SPSSX Discussion

Estimating actual earnings from earnings groups

Classic

List

Threaded

8 messages Options

John F Hall

Estimating actual earnings from earnings groups

I have a variable "Gross earnings from paid work" coded in 11 to 20 groups for each year in which it is used. The coding scheme used was changed twice, so the earnings intervals are not the same for each year.

In some years the variable [REarn] is not used: [REarnQ] "Earnings quartile group" is used instead. For years when [REarnQ] is not calculated I can get an approximation by recoding [REarn] ~ ~ INTO [REarnQ] but I'd like to estimate a mean/median for actual gross earnings for each category of [REarn] by something like:

DO IF (year = 1983) .

RECODE (<value(s)> = 1) (<value(s)> = 2) etc (else = copy)

INTO newrearn.

ELSE IF (year = 1984) etc.

~ ~ ~

ELSE IF (year = 2015)

~ ~ ~

END IF.

However, I would like to take account of the distribution of earnings for each year. For the centre of the distribution I can take a mid-point, but what should I do about the intervals further away from the centre, especially the outer ends "Under £x" and "£x or more"?

Oh, and variable [year] is not present for every year, so that has had to be inserted manually.

Sorry if there's too much detail below, but it gives a clearer idea of the problem.

Thanks in advance for any suggestions.

John F Hall MA (Cantab) Dip Ed (Dunelm)

[Retired academic survey researcher]

Email: [hidden email]

Website: Journeys in Survey Research

Course: Survey Analysis Workshop (SPSS)

Research: Subjective Social Indicators (Quality of Life)

Value	1983
0	SKIPPED
1	< 2000 POUNDS
2	2000-2999 POUNDS
3	3000-3999 POUNDS
4	4000-4999 POUNDS
5	5000-5999 POUNDS
6	6000-6999 POUNDS
7	7000-7999 POUNDS
8	8000-9999 POUNDS
9	10000-11999 POUNDS
10	12000-14999 POUNDS
11	15000+ POUNDS
12	~ ~ ~ ~
13	~ ~ ~ ~
98	DK
99	NA

[NB: No £ sign in 1983]

Value	1996
	~New letter codes introduced~
-1	Skp,not 10+hr employed
0	~Dropped ~
1	~Dropped ~
2	~Dropped ~
3	Q ls thn 3999
4	~Dropped ~
5	T 4000- 5999
6	~Dropped ~
7	O 6000- 7999
8	K 8000- 9999
9	L 10000- 11999
10	B 12000- 14999
11	Z 15000- 17999
12	M 18000- 19999
13	F 20000- 22999
14	J 23000- 25999
15	D 26000- 28999
16	H 29000- 31999
17	C 32000- 34999
18	G 35000- 37999
19	P 38000- 40999
20	N 41000 or more
21	~ ~ ~ ~
98	Don't know
99	Not answered

Value	2002

-1	Skp,not paid work

1	Q ls thn 3999
2	T 4000- 5999
3	O 6000- 7999
4	K 8000- 9999
5	L 10000- 11999
6	B 12000- 14999
7	Z 15000- 17999
8	M 18000- 19999
9	F 20000- 22999
10	J 23000- 25999
11	D 26000- 28999
12	H 29000- 31999
13	A 32000- 37999
14	W 38000- 43999
15	G 44000- 49999
16	N 50000- 55999
17	E 56000 or more
97	Refused information
98	Don't know
99	Not answered

Value	2006
	Code letters dropped
-1	Not work/wait work
1	ls thn 4000
2	4000- 5999
3	6000- 7999
4	8000- 9999
5	10000- 11999
6	12000- 14999
7	15000- 17999
8	18000- 19999
9	20000- 22999
10	23000- 25999
11	26000- 28999
12	29000- 31999
13	32000- 37999
14	38000- 43999
15	44000- 49999
16	50000- 55999
17	56000 or more

97	Refused information
98	Don't know
99	Not answered

	2009-2010
Value	Coding changed to ££ per month

-1	Skp,not paid work
1	Less than £270 p.m
2	£271 - 430 p.m
3	£431 - 590 p.m
4	£591 - 780 p.m
5	£781 - 930 p.m
6	£931 - 1,100 p.m
7	£1,101 - 1,200 p.m
8	£1,201 - 1,300 p.m
9	£1,301 - 1,400 p.m
10	£1,401 - 1,600 p.m
11	£1,601 - 1,700 p.m
12	£1,701 - 1,800 p.m
13	£1,801 - 2,000 p.m
14	£2,001 - 2,200 p.m
15	£2,201 - 2,500 p.m
16	£2,501 - 2,700 p.m
17	£2,701 - 3,100 p.m
	£3,101 - 3,600 p.m
	£3,601 - 4,900 p.m
	£4,901 or more p.m
97
98	Refused information
99	Don't know
	~ ~ ~ ~

