0.0.0.0/0 as the source to allow all web traffic to see your site.

Default versions of stuff I use:

• Python 2.6.6
• ruby 1.8.7 (2010-12-23 patchlevel 330) [i386-linux] (You’ll need to install Rubygems yourself)

Here’s the ps -ef output right after startup:

UID PID PPID C STIME TTY TIME CMD
root 1 0 0 15:40 ? 00:00:00 /sbin/init
root 2 0 0 15:40 ? 00:00:00 [kthreadd]
root 3 2 0 15:40 ? 00:00:00 [ksoftirqd/0]
root 4 2 0 15:40 ? 00:00:00 [migration/0]
root 5 2 0 15:40 ? 00:00:00 [watchdog/0]
root 6 2 0 15:40 ? 00:00:00 [events/0]
root 7 2 0 15:40 ? 00:00:00 [cpuset]
root 8 2 0 15:40 ? 00:00:00 [khelper]
root 11 2 0 15:40 ? 00:00:00 [netns]
root 12 2 0 15:40 ? 00:00:00 [async/mgr]
root 17 2 0 15:40 ? 00:00:00 [xenwatch]
root 18 2 0 15:40 ? 00:00:00 [xenbus]
root 64 2 0 15:40 ? 00:00:00 [sync_supers]
root 66 2 0 15:40 ? 00:00:00 [bdi-default]
root 67 2 0 15:40 ? 00:00:00 [kintegrityd/0]
root 69 2 0 15:40 ? 00:00:00 [kblockd/0]
root 76 2 0 15:40 ? 00:00:00 [kseriod]
root 184 2 0 15:40 ? 00:00:00 [khungtaskd]
root 185 2 0 15:40 ? 00:00:00 [kswapd0]
root 186 2 0 15:40 ? 00:00:00 [ksmd]
root 238 2 0 15:40 ? 00:00:00 [aio/0]
root 242 2 0 15:40 ? 00:00:00 [crypto/0]
root 253 2 0 15:40 ? 00:00:00 [khvcd]
root 306 2 0 15:40 ? 00:00:00 [khubd]
root 310 2 0 15:40 ? 00:00:00 [jbd2/xvda1-8]
root 311 2 0 15:40 ? 00:00:00 [ext4-dio-unwrit]
root 373 1 0 15:40 ? 00:00:00 /sbin/udevd -d
root 535 2 0 15:40 ? 00:00:00 [kstriped]
root 600 2 0 15:40 ? 00:00:00 [kauditd]
root 748 1 0 15:40 ? 00:00:00 /sbin/dhclient -1 -q -lf /var/lib/dhclient/dhclient-eth0.leases -pf /var/run/dhclient-eth0.pid eth0
root 785 1 3 15:40 ? 00:00:01 /sbin/rsyslogd -c 4
root 796 2 0 15:40 ? 00:00:00 [flush-202:1]
root 835 1 0 15:40 ? 00:00:00 /usr/sbin/sshd
ntp 842 1 0 15:40 ? 00:00:00 ntpd -u ntp:ntp -p /var/run/ntpd.pid -g
root 849 1 0 15:40 ? 00:00:00 crond
root 859 1 0 15:40 ? 00:00:00 /usr/sbin/atd
root 881 1 0 15:40 hvc0 00:00:00 /sbin/agetty /dev/hvc0 38400 vt100-nav
root 882 1 0 15:40 tty1 00:00:00 /sbin/mingetty /dev/tty1
root 884 1 0 15:40 tty2 00:00:00 /sbin/mingetty /dev/tty2
root 886 1 0 15:40 tty3 00:00:00 /sbin/mingetty /dev/tty3
root 888 1 0 15:40 tty4 00:00:00 /sbin/mingetty /dev/tty4
root 890 1 0 15:40 tty5 00:00:00 /sbin/mingetty /dev/tty5
root 894 373 0 15:40 ? 00:00:00 /sbin/udevd -d
root 895 373 0 15:40 ? 00:00:00 /sbin/udevd -d
root 896 1 0 15:40 tty6 00:00:00 /sbin/mingetty /dev/tty6
root 897 1 0 15:40 ? 00:00:00 /usr/bin/python -tt /usr/sbin/yum-updatesd
root 899 1 0 15:40 ? 00:00:00 /usr/libexec/gam_server
root 900 835 0 15:41 ? 00:00:00 sshd: ec2-user [priv]
ec2-user 902 900 0 15:41 ? 00:00:00 sshd: ec2-user@pts/0
ec2-user 903 902 0 15:41 pts/0 00:00:00 -bash
ec2-user 922 903 0 15:41 pts/0 00:00:00 ps -ef

