Reading Statistics: The SPSSPC Frequencies Printout

Reading Statistics:

The SPSSPC "Frequencies" Printout

Frequencies tell you both the number and the percentage of all parents who selected each

response. Following is an example:

A19

Caused Less Angry

Frequency

Valid

Missing

Total

No

Yes

Total

System

88

121

209

6

215

Percent

40.9

56.3

97.2

2.8

100.0

Valid

Percent

42.1

57.9

100.0

Cumulative

Percent

42.1

100.0

? The column labeled "frequency" lists the actual number of parents who selected each

response. In the example, 88 parents answered "no" and 121 answered "yes."

? The column labeled "percent" lists the actual percentages of the total sample who

answered either "yes" or "no."

? "Valid percent" is the percent when missing data are excluded from the calculations.

In other words, these are the percentages of parents who selected each response after

we disregard missing responses.

? The row labeled "missing system" shows the number and the percentage of parents

who either did not answer the question, or whose answer was invalid. In the example,

6 parents gave no answer (2.8% of the total).

What numbers do you report?

? For percentages in "yes/no" questions, such as the one above from the Dane County

evaluation, report the "percent." This makes the assumption that people who didn't

answer are really giving a "no" answer. Thus we are using the total sample.

Just the "percent" of people who answered "yes" should be reported. This is how a

report of these results might read: "Over half of the parents (56%) reported that

reading the newsletter caused me to be less angry when my baby is difficult." The

bar graph (labeled "Figure 7") shows the percentage of respondents answering "yes"

to the questions of self-reported effects of reading the newsletter. The third bar

represents the 56% of "yes" respondents from the table above.

? For questions 1 - 14 you should also report the "percent" since we want to know

the percentage of the total sample who selected each response.

? For the percentages of socio-demographic variables (A24-32), you should use the

entries under the column "valid percent." Valid percent is used for these variables,

rather than percent, because we assume that the missing cases are distributed

proportionately among all the response categories. This allows us to make pie charts

that add up to 100%.

A32

Family Income

Frequency

Valid

Missing

Total

Less than 10 K

10K - 19,999

20K - 29,999

30K - 49,999

50K or more

Total

System

17

20

29

65

74

205

10

215

Percent

7.9

9.3

13.5

30.2

34.4

95.3

4.7

100.0

Valid

Percent

8.3

9.8

14.1

31.7

36.1

100.0

Cumulative

Percent

8.3

18.1

32.2

63.9

100.0

? In the frequency table above, 205 of the 215 parents had valid data, while 10 families

(4.7%) had missing data. You can see the "valid percents" are slightly higher than the

"percents" because the 10 missing cases have been removed from consideration. The

bar chart labeled "Figure 3" demonstrates how these percentages would be graphed in

the final report.

? Note that sometimes you need to do some recalculating on your own. Here is an

example from the Frequencies table above: "Two-thirds of respondents (68%)

reported their family income to be $30,000 or more per year." The 68% figure comes

from adding the valid percents from two rows (31.7 for those from 30K to 49,999 and

36.1 for those with 50K or more equals 67.8% which rounds to 68%).

? In the final report, press releases, and other public documents, we recommend

rounding percentages to the nearest whole number. We lose a little accuracy this

way, but avoid frightening math-phobic people. So we would round 14.4 to 14 and

14.6 to 15. But which way would you round 14.5? Here is the rule of thumb we

follow: round .5 to the nearest even number. Therefore, 14.5 would round down to

14, but 15.5 would round up to 16, the nearest even number. In this way there is no

systemic bias upward or downward.

Figure 3

Family Incomes, 1992

36%

$50,000 or more

32%

$30,000 to 49,999

14%

$20,000 to 29,999

10%

$10,000 to 19,999

Less than $10,000

0%

8%

10%

20%

30%

40%

Percentage of Respondents

50%

Figure 7

Self-Reported Behavior Change

"Reading the newsletters caused me to...

Provide more things for my baby

to feel, see, listen to, smell."

74%

62%

Talk to my baby more."

56%

Be less angry when my baby is difficult."

Respond more quickly when my baby cried."

Smile, kiss & hug my baby more."

0%

46%

39%

50%

100%

Percentage of Respondents

Reading Statistics:

The SPSSPC "Crosstabs" Printout

Cross-tabulations are called "crosstabs" for short, and literally tabulate one

variable across another. In the example below we have cross-tabulated parents' selfreport responses of whether reading the newsletter caused them to be less angry when

their baby was difficult with the variable called RISK2, that indicates whether a parent

could be considered as having one of six risk factors. If a parent reported being

primiparous, teenage, in a non-two parent household, low education, low income, or

isolated, then this parent was defined as at risk. Otherwise they were defined in the nonrisk group.

Every statistical software system prints out a crosstab a little differently, so don't

be surprised if you need to relearn how to read this. We'll walk you through it here.

RISK2

2-LEVEL RISK VAR * CAUSED LESS ANGRY

Crosstab

Caused Less Angry

Total

2-LEVEL

RISK VAR

NON-RISK

PRIMI AND

RISK GRP

Total

Count

% within 2-LEVEL

RISK VAR

Count

% within 2-LEVEL

RISK VAR

Count

% within 2-LEVEL

RISK VAR

No

42

Yes

34

55.3%

46

44.7%

87

100.0%

133

34.6%

65.4%

100.0%

121

88

42.1%

57.9%

76

209

100.0%

Chi-Square Tests

Value

Pearson Chi-Square

8.48214

Continuity Correction 7.65513

Df

1

1

Asymp.Sig

(2-sided)

.00359

.00566

Exact Sig

(2-sided)

Exact Sig

(1-sided)

? First, note that it tells you right at the top which two variables are being cross

tabulated (i.e. 2-LEVEL RISK VAR * CAUSED LESS ANGRY).

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download