2011-2014

Gross earnings not used

Quartiles only

Value	2015

-2	Schedule not applicable
-1	Item not applicable
1	£271 - 430 pcm
2	£431 - 590 pcm
3	£591 - 780 pcm
4	£781 - 930 pcm
5	£931 - 1,100 pcm
6	£1,101 - 1,200 pcm
7	£1,201 - 1,300 pcm
8	£1,301 - 1,400 pcm
9	£1,401 - 1,600 pcm
10	£1,601 - 1,700 pcm
11	£1,701 - 1,800 pcm
12	£1,801 - 2,000 pcm
13	£2,001 - 2,200 pcm
14	£2,201 - 2,500 pcm
15	£2,501 - 2,700 pcm
16	£2,701 - 3,100 pcm
17	£3,101 - 3,600 pcm
18	£3,601 - 4,900 pcm
19	£4,901 or more pcm

97	Refused information
98	(Don't know)
99	(Refusal)

===================== To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

Rich Ulrich

Re: Estimating actual earnings from earnings groups

Can't you use RANK and its options to get quartiles by year,

regardless of the units for that year?

Rich Ulrich

From: SPSSX(r) Discussion <[hidden email]> on behalf of John F Hall <[hidden email]>
Sent: Saturday, October 13, 2018 4:18 AM
To: [hidden email]
Subject: Estimating actual earnings from earnings groups

DO IF (year = 1983) .

RECODE (<value(s)> = 1) (<value(s)> = 2) etc (else = copy)

INTO newrearn.

ELSE IF (year = 1984) etc.

~ ~ ~

ELSE IF (year = 2015)

~ ~ ~

END IF.

Oh, and variable [year] is not present for every year, so that has had to be inserted manually.

Sorry if there's too much detail below, but it gives a clearer idea of the problem.

Thanks in advance for any suggestions.

John F Hall MA (Cantab) Dip Ed (Dunelm)

[Retired academic survey researcher]

Email: [hidden email]

Website: Journeys in Survey Research

Course: Survey Analysis Workshop (SPSS)

Research: Subjective Social Indicators (Quality of Life)

Value	1983
0	SKIPPED
1	< 2000 POUNDS
2	2000-2999 POUNDS
3	3000-3999 POUNDS
4	4000-4999 POUNDS
5	5000-5999 POUNDS
6	6000-6999 POUNDS
7	7000-7999 POUNDS
8	8000-9999 POUNDS
9	10000-11999 POUNDS
10	12000-14999 POUNDS
11	15000+ POUNDS
12	~ ~ ~ ~
13	~ ~ ~ ~
98	DK
99	NA

[NB: No £ sign in 1983]

Value	1996
	~New letter codes introduced~
-1	Skp,not 10+hr employed
0	~Dropped ~
1	~Dropped ~
2	~Dropped ~
3	Q ls thn 3999
4	~Dropped ~
5	T 4000- 5999
6	~Dropped ~
7	O 6000- 7999
8	K 8000- 9999
9	L 10000- 11999
10	B 12000- 14999
11	Z 15000- 17999
12	M 18000- 19999
13	F 20000- 22999
14	J 23000- 25999
15	D 26000- 28999
16	H 29000- 31999
17	C 32000- 34999
18	G 35000- 37999
19	P 38000- 40999
20	N 41000 or more
21	~ ~ ~ ~
98	Don't know
99	Not answered

Value	2002

-1	Skp,not paid work

1	Q ls thn 3999
2	T 4000- 5999
3	O 6000- 7999
4	K 8000- 9999
5	L 10000- 11999
6	B 12000- 14999
7	Z 15000- 17999
8	M 18000- 19999
9	F 20000- 22999
10	J 23000- 25999
11	D 26000- 28999
12	H 29000- 31999
13	A 32000- 37999
14	W 38000- 43999
15	G 44000- 49999
16	N 50000- 55999
17	E 56000 or more
97	Refused information
98	Don't know
99	Not answered

Value	2006
	Code letters dropped
-1	Not work/wait work
1	ls thn 4000
2	4000- 5999
3	6000- 7999
4	8000- 9999
5	10000- 11999
6	12000- 14999
7	15000- 17999
8	18000- 19999
9	20000- 22999
10	23000- 25999
11	26000- 28999
12	29000- 31999
13	32000- 37999
14	38000- 43999
15	44000- 49999
16	50000- 55999
17	56000 or more