And the output of ls /usr/bin (yes, it’s messy, just do a Cmd+F and search for what you want):

[ db_verify gtbl lzdiff pathchk runcon timeout
a2p dc gtroff lzegrep pax run-parts tload
ac ddate gunzip lzfgrep pcregrep rvim tmpwatch
addftinfo deallocvt gzexe lzgrep pcretest s2p tnameserv
addr2line diff gzip lzless peekfd scp toe
afs5log diff3 h2ph lzma perl script top
amuFormat.sh diff-jars head lzmadec perl5.10.1 scriptreplay tput
apropos dig HEAD lzmainfo perlbug sdiff tr
ar dir hexdump lzmore perldoc sechecker troff
as dircolors host m4 perlthanks secon truncate
aserver dirname hostid Mail pfbtops sediff tset
at dislocate hoststat mailq pgawk sedismod tsort
atq djpeg hpftodit mailq.sendmail pgrep sedispol tty
atrm dprofpp i386 mailstat pic seinfo tzselect
attr du iconv makemap pic2graph semodule_deps ul
authconfig easy_install id man piconv semodule_expand unbuffer
authconfig-tui easy_install-2.6 idn man2html pinentry semodule_link unexpand
autoexpect env igawk manpath pinentry-curses semodule_package uniq
awk eqn implantisomd5 mapscrn pinky seq unlzma
base64 eqn2graph indexcon markdown pkill servertool unpack200
bashbug-32 erb indxbib mattrib pl2pm sesearch unshare
batch ex info mbadblocks pmap setarch unxz
bc expand infocmp mcat pod2html setfacl unzip
berkeley_db_svc expect infokey mcd pod2latex setfattr unzipsfx
build-classpath expr infotocap mcheck pod2man setkeycodes uptime
build-classpath-directory factor install mclasserase pod2text setleds urlgrabber
build-jar-repository faillog ionice mcomp pod2usage setmetamode users
bunzip2 fallocate ipcmk mcookie podchecker setsid utmpdump
bzcat fgconsole ipcrm mcopy podselect setterm uuidgen
bzcmp file ipcs md5pass POST setup-nsssysinit.sh uz
bzdiff find isohybrid md5sum post-grohtml sftp vdir
bzgrep find2perl isosize mdel ppmtolss16 sg vim
bzip2 findcon java mdeltree pr sha1hmac vimdiff
bzip2recover find-jar javaws mdir pre-grohtml sha1pass vimtutor
bzless fipscheck join mdu printenv sha1sum vmstat
bzmore fipshmac jpegtran memhog printf sha224sum w
c2ph flock jvmjar mesg procmail sha256hmac wall
cal floppy k5srvutil mformat psed sha256sum watch
captoinfo fmt kadmin migratepages psfaddtable sha384hmac watchgnupg
catchsegv fold kbdrate migspeed psfgettable sha384sum wc
c++filt formail kdestroy minfo psfstriptable sha512hmac weather
chacl free keytab-lilo mkdiskimage psfxtable sha512sum wget
chage ftp-rfc keytool mkfifo pstree shasum whatis
chattr funzip kibitz mkmanifest pstree.x11 showconsolefont whereis
chcon g32pbm kill mkpasswd pstruct showkey which
check-binary-files g3cat killall mlabel ptx shred whiptail
checkisomd5 gawk kinit mmd purgestat shuf who
checkmodule gencat klist mmount pwdx size whoami
checkpolicy geqn kpasswd mmove pxelinux-options skill write
cheetah GET ksu mpartition pydoc slabtop write-mime-multipart
cheetah-analyze getconf ktutil mrd pydoc2 slogin wrjpgcom
cheetah-compile getent kvno mren pydoc2.6 snice xargs
chfn getfacl last mshowfat pygmentize soelim xkibitz
chrt getfattr lastb mtools python splain xmlcatalog
chsh gethostip lastcomm mtoolstest python2 split xmllint
chvt getkeycodes lastlog mtype python26 sprof xmlwf