97	Refused information
98	Don't know
99	Not answered

	2009-2010
Value	Coding changed to ££ per month

-1	Skp,not paid work
1	Less than £270 p.m
2	£271 - 430 p.m
3	£431 - 590 p.m
4	£591 - 780 p.m
5	£781 - 930 p.m
6	£931 - 1,100 p.m
7	£1,101 - 1,200 p.m
8	£1,201 - 1,300 p.m
9	£1,301 - 1,400 p.m
10	£1,401 - 1,600 p.m
11	£1,601 - 1,700 p.m
12	£1,701 - 1,800 p.m
13	£1,801 - 2,000 p.m
14	£2,001 - 2,200 p.m
15	£2,201 - 2,500 p.m
16	£2,501 - 2,700 p.m
17	£2,701 - 3,100 p.m
	£3,101 - 3,600 p.m
	£3,601 - 4,900 p.m
	£4,901 or more p.m
97
98	Refused information
99	Don't know
	~ ~ ~ ~

2011-2014

Gross earnings not used

Quartiles only

Value	2015

-2	Schedule not applicable
-1	Item not applicable
1	£271 - 430 pcm
2	£431 - 590 pcm
3	£591 - 780 pcm
4	£781 - 930 pcm
5	£931 - 1,100 pcm
6	£1,101 - 1,200 pcm
7	£1,201 - 1,300 pcm
8	£1,301 - 1,400 pcm
9	£1,401 - 1,600 pcm
10	£1,601 - 1,700 pcm
11	£1,701 - 1,800 pcm
12	£1,801 - 2,000 pcm
13	£2,001 - 2,200 pcm
14	£2,201 - 2,500 pcm
15	£2,501 - 2,700 pcm
16	£2,701 - 3,100 pcm
17	£3,101 - 3,600 pcm
18	£3,601 - 4,900 pcm
19	£4,901 or more pcm

97	Refused information
98	(Don't know)
99	(Refusal)

John F Hall

Re: Estimating actual earnings from earnings groups

Rich

Thanks for the suggestion. Tried that on the cumulative file:

RANK variables = rearn /ntiles (4) into xrearnq .

CRO year by newrearnq.

*Year of Interview Percentile Group of REarn Crosstabulation**
% within Year of Interview
		Percentile Group of REarn				Total
		1	2	3	4	Total
Year of Interview	1983	55.9%	31.1%	13.0%		100.0%
	1984	52.0%	30.7%	17.3%		100.0%
	1985	47.8%	30.9%	21.3%		100.0%
	1986	43.8%	28.5%	21.9%	5.8%	100.0%
	1987	39.1%	28.9%	25.5%	6.5%	100.0%
	1989	30.0%	26.9%	30.6%	12.5%	100.0%
	1990	25.8%	24.4%	34.0%	15.8%	100.0%
	1991	24.9%	22.4%	34.2%	18.5%	100.0%
	1993	19.7%	21.2%	37.1%	21.9%	100.0%
	1994	18.0%	18.9%	35.7%	27.4%	100.0%
	1995	16.4%	18.4%	36.2%	29.0%	100.0%
	1996	16.2%	17.8%	34.1%	32.0%	100.0%
	1997	17.7%	15.2%	34.1%	32.9%	100.0%
	1998	18.1%	16.6%	31.1%	34.3%	100.0%
	1999	14.0%	16.3%	33.0%	36.7%	100.0%
	2000	14.6%	12.6%	32.2%	40.6%	100.0%
	2001	11.6%	12.2%	30.2%	45.9%	100.0%
	2002	28.9%	26.7%	22.0%	22.3%	100.0%
	2003	28.9%	27.8%	21.7%	21.5%	100.0%
	2004	28.1%	25.9%	21.6%	24.4%	100.0%
	2005	27.0%	28.1%	20.8%	24.2%	100.0%
	2006	23.3%	25.1%	23.4%	28.2%	100.0%
	2007	25.1%	24.7%	22.0%	28.2%	100.0%
	2008	22.6%	23.2%	22.9%	31.3%	100.0%
	2009	23.4%	24.5%	22.6%	29.5%	100.0%
	2010	24.4%	17.4%	17.6%	40.6%	100.0%
	2015	20.5%	13.7%	16.0%	49.8%	100.0%
Total		25.3%	22.4%	26.1%	26.2%	100.0%

It's done the same for all years. Got round that with:

SPLIT FILE by year.

RANK variables = rearn /ntiles (4) into newrearnq .

SPLIT FILE off.

CRO year by newrearnq /cel row.

*Year of Interview Percentile Group of REarn Crosstabulation**
% within Year of Interview
		Percentile Group of REarn				Total
		1	2	3	4	Total
Year of Interview	1983	21.1%	34.7%	20.1%	24.1%	100.0%
	1984	26.1%	25.9%	19.8%	28.2%	100.0%
	1985	26.8%	21.0%	30.9%	21.3%	100.0%
	1986	24.4%	27.5%	20.3%	27.8%	100.0%
	1987	29.3%	20.1%	29.3%	21.3%	100.0%
	1989	22.6%	22.6%	35.1%	19.6%	100.0%
	1990	25.8%	24.4%	24.3%	25.5%	100.0%
	1991	24.9%	22.4%	24.3%	28.3%	100.0%
	1993	29.6%	23.9%	24.6%	21.9%	100.0%
	1994	27.2%	21.0%	24.4%	27.4%	100.0%
	1995	25.8%	20.5%	31.0%	22.6%	100.0%
	1996	25.2%	19.3%	29.7%	25.8%	100.0%
	1997	26.0%	18.9%	29.3%	25.8%	100.0%
	1998	26.1%	29.6%	16.5%	27.8%	100.0%
	1999	22.3%	30.0%	24.0%	23.8%	100.0%
	2000	27.2%	22.6%	22.9%	27.3%	100.0%
	2001	23.8%	30.2%	21.8%	24.1%	100.0%
	2002	21.2%	29.0%	27.5%	22.3%	100.0%
	2003	28.9%	21.2%	23.0%	26.9%	100.0%
	2004	28.1%	19.8%	27.7%	24.4%	100.0%
	2005	27.0%	21.7%	27.2%	24.2%	100.0%
	2006	23.3%	25.1%	28.8%	22.8%	100.0%
	2007	25.1%	24.7%	27.1%	23.1%	100.0%
	2008	22.6%	23.2%	28.3%	25.9%	100.0%
	2009	23.4%	24.5%	27.6%	24.5%	100.0%
	2010	24.4%	22.9%	27.2%	25.6%	100.0%
	2015	25.2%	25.1%	26.7%	23.1%	100.0%
Total		25.3%	24.0%	26.1%	24.6%	100.0%

Quartile points calculated within years. Exactly what I wanted. Just got to slot in existing [rearnq] for years 2011 – 2014.

*Year of Interview Respondent earnings quartiles (dv) Crosstabulation**
% within Year of Interview
		Respondent earnings quartiles (dv)				Total
		Q1	Q2	Q3	Q4	Total
Year of Interview	2008	22.6%	23.2%	28.3%	25.9%	100.0%
	2009	23.4%	24.5%	27.6%	24.5%	100.0%
	2010	24.4%	30.3%	25.4%	19.9%	100.0%
	2011	25.0%	28.6%	24.9%	21.5%	100.0%
	2012	26.7%	26.6%	24.5%	22.2%	100.0%
	2013	25.3%	28.4%	22.4%	23.9%	100.0%
	2014	22.6%	27.8%	24.7%	25.0%	100.0%
	2015	22.5%	26.6%	26.0%	25.0%	100.0%
Total		23.9%	26.7%	25.7%	23.7%	100.0%

*Check.

CRO rearnq by newrearnq.

*Respondent earnings quartiles (dv) Percentile Group of REarn Crosstabulation**
Count
		Percentile Group of REarn				Total
		Q1	Q2	Q3	Q4	Total
Respondent earnings quartiles (dv)	Q1	1697	0	0	0	1697
	Q2	102	1688	110	0	1900
	Q3	0	119	1776	83	1978
	Q4	0	0	183	1590	1773
Total		1799	1807	2069	1673	7348

CORR rearnq xrearnq newrearnq.

Correlations
		Respondent earnings quartiles (dv)	Percentile Group of REarn	Percentile Group of REarn
Respondent earnings quartiles (dv)	Pearson Correlation	1	0.946	0.966
	Sig. (2-tailed)		0.000	0.000
	N	12975	7348	7348
Percentile Group of REarn	Pearson Correlation	0.946	1	0.889
	Sig. (2-tailed)	0.000		0.000
	N	7348	42208	42208
Percentile Group of REarn	Pearson Correlation	0.966	0.889	1
	Sig. (2-tailed)	0.000	0.000
	N	7348	42208	42208

John F Hall MA (Cantab) Dip Ed (Dunelm)

[Retired academic survey researcher]

Email: [hidden email]

Website: Journeys in Survey Research

Course: Survey Analysis Workshop (SPSS)

Research: Subjective Social Indicators (Quality of Life)

From: SPSSX(r) Discussion <[hidden email]> On Behalf Of Rich Ulrich
Sent: 13 October 2018 19:40
To: [hidden email]
Subject: Re: Estimating actual earnings from earnings groups

Can't you use RANK and its options to get quartiles by year,

regardless of the units for that year?