cjpeg get_module lchfn mxtar python2.6 sqlite3 xxd
cksum getopt lchsh mzip quota ssh xz
clean-binary-files get_reference_source ld nail ranlib ssh-add xzcat
clear gindxbib ldd namei rdist ssh-agent xzcmp
cloud-init glookbib lddlibc4 nano rdistd ssh-copy-id xzdec
cloud-init-cfg gneqn less nc rdjpgcom ssh-keygen xzdiff
cloud-init-run-module gnroff lessecho neqn readelf ssh-keyscan xzegrep
cmp gpasswd lesskey newaliases readlink stat xzfgrep
col gpg lesspipe.sh newaliases.sendmail rebuild-jar-repository stdbuf xzgrep
colcrt gpg2 linux32 newgrp rebuild-security-providers strings xzless
colrm gpg-agent linux64 nl refer strip xzmore
column gpgconf lkbib nm rename sudo yes
comm gpg-connect-agent loadunimap nohup renice sudoedit yum
conman gpg-error locale nproc replcon sum zcmp
conmen gpgkey2ssh localedef nroff reset symlinks zdiff
create-jar-links gpgparsemail lockfile nslookup resizecons syslinux zegrep
crontab gpgsplit logger nsupdate rev syslinux2ansi zfgrep
csplit gpgv logname ntpstat rftp systool zforce
curl gpgv2 look numactl rhino tabs zgrep
cut gpg-zip lookbib numademo rlogin-cwd tac zip
cytune gpic lpunlock numastat rmail tail zipcloak
db_archive gprof lsattr objcopy rmail.sendmail tailf zipgrep
db_checkpoint grefer lscpu objdump rmid tbl zipinfo
db_codegen grn lss16toppm od rmiregistry tclsh zipnote
db_deadlock grodvi lua oldfind rnano tclsh8.5 zipsplit
db_dump groff luac oldrdist rngtest tee zless
db_dump185 groffer lwp-download open rpcgen test zmore
db_hotbackup grog lwp-dump openssl rpm2cpio testrb znew
db_load grolbp lwp-mirror openvt rpmdb tfmtodit zsoelim
db_printlog grolj4 lwp-request orbd rpmquery tgz
db_recover grops lwp-rget pack200 rpmsign tic
db_stat grotty lz passmass rpmverify time
db_upgrade groups lzcat passwd rsync timed-read
dbus-binding-tool gsoelim lzcmp paste ruby timed-run

Instructions

First, you need to generate an SSH key pair. The AWS Management Console makes this easy: Choose “Key Pairs” from the left menu bar under EC2. Follow the instructions, and download your *.pem somewhere you’ll remember.

Then, start your EC2 instance with that key pair and associate your elastic IP with that instance.

Once you have an EC2 instance up and running with your designated key pair, you should be able to SSH in just like you do to other servers:

ssh -i ec2-keypair ec2-user@ec2-elastic-ip

(Or, use ssh-add ec2-keypair.pem before ssh so you can drop the -i option)

Common problems

Problem: I get Permission denied (publickey).
Solutions:

1. Make sure you’re using the right public key (check your EC2 Management Console. The Key Pair Name should match the *.pem you’re using.
2. Use the full path to the *.pem file for ec2-keypair.
3. Make sure you’re using the right username, should be ec2-user

Notes

Many forums recommend Elasticfox Firefox extension for managing your AWS instances. That’s overkill if you just want to SSH in and get something started. Clicking the “Connect to Instance” button in Elasticfox just opens terminal with the SSH command above anyway. So, if you’re a happy Chrome user like I am, you can continue to avoid Firefox and connect happily on your own via the command line.