Rich Ulrich

DO IF (year = 1983) .

RECODE (<value(s)> = 1) (<value(s)> = 2) etc (else = copy)

INTO newrearn.

ELSE IF (year = 1984) etc.

~ ~ ~

ELSE IF (year = 2015)

~ ~ ~

END IF.

Oh, and variable [year] is not present for every year, so that has had to be inserted manually.

Sorry if there's too much detail below, but it gives a clearer idea of the problem.

Thanks in advance for any suggestions.

John F Hall MA (Cantab) Dip Ed (Dunelm)

[Retired academic survey researcher]

Email: [hidden email]

Website: Journeys in Survey Research

Course: Survey Analysis Workshop (SPSS)

Research: Subjective Social Indicators (Quality of Life)

Value	1983
0	SKIPPED
1	< 2000 POUNDS
2	2000-2999 POUNDS
3	3000-3999 POUNDS
4	4000-4999 POUNDS
5	5000-5999 POUNDS
6	6000-6999 POUNDS
7	7000-7999 POUNDS
8	8000-9999 POUNDS
9	10000-11999 POUNDS
10	12000-14999 POUNDS
11	15000+ POUNDS
12	~ ~ ~ ~
13	~ ~ ~ ~
98	DK
99	NA

[NB: No £ sign in 1983]

Value	1996
	~New letter codes introduced~
-1	Skp,not 10+hr employed
0	~Dropped ~
1	~Dropped ~
2	~Dropped ~
3	Q ls thn 3999
4	~Dropped ~
5	T 4000- 5999
6	~Dropped ~
7	O 6000- 7999
8	K 8000- 9999
9	L 10000- 11999
10	B 12000- 14999
11	Z 15000- 17999
12	M 18000- 19999
13	F 20000- 22999
14	J 23000- 25999
15	D 26000- 28999
16	H 29000- 31999
17	C 32000- 34999
18	G 35000- 37999
19	P 38000- 40999
20	N 41000 or more
21	~ ~ ~ ~
98	Don't know
99	Not answered

Value	2002

-1	Skp,not paid work

1	Q ls thn 3999
2	T 4000- 5999
3	O 6000- 7999
4	K 8000- 9999
5	L 10000- 11999
6	B 12000- 14999
7	Z 15000- 17999
8	M 18000- 19999
9	F 20000- 22999
10	J 23000- 25999
11	D 26000- 28999
12	H 29000- 31999
13	A 32000- 37999
14	W 38000- 43999
15	G 44000- 49999
16	N 50000- 55999
17	E 56000 or more
97	Refused information
98	Don't know
99	Not answered

Value	2006
	Code letters dropped
-1	Not work/wait work
1	ls thn 4000
2	4000- 5999
3	6000- 7999
4	8000- 9999
5	10000- 11999
6	12000- 14999
7	15000- 17999
8	18000- 19999
9	20000- 22999
10	23000- 25999
11	26000- 28999
12	29000- 31999
13	32000- 37999
14	38000- 43999
15	44000- 49999
16	50000- 55999
17	56000 or more

97	Refused information
98	Don't know
99	Not answered

	2009-2010
Value	Coding changed to ££ per month

-1	Skp,not paid work
1	Less than £270 p.m
2	£271 - 430 p.m
3	£431 - 590 p.m
4	£591 - 780 p.m
5	£781 - 930 p.m
6	£931 - 1,100 p.m
7	£1,101 - 1,200 p.m
8	£1,201 - 1,300 p.m
9	£1,301 - 1,400 p.m
10	£1,401 - 1,600 p.m
11	£1,601 - 1,700 p.m
12	£1,701 - 1,800 p.m
13	£1,801 - 2,000 p.m
14	£2,001 - 2,200 p.m
15	£2,201 - 2,500 p.m
16	£2,501 - 2,700 p.m
17	£2,701 - 3,100 p.m
	£3,101 - 3,600 p.m
	£3,601 - 4,900 p.m
	£4,901 or more p.m
97
98	Refused information
99	Don't know
	~ ~ ~ ~

2011-2014

Gross earnings not used

Quartiles only

Value	2015

-2	Schedule not applicable
-1	Item not applicable
1	£271 - 430 pcm
2	£431 - 590 pcm
3	£591 - 780 pcm
4	£781 - 930 pcm
5	£931 - 1,100 pcm
6	£1,101 - 1,200 pcm
7	£1,201 - 1,300 pcm
8	£1,301 - 1,400 pcm
9	£1,401 - 1,600 pcm
10	£1,601 - 1,700 pcm
11	£1,701 - 1,800 pcm
12	£1,801 - 2,000 pcm
13	£2,001 - 2,200 pcm
14	£2,201 - 2,500 pcm
15	£2,501 - 2,700 pcm
16	£2,701 - 3,100 pcm
17	£3,101 - 3,600 pcm
18	£3,601 - 4,900 pcm
19	£4,901 or more pcm

97	Refused information
98	(Don't know)
99	(Refusal)

David Marso-2

Re: Estimating actual earnings from earnings groups

In reply to this post by John F Hall

Note that RANK has a BY subcommand.

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

John F Hall

Re: Estimating actual earnings from earnings groups

Tried all manner of variations, but BY doesn't work for me.

rank VARIABLES rearn /ntiles (4) into xxxx by year .

rank VARIABLES rearn (a) /ntiles (4) into xxxx by year .

Warnings

Text: VARIABLES Command: rank

An undefined variable name, or a scratch or system variable was specified in a variable list which accepts only standard variables. Check spelling and verify the existence of this variable.

Execution of this command stops.

No analysis is performed because of an invalid input variable list.

Expected A or D keyword on the VARIABLES subcommand was not found. Syntax checking begins with the next right parenthesis.

The number of variable names in the INTO variable list exceeds the number of variable names in the input variable list.

RANK syntax error. This is not a valid variable name. It either exceeds 64 characters, does not start with a letter, contains something besides letters, digits, "_", or ".", or is a reserved word.

An already existing variable name was used in the INTO variable list.

Not to worry, I got what I wanted with:

SPLIT FILE by year.

RANK variables = rearn /ntiles (4) into newrearnq .

John F Hall MA (Cantab) Dip Ed (Dunelm)

[Retired academic survey researcher]

Email: [hidden email]

Website: Journeys in Survey Research

Course: Survey Analysis Workshop (SPSS)

Research: Subjective Social Indicators (Quality of Life)

-----Original Message-----
From: SPSSX(r) Discussion <[hidden email]> On Behalf Of David Marso
Sent: 14 October 2018 17:31
To: [hidden email]
Subject: Re: Estimating actual earnings from earnings groups

Note that RANK has a BY subcommand.

=====================

To manage your subscription to SPSSX-L, send a message to [hidden email] (not to SPSSX-L), with no body text except the command. To leave the list, send the command SIGNOFF SPSSX-L For a list of commands to manage subscriptions, send the command INFO REFCARD

Rich Ulrich

Re: Estimating actual earnings from earnings groups

Your very first error message says that it objects to VARIABLES.

Your example that works has VARIABLES=

I'd say, apparently that "=" is required.

I'm not sure that's true for every command, but I never tested it.

Rich Ulrich

From: SPSSX(r) Discussion <[hidden email]> on behalf of John F Hall <[hidden email]>
Sent: Sunday, October 14, 2018 12:23 PM
To: [hidden email]
Subject: Re: Estimating actual earnings from earnings groups

Tried all manner of variations, but BY doesn't work for me.

rank VARIABLES rearn /ntiles (4) into xxxx by year .

rank VARIABLES rearn (a) /ntiles (4) into xxxx by year .

Warnings

Text: VARIABLES Command: rank

An undefined variable name, or a scratch or system variable was specified in a variable list which accepts only standard variables. Check spelling and verify the existence of this variable.

Execution of this command stops.

No analysis is performed because of an invalid input variable list.

Expected A or D keyword on the VARIABLES subcommand was not found. Syntax checking begins with the next right parenthesis.

The number of variable names in the INTO variable list exceeds the number of variable names in the input variable list.

RANK syntax error. This is not a valid variable name. It either exceeds 64 characters, does not start with a letter, contains something besides letters, digits, "_", or ".", or is a reserved word.

An already existing variable name was used in the INTO variable list.

Not to worry, I got what I wanted with:

SPLIT FILE by year.

RANK variables = rearn /ntiles (4) into newrearnq .

John F Hall MA (Cantab) Dip Ed (Dunelm)

[Retired academic survey researcher]

Email: [hidden email]

Website: Journeys in Survey Research

Course: Survey Analysis Workshop (SPSS)

Research: Subjective Social Indicators (Quality of Life)

Note that RANK has a BY subcommand.

=====================

John F Hall

Re: Estimating actual earnings from earnings groups

Looks as though DM's suggestion about keyword BY is the culprit.

I tried three variations, but none of them worked.

Here's one of them:

rank VARIABLES = rearn (a) /ntiles (4) into xxxx by year .

Warnings

RANK syntax error. This is not a valid variable name. It either exceeds 64 characters, does not start with a letter, contains something besides letters, digits, "_", or ".", or is a reserved word.

Execution of this command stops.

The number of variable names in the INTO variable list exceeds the number of variable names in the input variable list.

rank VARIABLES= rearn (a) /ntiles (4) into xxxx by year .

Warnings

RANK syntax error. This is not a valid variable name. It either exceeds 64 characters, does not start with a letter, contains something besides letters, digits, "_", or ".", or is a reserved word.

Execution of this command stops.

The number of variable names in the INTO variable list exceeds the number of variable names in the input variable list.

If I delete "by year" here's one that works, but it generates the same grouping for each year:

rank VARIABLES= rearn (a) /ntiles (4) into xxxx .

Created Variables^a
Source Variable	Function	New Variable	Label
REarn^c	Percentile Group^b	xxxx	Percentile Group of REarn
a. Mean rank of tied values is used for ties.
b. 4 groups are generated.
c. Ranks are in ascending order.

cro year by xxxx /cel row,.

*year Year of Interview xxxx Percentile Group of REarn Crosstabulation**
% within year Year of Interview
		xxxx Percentile Group of REarn				Total
		1	2	3	4	Total
year Year of Interview	1983	55.9%	31.1%	13.0%		100.0%
	1984	52.0%	30.7%	17.3%		100.0%
	1985	47.8%	30.9%	21.3%		100.0%
	1986	43.8%	28.5%	21.9%	5.8%	100.0%
	1987	39.1%	28.9%	25.5%	6.5%	100.0%
	1989	30.0%	26.9%	30.6%	12.5%	100.0%
	1990	25.8%	24.4%	34.0%	15.8%	100.0%
	1991	24.9%	22.4%	34.2%	18.5%	100.0%
	1993	19.7%	21.2%	37.1%	21.9%	100.0%
	1994	18.0%	18.9%	35.7%	27.4%	100.0%
	1995	16.4%	18.4%	36.2%	29.0%	100.0%
	1996	16.2%	17.8%	34.1%	32.0%	100.0%
	1997	17.7%	15.2%	34.1%	32.9%	100.0%
	1998	18.1%	16.6%	31.1%	34.3%	100.0%
	1999	14.0%	16.3%	33.0%	36.7%	100.0%
	2000	14.6%	12.6%	32.2%	40.6%	100.0%
	2001	11.6%	12.2%	30.2%	45.9%	100.0%
	2002	28.9%	26.7%	22.0%	22.3%	100.0%
	2003	28.9%	27.8%	21.7%	21.5%	100.0%
	2004	28.1%	25.9%	21.6%	24.4%	100.0%
	2005	27.0%	28.1%	20.8%	24.2%	100.0%
	2006	23.3%	25.1%	23.4%	28.2%	100.0%
	2007	25.1%	24.7%	22.0%	28.2%	100.0%
	2008	22.6%	23.2%	22.9%	31.3%	100.0%
	2009	23.4%	24.5%	22.6%	29.5%	100.0%
	2010	24.4%	17.4%	17.6%	40.6%	100.0%
	2015	20.5%	13.7%	16.0%	49.8%	100.0%
Total		25.4%	22.6%	26.4%	25.5%	100.0%

So try:

SPLIT FILE by year.

rank VARIABLES= rearn (a) /ntiles (4) into xxxx .

Warnings

An already existing variable name was used in the INTO variable list.

Execution of this command stops.

Oops!

Either

Highlight the row in the Data Editor:

xxxx Ordinal Percentile Group of REarn None None 0 Numeric 3 6 Right Input

Right click >> Clear

DELETE VARIABLES xxxx.

Here's the one that generates separate groupings for each year:

SPLIT FILE by year.

rank VARIABLES= rearn (a) /ntiles (4) into xxxx .

SPLIT FILE off.

cro year by xxxx /cel row.

*year Year of Interview xxxx Percentile Group of REarn Crosstabulation**
% within year Year of Interview
		xxxx Percentile Group of REarn				Total
		1	2	3	4	Total
year Year of Interview	1983	21.1%	34.7%	20.1%	24.1%	100.0%
	1984	26.1%	25.9%	19.8%	28.2%	100.0%
	1985	26.8%	21.0%	30.9%	21.3%	100.0%
	1986	24.4%	27.5%	20.3%	27.8%	100.0%
	1987	29.3%	20.1%	29.3%	21.3%	100.0%
	1989	22.6%	22.6%	35.1%	19.6%	100.0%
	1990	25.8%	24.4%	24.3%	25.5%	100.0%
	1991	24.9%	22.4%	24.3%	28.3%	100.0%
	1993	29.6%	23.9%	24.6%	21.9%	100.0%
	1994	27.2%	21.0%	24.4%	27.4%	100.0%
	1995	25.8%	20.5%	31.0%	22.6%	100.0%
	1996	25.2%	19.3%	29.7%	25.8%	100.0%
	1997	26.0%	18.9%	29.3%	25.8%	100.0%
	1998	26.1%	29.6%	16.5%	27.8%	100.0%
	1999	22.3%	30.0%	24.0%	23.8%	100.0%
	2000	27.2%	22.6%	22.9%	27.3%	100.0%
	2001	23.8%	30.2%	21.8%	24.1%	100.0%
	2002	21.2%	29.0%	27.5%	22.3%	100.0%
	2003	28.9%	21.2%	23.0%	26.9%	100.0%
	2004	28.1%	19.8%	27.7%	24.4%	100.0%
	2005	27.0%	21.7%	27.2%	24.2%	100.0%
	2006	23.3%	25.1%	28.8%	22.8%	100.0%
	2007	25.1%	24.7%	27.1%	23.1%	100.0%
	2008	22.6%	23.2%	28.3%	25.9%	100.0%
	2009	23.4%	24.5%	27.6%	24.5%	100.0%
	2010	24.4%	22.9%	27.2%	25.6%	100.0%
	2015	25.2%	25.1%	26.7%	23.1%	100.0%
Total		25.3%	24.0%	26.1%	24.6%	100.0%

John F Hall MA (Cantab) Dip Ed (Dunelm)

[Retired academic survey researcher]

Email: [hidden email]

Website: Journeys in Survey Research

Course: Survey Analysis Workshop (SPSS)

Research: Subjective Social Indicators (Quality of Life)

From: SPSSX(r) Discussion <[hidden email]> On Behalf Of Rich Ulrich
Sent: 15 October 2018 05:12
To: [hidden email]
Subject: Re: Estimating actual earnings from earnings groups

Your very first error message says that it objects to VARIABLES.

Your example that works has VARIABLES=

I'd say, apparently that "=" is required.

I'm not sure that's true for every command, but I never tested it.

Rich Ulrich

Tried all manner of variations, but BY doesn't work for me.

rank VARIABLES rearn /ntiles (4) into xxxx by year .

rank VARIABLES rearn (a) /ntiles (4) into xxxx by year .

Warnings

Text: VARIABLES Command: rank

An undefined variable name, or a scratch or system variable was specified in a variable list which accepts only standard variables. Check spelling and verify the existence of this variable.

Execution of this command stops.

No analysis is performed because of an invalid input variable list.

Expected A or D keyword on the VARIABLES subcommand was not found. Syntax checking begins with the next right parenthesis.

The number of variable names in the INTO variable list exceeds the number of variable names in the input variable list.

RANK syntax error. This is not a valid variable name. It either exceeds 64 characters, does not start with a letter, contains something besides letters, digits, "_", or ".", or is a reserved word.

An already existing variable name was used in the INTO variable list.

Not to worry, I got what I wanted with:

SPLIT FILE by year.

RANK variables = rearn /ntiles (4) into newrearnq .

John F Hall MA (Cantab) Dip Ed (Dunelm)

[Retired academic survey researcher]

Email: [hidden email]

Website: Journeys in Survey Research

Course: Survey Analysis Workshop (SPSS)

Research: Subjective Social Indicators (Quality of Life)

Note that RANK has a BY subcommand.

=====================

Bruce Weaver

Re: Estimating actual earnings from earnings groups

Administrator

Try this, John:

RANK VARIABLES = rearn (a) BY year /ntiles (4).

John F Hall wrote
> Looks as though DM's suggestion about keyword BY is the culprit.
>
> I tried three variations, but none of them worked.
>
> Here's one of them:
>
> rank VARIABLES = rearn (a) /ntiles (4) into xxxx by year .
>
> --- snip ---

-----
--
Bruce Weaver
[hidden email]
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

NOTE: My Hotmail account is not monitored regularly.
To send me an e-mail, please use the address shown above.

--
Sent from: http://spssx-discussion.1045642.n5.nabble.com/

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

--
Bruce Weaver
bweaver@lakeheadu.ca
http://sites.google.com/a/lakeheadu.ca/bweaver/

"When all else fails, RTFM."

PLEASE NOTE THE FOLLOWING:
1. My Hotmail account is not monitored regularly. To send me an e-mail, please use the address shown above.
2. The SPSSX Discussion forum on Nabble is no longer linked to the SPSSX-L listserv administered by UGA (https://listserv.uga.edu/).